Add files using upload-large-folder tool

processed/00004_W2114989862/document.md

@@ -0,0 +1,741 @@
+![Figure](figures/figure_000.png)
+
+Available online at www.sciencedirect.com
+
+![Figure](figures/figure_002.png)
+
+Research Policy 33 (2004) 1385-1404
+
+![Figure](figures/figure_004.png)
+
+www.elsevier.com/locate/econbase
+
+# Trademarks as an indicator of innovation and industrial change
+
+Sandro Mendonça$^{a,b,*}$, Tiago Santos Pereira$^{c}$, Manuel Mira Godinho$^{d}$
+
+a Dinamia, ISCTE University, Av. das Forças Armadas, 1649-026 Lisbon, Portugal
+
+b SPRU, The Freeman Centre, University of Sussex, Brighton, East Sussex, BN1 9QE, UK
+
+c CES — Centro de Estudos Sociais, Universidade de Coimbra, Colégio de S. Jerónimo, Apartado 3087, 3001-401 Coimbra, Portugal
+
+d ISEG/UTL and CISEP, Department of Economics, Rua Miguel Lupi, 20, 1200-725 Lisbon, Portugal
+
+Received 17 March 2004; received in revised form 2 September 2004; accepted 7 September 2004
+
+
+Available online 14 November 2004
+
+## Abstract
+
+As innovation becomes an ever more central issue for the development of firms and world economies, so the need for improved assessments of innovative performance grows more urgent. This paper suggests that trademark analysis can contribute in capturing relevant aspects of innovation phenomena and the process of industrial change. We propose trademarks as a complementary indicator in the portfolio of available empirical tools of innovation studies and industrial dynamics. Our empirical exploration is based on a study of community trade marks (CTM), an intellectual property right granted in the European Union, and draws on recent research on trademarking trends in Portugal. Quantitative as well as qualitative data, including survey data from a representative sample of Portuguese manufacturing and services firms, are used to identify the advantages and limitations of this indicator.
+
+© 2004 Elsevier B.V. All rights reserved.
+
+Keywords: Indicators; Trademarks; Innovation; Industrial dynamics
+
+## 1. Introduction
+
+The business of branding products has long been part of ordinary economic life. Trademarks are the outcome of establishing recognisable designations and symbols for goods and services, as well as firms' identi-
+
+ties. They play a crucial role in the process of marketing innovations, being instrumental in differentiating the attributes of goods and services in the marketplace. These characteristics make trademarks a potential indicator of product innovation and sectoral change. Moreover, recent developments in the institutions for the international regulation of trademarks, as well as the increasing availability of digital databases, have increased the case for using trademark statistics as a new source of information in industrial and innovation studies.
+
+* Corresponding author. Tel.: +351 21 7903007; fax: +351 21 7903933. E-mail addresses: sfm@iscte.pt (S. Mendonça), tsp@ces.uc.pt (T.S. Pereira), mgodinho@iseg.utl.pt (M.M. Godinho).
+
+0048-7333/$ – see front matter c 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.respol.2004.09.005
+
+---
+
+1386
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+Trademarks are of interest for social science research for at least three reasons: they confer the exclusive right to use a brand, therefore enhancing companies’ ability to appropriate the economic returns on new and existing products; they are an important aspect of contemporary culture world-wide; and they constitute a source of qualitative and quantitative information on socio-economic activities. This paper focuses on the third of these features. It does not address the more complex issue of the contribution of trademarks to welfare, which might be considered an urgent question for political economy in its own right. $^1$ More specifically, the paper assesses the possibilities and problems of using trademark data when analysing the introduction of new or improved products in competitive markets. Along with a methodological reflection, the paper offers a concrete empirical application of the indicator to the EU-15 countries $^2$ together with an in-depth study of an intermediate European economy, Portugal, for which we analyse: (i) statistical data on trademarks for the period since 1980; (ii) survey data collected from a representative sample of 724 firms in 2003; and (iii) information from thematic workshops held with entrepreneurs, managers and consultants. Data for the EU-15 countries was obtained from publicly available documents of the Office for Harmonization in the Internal Market (OHIM), which is responsible for managing community trade marks (CTMs). The lessons learned from the Portuguese case synthesise and elaborate on a study recently published by the Portuguese Patent and Trademark Office, INPI ( Godinho et al. , 2003 ) .
+
+We argue that trademark-based indicators provide a partial measure of the innovative output of profitoriented organisations. In its most simple formulation innovation can be understood as the introduction into the market of a new idea, product or production process. As an intellectual property right (IPR), trademarks are designed to differentiate certain products from those provided by other firms. In this context, the filing of new trademarks by economic actors partially reflects the introduction of new offerings aimed at persuading potential buyers that the range of their problems is not being solved by the supply of solutions currently avail-
+
+able in the market. In this way, since companies have to pay fees to register and renew their rights in national and international offices, the effort involved in filing for a new brand name or logo reveals an economic decision that is worth investigating. Furthermore, given the growing demand from governments, firms and academics for more reliable information on innovation, we find here an opportunity to test trademarks as a complementary indicator to the more traditional measures of innovative activity, namely R&D expenditure and patents. Trademarks are used by a wider set of business firms, capturing change in service activities as well as in small and medium-sized enterprises (SMEs).
+
+Trademark-based indicators show promise for advancing research agendas concerned with (i) the rates and directions of product innovations in different industrial sectors, (ii) international patterns of specialisation, (iii) links between technological and marketing activities; and (iv) the evolution of economic organisations and structures. However, simple counts of trademarks are affected by various sources of bias, such as difficulties in data consolidation (e.g. one brand can be protected simultaneously by a combination of words, symbols and 3D design), sectoral differences (the international trademark classification system follows the characteristics of the product and not of the industrial sector; cf. Appendix), and weaknesses in international comparability (given, for example, the different export markets and niches targeted by firms from different countries). All of these issues cannot be completely explored here. The full assessment of trademarks as indicators of innovative activity and industrial competitiveness requires further research, including econometric analysis and in-depth case studies.
+
+The paper is organised as follows. Section 2 discusses what a trademark is from both the IPR and economic perspectives. The third section addresses the conceptual and analytical issues that arise when we consider trademarks as an innovation indicator and a tool for assessing structural transformation. This is followed, in Section 4 , by an analysis of the use of CTMs in the EU-15 countries. Section 5 presents detailed data on the use of trademarks in Portugal, based on a survey of a representative sample of Portuguese firms. The final section concludes by summarising the main findings and identifying avenues for further research.
+
+1 For a recent book related to this subject see Klein (2000).
+
+2 The data, for the period 1996 – 2002, does not include the 10 new member states.
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1387
+
+## 2. Brands and logos in business life
+
+People show a tendency to label other beings and things surrounding them, as well as to exert ownership over them. It is thought that the first marks served to indicate the legitimate owner of livestock in Neolithic times, 7000 years ago. $^3$ Marks later evolved to designate the author of an object and to establish his or her obligation in ensuring the quality of the product. The first documented example of this economic use of trademarks is found in the Roman Empire: Roman bricks bore the stamp of the manufacturer, the date and the place of production. During the Middle Ages, corporations of craftsmen started to identify their workmanship with a mark. In the absence of modern means of advertising, the reputation of trade guilds was carried through marks inscribed in the merchandise. A century after the dawn of the industrial era, a series of explicit trademark laws were enacted in France (1857), the UK (1862) and the US (1870). In this field, as well as in others, framing institutions co-evolved with the actors' behaviour and goals. This section deals with the key institutional features of trademarks as a part of the IPR system and briefly examines the strategic rationale and the historical practice of their use by business firms.
+
+### 2.1. Trademarks as a property right
+
+According to the World Intellectual Property Organization (WIPO), a trademark is defined as a “distinctive sign, which identifies certain goods or services as those provided by a specific person or enterprise ” ( WIPO , 2004 ) . The two objectives of protection and dissemination built into this definition are practically indistinguishable. Like patents, a trademark affords the owner legal protection by granting the exclusive right to use it to identify goods or services, or to license its use to another entity in return for payment. Rights are granted at the national level but, unlike patents and copyrights, once trademarks are registered they can be renewed indefinitely on payment of additional fees. The common expectation in trademark regimes is that a registered
+
+trademark is used, otherwise it may be cancelled and applied for by another company after a period of grace. Its maintenance by economic agents can thus be seen as indicating the exercise of regular business activities; an unused trademark is implicitly regarded by IPR law as a barrier to economic activity.
+
+Trademarks are an IPR issued by an authorised national government agency following an examination process that is dependent on legal criteria and on a mix of limited human and technical resources administered by that agency. Once an application has been filed, examiners search available databases to detect any other marks in use that may come into conflict with that of the applicant. Whereas patents are granted to inventions on the basis of non-obviousness, inventiveness in the face of prior art and the potential for industrial application, a commercial sign, on the other hand, may be denied registration, but only if judged deceptive to consumers (e.g. if it can be confused with other marks, if it contains a misleading description of the character or quality of the goods or services, etc.), if it is deemed contrary to morality or if it denotes symbols reserved for the use of the state or public organisations. A successfully registered trademark is recognisable by having one of these two symbols attached: “ ® ” or “ TM ” . The lag between the trademark filing and its formal registration is much shorter than that for patents. For instance, while it normally takes up to a year to register a CTM, it can often take over 5 years to obtain a patent from the European Patent Office.
+
+The first international trademark settlement was reached at the Paris Convention of 1883, whereby the countries involved agreed to provide foreign applicants with the same protection regarding marks as that provided to nationals. In this context, the WIPO eventually emerged as the global coordinating institution promoting the development of IPR laws and facilitating the international registration of trademarks. This role stems from the 1891 “ Madrid Agreement Concerning the International Registration of Trademarks, ” which opened up the scope for the protection of marks beyond their market of origin. A more recent international development was the establishment of the CTM in Europe, which came into being with the establishment of the OHIM, a EU institution, in 1994.
+
+Today, trademarks may consist of one word or a combination of words, slogans, letters and numerals; they may also be drawings, symbols, three-dimensional
+
+3 http://www.lib.utexas.edu/engin/trademark/timeline/tmindex.
+
+html—website maintained by the Library of the University of Texas at Austin, which compiles and synthesises information on the history of trademarks from a wide collection of previous works on the topic.
+
+---
+
+1388
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+signs such as the shape and packaging of goods, audible signs such as musical or vocal sounds, or distinguishing fragrances, colours and holograms. Notwithstanding the traditional neglect by IPR analysts, trademarks constitute a crucial part of innovation and commercial processes, helping to attract scarce attention from the public and differentiate the nature and quality of products. In addition, as stated by Doern (1999, p. 72) , who carried out over 70 interviews with officials of seven agencies, trademarks generate the second highest volume of IPR revenue after patents.
+
+### 2.2. Brands and business strategy
+
+Brands are commercial signatures that firms create, maintain, protect and reinforce for a number of different marketing goals, in which the introduction of new products is a very relevant one. Brand names have multiple origins. Some are linked to the names of the company founders (Ford, Nestlé), others show a connection to a particular line of business (Microsoft, Airbus), whilst yet others are neologisms (Kodak, Xerox). Brands have become genuine cultural references, especially since the late 19th century with the rise of consumer product industries oriented toward mass markets in western societies. Brands such as Coca-Cola, Campbell’s Soup or McDonald’s are liked (or disliked) by many, but surely they mean something to everybody. Many brands have even become industrial eponyms, i.e., synonyms in many languages of iconic innovations in the 20th century (Hoover, Gillette, Black & Decker). Many others have established themselves as true national symbols (Ferrari, Nokia) or even come to signify the achievement of transnational partnerships (Airbus).
+
+For better or for worse, brands are with us, and their influence on our lives is not set to diminish in the near future. Judging from newspaper interviews with CEOs and popular business literature, branding is becoming an ever more central concern in corporate strategy. Evidence from everyday life tells us that hardly anything goes unbranded. Even fruits are branded, e.g., Chiquita bananas. But, where does all this interest in brands come from? Surely brands are obviously of interest to buyers. However, the ways in which the awareness and loyalty commanded by brands translates into a competitive asset are not completely clear.
+
+The rationale and influence of trademarks is a relatively neglected aspect in the study of the development of modern corporations. Economics has notoriously little to say about the activity of branding. From the point of view of conventional economics, prices and quantities are still the most important economic variables. The complex deliberation process concerning other attributes, such as quality, features, reputation, support services and so forth, has scarcely been dealt with in economic theory, not to mention applied research ( Trajtenberg , 1990 , p. 8) . Standard industrial organisation textbooks tend to group together issues such as product differentiation and advertising in the same chapter and focus on the relationship between market structure and the advertising-to-sales ratio. Brand decisions are implicitly considered part of advertising policy. $^4$
+
+The only paper that we found dealing explicitly with the “ economics of trademarks ” (an interesting, but seldom cited paper) gives us the usual cost-benefit perspective ( Economides , 1987 ) – a trademark should be filed when its expected value (probability of being accepted times the revenue associated with the brand) exceeds the opportunity cost of applying for it (including the present value of the fees needed to maintain the trademark). The author points out that the key reason why brands are indispensable for the efficient provision of products is because they are a device that facilitates consumer choice in the wide range of variety and quality combinations available in a modern economy. Nevertheless, the argument concedes, monopoly power embedded in the IPR is bound to cause inefficiencies and distortions in resource allocation, with informational benefits to consumers counterbalanced by the barriers to new competition.
+
+In management literature, the question “to brand or not to brand” is dealt with by marketing studies. In terms of the classic marketing framework, brands are
+
+4 Luís Cabral, a former editor of the Journal of Industrial Economics and author of a popular textbook in the field, makes an interesting remark on this subject. Under the heading “Price and non price strategies” of a teaching note available in the companion site to his book, he states that: “(a)dvertising is not one of the core topics of IO this chapter may therefore be omitted from a course directed primarily at economics majors,” and he goes on to admit that “(n)ot much economics research has been done on this topic. As a consequence, most of the points presented are rather tentative: more questions are raised than answered.” ( Cabral , 2000 ) .
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1389
+
+Table 1 The world's most valuable brands
+
+<table><tr><td></td><td>1990</td><td>1992</td><td>1996</td><td>1997</td><td>1999</td><td>2000</td></tr><tr><td>1</td><td>Coca-Cola</td><td>Marlboro</td><td>McDonald's</td><td>Coca-Cola</td><td>Coca-Cola</td><td>Coca-Cola</td></tr><tr><td>2</td><td>Kellogg's</td><td>Coca-Cola</td><td>Coca-Cola</td><td>Marlboro</td><td>Microsoft</td><td>Microsoft</td></tr><tr><td>3</td><td>McDonald's</td><td>Intel</td><td>Disney</td><td>IBM</td><td>IBM</td><td>IBM</td></tr><tr><td>4</td><td>Kodak</td><td>Kellogg's</td><td>Kodak</td><td>McDonald's</td><td>GE</td><td>GE</td></tr><tr><td>5</td><td>Marlboro</td><td>Nescafe</td><td>Sony</td><td>Disney</td><td>Ford</td><td>Nokia</td></tr></table>
+
+
+Sources: www.interbrand.com/, Kotler et al. (1999), Granstrand (1999).
+
+included in the Product variable of the marketing mix. The “ marketing mix ” refers to the conventional view of the set of tools that a firm blends to influence the demand for its product – it comprises the well-known “ four Ps ” : Product, Pricing, Promotion and Placement ( Kotler et al. , 1999 ) . In this literature, however, brands are not seen as a tactical tool for promotion, which instead refers to communication activities including advertising, the sales-force approach to customers, special promotions, public relations, etc. This is evidence that marketers regard brands as a key product characteristic that helps deliver the core benefits of the product to the consumers. The implication of this understanding is that brands do more than just transmit persuasive signals (spurious differentiation $^5$ ), they also have a deeper role in transforming customers' experiences of using the product (actual differentiation).
+
+Customer benefits such as trust and satisfaction have a direct translation into the strategic motivations that underlie trademark management. According to the marketing literature (e.g. Kotler et al. , 1999 ; Aaker , 1991 ) , there is a proliferation of strategic intentions when a company introduces new brands. Since there are no established taxonomies, we give our own incomplete list of the motives and strategies behind the uses of brands — see Box 1 .
+
+Naturally, as in the case of patents, trademarks are not all identical in value. The value of a patent is equated to the net money equivalent stemming from the exploration of the rights associated with a particular idea or innovation ( Griliches, 1990 , p. 1690) . That value is generally considered proportional to the magnitude of the innovation protected ( Trajtenberg, 1990 , p. 5) . In management literature, what counts in determining trademark equity is a set of characteristics, such as name awareness, customer loyalty, perceived
+
+quality and associations with the brand, that add value to the product being offered ( Aaker , 1991 ) . Trademark or brand equity influences the market value of a firm, namely of the large corporations acting on the global market. Interbrand, a specialised consultancy firm regularly releases reports in which it estimates which are the world's most important brands — see Table 1 .
+
+Box 1: Strategic motivations behind the creation of a brand.
+
+- • Building inelasticity around the product and
+achieving a premium pricing (differentiation,
+line extensions).
+• Improving the conditions for appropriating the
+returns on innovation whenever other means are
+not effective.
+• Extending the protection conferred by other
+IPRs after their expiry date (namely patents).
+• Opening up opportunities for entering new
+product segments or entirely new lines of busi-
+ness (brand-stretching or diversification).
+• Penetrating new geographical markets (geo-
+graphical market diversification).
+• Signalling changes in strategy or changes in
+corporate identity (internal and external mar-
+keting).
+• Entering the market for trademarks (licensing).
+• Saving on promotion expenditures (building
+loyalty).
+• Achieving greater bargaining power against
+suppliers (supply chain coordination).
+Summing up, in seeking and maintaining trademark rights, companies make important economic decisions. Companies decide to create or enhance a protected
+
+$^{5}$ See Carlton and Perloff (1994), p. 284.
+
+---
+
+1390
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+brand to distinguish themselves from other suppliers and to rise above the competition. By monitoring what happens to trademarks, we can learn a great deal about the entrepreneurial dynamics of firms or aggregates of firms, as well as gain a lever for making inferences about their innovative capacity and marketing capabilities. The central conjecture here is that the analysis of new brands, understood as the development of complementary assets to product innovations, might shed some light upon the innovation process. This is the reason why trademark statistics are interesting in spite of all the difficulties involved in their use and interpretation.
+
+## 3. Towards a trademark-based innovation indicator
+
+Trademarks match patents in terms of the accessibility and quantity of available data. Although they do not exhibit the same detail of explicit information (patents contain a description of a technological invention and provide the name of the individual inventors that contributed to it, together with citations made to prior art), it is important to understand what trademarks actually indicate. This section will argue that trademarks constitute a unique and underused data source for analysing product innovation and industrial evolution. We will briefly scan the panorama of innovation indicators, to concentrate afterwards on the analytical and empirical factors that lend support to our claim. Finally, some practical aspects of establishing the new indicator will be discussed.
+
+### 3.1. Indicators in innovation studies
+
+Indicators are means of obtaining quantitative information about certain aspects of a phenomenon. There are, of course, no perfect or complete indicators of innovation, just as there are no perfect indicators of other socio-economic phenomena. The particular problem with the innovation phenomenon is its multidimensional nature, involving qualitative changes in economic, strategic, organisational and institutional factors. Thus, in order to operationalise a trademark indicator, it is important to understand the ontological requisites for achieving reliable knowledge.
+
+Innovation indicators are expected to convey behavioural information on social entities. They detect and register the “levels” and “dynamics” of individuals, companies, institutions and countries. The imperfect nature of the information they afford is always present; they can only provide `indications'. Indicators capture, but only partially, some aspects of the object in question. One therefore needs to be aware of the intrinsic limitations of an indicator; it is not an objective, direct and complete measurement. Indicators are institutionally created and maintained and often turn out to be put to uses other than those originally intended ( Patel and Pavitt , 1995 ) . They are socially constructed, growing more out of practice than theory. $^6$
+
+Numbers can be as misleading as words. As Keith Smith ( 2004 ) stresses, indicators are not simple numbers. They tend to come with strings attached; they imply associations with given theories or views of the world that shape the way quantitative information is produced and/or interpreted.
+
+The study of the sources and patterns of technological change has progressed enormously since the mid1960s when the first internationally comparable statistics on R & D activities were published by the OECD, based on the work of Chris Freeman and others. Despite having such a long history, science and technology (S & T) indicators remain contested even today, one paramount reason being the emphasis on inputs rather than outputs and impacts ( Godin , 2003 ) .
+
+Hence, we believe that a trademark-based indicator has the potential for making an additional contribution to the understanding of innovation and industrial change. Trademarks have been used by firms as a means of reinforcing the differentiation of their products, and in this context they can emerge as strongly correlated with innovative efforts.
+
+But, in contrast to our view of innovation as an evolving and recursive process of interaction and feedback, we know that trademark data only refers to two specific events (filing and registration) while saying nothing about interactions, inputs, outcomes or differential impacts. This invites us to follow the recommendation of Martin and Irvine (1983) and take trademarks as a partial indicator of innovative performance, which
+
+6 It must be said that patents are subject to registration processes that are often as judgmental and context-dependent as trademarks ( Doem, 1999 ; Griliches, 1990 ) .
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1391
+
+should be combined with complementary data to produce reliable conclusions. The new data should therefore be treated seriously, together with such opportunities as they might contain for future research ( Pavitt , 1985 ) .
+
+### 3.2. Trademarks as an indicator in recent studies
+
+To our knowledge, the first explicit reference to trademarks as an indicator of innovation is the Germany’s Technological Performance 2001 Report, written on behalf of the Federal Ministry of Education and Research by a group of eight German research centres ( Velling , 2002 ) . The team of authors argue that trademarks can no longer be considered a subordinate industrial right. The fact that the number of trademark registrations filed in Germany tripled during the 1990s is considered a “clear indication that trademarks are being assigned considerably more importance than in the past” ( Velling , 2002 , p. 20) . Although novelty is not a requirement for registering a trademark, the authors say that one can safely assume that trademarks are filed primarily for new products and services. They also point out that commercial enterprises constitute the largest number of applicants. Another aspect emphasised is the extent to which the indicator keeps pace with the market; in Germany it takes just 6 months for a trademark to be entered into databases after application.
+
+In a recent study, Schmoch (2003) also highlighted the suitability of trademarks as an indicator of innovation. Specifically, he argues that “ they meet essential preconditions, in particular correlation to innovation, good data access by electronic databases, and the possibility of operationalising them in relevant dimensions of desegregation ” (p. 155).
+
+### 3.3. Why use trademark-based indicators in innovation studies?
+
+The use of trademarks as a complementary indicator in innovation studies can be justified on both analytical and empirical grounds. Research into innovation has shown that firms use different strategies to protect their innovations. Large-scale industrial surveys carried out over recent decades ( Levin et al. , 1987 ; Cohen et al. , 1996 , 2000 ) have made it clear that firms improve the conditions for appropriating the returns on
+
+their innovations through different channels, including lead time and moving quickly down the learning curve, secrecy, exploiting their reputation and implementing sales and services efforts, or using patents. The ranking of these strategies varies according to the sector of the firm, as well as between product and process innovations. A common finding of these studies is that, as a means of appropriating innovation returns, patents tend to rank lower in these hierarchies, with the exception of a few industries in which they play a strategic role. In contrast, marketing activities and assets tend to play a wider and more significant role.
+
+It was also pointed out long ago that “ R & D activities typically account for half of the expenditures of launching an innovation (excluding normal investment expenditures), the other half being spent on production engineering and marketing ” ( Pavitt , 1985 , p. 81) . These results have been reiterated by many other more recent studies, including the European Community Innovation Surveys. These sorts of findings only confirm the importance of marketing and its tools in connection with innovation activities.
+
+The critical role ascribed to marketing activities and assets within innovation research does not, however, mean that every new trademark is necessarily connected to a new innovative product. As is known, some trademarks are filed to protect products that have no substantive differences in relation to their competitors. But we believe that such applications represent only a minority in the overall demand for new trademarks. This happens because firms would not be able to sustain a trademark (with its associated costs of renewal fees, etc.) if their products had no distinctive advantages or attributes in relation to other offerings in the market. $^7$
+
+Many new trademarks are mainly associated with new consumer products, but they also play a role in the marketing of intermediate inputs and capital goods. As is well known, some renowned brands protect this sort of products (Airbus, Komatsu, Bosch). This variation
+
+7 A distinct situation is to be found when important product innovations are launched in the market together with protective trademarks. After many years of brand-building, these products may still be able to obtain extra profit even when the market has been invaded by clones. An example of this is Bayer aspirin, which has been able to maintain a price differential a long time after the original patent expired and when more than 400 brands of plain aspirin have joined it in the market ( Carlton and Perloff , 1994 , p. 284) .
+
+---
+
+1392
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+Table 2 Indicators of innovative activities
+
+<table><tr><td>Indicator</td><td>Type</td><td>Advantages</td><td>Limitations</td><td>Level of analysis</td><td>Sources</td></tr><tr><td rowspan="3">R&amp;D expenditure</td><td rowspan="3">Input</td><td rowspan="3">Availability Economic variable</td><td>Lack of detail</td><td>Country</td><td>OECD</td></tr><tr><td>Overestimates large electrical and chemical firms</td><td>Industry</td><td>National statistical offices</td></tr><tr><td></td><td></td><td>Firm</td></tr><tr><td rowspan="3">Technical personnel</td><td rowspan="3">Input</td><td rowspan="3">Captures different formal competencies Complement to R&amp;D statistics</td><td>Qualifications are not homogeneous</td><td>Country</td><td rowspan="3">OECD national statistical offices</td></tr><tr><td>Does not capture informal competencies</td><td>S\&amp;T field</td></tr><tr><td></td><td>Industry Firm</td></tr><tr><td rowspan="3">Patents</td><td rowspan="3">Output</td><td rowspan="3">Detailed Regular and very long-run</td><td>Uneven propensities to patent</td><td>Country</td><td>OECD</td></tr><tr><td>Underestimates small firms, design, mechanical, software and service activities</td><td>S\&amp;T field</td><td>National patent and trademark offices</td></tr><tr><td></td><td></td><td>Firm</td><td>European patent office</td></tr><tr><td rowspan="4">Trademarks</td><td rowspan="4">Output</td><td rowspan="3">Captures small firms</td><td>Product classes are considerably aggregated and heterogeneous</td><td>Country</td><td>OHIM</td></tr><tr><td>Captures innovations coming out of universities and public research organizations</td><td>Firm</td><td>WIPO</td></tr><tr><td></td><td></td><td>Product</td><td>National patent and trademark offices</td></tr><tr><td></td><td></td><td>Class</td></tr></table>
+
+
+among types of products and sectors shows some similarities to patenting patterns. But, in contrast to patents, trademarks seem to do particularly well in industries where patenting data provides no reliable information about innovation activities, as in many service sectors and also in low-tech industries where smaller firms contribute to most of the final output.
+
+In sum, we argue that trademarks appear to be highly complementary to other widely used innovation indicators (see Table 2 ). New trademarks are a critical instrument in helping to position new products in the market. When compared to patents, they are closer to commercialisation and cover a broader range of activities from manufacturing product classes to service classes.
+
+### 3.4. The link between innovation and trademarking
+
+There is also empirical evidence showing a correlation between innovation and the use of trademarks.
+
+In his study, Schmoch (2003) finds a highly significant correlation between innovation and trademarks, namely in the manufacturing sector. Focusing on service trademarks, he found considerable differences between sectors, namely in the case of knowledgeintensive services.
+
+The above-mentioned report on the use of IPRs by resident Portuguese firms ( Godinho et al. , 2003 , p. 154) also found sharp and statistically significant trademarking differences across manufacturing sectors according to their technological intensity. Trademark use is not randomly distributed across sectors at the 1 % significance level, with high technology-intensive manufacturing industries being heavy users of trademarks. As far as services are concerned, information-intensive services sectors were also found to be associated with a greater use of trademarks than the low informationintensive sectors at the 5 % significance level.
+
+The results from the Third Community Innovation Survey (CIS 3) offer further supporting evidence. The
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1393
+
+Table 3 CIS 3 results
+
+<table><tr><td></td><td colspan="2">Trademark use (%)</td><td colspan="2">Patent use (%)</td></tr><tr><td></td><td>Innovative firms</td><td>Non-innovative firms</td><td>Innovative firms</td><td>Non-innovative firms</td></tr><tr><td>Belgium</td><td>22</td><td>6</td><td>15</td><td>1</td></tr><tr><td>Denmark</td><td>25</td><td>8</td><td>14</td><td>1</td></tr><tr><td>Germany</td><td>21</td><td>6</td><td>21</td><td>2</td></tr><tr><td>Greece</td><td>23</td><td>6</td><td>6</td><td>0</td></tr><tr><td>Spain</td><td>15</td><td>4</td><td>12</td><td>2</td></tr><tr><td>France</td><td>34</td><td>9</td><td>27</td><td>5</td></tr><tr><td>Ireland</td><td>–</td><td>–</td><td>–</td><td>–</td></tr><tr><td>Italy</td><td>17</td><td>6</td><td>13</td><td>2</td></tr><tr><td>Luxembourg</td><td>19</td><td>10</td><td>8</td><td>1</td></tr><tr><td>The Netherlands</td><td>15</td><td>7</td><td>14</td><td>1</td></tr><tr><td>Austria</td><td>21</td><td>8</td><td>18</td><td>1</td></tr><tr><td>Portugal</td><td>18</td><td>7</td><td>6</td><td>3</td></tr><tr><td>Finland</td><td>25</td><td>5</td><td>20</td><td>2</td></tr><tr><td>Sweden</td><td>41</td><td>15</td><td>28</td><td>5</td></tr><tr><td>United Kingdom</td><td>37</td><td>14</td><td>14</td><td>1</td></tr><tr><td>Iceland</td><td>–</td><td>–</td><td>5</td><td>0</td></tr><tr><td>Norway</td><td>27</td><td>8</td><td>18</td><td>1</td></tr></table>
+
+
+Source: European Commission (2004).
+
+information collected on the use of different forms of protection, such as the registration of patents and trademarks, is presented in Table 3 . The table presents the proportion of firms, for the different EU countries, Iceland and Norway, which made use of patents or trademarks to protect their products. The results for each protection method are presented according to the innovative character of the responding firms.
+
+The CIS results indicate that the use of trademarks is higher than that of patents, which is not surprising. But what is relevant for our argument is that innovative firms consistently use more trademarks and patents. The differences in the use of patents and trademarks between innovative and non-innovative firms are evident. The fact that non-innovative firms report considerably less trademark use than innovative firms is reassuring news in relation to the value of trademarks as an innovation indicator. $^8$
+
+### 3.5. Advantages and disadvantages of trademarks as a product innovation indicator
+
+Given the experience with using patents as an output indicator of technological activities (e.g. Griliches , 1990 ) , we will take advantage of this accumulated knowledge and techniques to explore the potential of trademark data as an indicator of product innovation. As with patents, trademark statistics have the advantage of a reasonably unambiguous legal definition, being collected and classified by (the same) specialised institutions in accordance with international agreements, and long time-series are also available. The basic classification system of trademarks follows from the 1957 Nice Agreement Concerning the International Classification of Goods and Services for the Purposes of the Registration of Marks. The Nice classification system distinguishes between goods and services. It is regularly revised and is now in its 8th edition, which has been in force since January 1, 2002; it has 34 classes of manufactured goods and 11 classes of services (three new classes of services were added in the last edition). One difficulty, however, is that these classes do not have a direct connection with sectoral nomenclatures such as NACE (Statistical Classification of Economic Activities in the European Community). For the purpose of the analysis of innovation and industrial dynamics, the greatest limitation of this classification is that the different classes are highly aggregated. $^9$
+
+Another characteristic of trademark data is that a given trademark for a word or symbol can be requested for either just one or several or even all Nice classes. This means that the number of counts in all classes will be much higher than the total number of trademarks applied for, even if it is possible to identify such multiple classifications. This is a limitation for crosssectoral analysis, and it is different in this respect from other output indicators. Furthermore, trademark applications are not classified according to the main product line or productive sector of the applicant company. Conversely, a given product or supplier can also be protected by more than one trademark. For instance, the Coca-Cola beverage is protected by the word mark
+
+8 The fact that a higher proportion of non-innovative firms stated that they used trademarks (4–15%) rather than patents (0–5%) does not contradict our point, since we are arguing that what should be taken as an indicator is the flow of new trademarks and not the total stock of existing trademarks, which naturally includes many older products that can no longer be considered as “innovations”.
+
+9 For example, Class 5 covers the product categories of pharmaceutical, medical and veterinary preparations, being mixed together with dental wax, disinfectants, fungicides, herbicides and even baby food.
+
+---
+
+1394
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+## Box 2: Examples of overlapping protection us- ing trademarks and other IPRs.
+
+According to the Coca-Cola corporation, its flagship brand is the most widely recognised trademark in the world, with 94 % of the world's population recognising it, and it is also the most widely recognised word after “OK”. The trademark for the word “Coca-Cola” was obtained in 1887. The type of print used for the logo is called “Spencerian Script”. In an effort to safeguard the shape of the bottle, it was registered in 1977 as a threedimensional mark. The argument was that the packaging distinguished the beverage from others.
+
+In another example, Intel has also registered its well-known jingle. The musical notation describing Intel's registered jingle is pictured below.
+
+The relationship between trademarks and other IPRs, namely industrial design and copyright, is gradually evolving. Another aspect of these times of change is the institutional tension created by the disputes between companies over trademarks and domain names on the Internet.
+
+![Figure](figures/figure_006.png)
+
+Sources: www.coca-cola.pt, www.dww.com/ articles/protect.htm
+
+“ Coca-Cola ” , by its distinguishing logo (the stylised letters composing the word underlined by a ribbon) and by a three-dimensional mark protecting the distinctive shape of the bottle. Intel Corporation did not only protect its Intel name with a word mark and a corresponding logo, it also applied for a sound mark (see Box 2 ). These considerations imply that there is no one-to-one correspondence between a new product and a new trademark. Raw data can lead us to overestimate the patterns observed. However, the multiplica-
+
+tion of trademarks in certain product categories surely constitutes evidence of increased dynamic competition, whether through horizontal and/or vertical differentiation.
+
+A further limitation is that there are many unregistered brands in use in the market place, for instance many small firms such as shops, restaurants and the like work under the official firm designation and do not register it as a brand name. This problem is similar to the one that is found in patents: not all inventions can be patented and not all patentable inventions are patented. In the case of trademarks, brands take the place of inventions. Unlike inventions, however, a given brand might be protected by many trademarks (words, logos, 3D mark, sound, etc.) whereas a novel device is supposed to be protected by just one patent. The effect of this on different companies, product categories, industries and countries is not yet fully clear, and other limitations are possibly not yet identified. One lesson to be drawn from patent analysis is that decisions to file an IPR vary among different companies, technologies, industries and countries. Likewise, there is no reason why decisions to trademark should not vary as well.
+
+On the positive side, the large and increasing numbers of trademarks allow us to remain confident that many aspects of corporate commercial activities can be revealed through this indicator. Because they are cheaper and do not require a technological breakthrough, a much wider range of SMEs are likely to be involved in applying for trademark rights compared to patent rights. The nature of the products offered by service companies also make them more appropriate for trademark protection than for patent protection. This allows us to cover a wide range of traded products and a broad spectrum of the industrial structure.
+
+### 3.6. The sources of trademark information
+
+Only with methodological care can trademarks be used as an indirect measure of innovation and a tool for assessing structural change in the economy. So far, trademark data has not been widely used as an indicator of innovative and economic activity. As such, the sources for trademark information have to be found in places other than amongst the standard producers of S & T indicators, like, for example, national institutions, such as statistical offices, or international institutions, such as the OECD. Amongst the national sources, the
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1395
+
+Patent and Trademark Offices are the public agencies traditionally entrusted with registering and keeping the records of trademarks. The problem, however, is that the data supplied by those entities is not easily accessible or ready for statistical treatment, since providing data is not the main mission that these institutions are entrusted with. The major international public agencies responsible for trademarks are WIPO and OHIM. OHIM has published monthly and annual surveys of trademark activity since 1996, broken down according to the applicant's country, year of filing and registration, trademark classes of goods and services and trademark types (word mark, figurative mark, three-dimensional mark, etc). Finally, amongst private sources, there is an increasing variety of firms compiling and selling databases on trademarks, mostly for consultancy purposes.
+
+In this paper, we use CTMs as a source of internationally comparable trademark statistics. Why use OHIM data? This EU trademark system is certainly biased towards the member states' commercial activities, thereby giving rise to an over-representation of European trademarks. However, as the enlarging EU is the world's largest market, it is certainly an important space for protecting brands. Moreover, given the fact that CTMs are attractive to non-EU applicants and that applications from all countries are considered on an equal footing, we have, henceforth, a useful basis for international comparisons. Furthermore, different trademark applications are judged by precisely the same criteria, using the same database. From an instrumental perspective, OHIM presents the researcher with a wealth of comparable trademark data and reports are freely accessible online. It should also be noted that we will focus on the flux of new trademarks and not on the stock of existing trademarks. This perspective will gear our study towards capturing dynamic behaviour and identifying the evolution of the trademarking performances of firms from different countries.
+
+## 4. Empirical exploration of CTM data
+
+This section addresses two main questions: (1) can innovation and industrial change be assessed through trademark data? and (2) can significant differences and distinct trends be identified in the trademarking perfor-
+
+mance of individual countries? We will answer both questions affirmatively. Although the methodological understanding of the use of trademarks is far from mature, our results show that this indicator can generate useful insights for researchers, policy-makers and managers.
+
+This sub-section presents data on the rate and direction of CTM applications. We start with the basic observation of aggregate trademark applications. Fig. 1 shows the total number of CTM applications since the EU Community Trademark was instituted. In 7 years, 294,625 CTMs were applied for by companies from the EU and around the world, corresponding to an average of about 42,000 trademarks a year. Of this total, 65 % were word marks and 34 % were figurative marks (logos).
+
+The high level of trademark applications in the first year (1996) is explained by the start of the new registration system, since many firms were waiting for its establishment. If we compute the average rate of growth for the period 1996 – 2002, we have a 0.6 % change. But if we disregard the first year figure, there has been a growth of 8.7 % in annual applications. Later years reveal a downward trend in the number of applications, a phenomenon probably caused by the general macroeconomic downturn and the readjustment of business investment expectations.
+
+What about the territorial origin of those applications? As expected, Table 4 shows that applications are dominated by the EU-15 member states.
+
+Within the EU-15 countries, trademark applications are heavily concentrated ( Table 5 ). Germany, the UK and Italy alone account for 59 % of the EU-15 total applications in the period from 1996 to 2002. The first five countries in trademark applications correspond to the largest European economies and are responsible for 80 % of all applications. Outside the EU-15 group, the
+
+Table 4 Origin of CTM applications, 1996-2002
+
+<table><tr><td></td><td colspan="2">Annual average 1996–2002</td></tr><tr><td></td><td>Number of applications</td><td>Share (%)</td></tr><tr><td>EU-15</td><td>26,216</td><td>62</td></tr><tr><td>Non-EU countries</td><td>15,874</td><td>38</td></tr><tr><td>Total</td><td>42,089</td><td>100</td></tr></table>
+
+
+Source: OHIM (2003).
+
+---
+
+1396
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+![Figure](figures/figure_002.png)
+
+Fig. 1. CTM applications, all countries, 1996-2002. Source: OHIM (2004).
+
+US represents 62% of the applications made in this period. The US is the biggest single user of the CTM system, with 25% of all trademark applications in the period under analysis.
+
+The normalisation of these figures in keeping with population size and the size of the economy provides yet further information. Besides Luxembourg, where a high number of MNC headquarters are located, a number of small countries appear at the top, including the Nordic countries, Ireland and Austria. In the European context, their pattern of business activity appears to be trademark-intensive relative to their population size
+
+and GDP. The UK, and to some extent Germany, perform particularly well among the largest countries. $^10$
+
+Table 5
+
+
+Trademark performance among the EU-15, 1996–2002
+
+<table><tr><td colspan="2">1996–2002</td><td colspan="2">CTM/population</td><td colspan="2">CTM/GDP</td></tr><tr><td>DE</td><td>48,667</td><td>LU</td><td>3,383</td><td>LU</td><td>72</td></tr><tr><td>GB</td><td>38,524</td><td>DK</td><td>870</td><td>DK</td><td>29</td></tr><tr><td>IT</td><td>21,432</td><td>IE</td><td>732</td><td>SE</td><td>29</td></tr><tr><td>ES</td><td>19,438</td><td>SE</td><td>714</td><td>GB</td><td>26</td></tr><tr><td>FR</td><td>18,355</td><td>GB</td><td>645</td><td>IE</td><td>25</td></tr><tr><td>NL</td><td>7,641</td><td>DE</td><td>592</td><td>ES</td><td>24</td></tr><tr><td>SE</td><td>6,331</td><td>AT</td><td>545</td><td>DE</td><td>23</td></tr><tr><td>DK</td><td>4,642</td><td>FI</td><td>538</td><td>FI</td><td>21</td></tr><tr><td>AT</td><td>4,417</td><td>ES</td><td>493</td><td>AT</td><td>20</td></tr><tr><td>BE</td><td>4,253</td><td>NL</td><td>482</td><td>NL</td><td>17</td></tr><tr><td>FI</td><td>2,785</td><td>BE</td><td>414</td><td>BE</td><td>16</td></tr><tr><td>IE</td><td>2,773</td><td>IT</td><td>375</td><td>IT</td><td>15</td></tr><tr><td>PT</td><td>1,921</td><td>FR</td><td>312</td><td>FR</td><td>13</td></tr><tr><td>LU</td><td>1,485</td><td>PT</td><td>192</td><td>PT</td><td>11</td></tr><tr><td>GR</td><td>845</td><td>GR</td><td>80</td><td>GR</td><td>5</td></tr><tr><td>EU</td><td>183,509</td><td>EU</td><td>488</td><td>EU</td><td>20</td></tr></table>
+
+
+Sources : OHIM (various years); OECD (population and GDP). Note : BE—Belgium; DK—Denmark; DE—Germany; GR— Greece; ES—Spain; FR—France; IE—Ireland; IT—Italy; LU— Luxembourg; NL—The Netherlands; AT—Austria; PT—Portugal; Fl—Finland; SE—Sweden; GB—United Kingdom.
+
+In dynamic terms, the overall EU performance has been ahead of the “rest of the world”. The annual average of CTM applications, for the EU-15, during the last 3 years of the period analysed (2000–2002) was 55.8 % greater than the average for the first 3 years (1996–1998), as revealed in Table 6 . Within the EU, Greece, Luxembourg and Ireland, followed by Portugal, France, Italy and Spain have been increasingly active in applying for new CTMs. Belgium and The Netherlands have had the lowest growth rates during this period.
+
+Heterogeneity is also evident in terms of the distribution of applications by Nice classes. Table 7 shows the ten most “ trademarked ” product classes between 1996 and 2002. These accounted for 53.2 % of total applications. In 2002, the most sought after product categories were: instruments (Class 9), research (Class 42), business consultancy (Class 35), paper products (Class 16) and education (Class 41). It is interesting to note that service categories account for half of the most trademarked classes. Among these, the classes with the highest level of use broadly correspond to what Miles
+
+10 We should stress that European patent data points to a similar pattern ( Eurostat , 2002 ) . Within the EU, Germany accounts for the largest share of all patent applications at the European Patent Office (EPO) with 42.4 % . France and the UK accounted for 14.4 % and 12.9 % , respectively, showing that European patent applications at the EPO are largely skewed towards the large economies. However, when population size is taken into consideration, we again have the small Scandinavian economies outperforming the larger economies, with Sweden and Finland displaying the highest rates.
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1397
+
+Table 6 Detailed CTM annual data for the EU-15, 1996–2002
+
+<table><tr><td></td><td>1996</td><td>1997</td><td>1998</td><td>1999</td><td>2000</td><td>2001</td><td>2002</td><td>Growth (%) between annual averages 2000–2002/1996–1998</td></tr><tr><td>BE</td><td>643</td><td>418</td><td>562</td><td>574</td><td>767</td><td>687</td><td>598</td><td>26.4</td></tr><tr><td>DK</td><td>706</td><td>473</td><td>526</td><td>621</td><td>813</td><td>767</td><td>727</td><td>35.3</td></tr><tr><td>DE</td><td>7,714</td><td>3,911</td><td>4,811</td><td>7,040</td><td>9,935</td><td>8,285</td><td>7,113</td><td>54.1</td></tr><tr><td>GR</td><td>85</td><td>65</td><td>78</td><td>111</td><td>163</td><td>148</td><td>197</td><td>122.8</td></tr><tr><td>ES</td><td>2,794</td><td>1,468</td><td>1,965</td><td>2,609</td><td>3,466</td><td>3,426</td><td>3,759</td><td>71.0</td></tr><tr><td>FR</td><td>1,610</td><td>1,891</td><td>2,131</td><td>2,735</td><td>3,402</td><td>3,171</td><td>3,411</td><td>77.2</td></tr><tr><td>IE</td><td>282</td><td>217</td><td>289</td><td>436</td><td>640</td><td>484</td><td>429</td><td>97.1</td></tr><tr><td>IT</td><td>2,211</td><td>2,070</td><td>2,378</td><td>3,263</td><td>4,227</td><td>3,570</td><td>3,719</td><td>72.9</td></tr><tr><td>LU</td><td>137</td><td>112</td><td>148</td><td>227</td><td>281</td><td>279</td><td>233</td><td>99.7</td></tr><tr><td>NL</td><td>1,069</td><td>908</td><td>877</td><td>994</td><td>1,518</td><td>1,106</td><td>1,055</td><td>28.9</td></tr><tr><td>AT</td><td>678</td><td>393</td><td>537</td><td>615</td><td>739</td><td>711</td><td>717</td><td>34.8</td></tr><tr><td>PT</td><td>161</td><td>183</td><td>232</td><td>274</td><td>350</td><td>297</td><td>378</td><td>78.0</td></tr><tr><td>FI</td><td>358</td><td>282</td><td>330</td><td>387</td><td>513</td><td>483</td><td>438</td><td>47.8</td></tr><tr><td>SE</td><td>855</td><td>713</td><td>787</td><td>870</td><td>1,375</td><td>959</td><td>801</td><td>32.2</td></tr><tr><td>GB</td><td>5,705</td><td>3,659</td><td>4,234</td><td>5,303</td><td>7,930</td><td>6,141</td><td>5,860</td><td>46.6</td></tr><tr><td>EU</td><td>25,008</td><td>16,763</td><td>19,885</td><td>26,059</td><td>36,119</td><td>30,514</td><td>29,435</td><td>55.8</td></tr></table>
+
+
+Source: OHIM.
+
+(2004) and Godinho et al. (2003) , respectively classify as knowledge-intensive business services (KIBS) and information-intensive service sectors.
+
+How do intangible products measure up to tangible ones, on the whole? Table 8 shows that, while classes of goods still represent the major segment, services have generally been increasing in importance. This might be interpreted as evidence of structural change in the EU
+
+economies. 11 The study by Velling and his colleagues (2002) finds a similar pattern for German home trademark applications. Godinho et al. (2003) report similar conclusions for the Portuguese case (cf. the next section). 12
+
+Table 7 Most “ trademarked ” Nice product classes in CTMs
+
+<table><tr><td></td><td>Nice class</td><td>Accumulated applications 1996–2002</td><td>Share (%)</td></tr><tr><td>1</td><td>9. Instruments</td><td>92,335</td><td>11.9</td></tr><tr><td>2</td><td>42. Research and other services</td><td>72,661</td><td>9.4</td></tr><tr><td>3</td><td>16. Paper products</td><td>52,956</td><td>6.8</td></tr><tr><td>4</td><td>35. Advertising and business consultancy</td><td>48,476</td><td>6.2</td></tr><tr><td>5</td><td>41. Education</td><td>39,172</td><td>5.0</td></tr><tr><td>6</td><td>25. Clothing and footwear</td><td>35,340</td><td>4.6</td></tr><tr><td>7</td><td>38. Telecommunications</td><td>34,089</td><td>4.4</td></tr><tr><td>8</td><td>5. Pharmaceutical and hygiene products</td><td>25,638</td><td>3.3</td></tr><tr><td>9</td><td>36. Finance</td><td>24,415</td><td>3.1</td></tr><tr><td>10</td><td>3. Detergents and cosmetics</td><td>21,271</td><td>2.7</td></tr><tr><td>Total</td><td></td><td></td><td>57.5</td></tr></table>
+
+
+Source: OHIM.
+
+If services are a dynamic category, how have they changed in terms of structure? A look at the individual service classes can give further insight into the evolution of service industries and their relative dynamics. As Table 9 shows, the service industries with most trademark applications are all part of the knowledgebased services: research (Class 42); business consultancy and advertising (Class 35); education (Class 41); and telecommunications (Class 38). The Nice classes growing above the average of the services sector are also part of knowledge-based services: business consultancy and advertising (Class 35) and telecommunications (Class 38).
+
+Two difficulties do, however, limit our capacity to produce more precise conclusions. On one hand, as
+
+11 The latest OECD figures show that the service sectors now account for 70 % of the OECD's GDP ( OECD , 2003 ) .
+
+12 A suggestion advanced in the latter study is that a contributory factor to the spectacular increase in service trademarks in the 1990s was the rise in this type of trademark application amongst manufacturing companies, which it is not possible to confirm with these data.
+
+---
+
+1398
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+Table 8
+
+
+Sectoral dynamics of CTM “ trademarking ”
+
+<table><tr><td></td><td colspan="7">Share (%)</td><td rowspan="2">Growth (%) between annual averages 2000–2002/1996–1998</td></tr><tr><td></td><td>1996</td><td>1997</td><td>1998</td><td>1999</td><td>2000</td><td>2001</td><td>2002</td></tr><tr><td>Goods</td><td>76</td><td>72</td><td>71</td><td>66</td><td>57</td><td>62</td><td>66</td><td>37</td></tr><tr><td>Services</td><td>24</td><td>28</td><td>29</td><td>34</td><td>43</td><td>38</td><td>34</td><td>150</td></tr></table>
+
+
+Source: OHIM.
+
+Table 9
+
+Composition of CTM service class applications
+
+<table><tr><td>Nice class</td><td>Descriptor</td><td>1996–2002</td><td>Share of service classes (%)</td><td>Growth (%) between annual averages 2000–2002/1996–1998</td></tr><tr><td>35</td><td>Business consultancy</td><td>49,564</td><td>18.7</td><td>218</td></tr><tr><td>36</td><td>Finance</td><td>24,025</td><td>9.1</td><td>144</td></tr><tr><td>37</td><td>Construction</td><td>17,814</td><td>6.7</td><td>66</td></tr><tr><td>38</td><td>Telecommunications</td><td>33,654</td><td>12.7</td><td>244</td></tr><tr><td>39</td><td>Transportation</td><td>17,346</td><td>6.6</td><td>94</td></tr><tr><td>40</td><td>Processes</td><td>6,607</td><td>2.5</td><td>94</td></tr><tr><td>41</td><td>Education</td><td>39,172</td><td>14.8</td><td>124</td></tr><tr><td>42</td><td>Research</td><td>72,661</td><td>27.4</td><td>136 a</td></tr><tr><td>43</td><td>Restaurants and hotels</td><td>1,883</td><td>0.7</td><td>–</td></tr><tr><td>44</td><td>Medical care</td><td>1,501</td><td>0.6</td><td>–</td></tr><tr><td>45</td><td>Personal services</td><td>506</td><td>0.2</td><td>–</td></tr><tr><td>Total services</td><td></td><td>264,733</td><td>100</td><td>150</td></tr></table>
+
+
+Source: OHIM.
+
+a “Trend” figures for research (Class 42) include the 2002 data for Classes 43–45, which have only existed as a breakdown of the former Class 42 since 2002.
+
+b Only 2002 data.
+
+already mentioned, Nice classes are highly aggregated, containing many different kinds of products under the same heading. On the other hand, the short life of this EU IPR regime calls for caution in the discussion of the trends and comparative dynamics of product classes.
+
+## 5. Patterns of use of national trademarks by Portuguese firms13
+
+In this section, we explore the data for the Portuguese case in greater detail, drawing on an extensive
+
+study on IPR use in Portugal ( Godinho et al. , 2003 ) . The information provided by this study has the distinct advantage of combining publicly available statistics of Portuguese national trademarks with data from a firmbased survey of business attitudes and behaviour towards IPRs. In this section, we will concentrate on a relatively recent period, from 1980 to 2001, and on the use of national trademarks (those that provide protection exclusively within the Portuguese territory).
+
+With a yearly growth of about 10 % a year, trademarks granted to resident companies by the Portuguese Patent and Trademark Office (INPI) surpassed annual GDP growth. Portuguese trademarkers are mostly business firms (80 % ), the rest being private individuals (most of whom are likely to be owners of small firms). During this period, concession rates (trademarks registered on the base of trademarks applied for) were 86 % for residents and 96 % for non-residents. Nonresidents tend to have trademarks that last longer: the total dropout rate for this group in the period men-
+
+13 Trademark use in Portugal has a long history. The oldest trademark in Portugal still in use today was registered on February 20, 1890. The trademark Real Companhia Vintcola do Norte de Portugal was given to the company with the same name for the specific use of stamping it on Port wine casks (this trademark was registered as an International Trademark in 1925). Among the 10 oldest registered trademarks, the majority of them correspond to the wine and oliveoil businesses, reflecting the productive profile of the Portuguese economy at the time.
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1399
+
+Table 10 Distribution of Portuguese national trademark filings
+
+<table><tr><td></td><td colspan="3">Residents</td><td colspan="3">Non-residents</td></tr><tr><td></td><td>Goods (%)</td><td>Services (%)</td><td>Total</td><td>Goods (%)</td><td>Services (%)</td><td>Total</td></tr><tr><td>1980–1984</td><td>89.7</td><td>10.3</td><td>4,665</td><td>87.8</td><td>12.2</td><td>4,435</td></tr><tr><td>1985–1989</td><td>87.1</td><td>12.9</td><td>9,285</td><td>87.4</td><td>12.6</td><td>9,971</td></tr><tr><td>1990–1994</td><td>80.5</td><td>19.5</td><td>22,669</td><td>84.6</td><td>15.4</td><td>22,563</td></tr><tr><td>1995–1999</td><td>66.3</td><td>33.7</td><td>31,279</td><td>78.9</td><td>21.1</td><td>17,700</td></tr><tr><td>2000–2002 a</td><td>51.0</td><td>49.0</td><td>27,751</td><td>72.4</td><td>27.6</td><td>7,575</td></tr><tr><td>Total</td><td>68.4</td><td>31.6</td><td>95,649</td><td>82.2</td><td>17.8</td><td>62,244</td></tr></table>
+
+
+Source: Godinho et al. (2003). a Until June 2002.
+
+tioned was 11 % , while for residents it was 19 % . Nonresidents, however, have a lower growth rate of trademark filings, as the curve of their applications peaked just after Portugal became a member of the European Community in 1986 and then flattened out. Among non-residents, USA is the largest applicant with 39 % of all non-resident applications, followed by the UK (15 % ), Spain (12 % ), Japan (6 % ), France (4 % ), Germany (3 % ) and Brazil (3 % ). On the whole, EU-15 filings account for 42 % of the total non-resident trademarking. From the mid-1990s onwards, Spanish and Brazilian companies became the most dynamic foreign trademarkers.
+
+Portuguese national trademarks are also classified on the basis of the Nice Agreement. The trademark classes most sought after by residents are pharmaceutical and hygiene products (Class 5), instruments (Class 9), detergents and cosmetics (Class 3), paper products (Class 16), and clothing and footwear (Class 25). This pattern reveals some differentiation by trademark filings between residents and non-residents and, above all, an overall predominance of goods classes in relation to services. However, over the last decade, some changes have been witnessed, with several service classes showing a much higher demand, namely Class 41 (education), Class 42 (research) and Class 35 (business consultancy). As in the case of the CTM applications observed in the previous section, these transformations are essentially characterised by a gradual fall in the proportion of tangible goods and by increases in service products, with trademark filings now divided equally between goods and services. Table 10 shows this pattern. Also as expected, non-residents do not trademark as much in service classes as residents. This
+
+is due to the fact that a good deal of service products are non-tradable. $^14$
+
+The information about the use of IPRs by a representative sample of 724 resident firms, collected by a survey carried out in 2003, shows that the interest in and actual use of IPRs is still very limited. Only 39 % of business firms claim to have any knowledge of INPI's activities, while a smaller proportion (19 % ) acknowledges actual use of IPRs or the intention to use them in the near future. As indicated in Godinho et al. ( 2003 ) , these values have a statistically significant variance across sectors and display a positive association with firm size. These results are not surprising. They stem from a variety of reasons, the most important ones being: (1) the structure of the Portuguese economy (a relatively low weight of both high-tech industries and information-intensive services and a virtually complete absence of large firms acting on global markets); (2) a low supply of critical competencies (required to explore IPRs commercially and manage complementarities within IPR portfolios); (3) low access both to lead users and to technology markets. This last hurdle inhibits the capacity to commercially exploit any IPR undertaken by firms, inventors or public research organisations.
+
+In terms of strategic attitudes towards IPR, more than half of the surveyed firms expressed a “low interest” in IP issues with responses varying between 54.0 % and 86.7 % for different IPR types. Trademarks are the most attractive IPR instrument, with 17.8 % of the
+
+14 This pattern is likely to see some changes in the future given that negotiations are now underway for the expansion of the General Agreement on Trade in Services (GATS).
+
+---
+
+1400
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+firms showing “ high ” interest and an additional 28.2 % “ medium ” interest. In terms of actual use, 4.2 % of the surveyed firms state that they have applied for patents or utility models, while 17.7 % applied for trademarks and 2.6 % for protection related to technical designs and industrial models.
+
+As reported in Godinho et al. ( 2003 ) , a number of variables reveal a statistically significant degree of association with the importance attributed to trademarks or their effective use in the past. These factors include the existence of an autonomous marketing department in the company, the relationship with specialised consultancies, the size of marketing and R & D budgets, or the knowledge of the national IPR agency's services and role. However, whether a firm is geared to intermediary products or final consumer products does not discriminate the firms in terms of the importance attributed to trademarks, but only in terms of their actual use: consumer goods firms tend to use more trademarks.
+
+Within each sector, the results generally indicate a positive correlation between the use of patents and the use of trademarks, which is in line with the hypothesis of using trademark filings as a complementary indicator of innovation. The results from the survey mentioned above reveal that the OECD technological intensity classification discriminates particularly well in relation to the use of trademarks by Portuguese firms. For manufacturing as a whole, the differences between the OECD sectors are statistically significant for the importance attributed to, as well as the actual employment of, both patents and, particularly, trademarks during the past 10 years. Analysing the sectors pair-wise, the differences are found to be significant for the use of trademarks in manufacturing industries (with the exception of the difference in the use of trademarks between medium-low-tech and low-tech manufacturing sectors). This means that “ higher-tech ” sectors tend to care about, and actually use, more trademarks in the course of their business activities than “ lower-tech ” sectors. In spite of the fact that the “ higher-tech ” equivalence to innovative industries is not at all unproblematic, $^15$ the
+
+fact that “ higher-tech ” sectors tend to have their core competencies in fast-growing technologies lends additional support to the claim that trademarks can be used as an ancillary yardstick in innovation studies.
+
+As far as the service sectors are concerned, highly information-intensive services, such as consultancies, telecommunications and banking are also found to make greater use of trademarks in their business strategies than low information-intensive sectors (at 10 % level). It should be noted that the analysis does not provide the same conclusion regarding differences in the use of patents between the high and low informationintensive service sectors. This suggests that trademarks can reveal aspects of innovation and industrial dynamics that are not fully apprehended by patent analysis, namely within the service sector.
+
+## 6. Concluding remarks
+
+The question this paper set out to answer was “what can we learn from trademarks as an indicator of innovation and industrial evolution?” The objective was to test trademarks as an indicator of product innovation activity and as a measure correlated with structural change in contemporary economies.
+
+Brands are a very important part of firms' marketing plans and strategies. They are used to protect firms' products and business identity, but also for other purposes, such as product differentiation and business diversification. Firms make a huge (and increasing) use of brands as a (dynamic) competition tool. As a result, applications for service trademarks boomed in the 1990s, and this trend was led by information-(or knowledge)based services such as Business Consultancy, Telecommunications and Education. Within the EU-15 group, a number of small countries seem to exhibit very strong marketing capabilities, namely the Nordic countries, plus Ireland and Austria.
+
+The analysis of more detailed data on Portugal revealed that the country is lagging behind in terms of marketing capabilities, which are critical for supporting innovation and trade competitiveness in external markets. However, Portugal is showing clear signs of
+
+15 Recent contributions have questioned the OECD classification, mainly on the grounds that “ lower-tech ” sectors should not be regarded as incapable of innovating ( Smith , 2002 ; von Tunzelmann and Acha , 2004 ) . Hence, “ high-technology industries ” and “ hightechnology ” should not be confused. However, the fact that “ lowertech ” sectors are increasingly competent in the new technologies
+
+of the emergent techno-economic paradigm does not mean that the “ higher-tech ” sectors are acquiring knowledge in older and more mature areas ( Mendonça and von Tunzelmann , 2004 ) .
+
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1401
+
+structural change. It has been following the general trend towards an increase in service trademarking. This observation is compatible with the evidence from the CTM database on the dynamism of knowledge services in recent years. Moreover, the composition of this trend is biased towards education, research and business consultancy categories. Evidence from the Portuguese case suggests that companies which tend to use one kind of IPR also use other IPRs. This is in keeping with the CIS results that were analysed in Section 3 . And it further implies that high-technology sectors, which use more patents, also make a more intensive use of trademarks. The data from a survey of Portuguese firms also shows that the service industries usually classified as intensive users of information are the ones that use most trademarks.
+
+Combined with the increasing availability of electronic data, these results indicate an interesting opportunity for using trademarks as indicators of innovation and industrial change. We have argued that trademark data can serve the purpose of acting as a partial output indicator of innovations introduced into the goods and services markets and can therefore be used as an empirical yardstick for measuring overall changes in the patterns of economic activity. This can be especially useful for advancing research in innovation studies, industrial dynamics and international economics, as well as in economic and business history. We therefore conclude that new knowledge about innovation and industrial change can be acquired by including trademark analysis in the box of empirical tools. However, more work needs to be done in order to better understand the potential uses and limitations of this new source of data. The fact that much still remains to be learned is the best of the good news.
+
+There are a variety of ways in which trademark data can be further analysed. Aspects that can be researched include the life cycle of brands, as well as factors and trends in trademark licensing. The relationship between trademarks and other IPRs in the context of integrated IPR management can be further explored using quantitative data. The cross analysis of trademark data, regarding the ways in which different trademarks (word marks, logos, 3D, sound, etc.) are combined in order to protect given products or applications across different Nice classes (requiring extensive data cleaning and database management) is promising, but not very clearly understood. Seen from this perspective,
+
+trademarks can also be taken as indicators of the marketing capabilities of profit-oriented firms. Studying how firms exploit trademark positions in certain classes in order to make inroads into other classes could also yield valuable insights into the dynamics of product diversification. Trademark analysis could also be of help for those researching the innovative performance of traditional industries and intermediate industries. Since “lower-tech” firms are found to be patenting in “high-technologies”, they may also be found to be trademarking in “higher-tech” product classes. In a parallel way, scholars studying the dynamics of service innovation may find it fruitful, for instance, to investigate those long-established manufacturing firms that are increasingly active in service trademarking. Finally, the extent to which trademark data can contribute towards our understanding of why some countries grow faster than others constitutes a research challenge that is worth exploring in the future. In sum, our paper claims that trademarks display a high potential for revealing new stylised facts and for illuminating puzzles about innovation that are still in need of explanation. This potential needs to be realised with further theoretical, methodological and empirical work.
+
+## Acknowledgements
+
+The authors are grateful to the Portuguese Patent and Trademark Office (INPI) for permission to use data generated for a study on the use of industrial property in Portugal. Earlier versions of this paper were presented at the Conferences “What Do We Know About Innovation?—Conference in honour of Keith Pavitt”, SPRU, University of Sussex, 13–15 November 2003; “Economic Policies in the New Millennium”, Coimbra University, Portugal, 16–17 April 2004; Schumpeter Society Conference, “Innovation, Industrial Dynamics and Structural Transformation: Schumpeterian Legacies”, Bocconi University, Italy, 9–12 June 2004. Comments and suggestions by participants at these conferences, especially Scott Stern, our panel discussant at the Pavitt Conference, by three anonymous referees and by Richard Nelson, Virginia Acha and Orietta Marsili, the editors of this special issue, were extremely valuable in improving the paper. Any remaining errors or omissions are entirely our own responsibility.
+
+---
+
+1402
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+## Appendix A. Nice classification system
+
+- 1. Chemicals used in industry, science and pho-
+tography, as well as in agriculture, horticulture
+and forestry; unprocessed artificial resins, unpro-
+cessed plastics; manures; fire extinguishing com-
+positions; tempering and soldering preparations;
+chemical substances for preserving foodstuffs;
+tanning substances; adhesives used in industry.
+2. Paints, varnishes, lacquers; preservatives against
+rust and against deterioration of wood; colorants;
+mordants; raw natural resins; metals in foil and
+powder form for painters, decorators, printers and
+artists.
+3. Bleaching preparations and other substances for
+laundry use; cleaning, polishing, scouring and
+abrasive preparations; soaps; perfumery, essential
+oils, cosmetics, hair lotions; dentifrices.
+4. Industrial oils and greases; lubricants; dust absorb-
+ing, wetting and binding compositions; fuels (in-
+cluding motor spirit) and illuminants; candles and
+wicks for lighting.
+5. Pharmaceutical and veterinary preparations; san-
+itary preparations for medical purposes; dietetic
+substances adapted for medical use, food for ba-
+bies; plasters, materials for dressings; material for
+stopping teeth, dental wax; disinfectants; prepa-
+rations for destroying vermin; fungicides, herbi-
+cides.
+6. Common metals and their alloys; metal building
+materials; transportable buildings of metal; mate-
+rials of metal for railway tracks; non-electric ca-
+bles and wires of common metal; ironmongery,
+small items of metal hardware; pipes and tubes of
+metal; safes; goods of common metal not included
+in other classes; ores.
+7. Machines and machine tools; motors and engines
+(except for land vehicles); machine coupling and
+transmission components (except for land vehi-
+cles); agricultural implements other than hand-
+operated; incubators for eggs.
+8. Hand tools and implements (hand operated); cut-
+lery; side arms; razors.
+9. Scientific, nautical, surveying, photographic, cin-
+ematographic, optical, weighing, measuring, sig-
+nalling, checking (supervision), life-saving and
+teaching apparatus and instruments; apparatus and
+instruments for conducting, switching, transform-
+- ing, accumulating, regulating or controlling elec-
+tricity; apparatus for recording, transmission or re-
+production of sound or images; magnetic data car-
+riers, recording discs; automatic vending machines
+and mechanisms for coin-operated apparatus; cash
+registers, calculating machines, data processing
+equipment and computers; fire-extinguishing ap-
+paratus.
+
+10. Surgical, medical, dental and veterinary apparatus
+and instruments, artificial limbs, eyes and teeth;
+orthopaedic articles; suture materials.
+11. Apparatus for lighting, heating, steam generating,
+cooking, refrigerating, drying, ventilating, water
+supply and sanitary purposes.
+12. Vehicles; apparatus for locomotion by land, air or
+water.
+13. Firearms; ammunition and projectiles; explosives;
+fireworks.
+14. Precious metals and their alloys and goods in pre-
+cious metals or coated therewith, not included in
+other classes; jewellery, precious stones; horolog-
+ical and chronometric instruments.
+15. Musical instruments.
+16. Paper, cardboard and goods made from these
+materials, not included in other classes; printed
+matter; bookbinding material; photographs; sta-
+tionery; adhesives for stationery or household pur-
+poses; artists' materials; paint brushes; typewriters
+and office requisites (except furniture); instruc-
+tional and teaching material (except apparatus);
+plastic materials for packaging (not included in
+other classes); printers' type; printing blocks.
+17. Rubber, gutta-percha, gum, asbestos, mica and
+goods made from these materials and not included
+in other classes; plastics in extruded form for use
+in manufacture; packing, stopping and insulating
+materials; flexible pipes, not of metal.
+18. Leather and imitations of leather, and goods made
+of these materials and not included in other classes;
+animal skins, hides; trunks and travelling bags;
+umbrellas, parasols and walking sticks; whips, har-
+ness and saddlery.
+19. Building materials (non-metallic); non-metallic
+rigid pipes for building; asphalt, pitch and bitumen;
+non-metallic transportable buildings; monuments,
+not of metal.
+20. Furniture, mirrors, picture frames; goods (not in-
+cluded in other classes) of wood, cork, reed, cane,
+---
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+1403
+
+wicker, horn, bone, ivory, whalebone, shell, amber, mother-of-pearl, meerschaum and substitutes for all these materials, or of plastics.
+
+21. Household or kitchen utensils and containers (not of precious metal or coated therewith); combs and sponges; brushes (except paint brushes); brushmaking materials; articles for cleaning purposes; steel wool; unworked or semi-worked glass (except glass used in building); glassware, porcelain and earthenware not included in other classes.
+
+22. Ropes, string, nets, tents, awnings, tarpaulins, sails, sacks and bags (not included in other classes); padding and stuffing materials (except of rubber or plastics); raw fibrous textile materials.
+
+23. Yarns and threads, for textile use.
+
+24. Textiles and textile goods, not included in other classes; bed and table covers.
+
+25. Clothing, footwear, headgear.
+
+26. Lace and embroidery, ribbons and braid; buttons, hooks and eyes, pins and needles; artificial flowers.
+
+27. Carpets, rugs, mats and matting, linoleum and other materials for covering existing floors; wall hangings (non-textile).
+
+28. Games and playthings; gymnastic and sporting articles not included in other classes; decorations for Christmas trees.
+
+29. Meat, fish, poultry and game; meat extracts; preserved, dried and cooked fruits and vegetables; jellies, jams, fruit sauces; eggs, milk and milk products; edible oils and fats.
+
+30. Coffee, tea, cocoa, sugar, rice, tapioca, sago, artificial coffee; flour and preparations made from cereals, bread, pastry and confectionery, ices; honey, treacle; yeast, baking-powder; salt, mustard; vinegar, sauces (condiments); spices; ice.
+
+31. Agricultural, horticultural and forestry products and grains not included in other classes; live animals; fresh fruits and vegetables; seeds, natural plants and flowers; foodstuffs for animals, malt.
+
+32. Beers; mineral and aerated waters and other nonalcoholic drinks; fruit drinks and fruit juices; syrups and other preparations for making beverages.
+
+33. Alcoholic beverages (except beers).
+
+34. Tobacco; smokers' articles; matches.
+
+35. Advertising; business management; business administration; office functions.
+
+36. Insurance; financial affairs; monetary affairs; real estate affairs.
+
+37. Building construction; repair; installation services.
+
+38. Telecommunications.
+
+39. Transport; packaging and storage of goods; travel arrangement.
+
+40. Treatment of materials.
+
+41. Education; providing of training; entertainment; sporting and cultural activities.
+
+42. Scientific and technological services and research and design relating thereto; industrial analysis and research services; design and development of computer hardware and software; legal services.
+
+43. Services for providing food and drink; temporary accommodation.
+
+44. Medical services; veterinary services; hygienic and beauty care for human beings or animals; agriculture, horticulture and forestry services.
+
+45. Personal and social services rendered by others to meet the needs of individuals; security services for the protection of property and individuals.
+
+## References
+
+Aaker, D.A., 1991. Managing Brand Equity: Capitalizing on the Value of a Brand. The Free Press, New York.
+
+Cabral, L., 2000. www.huiscabral.com.
+
+Carlton, D., Perloff, J., 1994. Modern Industrial Organization, second ed. HarperCollins, New York.
+
+Cohen, W., Nelson, R., Walsh, J., 1996. Appropriability conditions and why firms patent and why they do not in the American manufacturing sector. Paper Presented to the OECD Conference "New S&T indicators for the Knowledge-Based Economy", Paris, 19-21 June.
+
+Cohen, W., Nelson, R., Walsh, J., 2000. Protecting their intellectual assets: appropriability conditions and why U.S. manufacturing firms patent or not. NBER Working Paper Series 7552.
+
+Doern, G.B., 1999. Global Change and Intellectual Property Agencies. Pinter, London.
+
+Economides, N., 1987. The economics of trademarks. TradeMark Register 78, 523-539.
+
+European Commission, 2004. Innovation in Europe: Results for the EU, Iceland and Norway, Data 1998–2001, Panorama of the European Union, 2004 Edition. Office for Official Publications of the European Communities, Luxembourg.
+
+Eurostat, 2002. Statistics in Focus: Science & Technology. Theme 9.1, European Communities, Brussels.
+
+Godin, B., 2003. The emergence of S&T indicators: why did governments supplement statistics with indicators? Research Policy 32, 679-691.
+
+---
+
+1404
+
+S. Mendonça et al. / Research Policy 33 (2004) 1385-1404
+
+Godinho, M.M., Pereira, T.S., Simões, V.C., Mendonça, S., Sousa, V., 2003. Estudo Sobre a Utilização da Propriedade Industrial em Portugal. INPI, Lisboa.
+
+Granstrand, Ö., 1999. The Economics and Management of Intellectual Property. Edward Elgar, Cheltenham, Northampton.
+
+Griliches, Z., 1990. Patent statistics as economic indicators: a survey. Journal of Economic Literature 27, 1661-1707.
+
+Klein, N., 2000. No Logo. Flamingo, London.
+
+Kotler, P., Armstrong, G., Saunders, J., Wong, V., 1999. Principles of Marketing, second European ed. Prentice Hall, Europe.
+
+Levin, R., Klevorick, A., Nelson, R., Winter, S., 1987. Appropriating the Returns from Industrial Research and Development. Brookings Papers on Economic Activity: Microeconomics 3, 783-831.
+
+Martin, B., Irvine, J., 1983. Assessing basic research: some partial indicators of scientific progress in radioastronomy. Research Policy 12, 61-90.
+
+Mendonça, S., von Tunzelmann, N., 2004. Brave old world: accounting for ‘high-tech’ knowledge in ‘low-tech’ industries. Presented at DRUID's Summer Conference 2004—Industrial Dynamics, Innovation and Development, Copenhagen, Denmark, June 14–16, 2004.
+
+Miles, I., 2004. Innovation in services. In: Fagerberg, J., Mowery, D., Nelson, R. (Eds.), Understanding Innovation—Oxford Handbook of Innovation. Oxford University Press, Oxford, forthcoming.
+
+OECD, 2003. STI Scoreboard 2003. OECD, Paris.
+
+OHIM, 2003. Statistics of Community Trade Marks. Situation at the end of 2002.
+
+OHIM, 2004. Statistics of Community Trade Marks. Situation at the end of 2002.
+
+Patel, P., Pavitt, K., 1995. Patterns of technological activity: their measurement and interpretation. In: Stoneman, P. (Ed.), Handbook of Economics of Innovation and Technical Change. Blackwell, Oxford, pp. 14-51.
+
+Pavitt, K., 1985. Patent statistics as indicators of innovative activities: possibilities and problems. Scientometrics 7 (1-2), 7799.
+
+Schmoch, U., 2003. Service marks as a novel innovation indicator. Research Evaluation 12 (2), 149-156.
+
+Smith, K., 2002. What is the ‘knowledge economy’? Knowledge intensity and distributed knowledge bases. INTECH Discussion Paper Series 2002-6.
+
+Smith, K., 2004. Measuring innovation. In: Fagerberg, J., Mowery, D., Nelson, R. (Eds.), Understanding Innovation—Oxford Handbook of Innovation. Oxford University Press, Oxford, forthcoming.
+
+Trajtenberg, M., 1990. Economic Analysis of Product Innovation: The Case of CT Scanners. Harvard University Press, Cambridge, MA.
+
+Velling, J. (Ed.), 2002. 2001 Germany's Technological Performance. Bundesministerium für Bildung und Forschung. Available at http://www.bmbf.de/pub/germanys_technological_performance_2001.pdf.
+
+von Tunzelmann, N., Acha, V., 2004. Innovation in ‘low-tech’ industries. In: Fagerberg, J., Mowery, D., Nelson, R. (Eds.), The Oxford Handbook of Innovation. Oxford University Press, Oxford, forthcoming.
+
+WIPO, 2004. What is intellectual property? Available at http:// www.wipo.int/freepublications/en/intproperty/450/wipo_pub_ 450.pdf .
+

processed/00011_W3122382941/document.md

@@ -0,0 +1,537 @@
+Strategic Management Journal Strat. Mgmt. J., 27: 1141-1158 (2006)
+
+Published online 13 September 2006 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.562
+
+Received 19 August 2004; Final revision received 7 April 2006
+
+![Figure](figures/figure_001.png)
+
+# THE LICENSING DILEMMA: UNDERSTANDING THE DETERMINANTS OF THE RATE OF TECHNOLOGY LICENSING
+
+ANDREA FOSFURI*
+
+Department of Business Administration, Universidad Carlos III de Madrid, Madrid, Spain
+
+The licensing of technology entails a trade-off: licensing payments net of transaction costs (revenue effect) must be balanced against the lower price—cost margin and/or reduced market share implied by increased competition (profit dissipation effect) from the licensee. We argue that the presence of multiple technology holders, which compete in the market for technology, changes such a trade-off and triggers more aggressive licensing behavior. To test our theory, we analyze technology licensing by large chemical firms during the period 1986—96 for 107 chemical products. We find that the rate of technology licensing displays an inverted U-shaped relationship with the number of potential technology suppliers and is negatively related to the licensor's market share and to the degree of technology-specific product differentiation. Copyright © 2006 John Wiley & Sons, Ltd.
+
+## INTRODUCTION
+
+The last two decades have witnessed an unprecedented growth in a variety of arrangements for the exchange of technologies and technological services (Rivette and Kline, 1999; Rigby and Zook, 2002) . In the United States alone, technology licensing revenues are estimated to account for U.S. $ 45 billion annually; worldwide, the figure is around U.S. $ 100 billion ( The Economist , 2005) . The biopharmaceutics, software, semiconductor and telecommunications sectors have made the licensing of intellectual property a way of life.
+
+It is not surprising that small technology-based firms license their intellectual property. Lacking the downstream manufacturing, distribution, and marketing capabilities, they have no other means
+
+of appropriating returns from innovation. More surprising is the active promotion of licensingbased strategies by large, established producers. Firms such as Union Carbide, Procter & Gamble, DuPont, Boeing, Hoechst, IBM, Texas Instruments, AT&T, and Phillips Petroleum are now explicitly considering licensing revenues as a part of the overall return from their technological investments (Rivette and Kline, 1999) . These firms are well established, have large market shares in the product markets, and are capable of exploiting the technology on their own.
+
+This paper examines the licensing strategies of large firms by focusing on some of the determinants of their rate of technology licensing. In particular, we aim to shed light on the relationship between competition in the market for technology and the licensing decisions of firms. We argue that a firm's rate of technology licensing can be explained by the interplay of two effects that licensing produces on the licensor's profits: the profit dissipation effect and the revenue effect (Arora and Fosfuri, 2003) . Indeed, technology
+
+Keywords: licensing; revenue effect; profit dissipation effect; chemical industry
+
+* Correspondence to: Andrea Fosfuri, Department of Business Administration, Universidad Carlos III de Madrid, Calle Madrid, 126, 28903 Getafe, Madrid, Spain. E-mail: fosfuri@emp.uc3m.es
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+![Figure](figures/figure_013.png)
+
+---
+
+1142 A. Fosfuri
+
+licensing forces a trade-off: licensing and royalty revenues net of transaction costs (the revenue effect) must be balanced against the lower price–cost margin and/or reduced market share implied by increased competition (the profit dissipation effect) from the licensee. The presence of multiple technology holders, which compete in the market for technology, changes such a trade-off and triggers more aggressive licensing behavior. We then focus on two other potentially important determinants of a firm's rate of technology licensing: the licensor's market share in the product market and the degree of technology-specific product differentiation.
+
+We test our theory on a sample of large chemical firms that possess technological competencies in a set of 107 chemical products by examining their licensing strategies over the period 1986 – 96. The chemical industry has a long tradition of technology licensing (Arora, 1997) that allows us to gather reliable data. In addition, as we highlight in the next section, there are often several technologies available from different licensors to produce the same chemical product, which helps us to underscore the effect of competition in the market for technology. There is an established market for polyolefin production processes, for instance, with about 25 polyethylene and 8 polypropylene technologies available. Large, established chemical producers like Dow Chemicals, BP-Amoco, Exxon-Mobil, Union Carbide, Univation, and Basell, along with independent technology suppliers like Novolen and Engelhard, compete face to face in the licensing market (Tullo, 2003).
+
+This research contributes to several streams of the licensing literature. First, using concepts from economic theory, we build a simple framework that helps us to predict empirically a firm's rate of technology licensing. It goes beyond the standard transaction costs theory used in the management literature to explain why licensing is or is not chosen for a given transaction (Williamson, 1991; Teece, 1986, 1988) . We follow an approach that considers the impact of licensing on the whole value chain of the licensor rather than narrowly focusing on economizing on every single transaction. Needless to say, transaction cost considerations are still important in shaping a firm's licensing strategy. Hence, our approach complements rather than substitutes the extant transaction costs theory.
+
+Second, and most important, this work contributes to the relatively underdeveloped empirical research on licensing. This lacuna is understandable. In many industries, licensing is a recent phenomenon, so available data are scattered. In addition, firms tend to conceal information about licensing deals, which are typically considered to be strategic decisions that are not publicly disclosed.
+
+Anand and Khanna (2000a) provide one of the few econometric investigations of the rate of licensing. Their study is aggregated at the level of the sector, and they do not attempt to explain inter-firm differences in the rate of technology licensing. Arora and Ceccagnoli (2006) analyze how patent effectiveness affects a firm's patenting behavior and its propensity to license. They do not look, however, at the interaction between licensing decisions and competition in the market for technology, which is crucial in this paper. Recently, several scholars have directed their attention to the analysis of licensing practices by universities (Mowery and Shane, 2002) . For instance, Sine, Shane, and Di Gregorio (2003) have shown that the rate of licensing by universities is an increasing function of their prestige. However, university licensing decisions differ considerably from those made by large firms, primarily because universities do not have stakes in the product market. $^1$
+
+Other empirical research on technology transfer has focused on the factors that determine a firm's choices among various organizational forms (e.g., Teece, 1986; Hill, 1992) . For instance, Gans, Hsu, and Stern (2002) study the determinants of commercialization strategy for start-up innovators. Within this tradition, the empirical literature has paid particular attention to foreign market entry choices (e.g., Kogut and Singh, 1988; Hill, Hwang, and Kim, 1990) . of these studies investigated the factors underpinning the rate of technology licensing among firms.
+
+The remainder of the paper is organized as follows. The next section briefly discusses licensing dynamics in two chemical products and explains the motivation behind our interest in the role of competition in the market for technology. Then,
+
+1 In the strategy literature, a great deal of research has been devoted to the analysis of alliances (e.g., Dussauge, Garette, and Mitchell, 2000) . Although licensing could be considered as a special type of alliance, the evidence has shown that licensing contracts behave quite differently, and therefore deserve ad hoc attention (Anand and Khanna, 2000b) .
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1143
+
+the article develops the theoretical framework for explaining the rate of technology licensing among firms and formulates some testable hypotheses. The description of the empirical methodology and data is followed by a discussion of the results and the overall conclusions.
+
+## TECHNOLOGY LICENSING IN THE CHEMICAL INDUSTRY
+
+Technology licensing in the chemical industry has a long tradition. Cross-licensing agreements were already being used at the beginning of the 20th century as a means of maintaining market shares and deterring entry into an international chemical market characterized by the presence of strong cartels (e.g., Arora, 1997) . It was only after World War II, however, that firms started to use licensing to profit from innovation, and a market for chemical technology began to arise. Beginning in the 1950s, an increasing number of chemical processes became available for licensing. In this section, we briefly describe licensing dynamics in two chemical markets, highlighting the role played by competition in the technology market in shaping the licensing strategies of firms. $^2$
+
+### Acrylic acid
+
+Acrylic acid is a common chemical compound, and the key ingredient in many household products: paints, textiles, adhesives, plastics, and detergents, for example. Although first prepared in 1843, it was not until the late 1920s that acrylic acid was produced commercially and not until the early 1950s that it was used in any major way. Several processes—acetylene-based, acrylonitrile hydrolysis, ketene process, ethylene cyanohydrin process—were employed to produce acrylic acid prior to the 1970s, when Nippon Shokubai, a Japanese chemical firm, developed the standard technology based on propylene oxidation. Until the mid 1980s, Nippon Shokubai licensed its technology, which was also available through another Japanese firm, Mitsubishi Chemicals, but other technology holders, like the German giant BASF, did not. In recent times, however, all the potential licensors
+
+are also producers, with symmetric stakes in the acrylic acid market and a strong interest in maintaining market stability. Thus the attitude toward licensing has become more conservative, even among Nippon Shokubai and Mitsubishi Chemicals, as firms tend to avoid direct and potentially destructive competition in the licensing business. Most licensing is now directed at markets with strong growth potential that could not be reached by any other means. Hence, a sort of tacit collusion in the market for acrylic acid technology seems to be in place nowadays. $^3$
+
+### Ethylene glycol
+
+Ethylene glycol is a clear, colorless, odorless, viscous liquid with a sweet taste that can produce dramatic toxicity. It is commonly found in antifreeze, automotive cooling systems, and hydraulic brake fluids and is used in industrial settings as a solvent and as raw material in a variety of processes.
+
+Although ethylene glycol was prepared as early as 1856, it was not until the 1920s that the U.S. firm, Union Carbide, began commercial production. The dominant technology for producing ethylene oxide is currently the direct oxidation of ethylene, a process developed by a French firm at the beginning of the 1930s and subsequently perfected by Union Carbide. By the end of the 1940s, two other companies—Shell and the engineering firm Scientific Design—had developed new technologies based on the oxidation of ethylene. Other technologies have been tried and abandoned by such firms as Dow Chemicals and DuPont, either because they were economically unviable or because they were environmentally unsound.
+
+Union Carbide, the world's largest producer of ethylene glycol, used its own process captively and did not license it. Scientific Design, which did not have a stake in the final market, licensed its process aggressively and now has the largest share of built capacity. The other major player, Shell, both used and licensed its own process. Recently, Union Carbide has also started to license its technology in a highly selective way. It is important to emphasize
+
+2 These two short case studies have been constructed using information assembled from a variety of sources, most notably specialized chemical journals, the business press, and the Internet.
+
+3 Economic theory suggests that tacit collusion is more likely when there are few symmetric firms. Although tacit collusion is assumed away in our theoretical framework, its inclusion would imply that aggressive licensing occurs when there are many technology holders with heterogeneous stakes in the product market, an implication that has the same flavor as our Hypotheses 1 and 2.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1144 A. Fosfuri
+
+the role of Scientific Design, which was the source of the technology, but not an active producer of the final product. The firm was not inhibited by concerns over its production position or by potential competition from new players. Its behavior has precipitated more proactive licensing strategies by Shell and Union Carbide—companies that may otherwise have shown greater concern for their market shares. Consequently, the industry structure of the ethylene glycol market is highly fragmented, with many players and strong competition in most regions of the world.
+
+## THEORY DEVELOPMENT AND HYPOTHESES
+
+A licensing contract is the less integrated, more market-based alternative allowing firms to profit from their innovation. In fact, licensing is positioned at the extreme of a continuum of governance structures ranging from a market mechanism to hierarchy, i.e., in-house exploitation of the innovation (Williamson, 1991) .
+
+The standard framework for analyzing licensing decisions is provided by transaction costs theory. This approach suggests that, given an absence of significant contracting hazards, an arm's-length contract such as licensing would be the most direct way to capture profits from a firm's intellectual asset, such as a process or an idea. However, writing and executing a reliable contract for the use of technology requires adequate specification of property rights, monitoring, and enforcement of contractual terms — any of which may be problematic. Teece (1988) identifies three major sources of transaction costs in technology transfer: (a) incomplete contracts that leave either party open to opportunistic behavior by the other; (b) transaction-specific investments that can give rise to switching costs and `lock-in' problems; and (c) leakages of valuable proprietary information to potential competitors.
+
+One weakness of the transaction costs theory in its analysis of licensing decisions is its narrow focus around isolated transactions (Williamson, 1999: 1102) . Each transaction is treated as an independent item, bearing almost no relationship to previous or future transactions or with the rest of the firm's activities (Argyres and Liebeskind, 1999; Nickerson, Hamilton, and Wada, 2001) . If
+
+interaction effects are missed or if holistic consequences are glossed over, transaction costs theory would suggest that licensing is chosen when the transaction costs of using a market-based mechanism for profiting from innovation are sufficiently small (Teece, 1988) . However, a licensing agreement might not take place, in spite of low transaction costs, if it does not fit within the firm's overall strategy because of negative effects on other sources of revenues that outweigh the profits from the transaction. The net balance is the important variable in deciding whether to license or not. This argument suggests that, rather than focusing on economizing at the level of single transactions, it is advisable to consider the effects of technology licensing on the whole value chain (i.e., economizing at the level of the firm): technology management cannot be performed in isolation from other value-creating activities, such as production or distribution.
+
+### Revenue effect vs. profit dissipation effect
+
+Below, we develop a framework that accounts for these subtleties and allows us to predict empirically the rate of technology licensing among firms. Our approach does not substitute the extant transaction costs theory, but rather complements it. We still argue that, other things held constant, smaller transaction costs stimulate licensing. One should bear in mind that the following discussion focuses on technology licensing and omits any reference to product licensing.
+
+Licensing decisions result from the interplay of two effects that licensing generates on the licensor’s profits: the profit dissipation effect and the revenue effect (Arora and Fosfuri, 2003) . The revenue effect is the present value of the flows of money accruing to the licensor in the form of licensing payments, net of all possible transaction costs that bear on the seller of the technology. It is the revenue effect on which firms focus when striking a licensing deal. $^4$ In the words of one of Dow's vice-presidents, `by both licensing and using the technology we could generate more cash . . . .' The revenue effect is positively related
+
+4 We are ignoring the possibility that licensing might be due to strategic reasons like the blockading of entry (Gallini, 1984) or in order to select competitors (Rockett, 1990) . Most importantly, in industries with strong network externalities, licensing may be a way to set and control industry standards (Shapiro and Varian, 1998) .
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1145
+
+to the gross profits that the licensee can extract from the licensed technology, negatively related to the transaction costs, and positively related to the bargaining power of the licensor. As predicted by the transaction costs theory, therefore, other things held constant, larger transaction costs imply a smaller revenue effect and, in turn, make licensing a less attractive strategy (Teece, 1986) . Needless to say, transaction costs, bargaining power, and gross profits earned by the licensee are functions of other variables. For instance, the licensor's bargaining power increases with the strength of intellectual property rights protection and with the number of suitable licensees.
+
+The profit dissipation effect is the reduction in the licensor's profits (i.e., all profits other than payments from the licensing agreement) that might occur as a consequence of an additional firm competing in the product market or of an existing firm becoming more aggressive. $^5$ Additional competition in the downstream market can reduce the price $-$ cost margin and/or erode market share. As a result, a licensor that also competes in the product market may encounter a reduction in the profits from directly producing and commercializing the final good. The danger of increased competition in the licensor's own market is echoed by industry players and is often reported as the main reason for not licensing out: `For our main chemical products, such as epoxy and polyketones, we just don't want to license them out because it would threaten our market' (a Shell spokesperson).
+
+Although the licensor has many strategies to limit the extent of this latter effect (for instance, the contract may impose quantity restrictions or exclusive territories or unit royalties may be set to control the licensee's output), an entrant is nevertheless a potential threat to the licensor. For instance, Hill et al. (1990) report the case of RCA, which once licensed its color TV technology to a number of Japanese companies for exclusive exploitation in Japan. The Japanese companies quickly assimilated RCA's technology, however, and used it to compete directly with RCA in the U.S. market.
+
+The profit dissipation effect depends on several factors, primary among them being the magnitude
+
+of the competitive pressure exerted by a new player in the downstream market. It is the careful comparison between the profit dissipation effect and the revenue effect that explains whether a firm is licensing or not, and, if it does, how much it is licensing.
+
+## The role of competition in the market for technology
+
+Let us consider a situation in which the profit dissipation effect dominates the revenue effect. This is typically the case of a monopolist in both the product and the technology market. In the absence of any threat in the market for technology (i.e., no other sources for the technology), the firm would optimally decide not to license. The presence of another producer (the licensee) would certainly put some pressure on the price that will fall below the monopoly level. Unless other reasons are introduced, the firm could not earn greater profits licensing than not licensing. We are implicitly assuming here that the demand for the final product is stable and that the incumbent monopolist has already made the necessary investment in order to satisfy such demand. Now suppose that another firm has developed the technology to produce the final product and can potentially license it to an entrant. For the sake of simplicity, assume that this potential licensor cannot produce the product itself. For instance, in the chemical industry, many process technologies are licensed by specialized engineering firms that have no stake in the product market and focus their business model around the design, engineering, licensing, and sometimes the construction of chemical plants (Arora, Fosfuri, and Gambardella, 2001) . Now assume that a potential entrant exists — one that needs a license to enter the market. What is the most plausible scenario if the monopolist does not license its technology? The potential entrant could strike a deal with the other technology owner and ultimately compete with the monopolist in the product market. As a result, the monopolist might suffer both eroded market share and reduced price — cost margin. Moreover, the monopolist does not collect any licensing payment because it has opted not to license its technology. In other words, the monopolist would have suffered from the profit dissipation effect in any case; but at least it would have benefited from the revenue effect, had it licensed out its technology to the potential entrant.
+
+5 Technology licensing might allow the entry of a firm that is currently outside the relevant market or might increase the efficiency of an incumbent firm when the licensor transfers its low-cost production technology. In both cases, competition in the product market is likely to increase.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1146 A. Fosfuri
+
+The argument would stand unchanged if the perspective licensee were a higher-cost incumbent seeking a more efficient production technology. The key assumption here is that all potential licensors have functionally equivalent technologies. Indeed, a company with proprietary lowcost production technology might still not want to license out its technology, despite the presence of multiple players that license similar but less efficient technology. However, the main idea remains unchanged: weaker competition in the market for technology implies less incentive to license, other things held constant.
+
+When a market for technology exists, as is typically the case for most chemical process technologies, a technology holder's refusal to license will not block entry into the product market, because the prospective licensee can obtain the technology from other potential licensors. Hence, the presence of multiple sources for a technology creates a strategic incentive to license.
+
+However, we do not expect the relationship between the rate of technology licensing and the number of potential technology suppliers to be monotonic everywhere. There are at least two arguments to suggest that, after a certain threshold, the rate of technology licensing might actually decrease as the number of potential technology suppliers increases. First, the number of potential licensees for a given process technology is bounded. In some cases, in fact, the search for suitable licensees turns out to be a long and costly process (Contractor, 1981) . If the number of potential licensees is fixed and bounded and the number of potential licensors keeps increasing, at a certain point, the number of effective licenses per licensor will hit the constraint. After that, a further increase in the number of potential licensors produces a reduction in the average number of licenses. Second, a larger number of technology suppliers means stronger competition in the market for technology. Licensors have weaker bargaining power vis-à-vis the prospective licensees. In other words, the revenue effect tends to be competed away when the number of potential licensors increases. When payments from licensing are insufficient to cover transaction costs, firms stop licensing.
+
+In sum, the presence of multiple sources for a technology tilts the balance between the profit dissipation effect and the revenue effect. Some competition in the market for technology makes profit dissipation considerations less important,
+
+sparking more aggressive licensing behavior. However, very strong competition in the market for technology annihilates the revenue effect, which, in turn, makes licensing less appealing.
+
+By combining the various arguments, we can formulate the following hypothesis:
+
+Hypothesis 1: There exists an inverted-U shaped relationship between the rate of technology licensing and the number of potential technology suppliers.
+
+In industries in which standards play a crucial role, firms may license aggressively because they would like their technology to be widely diffused. This alternative argument would also predict a positive relationship between the rate of technology licensing and the number of competing technologies. Battles over standards are more important early in the product life cycle (Teece, 1986) . Because we focus empirically on mature chemical products with well-established markets, we partially control for that possibility. However, one should be cautious about interpreting empirical findings.
+
+## Market share
+
+We focus here on the direct effect of the licensor's market share in the product market on its incentives to license, all other variables left unchanged. We analyze only the impact of market share on the profit dissipation effect, because the revenue effect is unrelated (at least directly) to market share. As noted previously, the profit dissipation effect is the erosion of profits due to additional competition in the product market. Hence, we claimed that a technology holder licenses if the net licensing revenues are greater than the loss in profits due to increased competition in the product market. Although all incumbent producers potentially lose from the increased competition, each licensor only internalizes the negative effect on its own profits. The smaller the profits the licensor obtains from direct production prior to licensing, the smaller this negative effect. This implies, in turn, that other things equal, firms with smaller market shares have stronger incentives to license, as they suffer from a much smaller profit dissipation effect. This argument can best be understood by fixing the quantity produced by each firm and assuming that entry simply reduces the price $-$ cost margin. Firms that sell larger quantities, i.e., firms
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1147
+
+with larger market shares, would suffer most from competition because the same reduction in the margin is multiplied by a higher volume of sales. Therefore, firms that center their business model on the pure supply of technology with no stakes in the product market have stronger incentives to license than do established producers that enjoy large market shares. This argument is exemplified by the various ways in which BP Chemicals has approached the acetic acid and polyethylene businesses. In acetic acid, BP has strong proprietary technology and a substantial market share. It licenses selectively, typically granting a license only in order to obtain access to markets it would otherwise be unable to enter. In contrast, BP's market share in polyethylene is small. Although it has good proprietary technology in polyethylene as well, there are many other sources of technology for making polyethylene. Thus, BP has licensed its polyethylene technology aggressively, competing with Union Carbide, the market leader in licensing polyethylene technology. We can therefore state the following hypothesis:
+
+Hypothesis 2: The higher the licensor's share in the product market, the smaller its rate of technology licensing.
+
+A key assumption behind our argument is that market share is independent of other firm-specific variables. However, market share is likely to be positively correlated with the licensor's stock of such complementary assets as distribution or manufacturing. A large endowment of complementary assets better insulates the licensor from the potential competition exerted by the licensees, thereby reducing the profit dissipation effect. However, the presence of a strong competitor reduces the licensees' expectations of future profits and, hence, their willingness to pay for the technology. The revenue effect shrinks as well, possibly offsetting the reduction in the profit dissipation effect. In addition, a larger market share might be due to a more efficient technology. Because the technology is more valuable, the licensor can extract a larger payment from the licensees. However, the licensor is also likely to face reduced competition in the market for technology, thereby having fewer incentives to license out its technology. As we argued above, this second effect is likely to dominate. Finally, a larger market share might be associated with a greater ability to extract rent from the
+
+licensees—because of stronger bargaining power in licensing negotiations, for instance—thereby increasing the revenue effect. This correlation is likely to mitigate the negative relationship between the rate of technology licensing and market share, as stated in Hypothesis 2.
+
+We acknowledge the possibility that anticipated market share, i.e., the potential market share the licensor expects to gain, may sometimes be more important than current market share in driving licensing decisions. This possibility can be easily tested, however: current market share should not show up significant in the empirical analysis.
+
+## Product differentiation
+
+Product differentiation implies that the price elasticities of demand for different varieties are not infinite at equal prices. Product differentiation is due to both tangible features (e.g., size, reliability, durability, or safety) and intangible features (e.g. image, status, or esthetic considerations). For the argument we develop below, the important factor is that product differentiation is technology-specific rather than firm-specific. We assume that two firms using identical technology would produce indistinguishable varieties of the product. By contrast, two firms using different technologies would produce differentiated varieties. $^6$ Hence, product differentiation is due to differences in the process technology rather than to simple branding. In turn, this feature fits quite well with most chemical products in our database — like ammonia, acetic acid, polypropylene — in which the major source of differentiation is the technology (e.g., type of catalyst, temperature, feedstock) rather than the firm.
+
+The role of technology-specific product differentiation is better understood by focusing on the profit dissipation effect alone. Let us do the following thought experiment. Consider the market for polyethylene, and assume that Mitsui, with its own proprietary technology, competes in the downstream market with several other producers. Assume that Mitsui has a 2 percent market share. Let us start with the case in which the final product, polyethylene, is perfectly homogeneous across
+
+6 If product differentiation is also firm-specific, the profit dissipation effect would shrink because licensor and licensee do not compete head to head. Hence, greater firm-specific product differentiation would imply more licensing. However, the correlation between technology-specific product differentiation and the rate of technology licensing would be qualitatively unchanged.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1148 A. Fosfuri
+
+all producers. Mitsui is a small player and, as we argued above, would not suffer undue loss if an additional firm were to step in and start producing polyethylene. Hence, the profit dissipation effect for Mitsui is relatively small and could well be smaller than the revenue effect. Mitsui might have strong incentives to license. Even if Mitsui has a good proprietary technology, in fact, it might be difficult for the firm to gain a larger share of the polyethylene market. Licensing is a quick and relatively risk-free alternative for accomplishing this task.
+
+Now, consider the opposite situation, in which each producer has a well-defined market niche. In other words, the polyethylene market is highly differentiated and each firm, because of its idiosyncratic technology, is producing its own not-easilysubstitutable variety. Although Mitsui is a small player in the overall polyethylene market, it is almost a monopolist in its niche. Licensing would create much stronger competition in this case, as Mitsui would allow the entry of another firm using the same technology and producing the same variety. The profit dissipation effect is much larger, and the firm might not find it profitable to license. In addition, other technology suppliers are less of a threat because, even if they license, they would not allow direct entry into Mitsui's own market niche. This reasoning leads to the following hypothesis:
+
+Hypothesis 3: The higher the degree of technology-specific product differentiation, the smaller the firm's rate of technology licensing.
+
+## METHODOLOGY AND DATA
+
+We hypothesize that the rate of technology licensing is generated by the function $y = f(x, \beta)$ , where $y$ is licensing counts, $x$ is a vector of explanatory variables that include those identified in our hypotheses (plus all available control variables), and $\beta$ is a vector of parameters to be estimated. Because the dependent variable is discrete and non-negative, with numerous zero entries, conventional linear regression models are inappropriate. The simplest model form to accommodate count data is the Poisson regression model. To guarantee non-negativity of $\lambda$ , we model the single parameter of the Poisson distribution function, $\lambda$ , as $E[y] = \lambda = \exp(x\beta)$ . However, our dependent variable has a variance that is four times
+
+larger than its mean, suggesting the presence of overdispersion. If this is the case, although the parameters will be consistently estimated, their standard errors will typically be underestimated, leading to spuriously high levels of significance. To address this problem, we use a negative binomial regression, which provides more efficient estimators (Hausman, Hall, and Griliches, 1984) . Another econometric concern is the large number of zero observations. As a robustness check, we estimate an equation using a zero inflated negative binomial regression (Greene, 2000) .
+
+Our data on licensing come from Chemintell, a commercial database compiled by Chemical Intelligence Services (Chem-Intell), a division of Reed Telepublishing Ltd, which is a member of the Reed Elsevier Plc Group. Chemintell purports to be comprehensive, covering the entire population of chemical plants in the period under study. Chemintell provides information on over 36,000 plants announced or constructed all over the world in 1980–96, in what is broadly defined as the chemical sector. It also reports, albeit incompletely, information on plants built prior to 1980. The database is organized by plant. It reports the name of the company that ordered the plant, the name of the licensor for that plant (or `staff' for in-house licensing), the city and country in which a plant is located, the name of the chemical process or the product, the date at which the investment was first reported in the specialized trade press, and other information. For about 40 percent of the plants, Chemintell also reports the total cost of investment in the plant in millions of U.S. dollars; and, for a larger number of the plants, it shows the actual or planned capacity.
+
+To test our hypotheses, we focused on a sample of large chemical firms from developed countries (Western Europe, United States, Canada, and Japan) that had, by the year 1988, more than U.S. $ 1 billion in aggregate sales (Aftalion, 1991). Of this set of firms, only 153 had at least one plant reported in Chemintell. These are the firms we used in our study: 67 U.S., 1 Canadian, 32 Japanese, and 53 European firms. (The complete list of firms is available upon request.) We restricted our attention to this sample of large firms because we had to collect firm-specific variables that our database did not provide. Moreover, the focus of this paper is on the licensing strategies of large corporations, which face the trade-off between revenue effect and profit dissipation effect
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1149
+
+more intensely than do smaller firms. Therefore, the licensing activity of specialized technology suppliers, small firms, and start-ups is omitted from our study.
+
+We look at the rate of technology licensing of our large chemical firms in a sample of 107 products. This sample includes all the most important products in our dataset (as defined by worldwide investment) and accounts for more than half of all plants listed in Chemintell. `Ammonia' and `acetic acid' are two examples of such products. Because we selected the products with large worldwide installed capacity, our sample is more representative of bulk chemicals than of specialties. A complete list of products along with their classification in chemical sub-sectors is available upon request.
+
+As a first step, we want to identify the firms that possess technological capabilities in a given product, and that can therefore license to others. Then, we can examine the licensing behavior of these firms, which obviously changes across products. The set of potential licensors of ammonia process technology is different from the set of potential licensors of acetic acid process technology, although the two sets might overlap.
+
+To address this issue, we exploited the richness of our dataset. We split the data into two periods — 1980 $-$ 85 and 1986 $-$ 96 — using the first period to identify which of our 153 firms possessed technological capabilities in any of our 107 products. As a criterion, we used the fact that the firm in question had either licensed the technology or built a plant in-house, using its own technology. As a sensitivity check, we also considered a different time break: 1980 $-$ 87 and 1988 $-$ 96. We also experimented with a more demanding criterion: two plants instead of one. Neither of these manipulations changed the qualitative results. 7 Finally, we used our second period of 1986 $-$ 96 to examine the licensing behavior of these firms.
+
+One might wonder why we did not exploit the time dimension of our dataset more efficiently by creating a panel rather than a cross-section averaged over 1986 – 96. There are three reasons for such aggregation. First, some variables, such as the number of potential licensors or the degree
+
+of product differentiation, vary little from year to year. Second, given the size and timing of the investment in a chemical plant, licensing decisions tend to be correlated to long-term rather than shortterm changes. Finally, and most important, we have few plants licensed by each sample firm per year, which would make our dependent variable highly skewed towards zero (the share of positive observations will be below 3 % ).
+
+A final important clarification concerns the geographical definition of the market. Although the chemical industry is dominated by large multinational firms, most chemical products included in our sample are characterized by significant transportation costs relative to their unit value. This is especially true for bulk chemicals — the vast majority of our products — and less so for specialties. $^8$ Hence, production tends to take place, with some notable exceptions, close to the final market. $^9$ To capture this localized feature of chemical markets and following the partition provided in Chemintell, we have divided the world into seven geographical areas: Africa, Western Europe, Eastern Europe, North America, South America, the Middle East, and South East Asia. As a robustness check, we have tried a finer partition, in which we isolated the big country markets (United States $-$ Canada, China, India, Russia, Mexico) and lumped other countries together using membership in trade blocks (EU, Mercosur, ASEAN) or geographical continuity (Middle East countries, Central European countries). This second partition produced 10 geographical areas. Although every definition of market can be criticized, similarity of results between these two groupings should make us confident about the robustness of our empirical findings.
+
+To summarize, we examine the rate of technology licensing of large chemical firms (indexed by $i$ ), in a set of product markets that are defined
+
+7 It is important to remind the reader that we are examining the licensing of the process technology used to fabricate a given chemical product. Our dataset does not report product licensing, which is the major subject of transaction in pharmaceuticals, for instance.
+
+8 In addition, some products are difficult to move because of their specific features. For instance, acrylic acid is highly corrosive to many metals and must be stored in stainless steel-, glass-, aluminum- or polyethylene-lined equipment. Also, many chemical acids degrade over time by polymerization.
+
+9 In 2002, the share of imports to apparent consumption in the United States was 3 percent for sulfuric acid and pharmaceutical preparations; 3.4 percent for chlorine gas; 21 percent for titanium dioxide; less than 1 percent for oxygen, nitrogen, and carbon dioxide; 1.1 percent for hydrogen; and less than 10 percent for paint products (US Census Bureau at http://www.census.gov). Trade in plastics tends to be higher.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1150 A. Fosfuri
+
+around chemical products (indexed by $j$ ) and geographical areas (indexed by $k$ ).
+
+## Variables
+
+### Rate of licensing
+
+This variable measures the number of licensed plants by firm $i$ , in product $j$ , and geographical area $k$ , over the period 1986 – 96. As discussed above, only a small subset of our 153 firms has technological competencies in any given product $j$ . Moreover this subset tends to change for each product.
+
+### Potential licensors
+
+We want to capture the presence of other sources of technological competencies. In other words, we would like to know how many other firms are capable of supplying the process for producing product $j$ . We do so by counting the number of firms (excluding firm $i$ ) that have licensed a given process technology for producing product $j$ in the period $1980-85$ or have built a plant in-house using their own technology. This is a good proxy for the number of potential licensors of that technology in the period $1986-96$ . There are two potential measurement errors that could affect this proxy: (1) some firms may not have been technologically active in the period $1980-85$ , but can potentially license their technology in the period $1986-96$ . For instance, they might not have developed technological capabilities until $1985$ ; (2) some firms may have been licensing their technology in the period $1980-85$ , but could not license in the period $1986-96$ — because their technology had become obsolete, for instance. Yet the average number of potential licensors is similar in magnitude to the average number of `real' licensors (i.e., the number of firms that have licensed in the period $1986-96$ ), with a correlation above 0.6. In approximately 40 percent of the cases, the two numbers coincide. In half of the remaining cases, the number of potential licensors is smaller than the number of real licensors, whereas the opposite is true for the rest of our observations. Finally, it is also plausible that this measure varies across geographical areas. The set of potential technology suppliers for polyethylene process technology in the United States may differ from the set of potential technology suppliers in China. There may
+
+be geographical idiosyncrasies that render a technology suitable for the economic, legal, and environmental conditions of one area, but unsuitable for another. We have estimated our regressions with the number of potential licensors varying across $j$ and across $jk$ . Because qualitative results do not change, however, we only show the latter in order to save space.
+
+### Market share
+
+We compute the market share of firm $i$ , in product $j$ , and geographical area $k$ as the ratio between the capacity built by firm $i$ in $jk$ and the total capacity in $jk$ . Because we did not have access to the market shares of firms in the mid 1980s, we had to reconstruct them using the information on capacity investment provided by our database. Although correlated with sales, installed capacity may not always be a good proxy for sales. However, there are no reasons to believe that the difference between installed capacity and sales is distributed across our sample of firms and products in a way that would bias our findings, although we cannot guarantee that this does not occur.
+
+### Product differentiation
+
+Chemintell does not provide a direct measure of technology-specific product differentiation at the level of the product $j$ . To operationalize this variable, we built a principal component of three different proxies. Although each proxy has its weaknesses, we think that our principal component measure captures the variability associated with technology-specific product differentiation. The first proxy, which we believe is the most accurate, is the number of different feedstocks that can be used to fabricate a given product, standardized by the size of the world market of that product. In many cases, different technologies use different feedstocks, and the physical properties of the final product might depend on the specific feedstock employed. The second proxy is computed at the sub-sector level. Chemintell classifies all products in 23 sub-sectors. We use plant counts at the product level to compute a Herfindahl index at the sub-sector level (Sutton, 1991) . Our index of product differentiation takes the value of 0 if the sub-sector has homogeneous products and the
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1151
+
+value of 1 if the products are totally differentiated. 10 Our third proxy of product differentiation is the average cost in U.S. dollars per unit of capacity installed in each product $j$ . To compute this measure, we have summed all investments (in U.S. dollars) in a given product and divided by the sum of the installed capacity in the same product. We expect that more homogeneous and basic products are produced in large-scale plants in which the cost per unit of capacity installed is lower. By contrast, more differentiated and sophisticated products are produced in small plants and tend to have, on average, a larger cost per unit of capacity installed.
+
+## Control variables
+
+### Demand growth
+
+This variable measures the growth potential in the demand for chemical plants in product $j$ and geographical area $k$ . We use the ratio between the total number of plants constructed in $jk$ during the periods 1991 $-$ 86 and 1980 $-$ 85. We expect to observe a positive relationship between the rate of technology licensing and demand growth, as a rising demand relaxes the negative effect on the licensor's profits due to increased competition.
+
+### Potential licensees
+
+The larger the number of potential licensees, the stronger the bargaining power of the licensor and, hence, the larger the revenue effect from licensing. Indeed, Contractor (1981) points out that, in some cases, the search for suitable licensees turns out to be a long and costly process. Hence, we posit a positive relationship between the rate of technology licensing and the number of potential licensees. The number of potential licensees is computed as the number of downstream chemical producers active in product $j$ and geographical market $k$ prior to 1986. This proxy is likely to underestimate the number of potential licensees,
+
+as it does not account for entry. The next control helps us to partially overcome this limitation. $^11$
+
+### Set-up costs
+
+This variable measures the average cost of setting up a chemical plant in product $j$ . Larger set-up costs imply that entry into the downstream market is more difficult, thereby limiting the number of potential licensees. Therefore this variable complements our proxy for the number of potential licensees.
+
+### Aggregate sales
+
+We include this variable to control for the effect of firm size on the rate of technology licensing. For instance, larger firms could have stronger bargaining power in the licensing negotiations or better options to profit from their technological competences. We use the natural log of aggregate sales of firm $i$ in 1988 to compensate for skewness (Aftalion, 1991) .
+
+### Multinational
+
+This variable complements aggregate sales to control for firm size. Multinational counts the number of different countries in which firm $i$ operates.
+
+### R&D intensity
+
+For each firm $i$ , R & D intensity captures the ratio between R & D expenditures and sales in 1988 (Aftalion, 1991). The sign of this variable is not theoretically clear a priori . Higher R & D intensity means that the firm is more likely to possess valuable technological assets to license out. On the other hand, firms tend to avoid giving away their state-of-the-art technology and typically prefer to license older vintage technologies. However, the reason for including this variable, as well as the aggregate sales and multinational variables, is to control for firm-specific sources of variation that might affect the rate of technology licensing.
+
+$^{10}$ We constructed the differentiation index as follows: $\text{DI}_n = \sum_{j=1}^J \left(\frac{\text{NBPLANT}_{jn}}{\text{NBPLANT}_n}\right)^2$ , where $m$ refers to the subsector and $j$ to the product. NBPLANT is the number of plants.
+
+11 As a robustness check, we also used the number of downstream chemical producers that were active in geographical market $k$ at the subsector level. This variable accounts for potential entry by diversification. Results do not change qualitatively.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1152 A. Fosfuri
+
+### Transaction costs
+
+Our theory suggests that the greater the transaction costs, the smaller the revenue effect and, in turn, the smaller a firm's rate of technology licensing. Therefore we must control as well as we can for different sources of transaction costs. We use the following variables. Experience is a dummy variable that takes the value of 1 if firm $i$ has undertaken at least one licensing deal prior to 1986 in product $j$ and geographical area $k$ . Experience in gathering information about prospective licensees, negotiating, and writing and enforcing contracts lowers the cost of licensing. Experience is therefore a good proxy for the transaction costs of licensing. We posit that, by reducing transaction costs, experience makes licensing a more appealing strategy. Patents reports the total number of patents granted at the U.S. Patent Office during the period 1976 $-$ 95 to the technology used to fabricate product $j$ . As several authors have noted (e.g., Teece, 1988; Von Hippel, 1994) , patents are more likely to be issued for technologies in which the underlying knowledge is sufficiently codifiable. Process technologies that have been granted few patents are more likely to rely on trade secrecy and are, hence, less likely to be transferred through contracts. Finally, we use a set of dummy variables for chemical sectors to help us to control for sector-specific differences in transaction costs. For instance, some chemical sectors may have better legal protection of intellectual property, thereby reducing transaction costs. Chemintell classifies all products into nine broad sectors: Oil Refining, Petrochemicals, Minerals and Metallurgy, Plastics and Rubber, Inorganic Chemicals, Agriculture, Gas, Organic Chemicals, and Miscellaneous. 12
+
+### Dummy variables for geographical areas
+
+This set of dummy variables is meant to control for sources of heterogeneity across locations. Some areas could have better conditions for technology licensing—better access to related technological services that make the transaction easier, for instance.
+
+### Dummy variables for licensor nationality
+
+This set of dummy variables helps to control for cross-country heterogeneity in the attitude towards licensing.
+
+Table 1 reports the means, standard deviations, and correlations among the dependent, independent, and control variables. The table employs our first definition of geographical market in which the world is divided into seven areas.
+
+## EMPIRICAL RESULTS
+
+Table 2 shows the results of the negative binomial regressions using Chemintell's partition of the world in seven geographical areas. Table 3 shows the same regressions for our second definition of geographical markets. As a robustness check, we have also performed Poisson estimations. Results, available from the author upon request, are similar to those reported here. However, the overdispersion parameter in the negative binomial regressions is significantly different from zero, thus reinforcing our conjecture that the Poisson distribution is inappropriate.
+
+Model 0 omits the core covariates, showing only the baseline model with the control variables. Models 1 and 2 differ only in the definition of market share. Whereas Model 1 reports the market share at the level of the product $-$ geographical area, Model 2 shows the worldwide market share at the product level. Our discussion above suggests that, for many chemical products, production tends to take place close to the final market; thus Model 1 is our preferred specification. An LR test comparing the sparser specification (Model 0) with the augmented models (Models 1 and 2) suggests that our main explanatory variables play a significant role on top of the controls.
+
+There are 1748 observations in Table 2. 13 Hypothesis 1 predicts that the rate of technology licensing is first increasing and then decreasing in the number of potential technology suppliers. To test this hypothesis, we add a square term to the number of potential licensors. If our hypothesis is correct, we should obtain a positive coefficient for the number of potential licensor and a negative coefficient for the square term. The parameter
+
+12 As mentioned above, Chemintell also provides a finer classification of products into 23 chemical subsectors.
+
+15 As we have explained above, we have a highly unbalanced panel: 153 firms, 107 products and seven geographical areas would total more than 100,000 observations in a balanced panel.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1153
+
+Table 1. Descriptive statistics
+
+<table><tr><td>Variable</td><td>Mean</td><td>S.D.</td><td>1</td><td>2</td><td>3</td><td>3a</td><td>4</td><td>5</td><td>6</td><td>7</td><td>8</td><td>9</td><td>10</td><td>11</td></tr><tr><td>1. Rate of technology licensing</td><td>0.324</td><td>1.125</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>2. Potential licensors</td><td>3.147</td><td>3.293</td><td>0.24</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>3. Market share in $\#$</td><td>0.029</td><td>0.085</td><td>−0.01</td><td>−0.09</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>3a. Market share in j</td><td>0.035</td><td>0.046</td><td>−0.01</td><td>−0.20</td><td></td><td>0.41</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>4. Product differentiation</td><td>0.009</td><td>8.811</td><td>−0.08</td><td>−0.14</td><td></td><td>−0.03</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>5. Set-up costs</td><td>12.09</td><td>0.811</td><td>−0.08</td><td>−0.14</td><td>−0.03</td><td>−0.14</td><td></td><td>0.18</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>6. Set-up costs</td><td>13.72</td><td>96.71</td><td>−0.03</td><td>0.08</td><td>−0.06</td><td>−0.13</td><td>−0.08</td><td>0.01</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>7. Demand growth</td><td>73.52</td><td>17.35</td><td>0.03</td><td>0.13</td><td>0.01</td><td>−0.00</td><td>−0.03</td><td>0.08</td><td>−0.03</td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>8. Aggregate sales (log)</td><td>8.971</td><td>0.749</td><td>−0.03</td><td>0.03</td><td>0.09</td><td>0.16</td><td>−0.04</td><td>0.01</td><td>−0.03</td><td>−0.03</td><td></td><td></td><td></td><td></td></tr><tr><td>9. Multinational</td><td>10.1</td><td>6.1</td><td>0.01</td><td>0.01</td><td>0.09</td><td>0.19</td><td>−0.05</td><td>0.00</td><td>0.08</td><td>−0.04</td><td>0.70</td><td></td><td></td><td></td></tr><tr><td>10. Reb intensity</td><td>0.055</td><td>0.028</td><td>−0.09</td><td>−0.03</td><td>0.03</td><td>0.04</td><td>0.25</td><td>0.06</td><td>−0.03</td><td>0.04</td><td>0.04</td><td>0.01</td><td></td></tr><tr><td>11. Experience (dummy)</td><td>0.343</td><td>0.475</td><td>0.39</td><td>0.30</td><td>0.31</td><td>0.06</td><td>−0.07</td><td>0.27</td><td>−0.01</td><td>0.08</td><td>0.04</td><td>0.07</td><td>−0.06</td></tr><tr><td>12. Parents (log)</td><td>3.829</td><td>0.944</td><td>0.06</td><td>0.13</td><td>−0.01</td><td>−0.01</td><td>0.25</td><td>0.21</td><td>0.09</td><td>−0.09</td><td>0.05</td><td>0.05</td><td>0.04</td></tr></table>
+
+
+a In millions of U.S. dollars
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1154 A. Fosfuri
+
+Table 2. Rate of technology licensing by firm $i$ in product $j$ and geographical area $k$ (seven geographical areas)
+
+<table><tr><td rowspan="2">Variable</td><td colspan="3">Negative binomial</td><td rowspan="2">Zero-inflated model</td></tr><tr><td>Model 0</td><td>Model 1</td><td>Model 2</td></tr><tr><td>Constant</td><td>−3.544***</td><td>−5.124***</td><td>−4.587***</td><td>−3.465***</td></tr><tr><td>Potential licensors</td><td></td><td>0.140***</td><td>0.164**</td><td>0.159***</td></tr><tr><td>Square term</td><td></td><td>−0.012***</td><td>−0.013***</td><td>−0.013***</td></tr><tr><td>Market share in jk</td><td></td><td>−6.506***</td><td></td><td>−6.496***</td></tr><tr><td>Market share in j</td><td></td><td></td><td>−5.609**</td><td></td></tr><tr><td>Product differentiation</td><td></td><td>−1.011***</td><td>−1.006***</td><td>−1.154***</td></tr><tr><td colspan="5">Controls</td></tr><tr><td>Potential licensees</td><td>0.037***</td><td>0.037***</td><td>0.041***</td><td>0.036***</td></tr><tr><td>Set-up costs</td><td>−0.001</td><td>−0.002*</td><td>−0.002*</td><td>0.002</td></tr><tr><td>Demand growth</td><td>0.097**</td><td>0.080**</td><td>0.091***</td><td>0.072*</td></tr><tr><td>Log of aggregate sales</td><td>−0.131</td><td>−0.070</td><td>−0.103</td><td>−0.063</td></tr><tr><td>Multinational</td><td>0.010</td><td>0.019</td><td>0.019</td><td>−0.028</td></tr><tr><td>R&amp;D intensity</td><td>−7.671**</td><td>−6.162**</td><td>−6.749***</td><td>−8.703***</td></tr><tr><td>Experience</td><td>4.253***</td><td>4.283***</td><td>4.171***</td><td>3.499***</td></tr><tr><td>Patents</td><td>−0.008</td><td>0.036</td><td>0.013</td><td>−0.066</td></tr><tr><td>Dummy variables for chemical sectors</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr><tr><td>Dummy variables for geographical areas</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr><tr><td>Dummy variables for licensor nationality</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr><tr><td>Number of observations</td><td>1748</td><td>1748</td><td>1748</td><td>1748</td></tr><tr><td>Log-likelihood</td><td>−790</td><td>−757</td><td>−768</td><td>−742</td></tr><tr><td>Chi-squared</td><td>623</td><td>689</td><td>667</td><td>135</td></tr></table>
+
+
+"p<0.1;""p<0.05;""p<0.01
+
+estimate for the number of potential licensors is positive and highly significant in all specifications reported in Table 2 . We also find a negative and significant coefficient for the square term, suggesting that the relationship between the rate of technology licensing and the number of technology suppliers displays an inverted-U shape. By using Model 1, for instance, one can show that the inflection point is around 6 (notice that the number of potential licensors varies between 0 and 17, with an average of about 3). If all other variables are kept at their mean value, moving from no potential licensors to 5 potential licensors would imply an increment in the rate of technology licensing of about 50 percent.
+
+Hypothesis 2 predicts that firms license more in product markets where they have smaller market shares. Indeed, as we argued above, the profit dissipation effect is positively correlated to market share. The sign of the parameter estimate for market share in Model 1 is negative and highly significant. If all other variables are kept at their mean value, moving from a 5 percent market share to a 10 percent market share would reduce the rate
+
+of technology licensing by about 40 percent. In Model 2, we use as a regressor the worldwide market share in product $j$ . Although the coefficient is still negative and significant, we observe a smaller magnitude and a much larger standard deviation. As an additional robustness check, we have also run a regression, using only the observations in which firm $i$ has a non-zero market share. Indeed, the decision to license by a firm with existing facilities in the market might be a quite different decision from that made by a firm with zero market share. Qualitative results remain unchanged, although coefficients display larger standard deviations, due to the reduced number of observations.
+
+Hypothesis 3 suggests that firms show a higher rate of technology licensing when the product market is sufficiently homogeneous. We used as a measure of technology-specific product differentiation a principal component extracted from three proxies. The coefficient is negative and significant, suggesting that technology licensing is less likely to occur in a differentiated product market. This finding seems to support Hypothesis 3. Keeping all other variables at their mean value, a standard
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1155
+
+Table 3. Rate of technology licensing by firm $i$ in product $j$ and geographical area $k$ (10 geographical areas)
+
+<table><tr><td rowspan="2">Variable</td><td colspan="3">Negative binomial</td><td rowspan="2">Zero-inflated model</td></tr><tr><td>Model 0</td><td>Model 1</td><td>Model 2</td></tr><tr><td>Constant</td><td>−1.901</td><td>−3.622***</td><td>−3.576***</td><td>−3.920***</td></tr><tr><td>Potential licensors</td><td></td><td>0.363***</td><td>0.369***</td><td>0.306***</td></tr><tr><td>Square term</td><td></td><td>−0.030***</td><td>−0.030***</td><td>−0.028***</td></tr><tr><td>Market share in jk</td><td></td><td>−2.992**</td><td></td><td>−6.716***</td></tr><tr><td>Market share in j</td><td></td><td></td><td>−0.243</td><td></td></tr><tr><td>Product differentiation</td><td></td><td>−0.981***</td><td>−0.940***</td><td>−1.231***</td></tr><tr><td colspan="5">Controls</td></tr><tr><td>Potential licensees</td><td>0.053***</td><td>0.036*</td><td>0.036*</td><td>0.044</td></tr><tr><td>Set-up costs</td><td>−0.003***</td><td>−0.003***</td><td>−0.003***</td><td>0.003**</td></tr><tr><td>Demand growth</td><td>0.254***</td><td>0.290***</td><td>0.292***</td><td>0.080*</td></tr><tr><td>Log of aggregate sales</td><td>−0.157</td><td>−0.102</td><td>−0.103</td><td>0.093</td></tr><tr><td>Multinational</td><td>0.038*</td><td>0.040*</td><td>0.036</td><td>0.001</td></tr><tr><td>R&amp;D intensity</td><td>−8.837***</td><td>−7.854**</td><td>−7.690***</td><td>−10.512***</td></tr><tr><td>Experience</td><td>0.794***</td><td>0.718***</td><td>0.717***</td><td>0.360***</td></tr><tr><td>Patents</td><td>0.021</td><td>0.072</td><td>0.055</td><td>0.136</td></tr><tr><td>Dummy variables for chemical sectors</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr><tr><td>Dummy variables for geographical areas</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr><tr><td>Dummy variables for licensor nationality</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr><tr><td>Number of observations</td><td>2294</td><td>2294</td><td>2294</td><td>2294</td></tr><tr><td>Log-likelihood</td><td>−970</td><td>−949</td><td>−951</td><td>−917</td></tr><tr><td>Chi-squared</td><td>274</td><td>316</td><td>312</td><td>122</td></tr></table>
+
+
+"p<0.1; **p<0.05; ***p<0.01
+
+deviation increase in the level of product differentiation would reduce the rate of technology licensing by 40 percent. As a robustness check, we have also entered our three proxies separately as measures of product differentiation. Results (not included) are qualitatively similar to the ones reported in Table 2 .
+
+The signs of the other variables are reasonable. Particularly interesting is the positive and highly significant coefficient of the number of potential licensees in a given product market $jk$ . A larger number of licensees creates greater bargaining power for the licensor. It also means that the licensor is much more likely to find a licensee that better suits the idiosyncrasies of the licensor's technology. In turn, this implies that the value generated from the transaction is higher, making licensing a more appealing option. Additionally, the average set-up cost for a chemical plant in product $j$ has a negative impact on the rate of technology licensing, thereby confirming that larger investment costs reduce the demand of technology. Demand growth has the positive expected sign, significant in all regressions. The findings seem to support the idea that higher transaction costs
+
+reduce the rate of technology licensing. Indeed, our dummy for experience in licensing in product $j$ and geographical area $k$ is positive and highly significant. However, the coefficient of the number of patents is not statistically different from zero. This finding is not surprising, in light of Cohen, Nelson, and Walsh's (2000) results; they found that patents are not good instruments for protecting process innovations. Our measures of firm size are not significant, whereas R & D intensity is negative, implying that firms with less R & D intensity have a higher rate of technology licensing.
+
+As we discussed in the previous section, one important econometric concern is the presence of a large number of zeros in our dependent variable. We have controlled for this potential problem through a zero-inflated negative binomial regression (Greene, 2000) . A zero-inflated negative binomial model, also known as hurdle model, assumes that the zeros are generated by a different process from the remaining counts. A binary probability model determines whether a zero or a nonzero outcome occurs; then, in the latter case, a (truncated) negative binomial distribution describes the positive outcomes. All
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1156 A. Fosfuri
+
+our explanatory variables have been used in the binary probability regression — except for the sector, nationality, and geographical area dummies, which we omitted because of convergence problems in the estimation. Results for the (truncated) negative binomial distribution are shown in the last column of Table 2 . All variables of theoretical interest maintain their sign and significance. A positive Vuong test shows that the inflated model performs slightly better than the standard negative binomial model. Moreover, many of the coefficients in the binary probability model are not statistically significant, which may imply that many zeros simply stand for the impossibility to license, and do not capture the deliberate outcome of a firm's decision.
+
+Table 3 reports our findings using the finer partition of the world into the 10 geographical areas described above. It is comforting to see that all variables keep their sign unchanged, although some coefficients display larger standard deviations. This suggests that our results are robust to different definitions of geographical markets. Model 2 additionally confirms our conjecture that the relevant market share is at the product–geographical area level, and not worldwide.
+
+## CONCLUSION, DISCUSSION, AND IMPLICATIONS
+
+There is increasing evidence that firms in some sectors hope to profit from their intellectual property not merely by embodying it in their own output, but also by licensing it to others, including potential competitors. Licensing does, however, entail a trade-off: licensing revenues must be balanced against the lower price–cost margin and/or reduced market share implied by increased competition from the licensees. In this paper we argued that competition in the market for technology could trigger more aggressive licensing behavior. Indeed, the trade-off between profit dissipation effect and revenue effect that normally guides licensing decisions should be adjusted for the presence of rival licensors. In particular, when there are multiple technology holders, they not only compete in the product market; they also compete in the market for technology. Because licensing partially substitutes for production, firms lacking adequate downstream commercialization (production and marketing) capabilities are naturally more
+
+aggressive licensors. Moreover, we have argued that increasing product differentiation, when driven by differences in the technologies, not only softens price competition in the product market, but also reduces the rate of licensing in the technology market.
+
+We tested this framework using an extensive dataset on worldwide investment in chemical plants. We examined licensing strategies by the use of a sample of large chemical producers in more than 100 products. Our results suggest that the presence of a market for technology plays a crucial role in creating incentives for a more proactive licensing behavior. Firms that normally would have not licensed their technology might be forced to do so because of the competitive pressure in the market for technology.
+
+Two contributions are particularly worth emphasizing here. First, this research underscores the crucial role of a market for technology in shaping firms' licensing strategy. This finding has implications both for the literature that has addressed the rationales behind a firm's decision to license (Gallini, 1984; Rockett, 1990; Anand and Khanna, 2000a) and for the literature on innovation and technology exploitation (Teece, 1986; Gans et al. , 2002) . In the former case, our paper suggests that in some industries it might be problematic, if not wrong, to analyze a firm's licensing strategy in isolation, abstracting from product and technology market dynamics. Indeed, most of this literature has assumed that the licensor is a monopolist technology holder, implying that the analysis of all potentially interesting interactions in the market for technology have not been considered. In the latter case, our paper offers an approach better suited to understanding a firm's rate of technology licensing rather than the extant transaction costs framework, although we still claim that transaction costs are important, ceteris paribus . Second, and perhaps most important, our paper provides one of the few large-scale studies of the determinants of the rate of technology licensing. This contribution is particularly valuable in light of the recent trend towards a more widespread use of licensing agreements for the exploitation of the firm's intellectual property.
+
+Our study has several limitations. From an empirical point of view, our database did not provide any measure of technology-specific product differentiation, forcing us to use proxies. Although we believe that these proxies represent the missing
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+The Licensing Dilemma 1157
+
+measure accurately, there is room for improvement. Ideally, this measure should capture the difference between the product produced with a firm's own process technology and the products fabricated by other incumbents. In this case, our empirical analysis could benefit from additional data collection, which might require an in-depth analysis of all process technologies available for each of our 107 products. In addition, although our theory predicts that stronger competition in the market for technology generates additional incentives to license because it changes the trade-off between the profit dissipation effect and the revenue effect, other explanations might generate similar findings. For instance, if standard setting is important, firms might license aggressively because they would like to have their technology widely diffused (Shapiro and Varian, 1998) . Although we focus on mature product markets in which standards battles are less crucial, empirically disentangling this alternative explanation is not easy. Hence, the interpretation of our findings should be made with caution. Theoretically our framework would clearly benefit from a deeper integration of our strategic positioning approach with the transaction costs approach. Ideally, one should be able to predict simultaneously the choice of the governance structure and the extent to which each governance structure is used. Finally, as far as it concerns the generality of our findings, one could easily contend that they are idiosyncratic to the chemical industry. As a partial defense to our work, we could point to empirical evidence showing that industries with large licensing activity, such as electronics, biotechnology, and semiconductors, are also those that have sufficiently well-functioning markets for technology (Arora et al. , 2001) . However, only future research can demonstrate whether our findings are industryspecific or more generally applicable.
+
+From a more applied point of view, our study suggests that firms exploiting licensing opportunities must ensure that the trade-off between licensing revenues and rent dissipation is well managed. First of all, this requires a close coordination between the various activities in the firm's value chain. Technology management cannot be performed in isolation from other value-creating activities such as production, distribution, and sales. Second, it becomes crucial to educate business managers about the net value added from sale of products vs. that from licensing. Licensing revenues are rarely comparable to the revenues from
+
+sales of products, but the cost of generating a dollar of licensing revenues is significantly lower than the cost of generating a dollar of product sales. Finally, it requires that managers have incentives consistent with those of the firm as a whole. Managers who are rewarded for sales growth or market share will tend to overlook licensing opportunities.
+
+A final cautionary remark is needed. The recent enthusiasm by many industry practitioners and independent consultants about the virtue of licensing might, in some cases, be misplaced. Licensing, especially when triggered by the presence of a market for technology, is a double-edged sword. Although some firms may benefit from aggressive licensing, the final outcome is likely to be increased competition in the product market and lower overall industry profits. Attempts to implicitly or explicitly collude in the licensing market, which are relatively common in the history of the chemical industry, confirm the relevance of this argument.
+
+## ACKNOWLEDGEMENTS
+
+I would like to thank Ashish Arora, Isabel Busón, Julio de Castro, Bronwyn Hall, Raffaele Oriani, Thomas Rønde, Giovanni Valentini, Oliver Williamson, two anonymous reviewers, and seminar participants at the 2003 Annual Meeting of the Academy of Management (Seattle), University Pompeu Fabra, Carnegie Mellon University, Copenhagen Business School, Sant'Anna School of Advanced Studies, Jornadas de Economía Industrial (Granada), University of Valencia and University of Zaragoza for helpful comments and suggestions on a previous draft. I gratefully acknowledge financial support from the European Commission (HPSE-CT-2002-00146), and the Spanish Ministry of Science and Technology (SEC2003-03797). The usual disclaimer applies.
+
+## REFERENCES
+
+Aftalion F. 1991. History of the International Chemical Industry. University of Pennsylvania Press: Pittsburgh, PA.
+
+Anand BN, Khanna T. 2000a. The structure of licensing contracts. Journal of Industrial Economics 48 (1): 103–135.
+
+Anand BN, Khanna T. 2000b. Do firms learn to create value? The case of alliances. Strategic Management Journal 21 (3): 295–315.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+
+---
+
+1158 A. Fosfuri
+
+Argyres NS, Liebeskind JP. 1999. Contractual commitments, bargaining power, and governance inseparability: incorporating history into transaction cost theory. Academy of Management Review 24 (1): 137–145.
+
+Arora A. 1997. Patent, licensing and market structure in the chemical industry. Research Policy 26: 391-403.
+
+Arora A, Ceccagnoli M. 2006. Patent protection, complementary assets, and firms: incentives for technology licensing. Management Science 52: 293–308.
+
+Arora A, Fosfuri A. 2003. Licensing the market for technology. Journal of Economic Behaviour and Organization 52: 277-295.
+
+Arora A, Fosfuri A, Gambardella A. 2001. Markets for Technology: Economics of Innovation and Corporate Strategy . MIT Press: Cambridge, MA.
+
+Cohen WM, Nelson RR, Walsh JP. 2000. Protecting their intellectual assets: appropriability conditions and why U.S. manufacturing firms patent (or not). Working paper 7552, National Bureau of Economic Research, Cambridge, MA.
+
+Contractor FJ. 1981. International Technology Transfer. D.C. Heath: Lexington, MA.
+
+Dussauge P, Garette B, Mitchell W. 2000. Learning from competing partners: outcomes and durations of scale and link alliance in Europe, North America, and Asia. Strategic Management Journal 21 (2): 99–126.
+
+Gallini NT. 1984. Deterrence through market sharing: a strategic incentive for licensing. American Economic Review 74: 931-941.
+
+Gans J, Hsu D, Stern S. 2002. When does start-up innovation spur the gale of creative destruction? RAND Journal of Economics 33 (4): 571–586.
+
+Greene WH. 2000. Econometric Analysis (4th edn). Prentice-Hall: Upper Saddle River, NJ.
+
+Hausman JA, Hall BH, Griliches Z. 1984. Econometric models for count data with an application to the patents-R&D Relationship. Econometrica 42(4): 909-938.
+
+Hill CWL. 1992. Strategies for exploiting technological innovations: when and when not to license. Organization Science 3: 428-441.
+
+Hill CWL, Hwang LP, Kim WC. 1990. An eclectic theory of the choice of international entry mode. Strategic Management Journal 11 (2): 117–128.
+
+Kogut B, Singh H. 1988. The effect of national culture on the choice of entry mode. Journal of International Business Studies 19: 411-432.
+
+Mowery DC, Shane S. 2002. Introduction to the Special Issue on University Entrepreneurship and Technology Transfer. Management Science 48 (1): v–ix.
+
+Nickerson JA, Hamilton BH, Wada T. 2001. Market position, resource profile, and governance: linking Porter and Williamson in the context of international courier and small package services in Japan. Strategic Management Journal 22(3): 251–273.
+
+Rigby D, Zook C. 2002. Open-market innovation. Harvard Business Review 80 (10): 5–12.
+
+Rivette KG, Kline D. 1999. Rembrandt in the Attic: Unlocking the Hidden Value of Patents. Harvard Business School Press: Boston, MA.
+
+Rockett K. 1990. Choosing the competition and patent licensing. RAND Journal of Economics 21 (1): 161–171.
+
+Shapiro C, Varian HR. 1998. Information Rules: A Strategic Guide to the Network Economy. Harvard Business School Press: Boston, MA.
+
+Sine WD, Shane S, Di Gregorio D. 2003. The halo effect and technology licensing: the influence of institutional prestige on the licensing of university inventions. Management Science 49(4): 478–496.
+
+Sutton J. 1991. Sunk Costs and Market Structure. MIT Press: Cambridge, MA.
+
+Teece DJ. 1986. Profiting from technological innovation. Research Policy 15 (6): 285–305.
+
+Teece DJ. 1988. Technological change and the nature of the firm. In Technological Change and Economic Theory, Dosi G, Freeman C, Nelson R, Silverberg G, Soete L (eds). Pinter: London; 256-281.
+
+The Economist. 2005. A market for ideas. 20 October: 48–51.
+
+Tullo AH. 2003. The next wave. CENEAR 81: 26-27.
+
+Von Hippel E. 1994. 'Sticky information' and the locus of problem solving: implications for innovation. Management Science 40 (4): 429-439.
+
+Williamson OE. 1991. Comparative economic organization: the analysis of discrete structural alternatives. Administrative Science Quarterly 36(4): 269-296.
+
+Williamson OE. 1999. Strategy research: governance and competence perspectives. Strategic Management Journal 20 (12): 1087–1108.
+
+Copyright c 2006 John Wiley & Sons, Ltd.
+
+Strat. Mgmt. J., 27: 1141-1158 (2006) DOI: 10.1002/smj
+

processed/00015_W3121800927/document.md

@@ -0,0 +1,656 @@
+© Oxford University Press 2003
+
+All rights reserved
+
+Oxford Economic Papers 55 (2003), 235-264 235
+
+# Does intellectual property protection spur technological change?
+
+By Sunil Kanwar* and Robert Evenson†
+
+* Department of Economics, Delhi School of Economics, University of Delhi, Delhi 110 007, India, and Yale University; E-mail: kanwar@cdedse.ernet.in
+
+†Department of Economics, Yale University
+
+Of the diverse factors motivating technological change, one factor that has received increasing attention in the recent past has been the protection of intellectual property rights. Given fairly recent changes in the international policy ethos where a regime of stronger intellectual property protection has become a fait accompli for most developing countries (and the developed too), it is of some significance to ask whether more stringent protection of intellectual property does indeed encourage innovation. And this is the question which this paper examines, utilising cross-country panel data on R&D investment, patent protection and other country-specific characteristics spanning the period 1981–95. The evidence unambiguously indicates the significance of intellectual property rights as incentives for spurring innovation.
+
+## 1. Introduction
+
+A distinctive characteristic of modern economic growth has been the significant role played by technological change. Of the diverse factors motivating technological change, one factor that has received increasing attention in the recent past has been the role of intellectual property protection. Given the shift in the ownership distribution of innovations away from individuals and towards large corporations in recent decades, 1 intellectual property protection has arguably become an even more important stimulus than hitherto; for such protection augments both the means and the incentive to undertake expensive innovation. While there has been an ongoing debate on whether strong(er) intellectual property protection encourages or retards the rate of technological change and, perhaps, that of economic growth, policy-makers have in any case moved towards a regime of stronger intellectual property protection in recent times, as evidenced by the agreement on Trade Related Intellectual Property issues (TRIPs) under the GATT-WTO 2 in April 1994.
+
+That the inventor be given protection for his invention was a principle recognized as early as the fifteenth century, when the very first patents were granted by the city state of Venice. The motives cited were the natural right of the inventor to
+
+1 Thus, in the United States, the share of all patents issued to individual inventors was 91% in 1901; but by the early 1980s this had reduced to under 19% (Scherer and Ross, 1990).
+
+$^{2}$ GATT-General Agreement on Tariffs and Trade; WTO-World Trade Organisation.
+
+---
+
+236 INTELLECTUAL PROPERTY PROTECTION
+
+the fruits of his labour, the benefits accruing therefrom to society at large, society's compensation (therefore) of the costs incurred by him, and the fillip this would provide to inventive propensities (Mandich, 1948). This is not to imply, however, that opposition to such protection did not also exist right from the earliest times. Some argued even then, for instance, that such protection mimicked prohibitive tariffs (insofar as the protected products could not be imported freely, hindering trade) and ought not to be granted (Schiff, 1971). $^3$ That this issue has continued to remain controversial into the present times would be understating the case; one of the important reasons for this being the lack of cumulative empirical evidence. Evenson (1990) categorises the attempts to examine the incentive effects of IPRs on innovation into two—studies of (firm) behaviour (see the following section; references in Evenson, 1990; Lanjouw and Cockburn, 2001; and Sakakibara and Branstetter, 2001); and studies on the intrinsic value of protection (see the references cited in Evenson, 1990; Lerner, 1994; and Schankerman, 1998). To these categories we may add (or else widen the first category to include) attempts to estimate aggregate or economy-wide response (Gould and Gruben, 1996; Park and Ginarte, 1997). The evidence is thin and mixed, to say the least. Evenson himself opines that the (weak) evidence ‘... does not imply that intellectual property rights are of little value as inducements to innovate. It attests to the relevance of the process of imitation’.
+
+This paper seeks to analyse empirically the influence that intellectual property protection might have on innovation and technological change. We study this relationship at the economy-wide level, using cross-country data on the strength of intellectual property rights, technological change, and other relevant countryspecific controls. Our evidence shows, unambiguously, that intellectual property protection (proxied by an index of patent rights) has a strong positive effect on technological change (proxied by R & D investment expenditures). This has obvious implications for developing countries, which have been confronted by an international policy ethos favouring the strengthening of intellectual property rights under the recent TRIPs agreement almost as a fait accompli . We provide some evidence that conforming with the agreement may not be undesirable in the longer run. Of course, needless to add, the strengthening of IPRs must occur not just on the statute books but in reality. Only then would stronger IPRs provide the right incentives for innovation.
+
+Section 2 briefly considers the pros and cons of weak protection versus strong protection in the global context, and the implications this might have for the relationship between intellectual property protection and technological change. Section 3 delineates the estimation model. Section 4 briefly discusses the data and samples used for estimation, and the appropriate estimation procedure in
+
+3 Thus, the Dutch government repealed the existing patent law in 1869, whereas the Swiss government could not enact a patent law because successive referendums in the late-19th century defeated such a proposal.
+
+---
+
+S. KANWAR AND R. EVENSON 237
+
+our context. Section 5 then presents the estimation results. Section 6 explores the possibility of a nonlinear relationship between technological change and protection. Section 7 refines the estimations carried out in Section 5, by purging the R&D data of its public sector component. Section 8 outlines some broad conclusions.
+
+## 2. Strong protection or weak protection?
+
+When considering the process of technological change, two important characteristics of innovations ought to be kept in mind. First, innovations are non-rival goods. That is, use of a particular innovation by a producer does not preclude other entrepreneurs from using it. Second, innovations are partially non-excludable goods. This implies that the innovator is often unable to completely prevent others from using the innovation without due authorization (Romer, 1990). It is these two properties of innovations that form the basis of the argument in favour of intellectual property protection, which serves to decrease the degree of nonexcludability of innovations by assigning to the inventor the property rights over his innovation for a given period of time. In return, disclosure by the inventor (in the patent application), adds to society's fund of knowledge once the property rights lapse. How strong should the protection given be, however, is a difficult and complex question to settle in practice. We briefly discuss below the various arguments for and against the strong protection of intellectual property.
+
+### 2.1 Arguments favouring weak protection
+
+One of the more obvious benefits supposed to accrue from the weak protection of intellectual property, especially for developing countries, is claimed to be the cheap acquisition of technology through imitation, 4 and the encouragement this provides to innovation. Underlying these arguments may 5 be the suppositions that protected products have no close substitutes or competition, that the menu of products available would be the same irrespective of the protection regime (i.e. strong or weak), and that there are no offsetting costs to the countries in question (Sherwood, 1990). Second, there are several instances of patent abuse involving inadequate disclosure in developing countries, where important bits of knowledge have been withheld from the patent applications, without which even those skilled in the art would not be able to replicate the inventions after the expiry of the patent. Thus, a system that is meant to add to society's fund of knowledge ends up thwarting technical change (Roffe, 1974). Third, strong protection, by creating a monopoly, may induce the producer to accumulate `sleeping patents' in an effort to preserve market share (Gilbert and Newbery, 1982). In which case, strong protection will only serve to limit the rate of technological change. Although, whether a producer
+
+$^{4}$ Which, in turn, keeps the market prices of products using the protected technology) low.
+
+5 This is not to deny the possibility that belief in the benefits of weak protection may not imply at all that one fails to recognize the drawbacks.
+
+---
+
+238 INTELLECTUAL PROPERTY PROTECTION
+
+would want to preserve market share with extant technology would depend upon whether or not he can do better by innovating (Park, 1997).
+
+Takalo and Kanniainen (2000) demonstrate that stronger protection, by enhancing the innovator's capacity to wait and see how viable the innovation turns out to be, delays the commercial exploitation of the innovation and hence inhibits technical change. Bessen and Maskin (2000) show that if innovation is sequential and complementary (as in certain industries like software, semiconductors and computers), stronger protection would limit imitation and thereby inhibit technological change. Helpman (1993) shows that strong protection will increase the rate of innovation only in the short term as it raises profitability; in the long term it lowers the innovation rate as the producers tend to produce the older products. In a global context, Deardorff (1992) demonstrates that as patent protection is extended (to developing countries, for instance), the gains from protection in the form of higher inventive activity (on account of higher profits), taper off as diminishing returns set in, and at some point could fall short of the costs of protection (due to monopoly pricing). Again, Chin and Grossman (1990) show that the protection of intellectual property enhances global incentives for R & D investment only when R & D productivity is large, but not when innovations are small. Grossman and Helpman (1991) show that stronger protection (or production subsidies) would cause the global rate of technical innovation to fall.
+
+### 2.2 Arguments favouring strong protection
+
+On the other hand, there is evidence which suggests that strong protection stimulates innovative activity, at least in some industries such as pharmaceuticals and chemicals. Thus, Levin et al. (1987), in a survey of 650 American R&D executives who had been asked to rank order alternative means of protecting products and processes across 130 industries, found that for the pharmaceuticals and chemicals industries the executives placed great emphasis on patent protection. This result is supported by Mansfield (1986), who questioned a hundred R&D executives across different US industries about what percentage of their inventions over the period 1981–3 would not have been made had it not been for the availability of patents. The survey responses revealed that 60 % of the inventions in pharmaceuticals and about 40 % in chemicals would not have been developed in those circumstances (see Evenson, 1990, for a summary of similar evidence). This evidence is in consonance with Segerstrom (1991) who shows that government incentives to innovation unambiguously increase inventive intensity. It is not, however, necessarily in agreement with Segerstrom et al. (1990), who show that stronger protection may either increase or lower inventive activity.
+
+Second, `strong' protection is often necessary, because even with this `strong' protection competition begins to gnaw away the profits fairly quickly, lowering innovation incentives. Lesser (2000) notes that the commercial life of a new (crop) variety in the US is only about seven years and has been declining due to competition (and is not the statutory 20 years). Mansfield (1985), using a survey of 100 US
+
+---
+
+S. KANWAR AND R. EVENSON 239
+
+firms spanning 13 manufacturing industries, reports that the average time lapse till the point where information is leaked to rival firms was only about a year for new products and processes, and 12 to 18 months for innovation decisions. 6
+
+Third, foreign direct investment and the transfer of technology from the North to the South may be adversely affected, if adequate protection is wanting in the South. Taylor (1993, 1994) shows that in such a scenario, weak protection in the South (to facilitate unintended technology transfers, for instance), would evoke a defensive response from inventors in the North; so that Southern industries which were importing technology would now have to employ less than the best practice technology, thereby lowering aggregate R&D. Lai (1998) qualifies this statement by showing that the relationship between protection in the South and innovation depends on whether imitation or foreign direct investment is the mode of production transfer from the North to the South. Stronger intellectual property protection in the South increases the rate of product innovation if foreign direct investment is the mode of production transfer, but decreases the rate of innovation if imitation is the mode of production transfer. Diwan and Rodrik (1991) demonstrate that stronger protection in any of the two regions (North and South) stimulates innovative activity, partly because of a better fit between the available and desired technologies in that region.
+
+Fourth, strong protection by the South may be necessary if it is to prevent rampant piracy of intellectual property and consequent retaliation by the North, which may adopt various means to stem the outflow of technology (especially frontier technology). This may, in turn, impede technological change in the South (for some piracy estimates see Gadbaw and Richards, 1988) . $^7$ It was the loss of this potential profit which originally motivated the North (particularly the United States) to bargain hard to bring intellectual property issues within the purview of the GATT/WTO; for the dispute settlement mechanism in this forum was reasonably effective, whereas there was no such mechanism in the WIPO $^8$ (Sherwood, 1990) . Moreover, by linking intellectual protection with trade, the North gained insofar as it could use trade sanctions to counter piracy. $^9$ This can be prevented, however, if the root cause of piracy — namely, weak protection — is remedied.
+
+Thus, we find that a host of arguments can be adduced for and against the strong protection of intellectual property rights. From the arguments per se, and their
+
+6 The empirical evidence cited in this paragraph as well as the previous one relates only to developed countries. While it would be preferable if similar evidence were also available from developing countries, the fact is that such studies are few. In fact, even within the developed countries, a lot of the evidence relates only to the US (see Kennedy and Thirlwall, 1973, who make this point).
+
+7 In assessing the phenomenon of `piracy' one must be mindful of the distinction between the perfectly legal phenomenon of imitation under weak protection laws, and the poor or non-enforcement of the laws that do exist. We thank Jenny Lanjouw for pointedly bringing this home to us.
+
+$^{8}$ World Intellectual Property Organisation.
+
+9 In fact, the US repeatedly used its trade laws 'super 301' and 'special 301' in its bilateral dealings with some nations.
+
+---
+
+240 INTELLECTUAL PROPERTY PROTECTION
+
+theoretical expositions and empirical evidence, 10 one is hard put to decide on the appropriate degree of intellectual property protection, specifically with respect to its influence on inventive activity. It is the purpose of this paper, therefore, to empirically examine the relationship between intellectual property protection and the rate of technological change.
+
+## 3. The estimation model
+
+In doing so, we do not wish to deduce the relationship from a more indirect relationship such as that between economic growth and protection (as estimated in Gould and Gruben, 1996) . The reason is, that the effect of intellectual property protection on economic growth involves, first, the effect of protection on innovation and, second, the effect of innovation on economic growth. So if we were to find an insignificant relationship between the strength of protection and economic growth, that need not imply an insignificant relationship between the strength of protection and innovation per se ; it may simply be the result of the fact that innovation accounted for an insignificant part of economic growth (in comparison to other growth-inducing factors such as input accumulation). And this does not rule out the possibility that, had there been relatively more innovation, there would have been more growth. In other words, the weak or zero relationship between economic growth and protection may send out the message that strengthening protection would not be useful whereas, in fact, that may not be the case — for strengthening protection would induce innovation, and if there is enough innovation, that would induce significant growth. It might be preferable, therefore, to dwell on the relationship between technological change per se and intellectual property protection. $^11$
+
+### 3.1 Capturing technological change
+
+To proceed with the estimation we must first clarify what constitutes technological change. At one level technological change may be interpreted to refer not to the change in technology in existence (i.e. known to society at any given point in time), but rather to the change in technology in use. Obviously, the latter cannot occur without the former. However, the former can and does occur independently of the latter. Thus, many new innovations may exist and yet not be commercially exploited because, for instance, they may be economically unviable. To the
+
+$^{10}$ Some of which is about marginal changes and some about absolute levels.
+
+11 Of course, if we are to draw any further implications of the effect of protection on economic growth in our case, we must refer to the wide literature on the relationship between technological change and economic growth (which, though positive, may be small or large depending on the particular economy or time period in question). The relevant literature is too vast to be exhaustively referenced here, but a partial listing would include Frantzen (2000), Van Ark et al. (2000), Yuhn and Kwon (2000), Engelbrecht (1997), Coe and Helpman (1995), Nishimizu and Robinson (1984), Norsworthy and Malmquist (1983), and Christensen and Cummings (1981).
+
+---
+
+S. KANWAR AND R. EVENSON 241
+
+extent that they are not yet employed, they do not cause any change in the extant technology of production, and hence do not influence economic growth. By the same token, only when they are exploited in the market can the technology of production be said to have changed, with possible ramifications for economic growth. Therefore, it is this phenomenon of the adoption or economic exploitation of innovations that ought ideally to be emphasized, rather than the phenomenon of the generation of innovations per se. In actual fact, however, this distinction may not be practicable.
+
+We propose to represent technological change, therefore, by research and development investment as a proportion of gross national product ( RDI ). 12 While all of research and development expenditure does not necessarily fructify into inventions and innovations (and, indeed, R & D productivity may decline with R & D intensity), such investment has the virtue of being more closely related to inventive activity than is, say, non-residential physical capital investment, while at the same time being more `all-encompassing' than still another possible proxy, namely patent applications. 13
+
+### 3.2 The explanatory variables
+
+Countries differ widely in the strength of protection that they provide to intellectual property. While a comprehensive evaluation of the relative strengths of protection would require dwelling on each of several instruments such as patents, trademarks, copyrights and trade secrets (South Centre, 1997), 14 differences in patent laws across countries are considered the most important and are perhaps the most dramatic. The protection offered may differ along several dimensions. Thus, patent laws across countries differ with respect to coverage. While some countries allow both product and process patents in a broad range of activities, some disallow product patents for innovations in certain areas such as pharmaceuticals, chemicals and food products (e.g. India). Second, patent laws differ with respect to the duration of protection. Patent duration in most western European nations has been 20 years, whereas in other countries it has varied between five and
+
+12 R&D data were available as proportions of GNP and not GDP; the discrepancy should be of the second order of smalls.
+
+13 Another drawback of using patent applications as the regressand would be, that they are partially, definitionally related to the strength of protection. Thus, if a country were to suddenly allow the patenting of pharmaceuticals, both the strength of protection as well as the number of patent applications would simultaneously go up. See Griliches (1990), who concludes that, ‘Among the major findings was the discovery of a strong relationship between patent numbers and R&D expenditures . . . ’, and that ‘ . . . the relationship between R&D and patents is close to proportional . . . ’, which should support our use of R&D data (just as much as measures based on patent data). See also Pavitt (1982) and Kortum (1997) for further discussion on this issue, and Trajtenberg (1990) for an alternative viewpoint.
+
+14 In addition, the protection of geographical indications, integrated circuits and industrial designs may also be considered subsequent to the TRIPs agreement of 1995 (South Centre, op. cit.); but this falls outside the sample period for this study.
+
+---
+
+242 INTELLECTUAL PROPERTY PROTECTION
+
+17 years (e.g. India, Pakistan, Sri Lanka, Jordan, US). Further, while some countries measure patent life from the patent application filing date (e.g. Nigeria, Jordan, Thailand), others measure it from the publication date (India, Korea, Austria, Australia), while still others measure it from the grant date (Pakistan, Mexico, Portugal, Canada, Iceland, US). Third, enforcement procedures may differ depending on whether or not countries allow for preliminary injunctions in the case of an alleged infringement, and on who carries the burden of proof. Fourth, while some countries allow pre-grant opposition to a patent application (e.g. Japan), others allow only post-grant opposition (e.g. western European countries). Fifth, in some countries patentees may face a higher risk of loss of protection (once granted) on account of `non-working' of the patent or else compulsory licensing. And so on. There are many other differences that one could list. While the TRIPs agreement of 1994 allows for a process of gradual harmonisation of the protection laws across the member countries, and this process is indeed underway, many of these differences still obtain in large measure and certainly apply to our data set. These differences are captured by Ginarte and Park (1997) in an index, which we use as the index of protection ( IP ) variable. $^15$ The index incorporates five aspects of patent laws: extent of coverage, membership in international patent agreements, duration of protection, provisions for loss of protection, and enforcement mechanisms. $^16$ It ranges from zero to five with higher values of the index indicating stronger patent protection. $^17$
+
+15 An alternative index is that constructed by Rapp and Rozek (1990), and used by Gould and Gruben (1996). The Rapp and Rozek index is based on a comparison of individual countries' patent laws with the guidelines proposed by the US Chamber of Commerce's Intellectual Property Task Force in Guidelines for Standards for the Protection and Enforcement of Patents. It ranks countries from zero to five, with the former signifying an absence of patent protection laws and the latter full conformity with the proposed standards. The Ginarte-Park index is superior in many respects, as it looks into various facets of patent protection in greater detail, and therefore makes for greater variation in the index of protection even amongst the developed countries.
+
+16 To some extent this index does allow for differences in effective protection across countries. Thus, the index attempts to capture differences in `enforcement mechanisms' insofar as countries allow for (i) preliminary injunctions, (ii) contributory infringement, and (iii) burden-of-proof reversal. These differences, however, pertain only to what the countries profess to do according to their statutes. It does not, by itself, tell us whether and to what extent countries actually use these features. At the present moment, however, it appears to me virtually impossible to carry out such an exercise. For instance, in the case of preliminary injunctions, one would have to consider all relevant litigation in each sample country in each time period (!) and compare, say, the percentage of cases in which this clause was actually used. This would be a monumental task, even if the data were available.
+
+17 It may, alternatively, be argued that a country with high R & D intensity would have an incentive to protect its intellectual property. Tackling this endogeneity is, however, beyond the scope of this paper. Let us clarify, however, that we mention this possibility more for completeness than for its likely importance. We personally do not think it to be potentially important, because changes in protection levels are mostly in the nature of legislative changes (either on the statutes or in their interpretation by the courts), and legislation does not react with any regularity to changes in R & D intensities. The reason for this, we feel, is that legislative changes follow from a complex decision process not quite determined by market forces.
+
+---
+
+S. KANWAR AND R. EVENSON 243
+
+Several control variables are considered. R & D investment, and investment in general, have been found to be pro-cyclical for various reasons — the availability of internal funds for activities for which loans are usually not available (Hall, 1992; Himmelberg and Petersen, 1994) , and demand pull forces indicating profitability being two of the most important (Geroski and Walters, 1995) . This may be captured by introducing (current) GDP as a regressor (Guellec and Ioannidis, 1997; Geroski and Walters, 1995; Barro, 1991; Kamien and Schwartz, 1982; Nadiri, 1980) . Using this variable as a regressor, however, does not tell us much about the mechanism of causation. Also, it is a nonstationary variable, and it does not make sense to use nonstationary variables to explain movements in stationary variables such as RDI . $^18$ To remedy this (at least relatively), $^19$ we propose to use instead two separate variables. To represent the internal funds available for R & D investment we use gross domestic savings as a proportion of GDP, lagged one period $(S_{t-1})$ , whereas to capture the demand-pull factors we use the ratio of current per capita GDP to per capita GDP lagged one period ( $\Delta GDPPC$ ). $^20$
+
+Countries that have factor bundles that contain relatively more human capital will tend to innovate and hence invest at a faster rate than countries having factor bundles with lower proportions of this resource, because this input is central to research and development (Romer, 1990) . In other words, it is an enabling factor. Given that in certain skills (e.g. in communications) the returns are higher if others are also skilled, increases in human capital tend to induce higher rates of investment (Barro, 1991) . In cross-country analysis we must also allow for the fact that more educated countries are better able to absorb the innovations made elsewhere (Nelson and Phelps, 1966) . We do not consider literacy rates apt for this purpose, because these are usually based on criteria such as whether the respondent can sign his name, or whether he can identify a given number of characters of some language etc., which are not particularly cogent indicators of his skill level. Nor would primary school enrolment rates serve the purpose, because too many countries have already achieved 100 % enrolment at this level and hence there would be insufficient variation in this variable (Barro, 1991) . Besides, ideally we ought to use a stock measure of human capital rather than a flow measure. Therefore, we represent the human capital variable by the Barro-Lee (2000) data on the average number of years of formal schooling of the population equal to or over age 15 ( EDU ).
+
+Political instability is an important factor influencing investment decisions. Countries with unstable political, and hence economic, climates witness a drying
+
+18 This is precisely another problem with the Gould and Gruben (1996) study, in that they try to explain a stationary variable such as the growth rate of GDP using a nonstationary variable such as GDP.
+
+19 Ideally, we ought to estimate a vector autoregression model to take care of the related endogeneity problem. This, however, would require long time series of data, which are just not available.
+
+20 To jump ahead for a moment, since all variables are measured in logs, the ratio of current to lagged per capita GDP will translate into a change in (log) GDP between periods; hence the use of the operator $\Delta$ for this variable.
+
+---
+
+244 INTELLECTUAL PROPERTY PROTECTION
+
+up of productive investment. Even localized political conflicts that tend to be long drawn out can stifle investment in those regions within countries or at the very least divert scarce resources away from R & D towards the control and resolution of those conflicts. We propose to represent this factor by a relatively comprehensive `state failure' dummy. Using data on genocides and politicides, revolutionary wars, ethnic wars, and abrupt regime changes towards autocratic rule compiled by the Center for International Development and Conflict Management (see Esty et al. , 1998) , we construct a political instability dummy ( ID ) for each country which equals 1 in years exhibiting one or more of these phenomena and equals 0 otherwise.
+
+The I-O literature points out that lending institutions are reluctant to lend for R&D activities simply because such ventures tend to be highly risky with uncertain expected rates of return. Therefore, R&D investment tends to get financed mainly from internal funds (Hall, 1992; Himmelberg and Petersen, 1994) . Hence, as we argued above, the importance of savings. Even so, the real lending rate of interest ( RLR ) may be used as an additional control variable, perhaps to reflect the opportunity cost of internal funds if not the actual cost of borrowed funds used for R&D (Guellec and Ioannidis, 1997) .
+
+Finally, it has been argued that the trade orientation of a country can be of importance in determining its propensity to innovate. The conjecture (sometimes implicit) is that relatively open economies tend to face relatively more competition, not having access to sheltered markets, and are compelled to invest relatively more in R & D (Edwards, 1992; World Bank, 1987; Krueger, 1978; Bhagwati, 1978; and Lewis, 1955) . Measuring the degree of openness of an economy, however, has proved a fairly thorny exercise. The alternative measures used in the literature have been exports shares, trade shares (i.e. exports plus imports as a share of GDP), effective tariff rates, real exchange rate distortions, black market exchange rate premiums and still other measures. All of these measures may be shown to be deficient in one respect or another. Some, however, are better than others and one that seems to be preferred is the black market exchange rate premium (Gould and Gruben, 1996) . We, therefore, use the black market exchange rate premia to construct a black market premium dummy ( BMPD ), which equals 0 for relatively open economies or those whose black market premium is less than the median of the sample countries, and 1 for the relatively closed economies or those whose black market premium is more than the median. $^21$
+
+The above discussion leads us to the following estimation model
+
+$$RDI_{t}=f(S_{t-1},\Delta GDPPC_{t},EDU_{t},BMPD_{t},ID_{t},RLR_{t},IP_{t})\quad (1)$$
+
+where the different variables are as defined above. What we are interested in is the `long term' relationship between intellectual property protection and research and
+
+21 The black market exchange rate premium itself is calculated as $(BMR-OR)/OR$ , where $BMR$ is the black market exchange rate and $OR$ is the official exchange rate (both measured in local currency per US dollar).
+
+---
+
+S. KANWAR AND R. EVENSON 245
+
+development investment, and it appears reasonable that yearly data would not be appropriate to capture it. Surely, yearly changes in research and development investment cannot be expected to reflect the response of innovation to changes in the strength of protection, not only because changes in the latter are rather occasional, but also because innovation decisions tend to be relatively long term decisions. It would be more plausible, therefore, to estimate this relationship using data averaged over longer periods than a single year. Such averaging would reduce, if not eliminate, the yearly variations in R & D investment on account of `short run' causes. Over how many periods should the data be averaged, however, is not a priori clear. Following the lead of earlier researchers studying long-term macrorelationships using panel data (e.g. Islam, 1995), we use quinquennial data to estimate this relationship. In part we are forced to do this, because data on some variables that we intend to use in this study — such as education ( EDU ) and the index of protection ( IP ) — are only available on a quinquennial basis (see the data section below).
+
+## 4. Data, samples, and estimation issues
+
+### 4.1 Data and samples
+
+Data on the variables discussed above were collected from several different sources (see Appendix 3 ). The maximum sample size was 32 countries (listed in Table 1 ). The major constraining factor for the sample period was the short series for R & D investment, 22 which was available only from 1981 to 1990 for a respectable number of countries; also, data on EDU and BMPD were only available till 1990 and the early 1990s, respectively. But for these constraints, the sample period could have been longer. While data on most variables were available on a yearly basis, those for EDU and IP were only available quinquennially. At least for variable IP this is understandable, because changes in patent laws and their enforcement are few and well-spaced so that it becomes difficult to capture changes in protection levels on a yearly basis. The summary statistics pertaining to the variables are listed in Appendix 1 .
+
+Data on all the variables were available for only 29 countries for the period 1981 – 1990. 23 Since we propose to use five-year averages, this implies that we have only two data points for each country, 1985 (or the average for 1981 – 5) and 1990 (or the average for 1986 – 90), for a total of 58 observations. Estimations using this sample constitute `Exercise 1'. If we are prepared to drop variable $RLR$ from the initial
+
+22 Given that the R&D data used include both private and public components, an underlying theoretical framework would require theories of both public and private investment in R&D. Further, the framework would need to address issues of `crowding out', complementarity/substitution between the two components, and sequencing (insofar as more basic public R&D feeds into more applied private R&D). This is well beyond the scope of this paper.
+
+23 Even so, for Austria, Mexico, and Indonesia (1981–85 only), we had to use the real deposit rate in lieu of the real lending rate.
+
+---
+
+246 INTELLECTUAL PROPERTY PROTECTION
+
+Table 1 Strength of intellectual property protection and R&D investment, 1981-90
+
+<table><tr><td colspan="2"> $0 ≤ IP^{\circ}&lt;1$ </td><td colspan="2"> $1 ≤ IP&lt;2$ </td><td colspan="2"> $2 ≤ IP&lt;3$ </td><td colspan="2"> $3 ≤ IP&lt;4$ </td><td colspan="2"> $4 ≤ IP ≤5$ </td></tr><tr><td>Country</td><td>RDI*</td><td>Country</td><td>RDI</td><td>Country</td><td>RDI</td><td>Country</td><td>RDI</td><td>Country</td><td>RDI</td></tr><tr><td>Indonesia</td><td>0.303</td><td>Venezuela Mexico India</td><td>0.352</td><td>Iceland Singapore</td><td>0.812 Singapore Canada</td><td>Nigeria 0.660</td><td>0.542 Sri Lanka</td><td>Austria Italy</td><td>1.276</td></tr><tr><td></td><td></td><td>Thailand Jordan Portugal Pakistan</td><td>0.772</td><td>Canada</td><td>1.419</td><td>Australia</td><td>1.222</td><td>Netherlands USA</td><td>2.081</td></tr><tr><td></td><td></td><td></td><td></td><td>Jamaica</td><td>0.058</td><td>Norway</td><td>1.516</td><td></td><td>2.719</td></tr><tr><td></td><td></td><td></td><td></td><td>Mauritius</td><td>0.352</td><td>Spain</td><td>0.612</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td>Finland</td><td>1.720</td><td>UK</td><td>2.229</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td>Finland</td><td>0.870</td><td>S. Africa</td><td>0.882</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td>Ireland</td><td>0.870</td><td>Sweden</td><td>2.822</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td><td>S. Korea</td><td>1.421</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td><td>Denmark</td><td>1.389</td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td></td><td>France</td><td>2.233</td><td></td><td></td></tr><tr><td>Av.RDI</td><td>0.303</td><td></td><td></td><td></td><td>Japan</td><td>2.704</td><td></td><td></td><td></td></tr><tr><td></td><td></td><td></td><td></td><td></td><td>Belgium</td><td>1.614</td><td></td><td></td><td>1.797</td></tr></table>
+
+
+Note: * IP-Index of protection; RDI-Share of R&D investment in GNP.
+
+---
+
+S. KANWAR AND R. EVENSON 247
+
+general model, on grounds that earlier work on R & D shows internal funds to be the really significant source of funds for such investment, then our sample size becomes 31 countries (now including Jordan and Pakistan). Estimations using this larger sample constitute `Exercise 2'. Many researchers (estimating various relationships) have used literacy rates ( LIT ) in lieu of the human capital variable that we have defined above. While we have reservations about the use of literacy rates to represent skill levels (which we briefly voiced above), doing so expands our sample of countries by another (Nigeria). Estimations based on this expanded sample of 32 countries constitute `Exercise 3'. Finally, in `Exercise 4' we limit ourselves to the OECD $^{24}$ countries included in Table 1 , for which R & D data were available for the longer time span 1981 – 95. In place of EDU , however, we had to use LIT for the human capital variable, because data on EDU are available only till 1990. Also, to construct the demand variable $\Delta GDPPC$ , we had to use World Bank data (World Bank, 2000) instead of the Penn World Table (Heston et al. , 2001) , because the latter reports data only till the very early 1990s (and not till 1995). Unfortunately, what we gain in terms of a longer sample period (one more time point for each country), we lose in terms of the smaller set of countries. $^25$
+
+### 4.2 Estimation issues
+
+A large menu of alternative methods obtains for the estimation of panel data models with individual country effects. A significant issue that arises in this context is whether such effects ought to be treated as `fixed' or `random'. In a lot many applications the former approach is adopted, which requires estimating the individual effects as parameters. It is not clear a priori , however, that it ought to be the automatically preferred approach. Indeed, it has been shown (Nerlove, 1967, 1971; Maddala, 1971; Nickell, 1981) that OLS estimates (either ignoring or) treating individual effects as constants to be estimated gave `exceedingly poor estimates' of the system parameters. These studies point out that the advantage of the `random effects' model follows from the fact that estimating a fixed effects model implies not only substantially fewer degrees of freedom but also rules out all information that may be available by directly comparing individual units. This is particularly egregious when the number of individual units in a panel substantially exceeds the number of time periods (as in our samples, where the number of countries outnumber the number of time periods by a factor of as much as 15); $^26$ for, in such a situation, we must make efficient use of the information
+
+$^{24}$ Organisation for Economic Co-operation and Development.
+
+25 The OECD sample includes Australia, Austria, Belgium, Canada, Denmark, Finland, France, Iceland, Ireland, Italy, Japan, Netherlands, Norway, Portugal, Spain, Sweden, United Kingdom, and the United States. We do not include Korea and Mexico because these joined the OECD only recently—1996 and 1994, respectively.
+
+$^{26}$ In the econometric studies cited above, the number of cross section units were assumed to exceed the number of time periods only by a factor of 2.5 .
+
+---
+
+248 INTELLECTUAL PROPERTY PROTECTION
+
+across individual units, to estimate that part of the behavioural relationship under study which contains variables that (are hypothesized to) differ substantially across the units. Even in samples of moderate size (e.g. $T \geqslant 2, N-K \geqslant 10$ ; where $T$ is the number of time periods, $N$ the number of cross-section units and $K$ the number of parameters), Taylor (1980) shows the random effects estimator to be relatively efficient. For these reasons we prefer to use a random effects model for studying our relationship. (As it turns out, a Hausman specification test also supports this choice.) Specifically, the econometric model estimated is of the type
+
+$$\begin{split}
+y_{it}=\overline{\beta}_{1}+\beta_{k}x_{kit}+\mu _{i}+\varepsilon_{it}&\quad i=1,...,N;t=1,...,T\\
+\mu _{i}&\sim N(0,\sigma_{\mu }^{2})\\
+\varepsilon_{it}&\sim N(0,\sigma_{\varepsilon}^{2})\\
+E\mu _{i}\varepsilon_{ijt}=0,&\quad \forall i,j\text{ and }t\\
+E\mu _{i}\mu _{j}=0,&\quad i\neq j\\
+E\varepsilon_{it}\varepsilon_{js}=0,&\quad i\neq j,t\neq s
+\end{split}$$
+
+where the regressand $y_{it}$ refers to R & D investment for the $i$ th country in the $t$ th year, while $x_{kit}$ refers to the $k$ th regressor for the $i$ th country in the $t$ th year. All variables are taken in (natural) logs. All variables are stationary. Estimation yields feasible GLS estimates of the model parameters, which are discussed below.
+
+## 5. Estimation results
+
+In Table 1 we categorize the list of countries in our data set into five groups, based on the average strengths of intellectual property protection obtaining in these countries over the period 1981 – 90. 27 We also note the share of GNP that they devote to R & D. 28 From these data there appears to be a monotonic positive relation between RDI and the strength of intellectual property protection. Thus, countries with the lowest level of protection invest less than 1/3 of 1 % of their GNP on research and development activities. At the other extreme, countries with the highest level of protection invest almost six times as much on R & D. Of course, no firm conclusion can be reached unless we allow for various control factors that impinge on R & D investment. This is what we do in the following exercises.
+
+### 5.1 Intellectual property protection and technological change: exercise 1
+
+Table 2 reports the `random effects' GLS estimates using a sample of 29 countries for which data were available for all the variables discussed above. Model (1) sets out the results of the `complete' model, i.e. eq. (1) above. Note, that in this analysis, we move from a relatively general model to a relatively specific one (Charemza and
+
+27 We could do the same for the expanded sample period 1981–95, but this would pertain to a much smaller sample of countries.
+
+$^{18}$ See note 12.
+
+---
+
+S. KANWAR AND R. EVENSON 249
+
+Table 2 Exercise 1-random effects estimates; dependent variable: RDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td></tr><tr><td> $S_{t-1}$ </td><td>0.628 (2.174)</td><td>0.669 (2.308)</td><td>0.623 (2.175)</td><td>0.616 (2.168)</td></tr><tr><td> $\triangle GDPPC$ </td><td>0.120 (0.534)</td><td>0.162 (0.711)</td><td>0.176 (0.783)</td><td></td></tr><tr><td>EDU</td><td>1.093 (2.435)</td><td>0.965 (2.193)</td><td>0.900 (2.063)</td><td>0.948 (2.205)</td></tr><tr><td>IP</td><td>0.600 (2.111)</td><td>0.579 (2.035)</td><td>0.506 (1.846)</td><td>0.475 (1.766)</td></tr><tr><td>BMPD</td><td>−0.182 (−1.120)</td><td>−0.210 (−1.202)</td><td>−0.222 (−1.368)</td><td>−0.236 (−1.452)</td></tr><tr><td>ID</td><td>0.521 (1.101)</td><td>0.448 (0.947)</td><td></td><td></td></tr><tr><td>RLR</td><td>0.167 (1.380)</td><td></td><td></td><td></td></tr><tr><td>Intercept</td><td>−5.473 (−4.723)</td><td>−4.674 (−4.684)</td><td>−4.293 (−4.657)</td><td>−4.315 (−4.712)</td></tr><tr><td>RMSE</td><td>2.024</td><td>1.656</td><td>1.274</td><td>1.203</td></tr><tr><td>SC</td><td>0.060</td><td>0.059</td><td>0.055</td><td>0.053</td></tr><tr><td> $F(H_{0}  : all slopes 0)$ </td><td>5.770 0.390</td><td>6.329 0.385</td><td>7.159 0.383</td><td>8.696 0.392</td></tr><tr><td>Observations</td><td>58</td><td>58</td><td>58</td><td>58</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+Deadman, 1997) . In doing so, we may use various model selection tests such as the `root mean squared error' (RMSE) and the `Schwarz criterion' (SC). We must emphasize, however, that one cannot afford to be doctrinaire about the model selection procedure adopted, for the simple reason that theory is just not welldefined enough to guide us in starting from the complete model. More often than not, the available data may be the binding constraint. Interestingly, this criticism applies as much to general-to-specific modelling as to specific-to-general modelling where researchers start off with a `base model' and then add variables one by one; for theory may not be particularly helpful in guiding us to the base model either. We bear these reservations in mind as we proceed.
+
+Model (1) results show that internal funds have a strong and significant positive influence on R & D. This supports earlier findings in the I-O literature which point to the importance of internal funds for investing in such uncertain ventures as research and development for which financial institutions are quite unwilling to lend. Relatively educated countries, ceteris paribus, invest significantly more in research and development; indeed, a skilled manpower base is a pre-requisite to R & D activities. The strength of intellectual property protection is positively and significantly associated with R & D, supporting the hypothesis that we had set out to
+
+---
+
+250 INTELLECTUAL PROPERTY PROTECTION
+
+test. Thus, countries which provided stronger protection tended to have larger proportions of their GDP devoted to R&D activities.
+
+Both the political instability dummy as well as the real lending rate, however, have the wrong sign, although insignificant. Perhaps the insignificance of the latter has to do with insufficient variation across the sample countries — the dummy being zero for all developed countries, which predominate in our sample. Model (2), therefore, drops RLR because theory suggested this to be relatively unimportant in explaining R & D (internal funds mostly financing such activity, as we pointed out). Variable ID still has the wrong sign, so model (3) drops this variable too. The results show that the conclusions we reached above about the importance of internal funds, education and the strength of intellectual property protection remain unaffected. Additionally, the trade orientation variable BMPD is now marginally significant (using a one-tail test) — economies which are relatively open tend to face greater competition and therefore spend more on R & D. In model (4) we further drop the demand variable $\Delta GDPPC$ , which had the right sign but was insignificant in model (3). The internal funds variable, education, the index of protection as well as the trade orientation variable all continue to be significant. 29 (All these conclusions would still be valid were we to reverse the model selection procedure and move from the `base model' (4) to a more general model (3); and indeed even (2) and (1), except that we should eschew (2) and (1) because in these models variables ID and RLR appear with the wrong signs.) Model (4) has the lowest root mean squared error and also minimizes the Schwarz criterion. 30 The ( $F$ -distribution variant of the) Hausman statistic or $m(F)$ is 1.307, with a $p$ -value of 0.281, supporting the estimation of a random-effects specification.
+
+### 5.2 Intellectual property protection and technological change: exercise 2
+
+The received literature on R&D emphasizes the significance of internal funds for such risky investments as R&D, for which creditors are generally unwilling to lend. If, therefore, we were to start with a model that excludes the variable $RLR$ , i.e.
+
+$$RDI_{t}=f(S_{t-1},\Delta GDPPC_{t},EDU_{t},BMPD_{t},ID_{t},IP_{t})\quad (3)$$
+
+our sample size rises to 31 countries (now including Jordan and Pakistan) or 62 observations. The `random effects' GLS estimates based on this larger sample are reported in Table 3 . Model (1) results reveal lagged savings to have a significant positive influence on R & D. The trade orientation variable is strongly negatively related to RDI , indicating the importance of international competition in encour-
+
+$^{29}$ The ( $F$-distribution variant of the) $LM$-statistic for the null hypothesis $H_0$ : $S_{t-1} = 0$ is $LMF = 30.504$ ; for the null hypothesis $H_0$ : $EDU = 0$ , $LMF = 30.242$ ; for the null hypothesis $H_0$ : $BMPD = 0$ , $LMF = 37.707$ , and for the null hypothesis $H_0$ : $IP = 0$ , $LMF = 35.098$ . In all four cases, the LM statistics comfortably exceed the critical value $F(1,53) = 7.161$ at the 1 % level.
+
+$^{30}$ It also happens to have the highest adjusted $R^2$ , but we do not use this statistic for model selection because it is quite unreliable in many situations. We report it here only for those who might be interested.
+
+---
+
+S. KANWAR AND R. EVENSON 251
+
+Table 3 Exercise 2-random effects estimates; dependent variable: RDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td><td>(5)</td></tr><tr><td> $S_{1-1}$ </td><td>0.344 (1.672)</td><td>0.372 (1.734)</td><td>0.377 (1.944)</td><td>0.338 (1.612)</td><td>0.344 (1.816)</td></tr><tr><td> $\triangle GDPPC$ </td><td>0.087 (0.443)</td><td>0.164 (0.807)</td><td>0.155 (0.804)</td><td></td><td></td></tr><tr><td>EDU</td><td>0.252 (0.842)</td><td>0.191 (0.634)</td><td></td><td>0.207 (0.692)</td><td></td></tr><tr><td>IP</td><td>0.943 (3.232)</td><td>0.806 (2.889)</td><td>0.890 (3.409)</td><td>0.806 (2.892)</td><td>0.897 (3.432)</td></tr><tr><td>BMPD</td><td>−0.546 (−2.208)</td><td>−0.375 (−1.617)</td><td>−0.381 (−1.676)</td><td>−0.394 (−1.715)</td><td>−0.400 (−1.780)</td></tr><tr><td>ID</td><td>0.857 (1.661)</td><td></td><td></td><td></td><td></td></tr><tr><td>Intercept</td><td>−2.796 (−3.506)</td><td>−2.602 (−3.298)</td><td>−2.354 (−3.113)</td><td>−2.492 (−3.212)</td><td>−2.224 (−3.010)</td></tr><tr><td>RMSE</td><td>1.218</td><td>0.932</td><td>0.766</td><td>0.866</td><td>0.701</td></tr><tr><td>SC</td><td>0.0498</td><td>0.052</td><td>0.045</td><td>0.0487</td><td>0.042</td></tr><tr><td> $F(H_{0}: all slopes 0)$ </td><td>4.156</td><td>4.566</td><td>5.061</td><td>5.361</td><td>6.151</td></tr><tr><td> $R^{2}$ </td><td>0.247</td><td>0.239</td><td>0.230</td><td>0.243</td><td>0.233</td></tr><tr><td>Observations</td><td>62</td><td>62</td><td>62</td><td>62</td><td>62</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+aging research and development activities. Intellectual property protection continues to have a strong positive association with R&D, as in the previous exercise.
+
+But the political instability dummy again has the wrong sign, although it is insignificant. Model (2), therefore, drops this variable. All of our previous results — from model (1) — remain unchanged. Models (3) and (4) further drop the education variable EDU and the demand variable $\Delta$ GDPPC , respectively, which have the right sign but are insignificant in model (2). We find that the sequence of dropping these variables (i.e. which variable we drop first) does not affect the qualitative results. Finally, model (5) drops both EDU and $\Delta$ GDPPC . The internal funds variable, the trade orientation variable and the index of protection all continue to be significant. $^31$ (Again, these conclusions would remain valid if we were to move from the `base model' (5) to the more general model (2) by adding variables. They would also be valid in the case of model (1), but as before we should eschew this model where variable ID appears with the wrong sign.) Model (5) has the lowest root mean squared error and also minimizes the Schwarz criterion. The F -distribution variant of the Hausman statistic is
+
+$^{31}$ The ( $F$-distribution variant of the) $LM$-statistic for the null hypothesis $H_0 : S_{t-1} = 0$ is $LMF = 16.742$ ; for the null hypothesis $H_0$ : $BMPD = 0$ , $LMF = 17.916$ , and for the null hypothesis $H_0$ : $IP = 0$ , $LMF = 7.478$ . In all three cases, the $LM$ statistics comfortably exceed the critical value $F(1,58) \approx 7.103$ at the 1 % level.
+
+---
+
+252 INTELLECTUAL PROPERTY PROTECTION
+
+$m(F)=1.249$, with a $p$-value of 0.301 , supporting the estimation of a randomeffects specification.
+
+### 5.3 Intellectual property protection and technological change: exercise 3
+
+Some researchers use literacy rates ( $LIT$ ) rather than the number of schooling years ( $EDU$ ) to define the human capital variable. While we prefer to use the latter as a measure of skill levels, using literacy rates instead expands our sample of countries by one more (Nigeria). The simple correlation coefficient between the `education' and `literacy' variables was 0.74, supporting the substitution of ` $LIT$ ' for ` $EDU$ ' only partially. Table 4 reports the random effects GLS estimates based on this larger sample of 32 countries or 64 observations. (In this exercise we continue to exclude variable RLR .) Model (1) results, based on the following equation
+
+$$RDI_{t}=f(S_{t-1},\Delta GDPPC_{t},LIT_{t},BMPD_{t},ID_{t},IP_{t})\quad (4)$$
+
+reveal lagged savings to have a strongly positive influence on R&D activities. So also do literacy levels, providing support to our results in exercise 1. The protection of intellectual property rights appears to strongly encourage R&D, as in both the previous exercises.
+
+The political instability dummy ID once again has the wrong sign, and is also significant. This variable is, therefore, omitted from model (2). The literacy variable
+
+Table 4 Exercise 3-Random effects estimates; dependent variable: RDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td></tr><tr><td> $S_{1-1}$ </td><td>0.629 (3.210)</td><td>0.889 (4.146)</td><td>0.885 (4.134)</td><td>0.939 (4.371)</td></tr><tr><td> $\Delta GDPPC$ </td><td>0.198 (1.030)</td><td>0.392 (1.762)</td><td>0.414 (1.871)</td><td>0.560 (2.551)</td></tr><tr><td>LIT</td><td>0.049 (1.976)</td><td>0.037 (1.252)</td><td>0.038 (1.268)</td><td></td></tr><tr><td>IP</td><td>1.132 (3.782)</td><td>0.795 (2.671)</td><td>0.876 (3.101)</td><td>0.843 (3.262)</td></tr><tr><td>BMPD</td><td>−0.070 (−1.025)</td><td>−0.073 (−0.904)</td><td></td><td></td></tr><tr><td>ID</td><td>1.368 (4.062)</td><td></td><td></td><td></td></tr><tr><td>Intercept</td><td>−3.945 (−5.413)</td><td>−4.304 (−5.373)</td><td>−4.396 (−5.542)</td><td>−4.373 (−5.622)</td></tr><tr><td>RMSE</td><td>0.815</td><td>0.833</td><td>0.804</td><td>0.766</td></tr><tr><td>SC</td><td>0.057</td><td>0.075</td><td>0.071</td><td>0.077</td></tr><tr><td> $F(H_{0}: all slopes 0)$ </td><td>9.403</td><td>7.055</td><td>8.350</td><td>10.829</td></tr><tr><td> $\overline{R}^{2}$ </td><td>0.467</td><td>0.345</td><td>0.348</td><td>0.362</td></tr><tr><td>Observations</td><td>64</td><td>64</td><td>64</td><td>64</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+---
+
+S. KANWAR AND R. EVENSON 253
+
+is now insignificant, although only marginally so (at the 10 % level using a one-tail test). This should not distract us from the plausible effects found for education in the previous two exercises, in view of the `modest' correlation between `literacy' and the `education' variables reported above. Also, the demand variable is now positively significant. Model (3) further omits the trade orientation variable BMPD , which has the right sign but is insignificant in model (2). This leads to no further changes in the results found in model (2). Finally, model (4) drops both LIT and BMPD . We find that the internal funds variable, the demand variable, and the index of protection all continue to be significant. $^32$ (Even if we move from the `reference model' (4) to the more general model (2) by progressively adding the relevant variables, the same conclusions hold.) Model (4) has the lowest root mean squared error, although not the Schwarz statistic. The Hausman statistic in its F distribution form is $m(F)=1.145$ , with an associated $p$ -value of 0.339, again supporting the estimation of a random-effects specification.
+
+### 5.4 Intellectual property protection and technological change: exercise 4
+
+In this exercise we restrict our sample to the set of OECD 33 countries from amongst those listed in Table 1 , although we are now looking at the expanded period 1981 – 95. As we noted above, we are forced to use $LIT$ as the human capital variable because data on $EDU$ are not available for 1995. Further, we drop the political instability dummy variable $ID$ from this set of regression exercises, because it equals 0 for all the OECD countries so that its inclusion would lead to the dummy variable trap. Table 5 reports the random effects GLS estimates based on this sample of 18 countries or 54 observations. Continuing to exclude variable $RLR$ , model (1) results based on the equation
+
+$$RDI_{t}=f(S_{t-1},\Delta GDPPC_{t},LIT_{t},BMPD_{t},IP_{t})\quad (5)$$
+
+reveal lagged savings to have a positive influence on R&D activities (that is significant using a one-tail test). Literacy levels appear to exert a strong positive effect on R&D, providing support to our results above. The protection of intellectual property rights appears to be strongly associated with R&D, as in our previous exercises.
+
+Both the demand variable $\Delta GDPPC$ as well as the trade orientation variable $BMPD$ , however, have the wrong signs, although they are insignificant. Most of the OECD countries are relatively open economies, so that there isn't much variation in this variable and one would expect it to be insignificant. Model (2), therefore, drops the former variable, while model (3) drops the latter. The results we noted in the previous paragraph continue to hold, and are unaffected by the sequence in
+
+$^{32}$ The ( $F$-distribution variant of the) $LM$-statistic for the null hypothesis $H_0$ : $S_{t-1} = 0$ is $LMF = 16.268$ ; for the null hypothesis $H_0$ : $\Delta GDPPC = 0$ , $LMF = 30.280$ ; and for the null hypothesis $H_0$ : $IP = 0$ , $LMF = 22.930$ . In all three cases, the $LM$ statistics comfortably exceed the critical value $F(1, 60) \approx 7.080$ at the 1 % level.
+
+$^{33}$ See note 24 above.
+
+---
+
+254 INTELLECTUAL PROPERTY PROTECTION
+
+Table 5 Exercise 4-random effects estimates; dependent variable: RDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td></tr><tr><td> $S_{1-1}$ </td><td>0.347 (1.557)</td><td>0.334 (1.517)</td><td>0.355 (1.629)</td><td>0.348 (1.606)</td></tr><tr><td> $\triangle GDPPC$ </td><td>−0.185 (−0.589)</td><td></td><td>−0.223 (−0.733)</td><td></td></tr><tr><td>LIT</td><td>6.102 (2.868)</td><td>6.038 (2.868)</td><td>5.979 (2.891)</td><td>5.852 (2.851)</td></tr><tr><td>IP</td><td>1.015 (3.965)</td><td>1.030 (4.075)</td><td>1.008 (3.985)</td><td>1.025 (4.088)</td></tr><tr><td>BMPD</td><td>0.031 (0.400)</td><td>0.043 (0.564)</td><td></td><td></td></tr><tr><td>Intercept</td><td>−29.982 (−3.148)</td><td>−29.693 (−3.151)</td><td>−29.427 (−3.180)</td><td>−28.874 (−3.143)</td></tr><tr><td>RMSE</td><td>95.444</td><td>93.351</td><td>90.111</td><td>88.723</td></tr><tr><td>SC</td><td>0.027</td><td>0.025</td><td>0.026</td><td>0.024</td></tr><tr><td> $F(H_{0}: all slopes 0)$ </td><td>12.326 0.464</td><td>15.003 0.472</td><td>15.353 0.476</td><td>19.203 0.481</td></tr><tr><td>Observations</td><td>54</td><td>54</td><td>54</td><td>54</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+which these variables are dropped. Finally, model (4) drops both these variables. We find that the internal funds variable, the human capital variable, and the index of protection all continue to be significant. $^34$ (Even if we move from the `reference model' (4) to the more general model (1) by progressively adding the relevant variables, the same conclusions hold.) Model (4) has the lowest root mean squared error as well as the lowest Schwarz statistic. The $F$ -variant of the Hausman statistic is $m(F) = 0.497$ , with a $p$ -value of 0.686, supporting the estimation of a randomeffects specification.
+
+## 6. Exploring possible nonlinearities
+
+A (small) part of the literature reviewed earlier (Bessen and Maskin, 2000; Takalo and Kanniainen, 2000; Helpman, 1993), pointed to the possibility of a (negative) nonlinear relationship between the rate of innovation and intellectual property protection. On various grounds, this literature suggested that ‘too strong’ a level of protection may lead to a decline in the rate of innovation. We test for such nonlinearities by introducing a quadratic transform of IP (the index of protection)
+
+$^{34}$ The $F$-variant of the $LM$-statistic for the null hypothesis $H_{0}: S_{t-1}=0$ is $LMF=24.665$; for the null hypothesis $H_{0}: LIT=0, LMF=21.300$; and for the null hypothesis $H_{0}: IP=0, LMF=17.490$. In all three cases, the $LM$ statistics comfortably exceed the critical value $F(1,50) \approx 7.425$ at the $1 \%$ level.
+
+---
+
+S. KANWAR AND R. EVENSON 255
+
+into the estimation model, and redo each of the exercises 1 to 4 above. The quadratic transform is defined as $IP2 = (IP-IP)^2 + 5$ , where $IP$ is the average value of $IP$ , so that $IP2$ measures `large' deviations in $IP$ from its mean value. 35 Given below are the `preferred' estimated equations for each of the four exercises; to avoid repetition, we do not present the complete `general-to-specific' routine for the exercises.
+
+$$\begin{array}{cccc} Exercise 1a:  R D I=-5.260+0.571 S_{t-1}+0.885 E D U-0.215 B M P D+1.316 I P+0.431 I P 2 \\ (-5.237) & (2.062) & (2.111) & (-1.356) & (2.660) & (2.003)\end{array}$$
+
+$$\begin{array}{r}
+ Exercise 2a:  R D I=-3.070+0.303 S_{t-1}-0.393 B M P D+1.723 I P+0.442 I P 2 \\
+(-4.046) \quad (1.696) \quad (-1.857) \quad (4.521) \quad (2.850)
+\end{array}$$
+
+$$\begin{array}{r}
+ Exercise 3a:  R D I=-4.343+0.895 S_{r-1}+0.418 \Delta G D P P C+1.013 I P+0.011 I P 2 \\
+(-5.665) \quad (4.196) \quad (1.799) \quad (3.695) \quad (1.675)
+\end{array}$$
+
+$$\begin{array}{rrrr}
+ Exercise 4a:  R D I=-31.856+0.437 S_{1-1}+6.402 L I T+1.164 I P+0.167 I P 2 \\
+(-3.187) & (1.951) & (2.909) & (4.172) & (1.077)
+\end{array}$$
+
+where exercise 1a corresponds to exercise 1 above etc., and the figures in parentheses are $t$-statistics.
+
+Interestingly, the overall results (even for the models not reported above to avoid repetition), are more or less in consonance with those discussed in the previous section, and need not be discussed here. In addition, we now find that variable IP2 is positively related to R & D expenditure, and significantly so in exercises 1a, 2a, and 3a, although only marginally in the lattermost. This evidence appears to be at odds with the claims made in some of the theoretical studies cited above in this section. It may, however, be reconciled with Bessen and Maskin's (2000) claim about a negative relationship between innovation and protection, by allowing for the fact that their study focuses on a specific type of innovations — those that are sequential and complementary — ignoring others. Takalo and Kanniainen (2000) themselves observe that `...while the effect of patents is to raise the rents on and thereby the potential amount of innovations, it also tends to slow down market introduction' insofar as it enhances the ability of the innovator to wait (our italics). As Park (1997) demonstrates, however, they miss the point that whether in fact the innovator finds it worthwhile to wait, really depends upon the relative profitability of the new technology compared to that in use. If indeed innovators stand to gain from the innovations (compared to their status quo profits), which is what Takalo and Kanniainen point out may happen, then they will innovate. As for Helpman (1993), a problem with the `North-South models' that such studies have used is, that they assume that innovation occurs only in the North. The point behind our study, however, is that (if an R & D sector exists in LDCs) then stronger protection will encourage innovation. In other words, Helpman does not really address the issue we attempt to — his model would have to be modified to allow innovation in
+
+$^{35}$ We could not use $I P^{2}$ itself on account of its high multicollinearity with variable $I P$. Further, the choice of scalar used to define $I P^{2}$ ( 5 in this case) does not appear to affect the results.
+
+---
+
+256 INTELLECTUAL PROPERTY PROTECTION
+
+the South. Further, even for the North he only claims that with stronger protection the rate of innovation first rises and then declines, but he does not show that the rate of innovation declines to a level lower than that from which it rose. Finally, the empirical analysis accounts for ‘sleeping patents’ and other abuses of patent rights implicitly, insofar as a positive relationship between R&D and protection may be the result of the incentive effects of protection outweighing the disincentive effects; for the latter need not be the result of ‘overstrong’ protection only, and could just as well follow from relatively weaker protection.
+
+Although our results seem to support the notion that the relationship between R & D and protection is probably not backward-bending, we are not fully confident that the relationship between R & D and the quadratic term IP2 is, on the other hand, positive. For while the latter is suggested by the results of exercises 1a and 2a, it is not quite supported by those of exercises 3a and 4a. Consequently, we have reservations about whatever insights the results of this section provide about the nonlinear aspect of the relationship between R & D expenditure and intellectual property protection. In other words, for the moment we repose greater confidence in the results that we obtained in the previous section.
+
+Table 6 Exercise 1b-random effects estimates; dependent variable: PRIVRDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td></tr><tr><td> $S_{t-1}$ </td><td>0.206 (2.135)</td><td>0.219 (2.269)</td><td>0.212 (2.242)</td><td>0.210 (2.274)</td></tr><tr><td> $\triangle GDPPC$ </td><td>0.150 (1.946)</td><td>0.168 (2.136)</td><td>0.170 (2.195)</td><td>0.169 (2.215)</td></tr><tr><td>EDU</td><td>0.458 (3.056)</td><td>0.415 (2.833)</td><td>0.405 (2.814)</td><td>0.403 (2.838)</td></tr><tr><td>IP</td><td>0.300 (3.189)</td><td>0.293 (3.141)</td><td>0.282 (3.161)</td><td>0.280 (3.232)</td></tr><tr><td>BMPD</td><td>0.019 (0.336)</td><td>0.009 (0.151)</td><td>0.007 (0.121)</td><td></td></tr><tr><td>ID</td><td>0.093 (0.587)</td><td>0.070 (0.442)</td><td></td><td></td></tr><tr><td>RLR</td><td>0.058 (1.401)</td><td></td><td></td><td></td></tr><tr><td>Intercept</td><td>−1.625 (−4.186)</td><td>−1.349 (−4.055)</td><td>−1.291 (−4.241)</td><td>−1.279 (−4.485)</td></tr><tr><td>RMSE</td><td>0.227</td><td>0.184</td><td>0.139</td><td>0.123</td></tr><tr><td>SC</td><td>0.007</td><td>0.007</td><td>0.0065</td><td>0.0061</td></tr><tr><td> $F(H_{0}: all slopes 0)$ </td><td>26.023 0.5152</td><td>30.110 0.5149</td><td>35.421 0.521</td><td>43.278 0.530</td></tr><tr><td>Observations</td><td>58</td><td>58</td><td>58</td><td>58</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+---
+
+S. KANWAR AND R. EVENSON 257
+
+## 7. Purging the effect of public sector investment in R&D
+
+A shortcoming of the R & D data used above is that it includes public expenditures on R & D. It may be argued that the public sector is not necessarily responsive to monetary incentives in the way the private sector is, and that it is unconcerned with the appropriability of public investments in R & D. Therefore, a more pertinent regressand would be some proxy of private sector R & D expenditure. Using cross-country data on the government's share of gross domestic R & D expenditure, we compute the ratio of private (or non-public) R & D expenditure to GNP ( PRIVRDI ), and redo all the four exercises once again. The estimation results (Exercises 1b to 4b) are presented in Tables 6 to 9 .
+
+The empirical results are in consonance with those derived above (using RDI as the regressand), and therefore need not be analysed in equal detail. It bears mentioning, however, that these results (using PRIVRDI ) as the regressand) are `stronger' than those obtained earlier. For one, the model fit is a whole lot better in each of the exercises 1b to 4b. Second, the demand side variable $\Delta GDPPC$ is now significant in the first three exercises. Third, the human capital variable is strongly and consistently significant in all the exercises. And finally, the index of protection variable is also more strongly significant in all the exercises. Thus, purging the R & D
+
+Table 7 Exercise 2b-Random effects estimates; Dependent variable: PRIVRDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td></tr><tr><td> $S_{1-1}$ </td><td>0.061 (0.845)</td><td>0.075 (1.016)</td><td></td><td>0.059 (0.800)</td></tr><tr><td> $\triangle GDPPC$ </td><td>0.099 (1.394)</td><td>0.116 (1.586)</td><td>0.101 (1.428)</td><td></td></tr><tr><td>EDU</td><td>0.246 (2.355)</td><td>0.252 (2.471)</td><td>0.283 2.987</td><td>0.271 (2.617)</td></tr><tr><td>IP</td><td>0.340 (3.582)</td><td>0.320 (3.652)</td><td>0.312 (3.547)</td><td>0.316 (3.602)</td></tr><tr><td>BMPD</td><td>−0.180 (−2.123)</td><td>−0.143 (−1.825)</td><td>−0.155 (−1.987)</td><td>−0.149 (−1.884)</td></tr><tr><td>ID</td><td>0.091 (0.528)</td><td></td><td></td><td></td></tr><tr><td>Intercept</td><td>−0.567 (−2.073)</td><td>−0.608 (−2.322)</td><td>−0.393 (−2.293)</td><td>−0.570 (−2.177)</td></tr><tr><td>RMSE</td><td>0.142</td><td>0.104</td><td>0.057</td><td>0.099</td></tr><tr><td>SC</td><td>0.0063</td><td>0.0066</td><td>0.0061</td><td>0.0067</td></tr><tr><td> $F(H_{0}: all slopes 0)$ </td><td>20.125 0.412</td><td>26.743 0.439</td><td>31.021 0.434</td><td>31.695 0.430</td></tr><tr><td>Observations</td><td>62</td><td>62</td><td>62</td><td>62</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+---
+
+258 INTELLECTUAL PROPERTY PROTECTION
+
+Table 8 Exercise 3b—Random effects estimates; Dependent variable: PRIVRDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td></tr><tr><td> $S_{t-1}$ </td><td>0.126 (1.933)</td><td>0.103 (1.637)</td></tr><tr><td> $\triangle GDPPC$ </td><td>0.139 (2.070)</td><td>0.120 (1.851)</td></tr><tr><td>LIT</td><td>0.040 (4.570)</td><td>0.042 (4.836)</td></tr><tr><td>IP</td><td>0.365 (4.034)</td><td>0.396 (4.490)</td></tr><tr><td>BMPD</td><td>−0.036 (−1.524)</td><td>−0.036 (−1.530)</td></tr><tr><td>ID</td><td>−0.126 (−1.132)</td><td></td></tr><tr><td>Intercept</td><td>−0.552 (−2.350)</td><td>−0.521 (−2.213)</td></tr><tr><td>RMSE</td><td>0.085</td><td>0.072</td></tr><tr><td>SC</td><td>0.007</td><td>0.006</td></tr><tr><td> $F(H_{0}$ : all slopes 0)</td><td>22.012</td><td>24.757</td></tr><tr><td> $\overline{R}^{2}$ </td><td>0.534</td><td>0.529</td></tr><tr><td>Observations</td><td>64</td><td>64</td></tr></table>
+
+
+Note: All variables are in (natural) logs; t-statistics are in parentheses.
+
+variable of its public expenditure component, in fact, strengthens the empirical results we obtained above.
+
+## 8. Conclusions
+
+This paper set out to establish an empirical relationship between the protection of intellectual property rights and technological change. We found evidence to support the claim that the former encourages the latter insofar as intellectual property protection was found to have a strong positive influence on R&D investment. This relationship continued to hold even when several pertinent control variables were allowed for. 36 Results may have been even more pronounced
+
+36 Note that this paper did not consider the question of which factor or combination of factors may be relatively desirable for achieving desired results. The fact that some variable (say, education) may be more `potent' (higher associated $t$ -statistic?) than some other variable (say, protection), does not allow us to draw conclusions about what is `best for the country', for a combination of factors may have to be employed anyway. Thus, education is a very important enabling factor. But while the availability of adequate skilled technical manpower enables R&D, this by itself may not be sufficient in motivating R&D. Such motivation may be expected from the provision of adequate profit opportunities, which are partly determined by adequate protection levels. Moreover, the relative potency of factors is not quite the
+
+---
+
+S. KANWAR AND R. EVENSON 259
+
+Table 9 Exercise 4b-random effects estimates; dependent variable: PRIVRDI
+
+<table><tr><td>Variable</td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td><td>(5)</td></tr><tr><td> $S_{1-1}$ </td><td>0.530 (1.304)</td><td>0.460 (1.162)</td><td>0.498 (1.265)</td><td>0.441 (1.144)</td><td></td></tr><tr><td> $\Delta GDPPC$ </td><td>−0.780 (−0.810)</td><td></td><td>−0.769 (−0.826)</td><td></td><td></td></tr><tr><td>LIT</td><td>10.363 (3.335)</td><td>10.461 (3.462)</td><td>10.147 (3.397)</td><td>10.228 (3.497)</td><td>11.241 (4.421)</td></tr><tr><td>IP</td><td>1.801 (4.048)</td><td>1.811 (4.156)</td><td>1.778 (4.212)</td><td>1.772 (4.266)</td><td>1.651 (4.171)</td></tr><tr><td>BMPD</td><td>0.001 (0.001)</td><td>0.043 (0.216)</td><td></td><td></td><td></td></tr><tr><td>Intercept</td><td>−52.272 (−3.864)</td><td>−52.565 (−4.000)</td><td>−51.102 (−3.938)</td><td>−51.335 (−4.043)</td><td>��54.369 (−4.747)</td></tr><tr><td>RMSE</td><td>193.983</td><td>182.180</td><td>178.540</td><td>170.087</td><td>137.783</td></tr><tr><td>SC</td><td>0.201</td><td>0.192</td><td>0.190</td><td>0.180</td><td>0.175</td></tr><tr><td> $F(H_{0}: all slopes 0)$ </td><td>25.603</td><td>29.940</td><td>30.254</td><td>37.022</td><td>46.190</td></tr><tr><td> $\overline{R}^{2}$ </td><td>0.548</td><td>0.552</td><td>0.558</td><td>0.562</td><td>0.563</td></tr><tr><td>Observations</td><td>54</td><td>54</td><td>54</td><td>54</td><td>54</td></tr></table>
+
+
+Note. All variables are in (natural) logs; t-statistics are in parentheses.
+
+if we had reliable, quantifiable evidence on the implementation aspect of intellectual property protection across countries, for many developing countries may appear to have strong protection laws on their statutes but are rather remiss in their implementation. 37 Further, there appears to be some basis for the claim that the relationship between R&D and protection may not be backward-bending; we would, however, prefer to hold this claim in abeyance till more data become available. Our results imply that the lack of an incentive structure can be a significant mitigating factor for technological change even when other constraints such as internal funds, availability of skills and trade orientation may not be binding.
+
+last word as regards policy. One could go on to conjecture about the ease with which specific factors may be manipulated to achieve desired ends. Although it may be relatively easier and faster to tighten protection laws (which is not to imply that this process is likely to be trivial), education levels may be much more difficult to change in developing countries (even over fairly long spans of time).
+
+Again, while it would no doubt be interesting to study ‘…which combination of factors would help stimulate technological change’, the specific combination of factors would, in general, vary depending on the sample of countries and time periods available. In any case, one would still not be able to comment upon the combination of factors that may be relevant in each specific case.
+
+37 To some extent, perhaps, this factor is counter-balanced by the fact that formal R&D (i.e. R&D explicitly undertaken as such) underestimates R&D relatively more in poor countries, where relatively more firms engage in informal inventive activity including `blue-collar' R&D (see Evenson, 1990; and the references cited therein).
+
+---
+
+260 INTELLECTUAL PROPERTY PROTECTION
+
+## Acknowledgements
+
+We are most indebted to T. Paul Schultz and Doug Gollin for their time and trouble. We would also like to express our gratitude to Peter Berck, Jim Chalfant, Jenny Lanjouw, William Nordhaus, Peter Phillips, Gus Ranis, Robert Solow, Stefan Klasen, the participants of the Development Economics discussion group at Yale University, seminar participants at the University of Munich and the IEA 13th World Congress (Lisbon), and the two anonymous referees of this journal. This research was conducted while the first author was a Fulbright Visiting Fellow at the Department of Economics at Yale University, and an earlier version is presented in the working paper Kanwar and Evenson (2001).
+
+## References
+
+Barro, R. (1991). 'Economic growth in a cross section of countries', Quarterly Journal of Economics, 106, 407-43.
+
+Barro, R. and Lee, J. (2000). 'Barro-Lee data set: international measures of schooling years and schooling quality', NBER website 'www.nber.org', National Bureau of Economic Research, Cambridge, MA.
+
+Bessen, J. and Maskin, E. (2000). 'Sequential innovation, patents, and imitation', Working Paper No. 00-01, Department of Economics, MIT, Cambridge, MA.
+
+Bhagwati, J. (1978). Anatomy and Consequences of Exchange Control Regimes, Ballinger, Cambridge, MA.
+
+Charemza, W. and Deadman, D. (1997). New Directions in Econometric Practice, Edward Elgar, Cheltenham.
+
+Chin, J. and Grossman, G. (1990). 'Intellectual property rights and North-South trade', in R.W. Jones and A.O. Krueger (eds) The Political Economy of International Trade: Essays in Honor of Robert E. Baldwin, Basil Blackwell, Cambridge.
+
+Coe, D. and Helpman, E. (1995). 'International R&D spillovers', European Economic Review, 39, 859-87.
+
+Christensen, L. and Cummings, D. (1981). 'Real product, real factor input and productivity in the republic of Korea, 1960-1973', Journal of Development Economics, 8, 285-302.
+
+Deardorff, A. (1992). 'Welfare effects of global patent protection', Economica, 59, 35-52.
+
+Diwan, I. and Rodrik, D. (1991). 'Patents, appropriate technology and North-South trade', Journal of International Economics, 29, 27-48.
+
+Edwards, S. (1992). 'Trade orientation, distortions and growth in developing countries', Journal of Development Economics, 39, 31-58.
+
+Engelbrecht, H. (1997). 'International R&D spillovers, human capital and productivity in OECD economies: an empirical investigation', European Economic Review, 41, 1479-88.
+
+---
+
+S. KANWAR AND R. EVENSON 261
+
+Esty, D., Goldstone, J., Gurr, T., Harff, B., Surko, P., Unger, A. and Chen, R. (1998). ‘The state failure project: early warning research for US foreign policy planning’, in J.L. Davies and T.R. Gurr (eds) Preventive Measures Building Risk Assessment and Crisis Early Warning Systems, Rowman and Littlefield, Boulder, CO.
+
+Evenson, R. (1990). 'Survey of empirical studies', in W.E. Siebeck (ed.) 'Strengthening Protection of Intellectual Property in Developing Countries: A Survey of the Literature', World Bank Discussion Paper 112, The World Bank, Washington, DC.
+
+Frantzen, D. (2000). 'R&D, human capital and international technology spillovers: a crosscountry analysis', Scandinavian Journal of Economics, 102, 57-75.
+
+Gadbaw, R. and Richards, T. (1988). Intellectual Property Rights: Global Consensus, Global Conflict?, Westview Press, Boulder, CO.
+
+Geroski, P. and Walters, C. (1995). 'Innovative activity over the business cycle', Economic Journal, 105, 916-28.
+
+Gilbert, R. and Newbery, D. (1982). 'Preemptive patenting and the persistence of monopoly', American Economic Review, 72, 514-26.
+
+Ginarte, J. and Park, W. (1997). 'Determinants of patent rights: a cross-national study', Research Policy, 26, 283-301.
+
+Gould, D. and Gruben, W. (1996). 'The role of intellectual property rights in economic growth', Journal of Development Economics, 48, 323-50.
+
+Griliches, Z. (1990). 'Patent statistics as economic indicators: a survey', Journal of Economic Literature, 28, 1661-707.
+
+Grossman, G. and Helpman, E. (1991). Innovation and Growth in the Global Economy, MIT Press, Cambridge, MA.
+
+Guellec, D. and Ioannidis, E. (1997). 'Causes of fluctuations in R&D expenditures: a quantitative analysis', OECD Economic Studies, 29, 123-39.
+
+Hall, B. (1992). 'Investment and R&D at the firm level: does the source of financing matter?', Working Paper No. 4096, National Bureau of Economic Research, Cambridge, MA.
+
+Helpman, E. (1993). 'Innovation, imitation and intellectual property rights, Econometrica, 61, 1247-80.
+
+Heston, A., Summers, R., Nuxoll, D. and Aten, B. (2001). 'Penn World Table Mark 5.6a', National Bureau of Economic Research, Cambridge, MA.
+
+Himmelberg, C. and Petersen, B. (1994). 'R&D and internal finance: a panel study of small firms in high-tech industries', Review of Economics and Statistics, 76, 38-51.
+
+Islam, N. (1995). 'Growth empirics: a panel data approach', Quarterly Journal of Economics, 110, 1127-70.
+
+Kamien, M. and Schwartz, N. (1982). Market Structure and Innovation, Cambridge University Press, Cambridge.
+
+Kanwar, S. and Evenson, R. (2001). 'Does intellectual property protection spur technological change?', Economic Growth Center Discussion Paper No. 831, Department of Economics, Yale University.
+
+Kennedy, C. and Thirlwall, A. (1973). 'Technical Progress', in Surveys of Applied Economics, vol. 1, The Royal Economic Society, Social Science Research Council, Macmillan, London.
+
+---
+
+262 INTELLECTUAL PROPERTY PROTECTION
+
+Kortum, S. (1997). 'Research, patenting and technological change', Econometrica, 65, 13891419.
+
+Krueger, A. (1978). Liberalization Attempts and Consequences, Ballinger, Cambridge, MA.
+
+Lai, E. (1998). 'International intellectual property rights protection and the rate of product innovation', Journal of Development Economics, 55, 133-53.
+
+Lanjouw, J. and Cockburn, I. (2001). 'New pills for poor people?: empirical evidence after GATT', World Development, 29, 265-89.
+
+Lerner, J. (1994). 'The importance of patent scope: an empirical analysis', Rand Journal of Economics, 25, 319-33.
+
+Lesser, W. (2000). ‘An economic approach to identifying an “effective sui generis system” for plant variety protection under TRIPs', in V. Santaniello, R. Evenson, D. Zilberman, and G. Carlson (eds) Agriculture and Intellectual Property Rights, CABI publishing, Wallingford.
+
+Levin, R., Klevorick, A., Nelson, R. and Winter, S. (1987), 'Appropriating the returns from industrial research and development', Brookings Papers on Economic Activity, 3, Brookings Institution, Washington, DC.
+
+Lewis, W. (1955) The Theory of Economic Growth, Allen and Unwin, London.
+
+Maddala, G. (1971). 'The use of variance components models in pooling cross section and time series data', Econometrica, 39, 341-58.
+
+Mandich, G. (1948). 'Venetian patents (1450-1550)', Journal of the Patent Office Society, 30, 166-224.
+
+Mansfield, E. (1985). 'How rapidly does industrial technology leak out?', Journal of Industrial Economics, 34, 217-23.
+
+Mansfield, E. (1986). 'Patents and innovation: an empirical study', Management Science, 32, 173-81.
+
+Nadiri, M. (1980). 'Contributions and determinants of research and development expenditures in the US manufacturing industries', in G.M. von Furstenberg (ed.) Capital, Efficiency and Growth, Ballinger, Cambridge, MA.
+
+Nelson, R. and Phelps, E. (1966). 'Investment in humans, technological diffusion, and economic growth', American Economic Review, 56, 59-75.
+
+Nerlove, M. (1967). 'Experimental evidence on the estimation of dynamic economic relations from a time series of cross sections', Economic Studies Quarterly, 18, 42-74.
+
+Nerlove, M. (1971). 'Further evidence on the estimation of dynamic economic relations from a time series of cross sections', Econometrica, 39, 359-87.
+
+Nickell, S. (1981). 'Biases in dynamic models with fixed effects', Econometrica, 49, 1417-26.
+
+Nishimizu, M. and Robinson, S. (1984). 'Trade policies and productivity change in semiindustrialized countries', Journal of Development Economics, 16, 177-206.
+
+Norsworthy, J. and Malmquist, D. (1983). 'Input measurement and productivity growth in Japanese and US manufacturing', American Economic Review, 73, 947-67.
+
+Park, W. (1997). 'A note on innovation and patent protection: intertemporal imitation-risk smoothing', Economics Letters, 54, 185-9.
+
+Park, W. and Ginarte, J. (1997). 'Intellectual property rights and economic growth', Contemporary Economic Policy, 15, 51-61.
+
+---
+
+S. KANWAR AND R. EVENSON 263
+
+Pavitt, K. (1982). 'R&D, patenting and innovative activities', Research Policy, 11, 33-51.
+
+Pick's Currency Yearbook, various years, Pick Publishing Corporation, New York.
+
+Rapp, R. and Rozek, R. (1990) 'Benefits and costs of intellectual property protection in developing countries', Journal of World Trade, 24, 74-102.
+
+Roffe, P. (1974), 'Abuses of patent monopoly: a legal appraisal', World Development, 2, 15-26.
+
+Romer, P. (1990) 'Endogenous technological change', Journal of Political Economy, 98, S71-102.
+
+Sakakibara, M. and Branstetter, L. (2001). 'Do stronger patents induce more innovation? Evidence from the 1988 Japanese patent law reforms', RAND Journal of Economics, 32, 77-100.
+
+Schankerman, M. (1998). 'How valuable is patent protection? Estimation by technology field', RAND Journal of Economics, 29, 77-107.
+
+Scherer, F. and Ross, D. (1990). Industrial Market Structure and Economic Performance, Houghton Mifflin, Boston.
+
+Schiff, E. (1971). Industrialization without National Patents: The Netherlands, 1869-1912; Switzerland, 1850-1907, Princeton University Press, Princeton.
+
+Segerstrom, P. (1991). 'Innovation, imitation and economic growth', Journal of Political Economy, 99, 807-27.
+
+Segerstron, P., Anant, T. and Dinopoulos, E. (1990). 'A Schumpeterian model of the product life cycle', American Economic Review, 80, 1077-91.
+
+Sherwood, R. (1990). Intellectual Property and Economic Development, Westview Press, Boulder, CO.
+
+South Centre (1997). The TRIPs Agreement: A Guide for the South, South Centre, Geneva.
+
+Takalo, T. and Kanniainen, V. (2000). 'Do patents slow down technological progress? Real options in research, patenting and market introduction', International Journal of Industrial Organization, 18, 1105-27.
+
+Taylor, M. (1993). 'TRIPs, trade and technology transfer', Canadian Journal of Economics, 26, 625-37.
+
+Taylor, M. (1994). 'TRIPS, trade and growth', International Economic Review, 35, 361-81.
+
+Taylor, W. (1980). 'Small sample considerations in estimation from panel data', Journal of Econometrics, 13, 203-23.
+
+Trajtenberg, M. (1990). 'A penny for your quotes: patent citations and the value of innovations', Rand Journal of Economics, 21, 172-87.
+
+United Nations (1999). UNESCO Yearbook 1999, United Nations, New York.
+
+Van Ark, B., Kuipers, S. and Kuper, G. (eds) (2000). Productivity, Technology and Economic Growth, Kluwer Academic, Boston.
+
+World Bank (1987). World Development Report 1987, Oxford University Press, New York.
+
+World Bank (2000). World Development Indicators 2000, The World Bank, Washington DC.
+
+World Currency Yearbook, various years, International Currency Analysis Inc., New York.
+
+Yuhn, K. and Kwon, J. (2000). 'Economic growth and productivity: a case study of South Korea', Applied Economics, 32, 13-23.
+
+---
+
+264 INTELLECTUAL PROPERTY PROTECTION
+
+## Appendix 1
+
+Table A1 Means and standard deviations of the variables, 1981-90
+
+<table><tr><td>Variable</td><td>mean</td><td>standard deviation</td></tr><tr><td>RDI</td><td>1.205</td><td>0.807</td></tr><tr><td>PRIVRDI</td><td>0.589</td><td>0.581</td></tr><tr><td> $S_{t-1}$ </td><td>20.701</td><td>8.762</td></tr><tr><td> $\Delta GDPPC$ </td><td>1.112</td><td>0.126</td></tr><tr><td>EDU</td><td>6.989</td><td>2.278</td></tr><tr><td>LIT</td><td>87.973</td><td>17.757</td></tr><tr><td>IP</td><td>2.940</td><td>0.977</td></tr><tr><td>BMPD</td><td>0.500</td><td>0.504</td></tr><tr><td>ID</td><td>0.159</td><td>0.367</td></tr><tr><td>RLR</td><td>3.953</td><td>7.966</td></tr></table>
+
+
+## Appendix 2
+
+### 1. Definitions of variables
+
+RDI: Research and development expenditure as a percentage share of GNP (%) $S_{t-1}$: Real savings share of GDP (at 1985 international prices), lagged one period (%) $\Delta$ GDPPC: Real GDP per capita (chain index in 1985 international prices) as a proportion of the previous period real GDP per capita (US $$) EDU: Average number of schooling years in population over 15 (years) LIT: Total literacy rate in population over 15 (%) IP: Index of patent protection BMPD: Black market exchange rate premium dummy ID: Political instability dummy RLR: Real lending rate of interest (%) GOVSHRD: Government share of gross domestic R&D expenditure (%)—used to compute PRIVRDI (see below) PRIVRDI: Private (or non-public) R&D expenditure as a proportion of GNP (%)
+
+## Appendix 3
+
+### Data sources
+
+RDI: World Bank (2000) $S_{t-1}$ : Heston et al. (2001) $\Delta$ GDPPC: Heston et al. (2001), and World Bank (2000) (for `Exercise 4' regressions). EDU: Barro and Lee (2000) LIT: World Bank (2000) IP: Ginarte and Park (1997) BMPD: Pick's Currency Yearbook and World Currency Yearbook (various years) ID: Esty et al. (1998) RLR: World Bank (2000) GOVSHRD: United Nations (1999)
+

processed/00022_W2238468911/document.md

@@ -0,0 +1,597 @@
+# Patent Citation Data in Social Science Research: Overview and Best Practices
+
+Adam B. Jaffe
+
+Motu Economic and Public Policy Research, Wellington 6011 New Zealand; Queensland University of Technology, and Te Pūnaha Matatini Centre of Research Excellence.
+
+E-mail: adam.jaffe@motu.org.nz
+
+Gaétan de Rassenfosse
+
+Ecole polytechnique fédérale de Lausanne, College of Management of Technology, CH-1015 Lausanne, Switzerland. E-mail: gaetan.derassenfosse@epfl.ch
+
+The last 2 decades have witnessed a dramatic increase in the use of patent citation data in social science research. Facilitated by digitization of the patent data and increasing computing power, a community of practice has grown up that has developed methods for using these data to: measure attributes of innovations such as impact and originality; to trace flows of knowledge across individuals, institutions and regions; and to map innovation networks. The objective of this article is threefold. First, it takes stock of these main uses. Second, it discusses 4 pitfalls associated with patent citation data, related to office, time and technology, examiner, and strategic effects. Third, it highlights gaps in our understanding and offers directions for future research.
+
+“ Knowledge flows [...] are invisible; they leave no paper trail by which they may be measured and tracked, and there is nothing to prevent the theorist from assuming anything about them that she likes. ”
+
+Paul Krugman (1991)
+
+## Introduction
+
+Eugene Garfield is one of the pioneers of the study of citation data. In his 1955 article, Garfield proposes to build a citation index for scientific articles in order to make it possi-
+
+ble for “the conscientious scholar to be aware of criticisms of earlier articles.” He further explains, “even if there were no other use for a citation index than that of minimizing the citation of poor data, the index would be well worth the effort required to compile it” (p. 108). It turns out that citation indices have been used in a variety of ways and for a variety of purposes. Two of the most notable uses are to assess the attributes of the idea embedded in a scientific article and to track its diffusion through time, space and technology domains. In fact, Garfield (1955) foresaw these two uses as he described the citation index as an “association-ofideas index” (p. 108) and as he explained that the citation index may “help the historian to measure the influence of the article—that is, its ‘impact factor’” (p. 111).
+
+Although the analogy with the broader field of bibliometrics may seem obvious, patent citations differ from scientific citations in substantial ways. Citations in patents are the results of a highly mediated process that involves multiple parties: the inventor, the patent attorney, and the patent examiner (Meyer, 2000) . These parties have different incentives for citing publications and may do so at different times and in different sections of the patent document (Cotropia, Lemley, & Sampat, 2013) . Much of the empirical research relies on U.S. citations, but there are important differences across jurisdictions in citation rules and practice. $^1$ This creates interesting opportunities for research on non-U.S. data, but also suggests a degree of caution in thinking about the global implications of results based solely on U.S. data.
+
+The widespread use of patent citations in social science research can be traced to the availability of patent statistics in digitally readable form in the late 1970s. 2 Zvi Griliches (1979), in his important manifesto for research on R&D and productivity growth, suggested that the frequency with
+
+Received August 14, 2015; revised January 4, 2016; accepted January 31, 2016
+
+© 2017 The Authors. Journal of the Association for Information Science and Technology published by Wiley Periodicals, Inc. on behalf of Association for Information Science and Technology • Published online 00 Month 2017 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/asi.23731
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY, 00(00):00-00, 2017
+
+This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is nonxcommercial and no modifications or adaptations are made.
+
+---
+
+which patents from different industries cite each other could be used as a measure of the technological proximity of industries. An early strand of research on patent citations was the work of Francis Narin and his associates at CHI Research, Inc. (Carpenter & Narin, 1983; Carpenter, Narin, & Woolf, 1981; Narin & Noma, 1985; Narin, Noma, & Perry, 1987) . An influential early demonstration of the potential utility of patent citation data in economic research was the PhD research of Griliches's student Manuel Trajtenberg (Trajtenberg, 1990a, 1990b) . The use of patent citation data has grown dramatically over the last two decades, as illustrated in Appendix A .
+
+What makes citations potentially useful is that they convey information about the cumulative nature of the research process, as well as information about the consequences. Although some inventors and research organizations pursue patents for motives of prestige or internal tracking of research success, most patent applications are made with the goal of securing commercial advantage, or at least preserving options for pursuit of commercial advantage. Another virtue of patent data for social science research is that patents reside in a nonmarket-based technological classification system, allowing one to place patents, inventors, and organizations in technology space in a way that is not derived from sales or other economic data that one may be trying to relate to invention. $^3$ Furthermore, the classification scheme is hierarchical so that technology categories can be very fine or relatively broad as desired. This feature, and others, has been combined with patent citation data to provide powerful indicators.
+
+This article provides an overview of the major uses of such data and the issues that arise in such research. Other authors have previously discussed the use of patent statistics in social science research (e.g., Griliches, 1990; Lerner & Seru, 2015) , and Gay and Le Bas (2005) provide a brief overview of the use of patent citations to measure invention value and knowledge flows. However, we are not aware of a broad survey on the use of patent citation data. $^4$ In order to identify the articles to include in this survey, we started from a limited number of references that we were aware of and complemented those using a keyword-based search on Google Scholar. We then expanded this core of references by looking at cited and citing references. Ultimately, we kept the most influential articles, either in terms of the number of citations received or in terms of relevance of the findings. The majority of articles are published in economics, management, and information science journals.
+
+Conceptually, we classify research using patent citations into two broad groups. One research line uses a variety of citation-based statistics to characterize the inventions, in terms of the magnitude and nature of their impact, as well as the nature and magnitude of the departure that they represent relative to the existing pool of knowledge. This work is discussed in the next section. The other research line focuses on the citations themselves, using them as proxies for knowledge linkages across inventors in order to explore the nature of knowledge flows and the factors that affect those
+
+flows. This research is discussed first with regard to relatively simple metrics of knowledge flow, and then with respect to attempts to map interactions in a more complex network framework. We then provide some brief comments on practical difficulties and pitfalls in using citation data. The last section concludes with opportunities for future research.
+
+## Citations as an Indicator of Invention Attributes
+
+There is no agreed-upon model of inventions and the inventive process, which leads to some ambiguity in how citation metrics are interpreted. Nonetheless it is possible to identify two broad aspects of the process that underlie citation-based inferences. First, we can think of all possible technologies as mapping onto a high-dimensional technology space, such that a given invention can be located in that space, and a patent represents the right to exclude others from marketing products that impinge upon a specified region (or regions) of that space. Second, the invention process is cumulative, that is, inventions build on those that came before and, in turn, facilitate those that come after. In this “ geometric ” interpretation, the patent claims delineate the metes and bounds of the region of technology space over which exclusivity is being granted, whereas the citations indicate previously marked-off areas that are in some sense built upon by or connected to the invention being granted.
+
+Thus the citations that appear in a patent (its “ backward ” citations) inform us about the technological antecedents of the patented invention. A patent that contains many citations corresponds to an invention with many antecedents; a patent whose citations are to technologically diverse previous patents has diverse antecedents; a patent whose citations are to old patents corresponds to an invention with old antecedents, and so forth. Conversely, the citations received by a patent from subsequent patents ( “ forward ” citations) inform us about the technological descendants of the patented invention. A patent that is never cited was a technological dead end. A patent with many or technologically diverse forward citations corresponds to an invention that was followed by many or technologically diverse descendants.
+
+Note that the discussion so far is entirely definitional. We have said nothing about the possibility of causal connections between these different attributes of inventions, or between any of these attributes and the private or social value of the invention. Ultimately, we are interested in whether, for example, patents with relatively few technological antecedents are more or less likely to spawn multiple lines of research or whether patents that generate many or diverse technological descendants correspond to inventions that generate large social benefits. It is in large part to be able to say something about these questions that citation metrics have been developed. In a very broad sense, citation analysis is predicated on an expectation that the extent and nature of an invention's antecedents tells us something about the novelty or “ radicalness ” of the invention, and the extent and nature of its descendants tell us something about both its
+
+2 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+technological impact and its economic value. But different authors propose or use different characterizations of citation information to elucidate these ideas.
+
+In practice, writers are not always clear on the underlying concept that a given metric is intended to measure, and given metrics are used in different contexts as proxies or indicators for different concepts. In some cases, researchers postulate a relationship between a given citation metric and an underlying concept, and then test hypotheses about the concept taking that relationship as a given. In other cases researchers attempt explicitly to validate the extent to which a given metric reflects a particular underlying conceptual attribute of inventions. We will consider these different approaches below in the context of specific articles, but for expositional purposes it is useful to consider five broad categories of approaches:
+
+- • Counts of forward citations as an indicator of subsequent
+technological impact;
+• Counts of backward citations as an indicator of the extent of
+reliance on previous technology;
+• Characterization of both backward and forward citations in
+terms of technological diversity and technological distance;
+• Examination of references to nonpatent literature as an indi-
+cator of science linkage; and
+• Use of citations as an indicator for private and social value.
+We consider each category in turn.
+
+## Forward Citations and Technological Impact
+
+Using the number of forward citations as a measure of technological impact of a patented invention can be motivated by direct analogy to the larger and pre-existing bibliometric literature starting with Garfield (1955) . Nonetheless, Trajtenberg, Henderson, and Jaffe (1997) undertook to demonstrate the validity of this (and other) metrics by comparing the citation rate to university patents and corporate patents, based on a maintained assumption that university patents are more “ basic ” and hence have, on average, greater technological impact. To incorporate the cumulative nature of invention into the metric, they proposed that the importance of an invention be characterized by the number of forward citations received, plus a fractional weight multiplied by the number of citations received by those citing patents. That is, important patents are those that are cited a lot, and are cited by patents that are themselves relatively highly cited. $^5$ The authors showed that importance by this definition is, indeed, higher for university patents than for corporate patents, using a sample of patents assigned to U.S. corporations, matched by patent class and grant date to patents assigned to U.S. universities. In addition, they discuss qualitatively the highest-importance patents in their sample, and argue that the citing patents can be seen as technological descendants, and these highly “ important ” patents are, indeed, subjectively very important in their respective fields.
+
+More recently, taking advantage of improvements in computing power, scholars have taken into account the
+
+whole stream of citations. For example, Lukach and Lukach (2007) have proposed computing importance by the PageRank score of patents. This method is directly inspired from Google's “ random surfer ” model and takes into account the fact that different citations weigh differently depending on the importance of the citing documents (Brin & Page, 1998). However, the authors are not able to validate their ranking using external measures such that the conditions under which the PageRank method is more appropriate than a straightforward citation count are unclear. This approach is a natural extension of earlier work, and begins to move this line of analysis towards the “ innovation network ” formulation discussed later in the text.
+
+Albert, Avery, Narin, and McAllister (1991) provide a validation study of the use of forward citations as an indicator of impact. They reported a strong correlation between the citation intensities of 77 Kodak silver halide patents and expert evaluations of technical impact and importance of the patents. Narin (1995) showed that patents that have attained the legal status of pioneering patents in the United States, as well as other prominent patents appearing in such patent office publications as “ Hall of Fame ” patents, are very highly cited. Czarnitzki, Hussinger, and Schneider (2011) relate a group of “ wacky ” patents to control groups and test the extent to which commonly used metrics are able to identify wacky patents from patents in the control group. Wacky patents are selected by an employee of the World Intellectual Property Organization “ for their futile nature, as they do not involve a high-inventive step or only marginally satisfy the `non obviousness' criterion ” (p. 131). They find that the number of forward citations is a good predictor of importance. However, other measures such as originality and generality (discussed below) were higher for wacky patents. Another interesting confirmation of patent citations as indicative of technological impact is Benson and Magee (2015). They identify 28 “ technological domains ” (e.g., “ Solar Photovoltaics ” or “ Genome Sequencing ” ) in which it is possible to identify a specific metric of the technological state of the domain (e.g., watts/ $ for Solar Photovoltaics). They take the exponential rate of improvement of these metrics across domains and across time as the dependent variable in regressions on various citation metrics of patents in the technology domain. They find that forward citations are positively related, and the average age of backward citations negatively related, to the rate of improvement of the technology over the subsequent 10-year period.
+
+## Backward Citations and Reliance on Previous Technology
+
+Although it seems clear that important inventions generate more forward citations, the opposite may hold for backward citations. That is, more trivial inventions are more extensively rooted in what has come before, whereas more basic inventions are less incremental in nature and thus have fewer identifiable antecedents (Trajtenberg et al., 1997) . Another way to think of this is that a patent will, to some
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017 DOI: 10.1002/asi
+
+---
+
+extent, tend to cite other patents all the way back along the inventive trajectory upon which it lies. Patents that are near the beginning of a trajectory are in this sense more basic, and may be expected to make fewer backward citations because they have less historical background.
+
+Empirical evidence is rather inconclusive. Trajtenberg et al. (1997) find that university patents (presumably more important than the average patent) do make fewer citations and cite patents that are themselves less highly cited. However, von Wartburg, Teichert, and Rost (2005) provide a different view. They correlate a measure of backward citations with expert ratings on the technological value added (in the form of technical scoring tables) of 107 patents related to four strokes internal combustion engines. Their backward citations measure counts first and second-generation's citations received. They obtain a statistically significant correlation coefficient of 0.38, implying that patents with higher technological value added build on more references. Liu et al. (2011) propose a more in-depth analysis of backward references and patent value. They correlate the number of backward references with the probability that a patent will stand up in court and find a statistically strong positive association. Overall, it is unclear whether the number of backward citations captures patent importance.
+
+## Technological Distance and Diversity
+
+As noted, one of the basic virtues of patent data is that they provide a nonmarket-based technological classification system for inventions. Looking at the way in which citations span the technology space defined by the classification scheme is a natural way to characterize the technological complexion of both an invention's roots and its impacts. Broadly speaking, there are two major aspects to be considered, whether looking forward or backward. One is pure distance: how technologically different are the patents connected by a citation link. For example, does a drug patent cite other patents for compounds in the same chemical class, or patents on other chemicals, or mechanical or electronic patents? The other is breadth or diversity: independent of whether that drug patent generally cites other patents that are close to or far from itself , are they all bunched together in technology space, or are they dispersed far from each other?
+
+Trajtenberg et al. (1997) implement a measure of technological distance using a three-level representation of the USPTO patent classification system. The lowest level used is the three-digit original patent class (e.g., Electric lamp and discharge devices); the next level is the set of two-digit categories (e.g., Electrical Lighting); the highest level is six very broad fields (e.g., Electrical and Electronic). The authors axiomatically set two patents in the same patent class at distance 0; two that are in different classes but the same category at distance 0.33; two that are in different categories but the same broad field as distance 0.66; and two that are not even in the same field as distance 1. They then calculate the average distance over both forward and
+
+backward citations for each patent in the university and corporate samples. As expected, they found that the forward citations received by university patents came, on average, from farther away in technology space, although the difference was small and not always statistically significant. For backward citations, there was no consistent pattern, that is, university patents did not systematically cite earlier patents that were, on average, technologically more distant by this metric.
+
+To measure technological dispersion or diversity, Trajtenberg et al. (1997) proposed 1 minus the HerfindahlHirschman Index (HHI) of concentration of the citations across patent classes, that is, 1 minus the sum of squared shares of citations in each class. This metric is equal to zero if all citations are in the same class, and it approaches unity as the citations are spread thinly across all classes. The authors dubbed this metric of diversity “ generality" when applied to forward citations, and “ originality" when applied to backward citations. 6,7 They conjectured that both measures should be larger for more basic inventions, and therefore expected to be larger for university patents than for corporate patents. This hypothesis was borne out in the data for generality measure, but not for originality.
+
+A concept related to generality is that of “ General Purpose Technology ” or GPT. GPTs are conceived as technologies that subsequently connect to many different application or development technologies to allow multiple lines of technology innovation and diffusion. Frequently mentioned examples are the electric motor in the late 19th and early 20th centuries, and digital information technology in the late 20th century. Hall and Trajtenberg (2006) use data from a selected sample of 780 most highly cited patents that were granted by the USPTO in the years 1967–1999 to construct generality, number of citations, and patent class growth, for both cited and citing patents, intended to identify GPTs in their early stages. The article finds that highly cited patents differ in almost all respects from the population of all patents (they take longer to be issued; have twice as many claims; are more likely to have a U.S. origin; are more likely to be assigned to a U.S. corporation; are more likely to have multiple assignees; have on average higher citation lags; have a higher generality; are in patent classes that are growing faster than average). The article concludes that the identified measures, although promising, give contradictory messages when taken separately and that it is not obvious how to combine those measures to choose a sample of GPT patents. $^8$ The fundamental difficulty is that we don't have measures of how general-purpose a technology is other than broad conceptions of GPT technologies. Thus, although it seems plausible that general-purposeness would be reflected in citation patterns, it is hard to pin such patterns down or test their validity. $^9$
+
+Youtie, Iacopetta, and Graham (2008) found that nanotechnology patents from 1990–1993 were more general than computer patents and much more general than drug patents, and interpret this result as evidence that nanotechnology is an emerging GPT. Moser and Nicholas (2004), however, found that electricity patents from the 1920s were less
+
+4 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+general and less highly cited than chemical and mechanical patents from the same period, suggesting that the relationship between the characteristics that make a technology a GPT and other characteristics of inventions is complex.
+
+Another concept related to technological distance and diversity is that of a “ radical ” or “ breakthrough ” invention. Ahuja and Lampert (2001) propose that radical inventions are simply the top 1 % of patents ranked on citations received in a given year. Dahlin and Behrens (2005) adopt a more sophisticated approach. They conceive a “ radical ” invention within a given technology domain (tennis rackets, in their application) to be one that recombines previous technology elements in a new and different way, but which is then imitated and so spawns subsequent patents that combine technology elements in a manner substantially similar to the radical invention. They construct a measure of the “ overlap ” in the respective sets of patents cited by two different patents, and show that the radical inventions (oversized and wide-body rackets, in their application) had little overlap with previous or contemporary patents, but significant overlap with patents that came after.
+
+## Linkage to Science
+
+As discussed, patents contain references to nonpatent documents, the overwhelming majority of which are scientific articles. On this basis, the number of nonpatent backward citations made by a patent, or the fraction of backward citations that these nonpatent citations represent, has been explored as a metric of the closeness of linkage between an invention and scientific research. $^10$
+
+Collins and Wyatt (1988) looked at citations to scientific articles from 366 genetics patents granted from 1980 to 1985, in order to trace linkages from basic research to genetics technology. The United States had the highest number of articles cited in patents, followed by the United Kingdom, Japan, Germany, and France. These figures were compared to the total output of genetics articles for those countries, showing some differences, which were interpreted as indicating that the United Kingdom produced more articles that were useful in developing patented technology than Germany, France or Japan. The number of citations from patents received per article was highest for the United Kingdom, followed by the United States and Germany.
+
+Callaert, Van Looy, Verbeek, Debackere, and Thijs (2006) characterizes nonpatent references in a sample of patents at the USPTO and the European Patent Office (EPO) from 1991–2001. Nonpatent references are found in 34 % of USPTO patents and 38 % of EPO patents, comprising about 17 % of all references (patent and nonpatent combined). For both the USPTO and EPO, more than half of nonpatent references are journal references. Of the remaining nonpatent references, many can be considered scientific in the broader sense (as they consist of conference proceedings, books, databases or other nonjournal scientific publications), or technology related. The article reports that at the USPTO at least 42 % of nonjournal nonpatent references can be
+
+considered scientific in broader sense, and 40 % relate to technological information. For the EPO sample these figures are 77 % and 20 % , respectively.
+
+Tijssen (2002) provides a note of caution on the use of nonpatent references. He found no relationship between the number of nonpatent references and the inventor-reported dependence on science in a small ( $<$ 100) sample of Dutch patents from 1998 – 99. Li, Chambers, Ding, Zhang, and Meng (2014) qualify this finding. They argue that nonselfcitations to scientific articles are a noisy measure of science linkage but that applicant self-citations to scientific articles are indeed informative of science linkage. Roach and Cohen (2013) matched patent citations to survey reports from R & D lab managers in the United States, with particular focus on the extent to which patent citations capture knowledge flows to commercial R & D from publicly funded research. They find that patent citations reflect codified knowledge. However, citations miss the reliance on private and contractbased science, as well as basic research. (The discussion in the section on citations as a measure of knowledge flows considers further whether nonpatent references are an indicator of science dependence.)
+
+## Economic Value
+
+As noted earlier, the (public or private) economic value of an invention is a distinct concept from its technological impact. Citations are, first and foremost, an indicator of technological impact. But it turns out that forward citation intensity is, in fact, correlated with economic value. There are, however, several different concepts of economic value. First, we can in principle think of the (gross) social value of an invention, that is, the total producers' and consumers' surplus associated with its use. In some cases this gross social value may be much greater than the net value, for which we would subtract off the lost rents that may be suffered by previous technologies made wholly or partially obsolete. The gross social value is greater than the private value, that is, the value to the owner of a patented invention; the net social value may be either greater or less than the private value, depending on the magnitude of the “rent stealing” effect. For any of these concepts, we can distinguish the value of the invention and the value of the patented invention , which differ by the value of the legal protection afforded by the patent grant. In practice, these different value concepts may or may not be distinguishable, and proxies for value are often used whose mapping onto these different value concepts may be ambiguous.
+
+An early strand of research on citations and economic value was the work of Francis Narin and his associates seeking to develop indicators based on patent data of companies' competitiveness or technological strength. Carpenter et al. (1981) showed that inventions identified in The Industrial Research Institute IR100 awards are much more highly cited than a random sample of matched patents. Narin et al. (1987) found that the average citation frequency of a company’s patent portfolio was associated with increases in
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017 DOI: 10.1002/asi
+
+---
+
+firms' profits and sales among publicly traded pharmaceutical companies.
+
+Trajtenberg (1990b) calculated the social welfare gains associated with successive generations of Computed Tomography (CT) scanners by estimating hedonic demand functions for the attributes. He then showed that the number of citation-weighted patents associated with each generation was statistically predictive of the magnitude of welfare gains, while the raw or unweighted count of patents was not correlated with surplus (sample of about 500 patents). This suggests that the gross social value of these inventions is associated with the citation intensity of the associated patents. Interestingly, the unweighted patent counts were correlated with the level of R & D expenditure. He interpreted these findings as suggesting that the number of patents is associated with the magnitude of research effort, but not indicative of research success. Counting citation-weighted patents then combines the scale of effort with a measure of such success and yields a measure of effective research output.
+
+Moser, Ohmstedt, and Rhode (2014) identified specific improvements in hybrid corn and gathered data on the magnitude of the yield improvement they allowed. They interpret this as measuring the “ inventive step ” associated with the patent, but as the measurement is in the use domain rather than strictly in the technology domain it seems more closely related to social value than to inventive step, per se . They found that there is, indeed, a strong correlation between yield improvements and citation intensities. Interestingly, they find that there are a small number of early patents that are routinely cited in almost all patents in the field. Excluding these citations enhances the correlation between yield and citation frequency.
+
+Hall, Jaffe, and Trajtenberg (2005) consider the relationship between citation intensity and the private value of patents by relating citation-weighted patents to the market value of the firm. They confirm that citation weighting greatly improves the information content of patent counts in terms of predicting market value. In addition, they find that citations from future patents assigned to the same firm as the original patentee have a larger associated market value than citations from others. $^11$ They also find that a disproportionate share of the value associated with patents is associated with a very small number of highly cited patents. Finally, they find that forward citations are associated with increases in market value at the time a patent is initially granted, suggesting that to a significant extent market participants can anticipate the eventual value of inventions at this early stage, and those expectations are (on average) then confirmed by subsequent citations.
+
+Lanjouw and Schankerman (2001) provide indirect evidence of the relationship between citations and value, by assuming that patents that are litigated are, on average, more valuable than those that are not, and comparing the citation patterns of litigated patents with a control sample of nonlitigated patents. They find that the probability of litigation rises with the number of claims and the number of forward
+
+citations per claim, whereas declining with the number of backward citations per claim. Allison, Lemley, Moore, and Trunkey (2003) undertake a similar approach. Consistent with expectations, they find that litigated patents are more highly cited. Interestingly, they find that litigated patents also have more backward citations.
+
+Harhoff, Scherer, and Vopel (2003) obtained estimates from patent holders of the private value of 772 patents with a 1977 German priority date, and that were maintained to full term. They then examined how that reported value correlated with publicly observable indicia of patent value, including patent citations (and also the number of four-digit IPC codes and family size). They found that both the number of forward citations and the number of backward references to the patent literature are significantly correlated with patent value (see also Harhoff, Narin, Scherer, & Vopel, 1999) . Interestingly, they also found that the number of citations made to nonpatent literature was predictive of value, particularly in drug and chemical patents. They note that the predictive value of backward citations (both patent and nonpatent) is quite useful, as this information is available at time of patent grant, while forward citations must be awaited. $^12$ It is unclear theoretically why backward citations are predictive of value. For nonpatent references, it is plausible that in some fields inventions linked to science are less incremental and hence more valuable. For backward patent citations, it may reflect some tendency for bigger, more complex patents to make more backward citations and also be more valuable on average. In addition, the positive correlation between the number of backward citations and value may simply arise from the fact that applicants have stronger incentives to search for prior art for more important patents (Sampat, 2010) .
+
+Gambardella, Harhoff, and Verspagen (2008) undertook a similar survey of inventors listed in patent applications at the EPO. They found that the number of forward citations is by far the best predictor of reported value, but that the fraction of the variance in reported value explained by any or all of the metrics was relatively low, consistent with a view of citation-weighted patents as an indicator of value, but one with substantial noise.
+
+Nicholas (2008) looked at patents granted to U.S. corporations between 1910 and 1939, and identified the citations to those historical patents from the period 1976 – 1999. He found that about 15 % of the patents from the 1910s received at least one citation from the recent patents, rising to almost 30 % for those from the 1930s. He then goes on to show that citation-weighted patents constructed in this way are correlated with firm market value. Thus, patents that are still cited after 40 to 60 years are more valuable than those that are not. What we cannot know from this exercise (since early citations have not been captured) is the extent to which valuable patents are simply more highly cited at all lag durations, or whether there is greater persistence in the sense that the rate of obsolescence is lower.
+
+Bessen (2008) related the value of patents, as indicated by both renewal information and firm financial data, to a
+
+6 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+number of patent characteristics, including forward citations received. He estimated that each additional citation is associated, on average, with an increase in value of about 1%. Nonetheless, the relationship is very noisy, so that even among very highly cited patents, a significant fraction appears to be of little value; 37% of the patents in the top decile in citation intensity from 1991 were not renewed.
+
+Recent work by Abrams, Akcigit, and Popadak (2013) also suggest an overall positive correlation between forward citations and patent value, but with an inverted-U-shaped relationship in which value falls at high citation rates. This finding is provocative, but it is unclear how robust it is, given the highly selected nature of the sample and the fact that the value of individual patents was estimated as the value of patent portfolios divided by the number of patents in the portfolio.
+
+The next section moves away from work focused on citations as indicators of invention characteristics, and discusses the use of citation data to capture geographic and temporal dimensions of the innovation process.
+
+## Citations as an Indicator of Knowledge Flows
+
+### Geographic Dimension of Knowledge Flows
+
+Jaffe, Henderson, and Trajtenberg (1993) took on the challenge identified by Krugman (1991) on the invisibility of knowledge flows. They suggested that patent citations could be used as a kind of “article trail” that could allow knowledge flows to be measured and tracked. They took a sample of patents from universities, large firms and other firms, and identified all of their citations. They then found, for every citing patent, a corresponding “control” patent, issued at the same time and in the same primary U.S. patent class as the citing patent, and compared the frequency with which citing patents were geographically proximate to the cited patents with the frequency with which the control patents were proximate. Looking at metropolitan statistical areas, states and countries, and eliminating citations that are “self-citations” from the same firm, they showed that citations are indeed more likely to be proximate. For example, at the level of metropolitan areas, 7–9 % of citations (depending on the nature of the cited patents) were from the same area, while only 1–4 % of the control patents were, and the differences were highly significant statistically.
+
+Thompson and Fox-Kean (2005) criticize the Jaffe, Henderson, and Trajtenberg methodology. They argue that selecting control patents based on the primary patent class of the citing patents is too rudimentary to capture the heterogeneity of technology. Patents in the same main patent class may be in different subclasses with inherently different technologies, and patents are assigned to multiple classes, again introducing heterogeneity not captured by the main patent classification. In response, Henderson, Jaffe, and Trajtenberg (2005) agree that it is possible that finer technological controls might be appropriate, but they point out that slicing things too finely minimizes the possibility for identifying knowledge flows across subclasses. Ultimately, the question
+
+comes down to the robustness of the localization effect under different identifying assumptions.
+
+A number of other authors have similarly used citation data to measure knowledge flows. Almeida and Kogut (1997) compare the patent citations of small and large semiconductor firms, and find that the citations made by small firms are more geographically localized. Hicks, Breitzman, Olivastro, and Hamilton (2001) show that U.S. companies' citations to university patents exhibit geographic localization, particularly to patents of nearby public universities. Almeida and Kogut (1999) examine citation patterns among semiconductor firms in the United States, including data on both the firms and the inventors. They show that a significant fraction of the geographic localization of the citations can be traced to specific engineers who move among firms, but are more likely to move to another nearby firm than to one that is farther away. Sonn and Storper (2008) show that, despite improvements in communications technologies, geographical localization has been increasing over time.
+
+Thompson (2006) compares the extent of localization in citations listed by the inventor to those added by the examiner. He finds localization at both the metropolitan area and state levels in both the examiner and inventor citations. Inventor citations are found to be about 20 % more likely to match the country of origin of the citing patent than are examiner citations. In a similar vein, Alcácer and Gittelman (2006) estimate the probability that a citation is generated by an examiner or an inventor, conditional on a set of variables that are frequently employed in the knowledge spillover literature. They find that examiner citations introduce bias for some variables only e.g., self-citations). They find no evidence that the degree of geographic proximity between citing and cited patents differs for inventor and examiner citations.
+
+A subtler pitfall in the use of citations to track knowledge flows relates to the intervention of law firms in the drafting of the patent document. Wagner, Hoisl, and Thoma (2014) show that patents by firms who rely on external agents are more likely to cite documents that are part of the law firm's knowledge repository. They take this result as evidence that law firms help overcome localization. However, a blunter interpretation is that external agents include citations that the firms were not aware of, further increasing the noise in patent citation data.
+
+Maurseth and Verspagen (2002) used data on citations among European patents to construct a region-by-region citation frequency matrix. They then looked at numerous variables to explain these frequencies. Geographical distance has a negative and substantial impact on knowledge flows. Controlling for distance, knowledge flows are greater between regions located within one country than between regions located in separate countries. The country effect remains even if regions share the same language, though sharing a language increases the amount of knowledge flows between two regions by up to 28 % . The study also suggests that knowledge flows are industry specific, and regions'
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017 7 DOI: 10.1002/asi
+
+---
+
+technological specialization is an important determinant of their technological interaction.
+
+### Temporal Dimension of Knowledge Flows
+
+Caballero and Jaffe (1993) and Jaffe and Trajtenberg (1999) developed a structured model of knowledge diffusion across space and time. They postulate that two competing forces dominate the citation process. Over time, knowledge gradually diffuses, so that the number of people potentially citing a given patent increases exponentially with time. But the relevance or usefulness of a bit of knowledge becomes obsolete, leading to a countervailing exponential depreciation in the likelihood of citations. The parameters of these two exponential functions can be estimated econometrically. If allowed to vary across different technologies, different kinds of research organizations, and different geographic locations, they then capture the rates of diffusion in different areas across organizations and across space. Jaffe and Trajtenberg show, for example, that the geographic localization of citations diminishes as time passes, and also that obsolescence (as captured by declining citation rates) is more rapid in electronic technologies than in chemical and mechanical technologies.
+
+Bacchiocchi and Montobbio (2009) used this doubleexponential function to look at knowledge flows from universities and public research organizations compared to flows from corporate patents in six countries: France, Germany, Italy, Japan, the United Kingdom, and the United States. They found that technology embodied in patents from universities and public research organizations diffuses more rapidly than that of firms. The diffusion rates are relatively homogenous across technological fields, but vary across countries: rapid in the United States and Germany, less so in France and Japan.
+
+Mehta, Rysman, and Simcoe (2010) have criticized this diffusion model on the ground that the age of a citation is computed as the citation year minus the application year, leading to an identification problem. Because citations received by a patent are rare before it is issued, the authors propose to use the lag between application year and grant year as a source of exogenous variation. They find that the citation peak occurs earlier than suggested by the doubleexponential function. However, their method does not alter differences in the mean citation ages across industries. They conclude that the double-exponential function provides a good approximation to the nonparametric age distribution.
+
+### Validation Studies
+
+Jaffe, Trajtenberg, and Fogarty (2000) report a survey of inventors to test the extent to which citations in those inventors’ patents correspond to the inventors’ perceptions of how their inventions depended on earlier knowledge, and how the rate of citation relates to inventors’ own perceptions of impact or importance. They find that citations are a valid but noisy indicator of knowledge flows: The likelihood of reported knowledge impact is significantly higher (both
+
+quantitatively and statistically) when a citation link exists, but a significant fraction of citations (perhaps as high as one half) do not correspond to any reported knowledge link.
+
+Duguet and MacGarvie (2005) tested the validity of patent citations as a measure of knowledge flows using data from French firms on their patents and citations, combined with survey responses regarding sources of knowledge. The total number of backward citations was correlated with survey answers about R & D and innovation, but this correlation was weakened by controlling for the number of patents held by the citing firm. Backward citation rates of French firms can reflect their R & D activities (if the technology is obtained from firms located in the EU), or purchases of equipment goods (if the source is located outside the EU). In general it can be understood that backward citations are correlated with learning through R & D collaboration, licensing of foreign technology, mergers and acquisitions and equipment purchases.
+
+In their analysis, Roach and Cohen (2013, discussed earlier) found evidence of both “ errors of omission ” (reported knowledge flows with no corresponding citations) and “ errors of commission ” (observed citations with no corresponding reported knowledge flows). They conclude that despite these sources of measurement error, patent citations are likely to reflect meaningful aspects of knowledge flows from public research. Interestingly, they found that references in patents to nonpatent publications (primarily scientific literature) are a better indicator of knowledge flow than are citations in commercial patents to the patents of universities and other public labs (cf. Tijssen, 2002) .
+
+The next section discusses a third category of citation data research, in which the focus shifts to using citation links to understand and characterize networks.
+
+## Citations as Links in Knowledge or Innovation Networks
+
+A natural way of representing citation data is in the form of a network. Researchers have used concepts from network theory to grasp the way the innovation system is structured and the way knowledge is formed. A first group of studies seek to map key components of the innovation system (patents, individuals, institutions, and regions). A second group of studies use the network of citations to map technological trajectories. We review these two applications in turn.
+
+### Mapping Patents, Individuals, Institutions, and Regions
+
+Huang, Chiang, and Chen (2003) rely on patent citation data to map Taiwan's electronic industry. The researchers identify USPTO patents belonging to 58 relevant Taiwanese companies as well as the citations made by these patents. They identify the strength of the relationship between companies by looking at the strength of bibliographic coupling. Bibliographic coupling is a method proposed by Kessler (1963) that involves identifying related documents through common cited references. The researchers then applied cluster analysis on the data produced to identify major sectors of
+
+8 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+the Taiwanese electronic industry. Although bibliographic coupling provides rich insights on the relatedness of patent documents, more recent studies make better use of network analysis theory and tools.
+
+Chen and Hicks (2004) study the citation “ degree ” distribution of 16 million citations made to the 3 million USPTO patents granted in the period from 1963 to 1999. The degree of a “ node ” (patent) is simply the number of “ connections ” (citations) received by the node. They estimate that the distribution follows a power law with an exponent of 2.89, which is very similar to the parameter obtained for scientific articles by Dorogovtsev and Mendes (2002) . $^13$ The fact that the degree distribution of the patent citation network follows a power-law is indicative of so-called scale-free networks, which can be seen as networks characterized by large hubs through which knowledge flows.
+
+Li, Chen, Huang, and Roco (2007) use a patent citation network to study the knowledge transfer process between entities. In particular they study the efficiency with which knowledge transfers within the network compared to a random network. Their measure of efficiency is the average path length between any pair of patents in the network. They focus on USPTO nanotechnology patents in the period from 1976 to 2004. They find that knowledge transfer across assignees in the citation network is more efficient than knowledge transfer that would occur in a random network. Knowledge flow across (assignee) countries is as efficient as a random network. However knowledge flow across technology fields is less efficient than knowledge flow that would occur in a random network. In other words, technological distance is a greater barrier to knowledge flows than geographic distance.
+
+Hung and Wang (2010) examine the characteristics of the citation network formed by RFID patents. They find that the network can be characterized as a “small-world” network, that is, a network in which most nodes can be reached from every other by a small number of steps. They also find that the network has a power-law connectivity distribution and exhibits preferential connectivity behavior. That is, a few key patents have a very large number of connections and the majority of patents have few connections. The authors conclude that only a limited number of patents play a key role in diffusing RFID technology. This approach provides a more system-based way of thinking about knowledge flows than simply counting citations: Key patents are not only highly cited patents, but also connect and integrate different technological trajectories. $^14$ More detailed analyses of technological trajectories using citation network are described in the next section.
+
+## Mapping of Technological Trajectories
+
+Scholars have recently used citation networks to identify technological trajectories that led to the advent of major technological breakthroughs. The main trajectory, or search path, is the sequence of links and nodes that is central to the development of a technology. It represents the main flow of
+
+ideas in the development of a technology. The method was pioneered by Hummon and Doreian (1989) on a citation network of scientific articles describing the development of DNA theory. This approach shifts the focus from the nodes of the network (looking at individual patents) to the connections that these nodes form. It allows identifying key patents through their structural connectivity in the network. Technologically important patents should belong to the main paths of the citation network and/or locate at particularly critical junctions within those paths.
+
+Mina, Ramlogan, Tampubolon, and Metcalfe (2007), Verspagen (2007), and Fontana, Nuvolari, and Verspagen (2009) applied the method to patent citation networks. Mina et al. (2007) use it to understand how medical knowledge emerges, grows, and evolves. They argue that the approach provides a dynamic view of innovation that recognizes the long-term, path-dependent, and complex nature of technology. Their case study is based on treatment for coronary artery disease and covers 5,136 USPTO patent documents granted between 1976 and 2003. The authors seek to identify the main path and “ islands ” of the network. Islands are small clusters of inventions whose internal connectedness is relatively superior to the strength of their outward connections within the global network. The authors argue that islands allow accounting for the variety of complementary and competing areas of technical expertise that contributed to the advancement of the technology. They report that the results form a consistent map of the major scientific and technological trajectories in the domain.
+
+Fontana et al. (2009) study the structural connectivity of the citation network formed by patents related to local area networks (LAN) technology. Innovation in such a systemic technology has three main features. First, innovation is distributed: it takes place at the level of individual components but these components all have to work together. Second, innovations in systems tend to be incremental and to occur around well-established technical designs. Third, innovations also tend to occur continuously. The authors argue that the classical approach of assessing the importance of patents by counting the number of citations they have received may have drawbacks in such systemic technologies. It may fail to identify concepts and principles that could act as “focusing devices” for a sequence of inventive activities. By contrast a structural analysis of the citation network would allow the identification of inventions that have played a major role in the evolution of LAN technology. They find that the main path they have identified displays a coherent economic and engineering logic, consistent with qualitative accounts of the evolution of the Ethernet standard.
+
+One of the most interesting insights of the article comes from the analysis of companies owning patents that lie on the main path. No company is “dominant” in the sense of claiming ownership of the majority of patents on the main path, which the authors take as evidence that no company is strategically placed along the main path of knowledge flow. Verspagen (2007) performs a similar analysis for citations among fuel cell patents. He finds that there are dominant
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017 DOI: 10.1002/asi
+
+---
+
+companies: A small number of organizations hold patents belonging to the main path. Study of the ownership structure of technologies on the main path provides a novel way of characterizing technology dominance. It is a promising avenue for research in industrial economics and strategic management.
+
+## Pitfalls and Best Practices in Use of Citation- Based Indicators
+
+We take the opportunity of this review to discuss potential pitfalls associated with patent data. We focus on four key challenges.
+
+### Office Effects
+
+Institutional differences across jurisdictions induce differences in citation practices across offices. We briefly summarize two main differences in citation practices between the EPO/Japan Patent Office (JPO) and the USPTO for illustrative purposes. More generally, researchers should get a clear understanding of citation practices in the office of interest before using citation-based indicators. 15
+
+A first difference is the “ duty of candor ” in U.S. patent law. Failure to report known relevant prior art may lead to subsequent revocation of the patent (inequitable conduct doctrine). There is no duty of candor in European patent law, and applicants do not have to submit a list of prior art. It follows that search reports at the EPO usually contain many fewer references than USPTO search reports. In fact, according to EPO philosophy, “ a good search report contains all the technically relevant information within a minimum number of citations ” (Michel & Bettels, 2001, p. 189) . In addition, since applicants at the EPO do not bear the same responsibility to disclose prior art as applicants at the USPTO, the citations come mostly from the examiner. This does not undermine their interpretation as indicators of impact or value; for example Harhoff et al. (2003, discussed earlier) find EPO citations to be predictive of value. It does suggest that EPO citations might be less indicative of knowledge flows; although we are not aware of any empirical analysis of this question comparable to the survey work of Jaffe et al. (2000) . In Japan, the patent law was revised in 2002 and imposed on applicants the obligation to disclose prior art. Although not complying with the disclosure requirement bears less severe consequence than in the United States, the reform led to a substantial increase in prior art disclosure by applicants. Takahiro, Nagaoka, and Naito (2015) find that about 8 % of citations came from applicants in the years following the reform, compared to around 4 – 5 % before the reform.
+
+A second important difference with the USPTO is that EPO patent examiners classify documents cited in particular citation categories (Schmoch, 1993) . A document that shows essential features of the invention or questions the inventive step of these features if taken alone is marked with the letter “ X. ” A document that questions the inventive step if combined with another document is marked with the letter “ Y ”
+
+(hence “ Y ” citations never occur singly). The letter “ A ” marks a document that shows the general state of the art. According to Schmoch (1993, p. 195) a patent document can be highly cited because it comprises “ a good description of the prior art from a didactic point of view. ” The classification provides opportunities for finer analyses. One may want to exclude class “ A ” citations for assessing the inventive step of patents, but class “ A ” citations are relevant for measuring technological proximity of patents. Additional classification codes exist; see Webb, Dernis, Harhoff, and Hoisl (2005) for a discussion. Examiners at the JPO also classify citations into categories. In particular, they flag whether citations are used as ground for rejection, similar to “ X ” and “ Y ” citations at the EPO, or whether they are used for assessing the application but do not serve as a basis for rejection, similar to “ A ” citations at the EPO (Goto & Motohashi, 2007) .
+
+The classification into categories opens the door to original uses of citation data. For example, von Graevenitz, Wagner, and Harhoff (2011) identify patent thickets at the EPO using X and Y citations. Their measure identifies constellations in which three firms each own patents that block patent applications of the other two firms (so-called triples). The authors show that density of triples in complex technology areas has risen steadily since the early 1980s, whereas the density of triples has been constant in discrete technology areas. Guellec, Martinez, and Zuniga (2012) use “ X" and “ Y" citations together with administrative information on the patent examination process (withdrawal and grant events) to identify defensive patents, that is, patent applications used to pre-empt others from getting their patents granted. Palangkaraya, Webster, and Jensen (2011) posit that patents with a higher inventive step will generate more “ X" and “ Y" citations, and use this information to proxy for the probability of grant ex-ante.
+
+Beyond institutional differences in the use of citations, researchers have also illustrated the presence of home bias in citation practices. Bacchiocchi and Montobbio (2010) analyze the geographic distribution of cited documents for a set of 657,151 equivalent patents filed at the EPO and the USPTO. In theory, distributions should be similar since they refer to the same invention. They find that the frequency of U.S.-cited patents at the USPTO exceeds 65 % , while the frequency at the EPO is less than 40 % . That examiners have a tendency to cite local documents does not come as a surprise. $^16$ However, it illustrates an important limitation of the use of citations for assessing cross-border knowledge flows.
+
+### Time and Technology Field Effects
+
+The number of citations received by a patent increases as time passes such that there are strong cohort effects. This issue can be dealt with in a straightforward manner by counting citations received in a fixed time interval (e.g., citations received up to 5 years after grant). A more serious concern is the increase over time of citations made per patent . Hall, Jaffe, and Trajtenberg (2001) report that the average
+
+10 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+patent issued in 1999 made over twice as many citations as the average patent issued in 1975 (10.7 vs. 4.7 citations). Although this issue does not affect the comparison of patents within a cohort, citation inflation makes it challenging to compare patents across cohorts. Analogously, citation practices and the intensity of activity vary by technology fields, so that what constitutes a high citation rate in one field may be modest or small for another field. $^17$ The authors discuss two econometric techniques to deal with citation inflation and varying intensities by field: scaling citation counts by “ dividing them by the average citation count for a group of patents to which the patent of interest belongs ” ; and identifying the multiple biases on citation rates via econometric estimation. Marco (2007) provides a recent illustration of the latter technique. He argues that by estimating a hazard rate based only on factors that are correlated to citation inflation rather than value, residuals can be used to measure latent patent value. For example, the ratio of observed citations to predicted citations may represent a proxy of patent value. Such an approach is an important step forward, although it is difficult to identify factors that are truly exogenous to value.
+
+A broader question, which has received little coverage in the literature, relates to differences in patenting and citation practices across technology fields. We know that the propensity to patent differs across fields (Cohen, Nelson, & Walsh, 2000) and that the relevance of patent data as innovation indicator therefore also varies across fields (e.g., Danguy, de Rassenfosse, & van Pottelsberghe, 2014) . However, to the best of our knowledge, no study has investigated in a systematic manner how differences across fields affect the relevance of patent citation data.
+
+### Examiner Effects
+
+Cockburn, Kortum, and Stern (2002) show that there is substantial examiner heterogeneity, for example, in terms of variations in tenure at the USPTO and in the average approval time per issued patent. Such heterogeneity translates into variations in outcomes of the examination process — such as in the volume and pattern of citations made. $^18$ Lemley and Sampat (2012) demonstrate the presence of an examiner effect, in the sense that more experienced examiners cite less prior art. Alcacer, Gittelman, and Sampat (2009) painted a picture of examiner-added citations across key strata of patent data. They report that the proportion of citations added by examiners is higher for patents: by foreign applicants to the USPTO; by applicants with a large patent portfolio; and by applicants in electronics, communications, and computer-related fields. Criscuolo and Verspagen (2008) perform a similar analysis for EPO patent data. They show that the share of inventor citations has been declining from about 14 % in 1985 to 9 % in 2000. In addition, there is also substantial variation across fields. More than 20 % of citations in organic chemistry patents were added by the inventor, while for information technology patents this share is 4 % . $^19$
+
+Examiner intervention may bias the information content of citations. It may undermine the use of citations as a measure of knowledge flow, since the inventors may not have even been aware of the patents cited by examiners at the time of invention. However, examiner citations may be taken as a valid reflection of technological and economic value. In this spirit, Hegde and Sampat (2009) show that examiner citations have a much stronger relationship with renewal probability (a measure of private value) than the number of applicant citations.
+
+### Strategic Effects
+
+Variations in the number of examiner-added citations may also come from differences in applicants' incentives to search for or disclose prior art. Recent research suggests that citing prior art (or not) is a strategic decision. Atal and Bar (2010) study firms' incentives to search for unknown prior art . Although applicants at the USPTO have a duty to disclose what they know, they have no duty to search for prior art and may be better off by remaining ignorant. The authors show theoretically that firms search more when R & D investment (a proxy for innovation quality) and patenting costs are higher. Sampat (2010) provides empirical data on when applicants search for prior art. He shows that applicants contribute more prior art for their more important inventions. He also shows that applicants are more likely to search for prior art in fields where individual patents are important for appropriating returns from R & D (chemicals and drugs) and less likely to do so in industries where firms tend to accumulate patent portfolios for other strategic reasons (computers and communications, electronics and electrical, and mechanical).
+
+Lampe (2012) focuses on applicants' decision to disclose known prior art . He identifies “ voluntary withholding" of citations to prior art material by looking only at citations that were present on prior patents issued to the same firm. He estimates that applicants withhold between 21 % and 33 % of relevant citations. The rate is higher for firms applying for computer and electronic patents (25 to 42 % ) and lower for firms applying for drug and chemical patents (8 – 22 % ). More generally, Lampe finds that the likelihood of citation is positively correlated with proxies of patent value (number of claims and forward citations) and negatively correlated with the size of applicant patent portfolios.
+
+## Conclusion
+
+The use of patent citation data in social science research has exploded in the last two decades. As just one indication, the frequency of appearance of the term “patent citation” in scientific documents listed in Google Scholar increased 10fold between 2000 and 2014 (Appendix A). As is often the case, this increase reflects increases in both supply and demand. On the supply side, the digitization of the patent office records, combined with the increased power of computers to analyze them, makes analyses possible today that simply could not have been undertaken 25 years ago. The
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017 11 DOI: 10.1002/asi
+
+---
+
+number of scientific documents referencing the National Bureau of Economic Research (NBER) patent citation data file is likewise continuously increasing (Appendix A). On the demand side, intangible assets are increasingly seen as a source—some would argue the dominant source—of economic returns. By definition, intangible assets are hard to track and measure, and so researchers interested in diverse questions about knowledge accumulation and diffusion, innovation, firm strategy and regional economic growth seek measures that convey information about the sources and consequences of these assets.
+
+Neither of these trends is likely to reverse, so interest in measures of this kind is likely to continue to grow. Recent developments in computational linguistics may allow for construction of measures that are conceptually related to citations but use all of the information contained in the patent text rather than relying solely on the links between patents that are explicitly identified via citation. It is now possible, for example, to identify connections between a patent and its antecedents by measuring the frequency with which important words are used in both patents, to measure novelty by identifying patents that use a certain technical term or combination of words in a particular phrase for the first time, and to measure impact by counting the number of subsequent patents that use such a phrase (e.g., Packalen & Bhattacharya, 2015) . Younge and Khun (2015) use more advanced techniques to develop a text-based pairwise similarity comparison of any and every two patents at the USPTO. These new approaches have not yet been subjected to the kind of validation that has demonstrated the economic significance of citations, but because they utilize more information, they offer the promise of a valuable broadening and deepening of the research possibilities.
+
+A more mundane, but equally important task, is to further validate citation indicators. This applies to both established and novel indicators, at both the USPTO and other offices. For example it is unclear whether the count of backward citations proxies for patent importance. Even the link between forward citations and economic value, one of the most established and used indicators, is not well understood. In a similar vein, little research exists on technology field differences on the relevance of patent citation data. The need for validation studies will grow more pressing as new indicators are being developed and more patent offices make their data available. Similarly, legislative changes affect citation practices in nontrivial ways, and conclusions drawn using data from one-time period are not necessarily valid in another time period. This calls for a continuous assessment of the validity of citation indicators.
+
+Another exciting area of research is the further application of network theory and analysis tools to the patent citation network. For example, the identification of key technologies and actors on the main knowledge path promises to greatly improve our understanding of industry dynamics and the knowledge creation process. A limitation of current research in the area is the insularity of two com-
+
+munities of scholars. Studies by scholars using advanced network analysis tools offer little practical implications, whereas studies by scholars looking at real-world implications use quite basic network analysis tools. A promising way forward is to better integrate the technical and the practical aspects of network analysis.
+
+Finally, researchers realize that the patent citation generation process is complex but more work needs to be done to understand it. The complexity of the patent citation generation process is a blessing and a curse. Whereas it may distort the reality in an undesirable fashion, it may also provide a window into the incentives faced by inventors, patent attorneys and examiners and serve as a source of econometric identification. The example of examineradded citations is a case in point. Whereas citations made by examiners arguably weaken the measurement of knowledge flows, they also strengthen the measurement of patent value.
+
+## Acknowledgments
+
+We thank David Schwartz for suggesting this project. We are grateful to two anonymous referees for valuable comments. We also received helpful comments from Bronwyn Hall, Dietmar Harhoff, Sadao Nagaoka, and Beth Webster, as well as participants at the 2015 Workshop on the Economics of Intellectual Property in Northwestern University. Jan Kozak provided valuable research assistance. We are solely responsible for all opinions or errors.
+
+## Endnotes
+
+1The present survey discusses evidence on citations at the European Patent Office whenever available.
+
+2 The earliest reference that we found is Clark (1976) . It presents statistics on the obsolescence of United States Patent and Trademark Office (USPTO) patents using citation data. Garfield (1966) discusses the use of patent citation searches to say something about the significance of a patent, but it does not present any systematic analyses or statistics. Kuznets (1962) did not specifically discuss citations, but did emphasize that patent documents are a rich and deep source of information on the inventive process, and urged that this richness be exploited in addition to researchers' simply counting patents
+
+3 Jacob Schmookler pointed out that in a patent subclass “Dispensing of semi-solid materials,” he found a patent for a manure spreader and another for a toothpaste tube (Schmookler, 1966).
+
+4Jaffe and Trajtenberg (2002) reprints 12 of the key articles on patent citations by them and their co-authors.
+
+5 The authors report “ forward importance ” as the number of citations received plus .5 times the number of citations received by the citing patents, and undertook sensitivity analysis varying this weight between 0.25 and 0.75. Extending this throughout the citation tree involves a geometrically declining weight, for example, if patent E cites patent D which cites patent C which cites patent B which cites patent A, we might consider patent B to contribute 1 to the importance of A, patent C 0.5, patent D 0.25 and patent E 0.125.
+
+$^{6}$ For a small number of citations, it is clear that this measure is heavily influenced by the number of citations, for example, a patent receiving only two citations cannot possibly have generality greater than 0.5. Whether or not this is a problem is largely a matter of interpretation; in some sense it is meaningful to say that a patent receiving only two citations cannot have a very diverse impact. A different interpretation is that every invention has a latent or unobserved generality that is
+
+12 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+randomly realized in the citations it happens to receive. Under this formulation, the distribution of citations across patent classes is multinomial, and the observed generality or originality is a biased estimator of the true parameter. Bronwyn Hall has derived a formula to correct for this bias (Hall et al., 2001) ; it produces a significant correction for patents with just a few citations.
+
+7Ziedonis (2004) has built on this idea to construct a measure of the fragmentation of ownership rights to a firm's complementary patents. Backward citations are stratified by assignee instead of technology class.
+
+$^8$ Hall and Trajtenberg (2006) explain that the generality measures suffer from the fact that they treat citations from patents in patent classes different from the cited patents in the same way, although some patent classes are very different and some are closely related. They suggest that the future research could construct a weighted generality measure, with weights inversely related to the overall probability that one class cites another class. To the best of our knowledge no one has implemented such an approach.
+
+$^9$ Hall and Trajtenberg (2006) also show that a disproportionate share of the patents in the extreme upper tail of the distribution for generality and total forward citations in the period 1967 – 1999 are information technology (IT) patents, suggesting that these metrics may be indicative of a GPT.
+
+$^{10}$ Lemley and Sampat (2012, footnote 12) find that the vast majority of references to nonpatent prior art at the USPTO come from applicants, not examiners, potentially making these a relevant measure of science dependence.
+
+$^{11}$ Since Trajtenberg (1990b) showed that total citations are correlated with social returns, the finding that self-citations have a stronger effect on market value than other citations suggest that self-citation is associated with the extent of appropriation of the social returns by the original patenting firm.
+
+12 Similarly, international family size is a measure predictive of value that is knowable soon after patent application.
+
+$^{13}$ Cf. Huang, Huang, Chang, Chen, and Lin (2014) who provide evidence that the distribution of patent citations is more concentrated than the distribution of citations in scientific articles.
+
+$^{14}$ Hu, Rousseau, and Chen (2012) provide another study on the importance of patents using their positions in the citation network. Other applications include, for example, Liu and Shih (2011) who use the network formed by patents to improve patent classification.
+
+15 For example, researchers interested in EPO citations should read the “Guidelines for Examination in the European Patent Office” available on the EPO website.
+
+16 For example, there is a substantial cost to including non-English references at the USPTO. When using a foreign language reference in a rejection, examiners should provide a translation of the entire document.
+
+17 Technology fields are tracked using the patent office classification systems. Historically, the United States has maintained its own classification (USPC), while other offices use the International Patent Classification (IPC). The USPTO has recently introduced a Cooperative Patent Classification (CPC) based on the IPC, and is phasing out the USPC.
+
+18 Alcácer and Gittelman (2006) estimate that examiners insert two thirds of citations on the average patent, and 40 % of all patents have all citations added by examiners.
+
+19 There are few country-specific studies. See Azagra-Caro, Mattsson, and Perruchas (2011) for Spanish evidence.
+
+## References
+
+Abrams, D., Akcigit, U., & Popadak, J. (2013). Patent value and citations: Creative destruction or strategic disruption? National Bureau of Economic Research Working Paper No. 19647.
+
+Ahuja, G., & Lampert, C.M. (2001). Entrepreneurship in the large corporation: A longitudinal study of how established firms create breakthrough inventions. Strategic Management Journal, 22(6-7), 521-543.
+
+Albert, M.B., Avery, D., Narin, F., & McAllister, P. (1991). Direct validation of citation counts as indicators of industrially important patents. Research Policy, 20(3), 251-259.
+
+Alcácer, J., & Gittelman, M. (2006). Patent citations as a measure of knowledge flows: The influence of examiner citations. The Review of Economics and Statistics, 88(4), 774–779.
+
+Alcacer, J., Gittelman, M., & Sampat, B. (2009). Applicant and examiner citations in US patents: An overview and analysis. Research Policy, 38(2), 415–427.
+
+Allison, J.R., Lemley, M.A., Moore, K.A., & Trunkey, R.D. (2003). Valuable patents. Georgetown Law Journal, 92, 435.
+
+Almeida, P., & Kogut, B. (1997). The exploration of technological diversity and the geographic localization of innovation. Small Business Economics, 9, 21–31.
+
+Almeida, P., & Kogut, B. (1999). Localization of knowledge and the mobility of engineers in regional networks. Management Science, 45(7), 905-917.
+
+Atal, V., & Bar, T. (2010). Prior art: To search or not to search. International Journal of Industrial Organization, 28(5), 507-521.
+
+Azagra-Caro, J.M., Mattsson, P., & Perruchas, F. (2011). Smoothing the lies: The distinctive effects of patent characteristics on examiner and applicant citations. Journal of the American Society for Information Science and Technology, 62(9), 1727–1740.
+
+Bacchiocchi, E., & Montobbio, F. (2009). Knowledge diffusion from university and public research. A comparison between US, Japan and Europe using patent citations. The Journal of Technology Transfer, 34(2), 169–181.
+
+Bacchiocchi, E., & Montobbio, F. (2010). International knowledge diffusion and home-bias effect: Do USPTO and EPO patent citations tell the same story?. The Scandinavian Journal of Economics, 112(3), 441-470.
+
+Benson, C.L., & Magee, C.L. (2015). Quantitative determination of technological improvement from patent data. PLoS One, 10(4), e0121635.
+
+Bessen, J. (2008). The value of US patents by owner and patent characteristics. Research Policy, 37(5), 932-945.
+
+Brin, S., & Page, L. (1998). The anatomy of a large-scale hypertextual Web search engine. Computer Networks and ISDN Systems, 30, 107-117.
+
+Caballero, R.J., & Jaffe, A.B. (1993). How high are the giants' shoulders: An empirical assessment of knowledge spillovers and creative destruction in a model of economic growth. In O.Blanchard and S. Fischer (Eds), NBER Macroeconomics Annual 1993 (Vol. 8, pp. 15– 86). Cambridge: MIT press.
+
+Callaert, J., Van Looy, B., Verbeek, A., Debackere, K., & Thijs, B. (2006). Traces of prior art: An analysis of non-patent references found in patent documents. Scientometrics, 69(1), 3-20.
+
+Carpenter, M.P., & Narin, F. (1983). Validation study: Patent citations as indicators of science and foreign dependence. World Patent Information, 5(3), 180–185.
+
+Carpenter, M.P., Narin, F., & Woolf, P. (1981). Citation rates to technologically important patents. World Patent Information, 3(4), 160-163.
+
+Chen, C., & Hicks, D. (2004). Tracing knowledge diffusion. Scientometrics, 59(2), 199-211.
+
+Clark, C.V. (1976). Obsolescence of the patent literature. Journal of Documentation, 32(1), 32-52.
+
+Cockburn, I.M., Kortum, S., & Stern, S. (2002). Are all patent examiners equal?: The impact of characteristics on patent statistics and litigation outcomes. National Bureau of Economic Research Working Paper 8980.
+
+Cohen, W.M., Nelson, R.R., & Walsh, J.P. (2000). Protecting their intellectual assets: Appropriability conditions and why U.S. manufacturing firms patent or not. National Bureau of Economic Research Working Paper 7552.
+
+Collins, P., & Wyatt, S. (1988). Citations in patents to the basic research literature. Research Policy, 17(2), 65-74.
+
+Cotropia, C.A., Lemley, M.A., & Sampat, B. (2013). Do applicant patent citations matter? Research Policy, 42(4), 844–854.
+
+Criscuolo, P., & Verspagen, B. (2008). Does it matter where patent citations come from? Inventor vs. examiner citations in European patents. Research Policy, 37(10), 1892–1908.
+
+Czarnitzki, D., Hussinger, K., & Schneider, C. (2011). "Wacky" patents meet economic indicators. Economics Letters, 113(2), 131-134.
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017 13 DOI: 10.1002/asi
+
+---
+
+Dahlin, K.B., & Behrens, D.M. (2005). When is an invention really radical?: Defining and measuring technological radicalness. Research Policy, 34(5), 717-737.
+
+Danguy, J., de Rassenfosse, G., & van Pottelsberghe de la Potterie, B. (2014). On the origins of the worldwide surge in patenting: An industry perspective on the R&D-patent relationship. Industrial and Corporate Change, 23(2), 535–572.
+
+Dorogovtsev, S.N., & Mendes, J.F. (2002). Evolution of networks. Advances in Physics, 51(4), 1079-1187.
+
+Duguet, E., & MacGarvie, M. (2005). How well do patent citations measure flows of technology? Evidence from French innovation surveys. Economics of Innovation and New Technology, 14(5), 375–393.
+
+Fontana, R., Nuvolari, A., & Verspagen, B. (2009). Mapping technological trajectories as patent citation networks. An application to data communication standards. Economics of Innovation and New Technology, 18(4), 311–336.
+
+Gambardella, A., Harhoff, D., & Verspagen, B. (2008). The value of European patents. European Management Review, 5(2), 69-84.
+
+Garfield, E. (1955). Citation indexes for science: A new dimension in documentation through association of ideas. Science, 122(3159), 108-111.
+
+Garfield, E. (1966). Patent citation indexing and the notions of novelty, similarity, and relevance. Journal of Chemical Documentation, 6(2), 63-65.
+
+Gay, C., & Le Bas, C. (2005). Uses without too many abuses of patent citations or the simple economics of patent citations as a measure of value and flows of knowledge. Economics of Innovation and New Technology, 14(5), 333-338.
+
+Goto, A., & Motohashi, K. (2007). Construction of a Japanese Patent Database and a first look at Japanese patenting activities. Research Policy, 36(9), 1431-1442.
+
+Griliches, Z. (1979). Issues in assessing the contribution of research and development to productivity growth. The Bell Journal of Economics, 92-116.
+
+Griliches, Z. (1990). Patent statistics as economic indicators: A survey. Journal of Economic Literature, 28(4), 1661-1707.
+
+Guellec, D., Martinez, C., & Zuniga, P. (2012). Pre-emptive patenting: Securing market exclusion and freedom of operation. Economics of Innovation and New Technology, 21(1), 1-29.
+
+Hall, B.H., Jaffe, A.B., & Trajtenberg, M. (2001). The NBER patent citation data file: Lessons, insights and methodological tools. National Bureau of Economic Research Working Paper 8498.
+
+Hall, B.H., Jaffe, A., & Trajtenberg, M. (2005). Market value and patent citations. RAND Journal of Economics, 36(1), 16-38.
+
+Hall, B.H. & Trajtenberg, M. (2006). Uncovering GPTs using patent data. In C. Antonelli, D. Foray, B.H. Hall, & W.E. Steinmuller (Eds.), New frontiers in the economics of innovation and new technology—Essays in honor of Paul A. David. Cheltenham: Edward Elgar Publishing.
+
+Harhoff, D., Narin, F., Scherer, F.M., & Vopel, K. (1999). Citation frequency and the value of patented inventions. Review of Economics and Statistics, 81(3), 511-515.
+
+Harhoff, D., Scherer, F.M., & Vogel, K. (2003). Citations, family size, opposition and the value of patent rights. Research Policy, 32(8), 1343-1363.
+
+Hegde, D., & Sampat, B. (2009). Examiner citations, applicant citations, and the private value of patents. Economics Letters, 105(3), 287–289.
+
+Henderson, R., Jaffe, A., & Trajtenberg, M. (2005). Patent citations and the geography of knowledge spillovers: A reassessment: Comment. American Economic Review, 95(1) 461–464.
+
+Hicks, D., Breitzman, T., Olivastro, D., & Hamilton, K. (2001). The changing composition of innovative activity in the US—A portrait based on patent analysis. Research Policy, 30(4), 681–703.
+
+Hu, X., Rousseau, R., & Chen, J. (2012). A new approach for measuring the value of patents based on structural indicators for ego patent citation networks. Journal of the American Society for Information Science and Technology, 63(9), 1834–1842.
+
+Huang, M.H., Chiang, L.Y., & Chen, D.Z. (2003). Constructing a patent citation map using bibliographic coupling: A study of Taiwan's hightech companies. Scientometrics, 58(3), 489-506.
+
+Huang, M.H., Huang, W.T., Chang, C.C., Chen, D.Z., & Lin, C.P. (2014). The greater scattering phenomenon beyond Bradford's law in patent citation. Journal of the Association for Information Science and Technology, 65(9), 1917-1928.
+
+Hummon, N.P., & Doreian, P. (1989). Connectivity in a citation network: The development of DNA theory. Social Networks, 11(1), 39-63.
+
+Hung, S.W., & Wang, A.P. (2010). Examining the small world phenomenon in the patent citation network: A case study of the radio frequency identification (RFID) network. Scientometrics, 82(1), 121–134.
+
+Jaffe, A.B., & Trajtenberg, M. (1999). International knowledge flows: Evidence from patent citations. Economics of Innovation and New Technology, 8(1–2), 105–136.
+
+Jaffe, A.B., & Trajtenberg, M. (2002). Patents, citations, and innovations: A window on the knowledge economy. Cambridge: MIT press.
+
+Jaffe, A.B., Trajtenberg, M., & Fogarty, M.S. (2000). Knowledge spillovers and patent citations: Evidence from a survey of inventors. American Economic Review , 90(2), 215–218; also published with additional detail as “The Meaning of Patent Citations: Report on the NBER/Case-Western Reserve Survey of Patentees,” in Jaffe and Trajtenberg (2002), op cit.
+
+Jaffe, A.B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as evidenced by patent citations. The Quarterly Journal of Economics, 108(3), 577–598.
+
+Kessler, M.M. (1963). Bibliographic coupling between scientific papers. American Documentation, 14(1), 10-25.
+
+Krugman, P.R. (1991). Geography and trade. Cambridge: MIT press.
+
+Kuznets, S. (1962). Inventive activity: Problems of definition and measurement. In The rate and direction of inventive activity: Economic and social factors (pp. 19-52). Princeton, NJ: Princeton University Press.
+
+Lampe, R. (2012). Strategic citation. Review of Economics and Statistics, 94(1), 320-333.
+
+Lanjouw, J., & Schankerman, M. (2001). Characteristics of patent litigation: A window on competition. RAND Journal of Economics, 32(1), 129-151.
+
+Lemley, M., & Sampat, B. (2012). Examiner characteristics and patent office outcomes. Review of Economics and Statistics, 94(3), 817-827.
+
+Lerner, J., & Seru, A. (2015). The use and misuse of patent data: Issues for corporate finance and beyond. Harvard Business School Mimeo. Cambridge: MA.
+
+Li, R., Chambers, T., Ding, Y., Zhang, G., & Meng, L. (2014). Patent citation analysis: Calculating science linkage based on citing motivation. Journal of the Association for Information Science and Technology, 65(5), 1007–1017.
+
+Li, X., Chen, H., Huang, Z., & Roco, M.C. (2007). Patent citation network in nanotechnology (1976–2004). Journal of Nanoparticle Research, 9(3), 337–352.
+
+Liu, Y., Hseuh, P.Y., Lawrence, R., Meliksetian, S., Perlich, C., & Veen, A. (2011). Latent graphical models for quantifying and predicting patent quality. In Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 1145–1153). August 21–24, 2011, San Diego, California, USA.
+
+Liu, D.R., & Shih, M.J. (2011). Hybrid-patent classification based on patent-network analysis. Journal of the American Society for Information Science and Technology, 62(2), 246-256.
+
+Lukach, R., & Lukach, M. (2007). Ranking USPTO patent documents by importance using random surfer method (pagerank). Available at SSRN: http://ssrn.com/abstract=996595.
+
+Marco, A.C. (2007). The dynamics of patent citations. Economics Letters, 94(2), 290–296.
+
+Maurseth, P.B., & Verspagen, B. (2002). Knowledge spillovers in Europe: A patent citations analysis. The Scandinavian Journal of Economics, 104(4), 531–545.
+
+Mehta, A., Rysman, M., & Simcoe, T. (2010). Identifying the age profile of patent citations: New estimates of knowledge diffusion. Journal of Applied Econometrics, 25(7), 1179–1204.
+
+Meyer, M. (2000). What is special about patent citations? Differences between scientific and patent citations. Scientometrics, 49(1), 93-123.
+
+Michel, J., & Bettels, B. (2001). Patent citation analysis. A closer look at the basic input data from patent search reports. Scientometrics, 51(1), 185-201.
+
+14 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY-Month 2017 DOI: 10.1002/asi
+
+---
+
+Mina, A., Ramlogan, R., Tampubolon, G., & Metcalfe, J.S. (2007). Mapping evolutionary trajectories: Applications to the growth and transformation of medical knowledge. Research Policy, 36(5), 789–806.
+
+Moser, P., & Nicholas, T. (2004). Was electricity a general purpose technology? The American Economic Review, 94(2), 388-394.
+
+Moser, P., Ohmstedt, J., & Rhode, P.W. (2014). Patent citations and the size of patented inventions - Evidence from Hybrid Corn (p. 48). Available at SSRN: http://ssrn.com/abstract=1888191.
+
+Narin, F. (1995). Patents as indicators for the evaluation of industrial research output. Scientometrics, 34(3), 489-496.
+
+Narin, F., & Noma, E. (1985). Is technology becoming science? Scientometrics, 7(3-6), 369-381.
+
+Narin, F., Noma, E., & Perry, R. (1987). Patents as indicators of corporate technological strength. Research Policy, 16(2), 143-155.
+
+Nicholas, T. (2008). Does innovation cause stock market runups? Evidence from the great crash. The American Economic Review, 1370-1396.
+
+Packalen, M., & Bhattacharya, J. (2015). New ideas in invention. National Bureau of Economic Research Working Paper 20922.
+
+Palangkaraya, A., Webster, E., & Jensen, P.H. (2011). Misclassification between patent offices: Evidence from a matched sample of patent applications. Review of Economics and Statistics, 93(3), 1063-1075.
+
+Roach, M., & Cohen, W.M. (2013). Lens or prism? Patent citations as a measure of knowledge flows from public research. Management Science, 59(2), 504–525.
+
+Sampat, B.N. (2010). When do applicants search for prior art? Journal of Law and Economics, 53(2), 399-416.
+
+Schmoch, U. (1993). Tracing the knowledge transfer from science to technology as reflected in patent indicators. Scientometrics, 26(1), 193-211.
+
+Schmookler, J. (1966). Invention and economic growth. Cambridge: Harvard University Press.
+
+Sonn, J.W., & Storper, M. (2008). The increasing importance of geographical proximity in knowledge production: An analysis of US patent citations, 1975-1997. Environment and Planning A, 40, 1020-1039.
+
+Takahiro, M., Nagaoka, S., & Naito, Y. (2015). Effects of stronger disclosure rule on applicants' behavior and on examination efficiency: Evidence from Japan. Paper presented at EPIP 2015 conference , 2–3 September 2015, University of Glasgow, UK.
+
+Thompson, P. (2006). Patent citations and the geography of knowledge spillovers: Evidence from inventor-and examiner-added citations. The Review of Economics and Statistics, 88(2), 383–388.
+
+Thompson, P., & Fox-Kean, M. (2005). Patent citations and the geography of knowledge spillovers: A reassessment. American Economic Review, 95, 450–460.
+
+Tijssen, R.J. (2002). Science dependence of technologies: Evidence from inventions and their inventors. Research Policy, 31(4), 509-526.
+
+Trajtenberg, M. (1990a). Economic analysis of product innovation: The case of CT scanners (Vol. 160). Harvard University PressCambridge: MA.
+
+Trajtenberg, M. (1990b). A penny for your quotes: Patent citations and the value of innovations. The Rand Journal of Economics, 21(1), 172–187.
+
+Trajtenberg, M., Henderson, R., & Jaffe, A. (1997). University versus corporate patents: A window on the basicness of invention. Economics of Innovation and New Technology, 5(1), 19-50.
+
+Verspagen, B. (2007). Mapping technological trajectories as patent citation networks: A study on the history of fuel cell research. Advances in Complex Systems, 10(1), 93–115.
+
+von Graevenitz, G., Wagner, S., & Harhoff, D. (2011). How to measure patent thickets—A novel approach. Economics Letters, 111(1), 6–9.
+
+von Wartburg, I., Teichert, T., & Rost, K. (2005). Inventive progress measured by multi-stage patent citation analysis. Research Policy, 34(10), 1591–1607.
+
+Wagner, S., Hoisl, K., & Thoma, G. (2014). Overcoming localization of knowledge—The role of professional service firms. Strategic Management Journal, 35(11), 1671–1688.
+
+Webb, C., Dernis, H., Harhoff, D., & Hoisl, K. (2005). Analysing European and international patent citations: A set of EPO patent database building blocks. OECD STI Working Paper 2005/09.
+
+Youtie, J., Iacopetta, M., & Graham, S. (2008). Assessing the nature of nanotechnology: Can we uncover an emerging general purpose technology? The Journal of Technology Transfer, 33(3), 315-329.
+
+Younge, K, & Kuhn, J. (2015). Patent similarity: A vector space model. Mimeo, Ecole polytechnique fédérale de Lausanne, Lausanne, Switzerland.
+
+Ziedonis, R.H. (2004). Don't fence me in: Fragmented markets for technology and the patent acquisition strategies of firms. Management Science, 50(6), 804-820.
+
+## Appendix A
+
+Figure A1 plots the yearly number of scientific articles listed in Google Scholar that contain the term “ patent citation ” (blue line), and the number of articles citing the NBER patent citation data file described in Hall et al. (2001) (red dashed line). One can reasonable assume that the latter group of articles forms a subset of the former group.
+
+![Figure](figures/figure_031.png)
+
+FIG. A1. Number of scientific articles listed in Google Scholar. Notes: HJT refers to Hall et al. (2001). [Color figure can be viewed at wileyonlinelibrary.com]
+
+JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2017
+
+
+DOI: 10.1002/asi
+
+
+1
+

processed/00024_W2166565848/document.md

@@ -0,0 +1,577 @@
+BU School of Law
+
+# THE PRIVATE AND SOCIAL COSTS OF PATENT TROLLS
+
+Boston University School of Law Working Paper No. 11-45 (September 19, 2011) Revision of November 9, 2011
+
+James Bessen Boston University School of Law
+
+Jennifer Ford Boston University School of Law
+
+Michael J. Meurer Boston University School of Law
+
+This paper can be downloaded without charge at:
+
+http://www.bu.edu/law/faculty/scholarship/workingpapers/2011.html
+
+Electronic copy available at: http://ssrn.com/abstract=1930272
+
+---
+
+# The Private and Social Costs of Patent Trolls
+
+Version: November 2011
+
+By James Bessen, Jennifer Ford and Michael J. Meurer*
+
+Abstract: In the past, non-practicing entities (NPEs) — firms that license patents without producing goods — have facilitated technology markets and increased rents for small inventors. Is this also true for today's NPEs? Or are they “ patent trolls ” who opportunistically litigate over software patents with unpredictable boundaries? Using stock market event studies around patent lawsuit filings, we find that NPE lawsuits are associated with half a trillion dollars of lost wealth to defendants from 1990 through 2010, mostly from technology companies. Moreover, very little of this loss represents a transfer to small inventors. Instead, it implies reduced innovation incentives and a net loss of social welfare.
+
+Keywords: patent, litigation, litigation cost, non-practicing entities, software patents
+
+JEL Classifications: 031, 034, K41
+
+*Boston University School of Law. Thanks to Colleen Chien, Dennis Crouch and Michael Risch for comments. Thanks to research assistance from Tim Layton, data from Patent Freedom and some support from the Coalition for Patent Fairness. A version of this working paper will be published in Regulation.
+
+Contact: jbessen@bu.edu
+
+Electronic copy available at: http://ssrn.com/abstract=1930272
+
+---
+
+2-Troll - 10/11
+
+## Executive Summary
+
+Firms that license patents without producing goods—“ non-practicing entities” (NPEs)— have historically facilitated technology markets and increased the profits that small inventors earn from their inventions.
+
+But a self-described new crop of NPEs has emerged that asserts patents and litigates them on an unprecedented scale, involving thousands of defendants every year in hundreds of lawsuits. Do these litigating NPEs improve markets for technology and increase incentives for small inventors? Or are they “patent trolls” who exploit weaknesses in the patent system?
+
+This paper makes several findings about this litigation. First, by observing what happens to a defendant's stock price around the filing of a patent lawsuit, we are able to assess the effect of the lawsuit on the firm's wealth, after taking into account general market trends and random factors affecting the individual stock. We find that NPE lawsuits are associated with half a trillion dollars of lost wealth to defendants from 1990 through 2010. During the last four years the lost wealth has averaged over $ 80 billion per year. These defendants are mostly technology companies who invest heavily in R & D. To the extent that this litigation represents an unavoidable business cost to technology developers, it reduces the profits that these firms make on their technology investments. That is, these lawsuits substantially reduce their incentives to innovate.
+
+Second, by exploring publicly listed NPEs, we find that very little of this loss of wealth represents a transfer to inventors. This suggests that the loss of incentives to the defendant firms is not matched by an increase in incentives to other inventors.
+
+Third, the characteristics of this litigation are distinctive: it is focused on software and related technologies, it targets firms that have already developed technology, and most of these lawsuits involve multiple large companies as defendants. These characteristics suggest that this litigation exploits weaknesses in the patent system. In our book Patent Failure , we argue that patents on software and business methods are litigated much more frequently because they have “ fuzzy boundaries. ” The scope of these patents is not clear, they are often written in vague language, and technology companies cannot easily find them and understand what they claim. It appears that much of the NPE litigation takes advantages of these weaknesses.
+
+We conclude that the loss of billions of dollars of wealth associated with these lawsuits harms society. While the lawsuits increase incentives to acquire vague, over-reaching patents, they decrease incentives for real innovation overall.
+
+Electronic copy available at: http://ssrn.com/abstract=1930272
+
+---
+
+3-Troll - 10/11
+
+## 1 Introduction
+
+In 2010, operating companies in the US found themselves in lawsuits initiated by nonpracticing entities (NPEs) more than 2,600 times, over five times more often than in 2004 (Patent Freedom 2011). Is this a good thing or a bad thing?
+
+NPEs are firms that do not produce goods, rather they acquire patents in order to license them to others. $^1$ In principle, NPEs can perform the socially valuable function of facilitating markets for technology. Some inventors lack the resources and expertise needed to successfully license their technologies or, if necessary, to enforce their patents. NPEs provide a way for these inventors to earn rents that they might not otherwise realize, thus providing them with greater incentives to innovate. For example, economic historians find evidence of a robust market for technology during the nineteenth century that allowed individual inventors to earn returns on their inventions in the era before the rise of the large R & D laboratories (Lamoreaux and Sokoloff 1999). $^2$ Optimists argue that the current crop of NPEs perform a similar function and should not be discouraged (Hosie 2008, McDonough 2006, Shrestha 2010, Myhrvold 2010, Morgan 2008).
+
+On the other hand, the recent surge in NPE-related litigation may be more insidious. Critics, including many technology firms, compare these NPEs to the mythical trolls who hide under bridges built by other people, unexpectedly popping up to demand payment of tolls (see, for example, Temple 2011). The critics call these NPEs “patent trolls,” claiming that they buy up vaguely worded patents that can be construed to cover established technologies and use them opportunistically to extract licensing fees from the real innovators. Indeed, there has been a general and dramatic rise in patent litigation that some analysts attribute to rapid growth in the number of patents with unclear or unpredictable boundaries (Bessen and Meurer 2008, FTC
+
+1A wide variety of non-practicing firms engage in patent markets including patent brokers, consultants, auctioneers, and more (see Yanagisawa and Guellec 2009 for an overview). Our focus is on non-practicing firms that assert and litigate patents.
+
+2 See also Magliocca (2007) about less socially beneficial activities of nineteenth century NPEs.
+
+---
+
+4-Troll - 10/11
+
+2011) . To the extent that the recent NPEs opportunistically assert “fuzzy patents” against real technology firms, they can decrease the incentives for these firms to innovate. Innovators deciding to invest in new technology have to consider the risk of inadvertent infringement as a cost of doing business. This risk reduces the rents they can expect to earn on their investment and hence decreases their willingness to invest..
+
+Using empirical evidence, this paper investigates the effect of the current crop of NPE litigation on innovation incentives and on social welfare. We begin by estimating the private losses to publicly listed companies who are defendants in NPE patent litigation by measuring the reaction of the defendant firm's share price during the days following the filing of the lawsuit. $^3$ Using a database of patent lawsuits collected by Patent Freedom (2011), we perform 4,114 of these event studies from 1990 through 2010. In theory, investors respond to the news of a lawsuit filing by reducing their expectations of future earnings for the defendant firm. This reduction should reflect all the costs the firm faces from the suit, including lost business, fees paid to settle the case, etc., depending on how investors expect the suit to be resolved. Investors also consider the loss or delay of profits from future opportunities. The total change in expected profits is reflected by a drop in the share price.
+
+Of course, other events also affect the share price on any given day, including events that affect the market generally and idiosyncratic events that affect the firm being studied. We use standard methods to control for the effect of the market and we average over a large number of lawsuits to filter out random idiosyncratic price changes. This allows us to estimate the average percentage change in the defendant's stock price for each lawsuit filing and the change in market capitalization of outstanding common stock. Aggregating the change in market capitalization over two decades, we find that the aggregate loss of wealth to these firms exceeds half a trillion dollars. Over the last four years, the loss of wealth exceeds $ 83 billion per year. It is possible, of
+
+3 In this paper, we use the term "defendant" to refer to the firm against which the NPE is asserting a patent. In some cases, this firm will technically be the plaintiff in a legal action seeking a declaratory judgment.
+
+---
+
+5-Troll- 10/11
+
+course, that the stock market initially overreacts to news of a lawsuit filing, possibly making our estimates overstated. We present evidence below to argue that this does not likely bias our results significantly.
+
+This private loss might seem surprisingly large, but it does not necessarily mean that this litigation harms society. The effect on society depends on two considerations. First, there is a static effect on net social welfare. To the extent that litigation involves socially wasteful activity, such as a diversion of firm resources from production to litigation support, it reduces social welfare. Such activity implies a ��� deadweight ” loss. On the other hand, to the extent that the losses just represent transfers of wealth from one party to another — perhaps from large defendants to independent inventors — then the static effect on social welfare could be neutral.
+
+Second, there is a dynamic effect: this litigation could increase or decrease innovation incentives overall, thus affecting future social welfare. The large private losses seem to imply a disincentive for the defendants, who are largely technology firms after all. But perhaps transfers to the patent holders constitute a positive incentive to them that more than compensates for the disincentives imposed on the defendant firms. Then the dynamic effect could be to increase innovation incentives overall.
+
+Some general evidence leans against such an optimistic evaluation. The literature on litigation commonly finds that the loss of wealth experienced by defendants is, in fact, largely a deadweight social loss; little of it flows to the plaintiffs (Bhagat and Romano 2002) . Moreover, the large magnitude of lost wealth in these patent cases seems hard to reconcile with a story of transfers to independent inventors — in recent years the losses comprise a significant fraction of total US R & D spending. If these losses were offset by massive transfers to independent inventors, we think we would have heard or read reports documenting this bonanza and a corresponding surge in research activities by small inventors. There is little evidence that NPE litigation has produced massive transfers to independent – or any other sort of – inventor.
+
+---
+
+6-Troll - 10/11
+
+Nevertheless it is helpful to look specifically at evidence of the wealth actually transferred to NPEs and to inventors as a result of NPE litigation. Using the financial statements of publicly listed NPE firms, we obtain upper bound estimates on these transfers. We find that relatively little of the wealth lost by defendant firms shows up as a transfer to NPEs and relatively little of the funds flowing to NPEs is transferred to outside inventors. It appears that there are a lot of big losers in NPE litigation and few beneficiaries.
+
+These findings allow us to draw some conclusions about the effect of the recent surge in NPE litigation on markets for technology, how the current crop of NPEs are different from those in the past, and how this affects innovation incentives.
+
+### 1.1 Literature Review
+
+Various sorts of NPEs have long played an important role in technology markets (see Arora et al. 2004). For example, Arora (1997) documents the pivotal role that specialized engineering firms play in the refining and petrochemical industries by licensing their technology. Several commentators have argued that today's NPEs play a similar role in facilitating markets (McDonough 2006, Hosie 2008, Morgan 2008, Myhrvold 2010). It is crucial to note, however, that the firms studied by Arora provided valuable technological information as well as patent and trade secret licenses to licensees. It appears that the current crop of NPEs usually offers naked patent licenses after the technology in question has already been developed (FTC 2011). Little empirical evidence has been advanced to show whether today's NPEs are providing enhanced incentives to small inventors or whether NPE litigation is inhibiting innovation.
+
+What evidence we have supports the view that today's NPE are different from the specialized engineering firms operating in chemical industries. Consider the following distinctive aspects of current NPE litigation:
+
+- 1. The scale of litigation. While Ball and Kesan (2008) show that NPEs accounted for
+---
+
+7-Troll - 10/11
+
+only about 5 % of patent litigation in 2000-2002, Chien finds (2009) that NPEs account for 17 % of high tech patent lawsuits. Patent Freedom (2011) finds that NPEs account for 16 % of all patent lawsuits in 2009. This amounts to hundreds of lawsuits per year.
+
+2. Many of these lawsuits involve multiple defendants (Chien 2009), making the effective impact greater. The lawsuits involve thousands of defendants per year.
+
+3. Much of this litigation concerns software patents, including business process patents. Chien finds that 90 % of the high tech NPE lawsuits involve software or finance patents. Allison et al. (2010) study patents litigated multiple times and find that software patents account for 94 % of the lawsuits.
+
+4. These lawsuits tend to happen long after the initial patent application. Allison et al. (2009) find that the patents in these lawsuits are much more likely to have multiple continuing applications, allowing for claims to be modified long after the initial application. Risch (2012) finds that the mean NPE lawsuit occurs 8 years after the patent was issued. $^4$ Love (2010) finds that NPEs are far more likely to enforce old patents than practicing entities. The long delays suggest that in many cases these patents are not asserted until other firms actually develop the technology.
+
+These findings suggest that today's NPEs are distinct in some ways, however, that does not really tell us much about their effect on innovation. Shrestha (2010) compares the characteristics of patents in NPE lawsuits to a sample of other patents (see also Allison et al. 2009, Risch 2012, Fischer and Henkel 2011). Finding, for example, that NPE patents receive more citations than other patents, Shrestha concludes that many NPEs hold “high value” patents and are therefore good for innovation. $^5$ Unfortunately, this conclusion does not logically follow. While it is true that higher value patents tend to receive more citations, this is a rather weak
+
+4 And this underestimates the lag because it only considers lawsuits filed by 2010.
+
+5 Shrestha also looks at non-self citations and indices of originality and generality.
+
+---
+
+8 - Troll - 10/11
+
+correlation and many factors other than value can influence citations received, including selfcitations (Bessen 2008) . This correlation does not imply that NPE litigated patents are more valuable just because they have more citations. $^6$ Moreover, even if these patents are valuable, it is important to remember that the ultimate question is whether or not enforcement of these patents provides a net incentive for innovation. $^7$ This paper looks at the actual transfer of wealth to inventors from NPE patent litigation.
+
+Shrestha also looks at win rates for those lawsuits that do proceed to a final judgment, finding that NPEs have similar win rates to other patent plaintiffs. Based on this, Shrestha concludes that these lawsuits are not “frivolous.” However, this finding is based on a very small sample and the lawsuits that proceed to final judgment are not necessarily representative of all of the lawsuits filed. Moreover, Risch (2012) finds much higher rates of invalidation of NPE patents brought to judgment (54 % had no valid claims while an additional 44 % had some invalid claims). Allison et al. (2010) look at win rates for a larger sample of the most-litigated patents and find that plaintiff win rates are much lower than for other patent litigation. But even so, this does not directly measure how harmful the litigation is to innovation or to social welfare. Using extensive event studies, this paper measures the private losses that result from NPE litigation and relates this to possible social losses.
+
+The event study methodology has been used before to study litigation, beginning with Cutler and Summers (1988) in the context of litigation over a merger. Several papers have performed event studies of patent litigation, both the event of the initial filing and the terminating event (settlement, judgment or verdict), including small sample studies by Bhagat, et al. (1994), Lerner (1995), Bhagat et al. (1998), Lunney (2004), Haslem (2005), and a large sample study by
+
+6 In effect, Shrestha is arguing: A) Valuable patents receive higher citations, and, B) NPE litigated patents receive higher citations, therefore, C) NPE litigated patents are valuable patents. This is a classic logical fallacy.
+
+7 Broad patents that can be credibly asserted against valuable technologies might have enormous private value and at the same time negative social value when they are not disclosed until after the technology was independently developed, and especially when they face a significant risk of invalidity. Such patents might attract a large number of citations, and might also retard innovation.
+
+---
+
+9-Troll - 10/11
+
+Bessen and Meurer (2007). None of these studies looked specifically at NPE litigation.
+
+## 2 Data and Methods
+
+### 2.1 Data Sources
+
+The data for this research comes from two primary sources. The first source is an extensive database of NPE lawsuits generously provided by Patent Freedom, an organization devoted to researching and providing information on NPE behavior and activities. Patent Freedom defines a non-practicing entities as companies that “ do not practice their inventions in products or service, or otherwise derive a substantial portion of their revenues from the sale of products and services in the marketplace. Instead, NPEs seek to derive the majority of their income from the enforcement of patent rights. ” Since we study litigation, we only focus on those NPEs who file lawsuits ( “ patent assertion entities ” ).
+
+The second data source is the Center for Research in Security Prices (CRSP) US Stock Database, a comprehensive collection of security information. Using these sources, a sample comprised of all instances in which a known NPE sued a publicly traded firm between 1990 and October 2010 was constructed. This was done by first matching defendant names with a previously constructed list of public domestic firms and subsidiaries using a software program, and then manually reviewing the resulting list and updating matches that had been either missed by or incorrectly assigned by the software. To assess the validity and coverage of the matches, a random sample of 100 parties was manually checked using corporate websites and CRSP's Company Code Lookup tool. For this sample, while 11 % of parties that were either public companies or their subsidiaries were left unmatched, there were no false positives.
+
+This process yielded a sample of 1,630 lawsuits filed by a NPE against one or more publicly listed defendants. Because many of these lawsuits were filed against multiple defendants, the total number of events in the sample was substantially higher than the number of
+
+---
+
+10 - Troll - 10/11
+
+suits, at 4,114 (for the sample using a 5 day window to measure the returns).
+
+Finally, we linked the data in our sample to Compustat and to data from Derwent Litalert to obtain information on firm characteristics and patents involved in the lawsuits. We also used financial information on publicly listed NPEs from Compustat.
+
+### 2.2 Estimating Cumulative Abnormal Returns
+
+To estimate the impact of a lawsuit filing on the value of a firm, we use event study methodology (see Mackinlay 1997 for a review). In particular, we use the dummy variable method described by Michael Salinger (1992). 8 This assumes that stock returns follow a market model,
+
+$$ (1)  \quad r_{t}=\alpha+\beta r_{t}^{m}+\epsilon_{t}$$
+
+where $r_t$ is the return on a particular stock at time $t$ , $\frac{m}{r_t}$ is the compounded return on a market portfolio, and $\epsilon_t$ is a stochastic error. If an event, such as a lawsuit filing, occurs on day $T$ , then there may be an “ abnormal return ” to the particular stock on that day. This can be captured using a dummy variable,
+
+$$ (2)  \quad r_{t}=\alpha+\beta r_{t}^{m}+\delta I_{t}+\epsilon_{t}$$
+
+where $I_t$ equals 1 if $t{=}T$ and 0 otherwise. Equation (2) can be estimated using OLS for a single event. In practice, this equation is estimated over the event period and also over a sufficiently long pre-event window. In this paper we use a 200 trading-day pre-event window. The coefficient estimate of $\delta$ obtained by this procedure is then an estimate of the abnormal return on this particular stock. For different stocks, the precision of the estimates of $\delta$ will vary depending on how well equation (2) fits the data. The estimated coefficient variance from the regression
+
+8 Salinger shows that this model is mathematically equivalent to the OLS market model described in Brown and Warner (1985) and widely used.
+
+---
+
+11 - Troll - 10/11
+
+provides a measure of the precision of the estimate of the abnormal return.
+
+We want to obtain a representative estimate of the abnormal returns from lawsuit filings for multiple stocks, under the assumption that these represent independent events and that they share the same underlying “ true ” mean. Previous papers estimating abnormal returns from patent lawsuits have simply reported unweighted means for the group of firms. Although the unweighted mean is an unbiased estimator, it is not efficient. Since we are concerned with obtaining the best estimate to use in policy calculations (and not just testing the sign of the mean), we use a weighted mean to estimate the “ average abnormal return, ” where the weight for each observation is proportional to the inverse of the variance of the estimate of $\delta$ for that firm. $^9$
+
+When we test our means against the null hypothesis that the true mean is zero, we report both the significance of $t$ -tests using the weighted mean and also the significance of the $Z$ statistic (see Dodd and Warner 1983), a widely used parametric test of significance that incorporates the variation in precision across events. $^10$ In any case, the significance test results are closely similar as are those of some non-parametric tests.
+
+Finally, (2) describes the abnormal return for a single day. It is straightforward to design dummy variables to estimate a “cumulative abnormal return” (CAR) over an event window consisting of multiple consecutive days. In the following, for instance, if the suit is filed on date $t{=}T$ , then we may use a window from day $T{-}1$ to $T{+}4$ .
+
+## 3 Empirical Findings
+
+### 3.1 Summary statistics
+
+Some characteristics of defendant firms in our sample are reported in Table 1. These are, on average, large firms. Almost two thirds of the firms are technology firms, including software
+
+9 In any case, we find that for our entire sample, the weighted mean is quite close to the unweighted mean and also to the median.
+
+10 The Z statistic is a joint test of the individual firm t-tests. We use a robust version described in Kramer (2001).
+
+---
+
+12 - Troll - 10/11
+
+and communications firms, and these firms, on average, spend a lot on R & D and have very substantial intangible assets. A significant number of financial, retail and wholesale firms are also represented. And these firms are typically subject to multiple NPE lawsuits.
+
+Table 2 shows that most of the NPE disputes involve multiple defendants, either in the same suits or from multiple suits filed by the NPE on the same day. $^11$ The number of publicly listed defendants mostly range between two and nine defendants (median of 5). Only 17 % of the defendants were the sole defendant listed. This contrasts sharply with other patent litigation where 85 % of defendants are solo (Bessen and Meurer 2007).
+
+Another difference is the distribution of these patents across technology classes. Looking at the main patent listed in Derwent, about 62 % of the patents are software patents, using the technology class categorization used in Bessen (2011). Using the NBER categorization (Hall et al. 2001), 75 % of the patents are in computer and communications technology. Thus this sample shows the same concentration of NPE litigation in software and related technologies as in earlier studies. Both this technological concentration and the prevalence of multiple defendants are important for interpreting the nature of the current crop of NPEs.
+
+### 3.2 Estimates of cumulative abnormal returns
+
+Table 3 reports basic estimates of cumulative abnormal returns (CARs) for the sample of NPE defendants. Columns 1 and 2 report the weighted mean (with standard error) and median values. The first row shows the results using a five day event window that starts one day before the lawsuit filing and continues through the fourth day after. The mean loss is 0.32 % and the median loss is 0.52 % .
+
+One concern is that this estimate of lost value might reflect a temporary over-reaction on the part of investors. Given that there are now hundreds of these troll lawsuits every year, it is
+
+11 The numbers are also high if we restrict the defendants to the same suit filings.
+
+---
+
+13 - Troll - 10/11
+
+hard to understand why investors would consistently over-react and never learn from their mistakes. Nevertheless, a persistent over-reaction would be noticed by arbitrageurs who would then come in, buy the artificially low stock and thus drive the price up to a more accurate level. If it took some time for arbitrageurs to enter, the price we observe during the five day event window might be artificially low, making our estimate of losses too high.
+
+One way to check this is to look at a longer event window to see if the stock showed evidence of recovery over the subsequent month or so. The second row reports results for a comparable analysis using a 25 day window. If it took some time before wiser investors arbitraged the stock, then we should see some evidence of a price correction within this longer window. Instead, the CARs in this row are slightly larger (more negative) than those in the five day window. This suggests that the initial loss of wealth was not an overreaction by investors that was subsequently corrected, at least not within 25 days. Because the longer window has larger standard errors as a result of the measurement technique, we use the sample with the five day event window for most of the remaining analysis.
+
+Perhaps, instead, the stock price stays artificially low until the lawsuit is resolved. This might be the case if investors react to the uncertainty of the lawsuit, demanding a higher return on investment until the uncertainty is resolved. If this were the case, then we should see an increase in the stock price at the announcement that the suit was settled. However, two event studies of lawsuit settlements find no such positive correction on average, suggesting that investors overall appear to anticipate settlement correctly, pricing it in to the share value. $^12$ Thus this theory, too, seems difficult to reconcile with the evidence. While we accept the idea that investors do not always act rationally, we have found no explanation consistent with the evidence for why investors should persistently over-react to lawsuit filings.
+
+The estimated CARs are substantially smaller than those found in the study of all patent
+
+12 Haslem (2005) finds a statistically significant decrease in the stock price on settlement. Bhagat et al. (1998) find a CAR that is not significantly different from zero.
+
+---
+
+14 - Troll - 10/11
+
+lawsuits involving publicly listed firms from 1984 to 1999 by Bessen and Meurer (2007). The third row shows the CARs for defendant firms from that study and the fourth row shows the CARs from solo defendant firms in that study. We parse out the results in the fourth row to provide the most relevant comparison to the NPE lawsuits in this study. Most NPE lawsuits in our current study have multiple defendants (83 % ). Most of the lawsuits in our earlier studied involved a single defendant (85 % ); we suspect that almost all of those lawsuits do not involve an NPE plaintiff. The mean CAR for all single-defendant lawsuits is nearly twice as large as the mean CAR reported for the five day window in the NPE sample. This difference is also statistically significant. $^13$
+
+The NPE CARs are also much smaller than those reported in the previous literature on patent litigation event studies. For example, Bhagat et al. (1998) study 33 defendants of patent lawsuits announced in the Wall Street Journal. They find a mean CAR of -1.50 % , nearly five times larger than the estimate here. Studying 26 biotech firms, Lerner (1995) found a 2.0 % reduction in the wealth of the defendants and plaintiffs combined.
+
+### 3.3 Why do NPE lawsuits cause smaller percentage losses?
+
+One clear reason that the NPE lawsuits have lower CARs than in previous studies is that the sample of defendants in the NPE lawsuits is very different from the samples in the earlier studies. Some of those studies found much larger losses but used highly select small samples of lawsuits that had been announced in the Wall Street Journal or Dow Jones News Service. Bessen and Meurer (2007) show that patent lawsuits announced in the Wall Street Journal tended to involve companies with greater capital per employee and higher stock market betas. These factors might be directly related to larger percentage losses on the announcement of a lawsuit.
+
+The large sample of lawsuits involving publicly listed firms in Bessen and Meurer (2007)
+
+13 Using one-tailed t-tests, allowing unequal variances between the sub-groups and calculating the degrees of freedom using Satterthwaite's approximation (1946), P = .070.
+
+---
+
+15 - Troll - 10/11
+
+were not necessarily announced, but these, too, show larger percentage losses than in the current sample of NPE lawsuits, although not so much larger. The NPE sample of public firms differs from that sample in two important ways: NPE lawsuits tend to involve larger defendants and multiple defendants.
+
+Although larger defendants tend to have smaller CARs (Bessen and Meurer 2007), size related differences cannot directly explain much of the difference in the CARs between the samples. The difference in the CARs between small and large firms is simply not large enough to account for the difference in the NPE sample and these small firms only make up 14 % of the NPE sample in any case. $^14$
+
+Nevertheless, the large size of the defendants in the NPE lawsuits and the fact that so many of these lawsuits involve multiple defendants changes the economics of litigation in an important way: in these circumstances, litigation might still be credible for plaintiffs who have a low probability of winning. A lawsuit only poses a credible threat if the plaintiff's expected gains from winning exceed the costs from litigating. The expected gains are the ex ante probability of winning times the conditional benefits of winning. Normally, a lawsuit with a low probability of winning does not pose a credible threat. However, when a patent has a chance of being interpreted broadly so that it reads on the business of multiple large companies, the payoff to winning might be so large that the threat of a lawsuit is credible even if the probability of winning is low.
+
+This provides another possible explanation for lower percentage losses found in NPE lawsuits: the plaintiffs in a substantial portion of NPE lawsuits might have low probabilities of winning at court, hence these lawsuits will cause smaller losses to defendants, all else equal. Because many of these suits might involve aggressive interpretations of patent scope, allowing the claims to read on many defendants, they might have lower probabilities of winning, but still
+
+14 In an unreported result from the 2008 study, the mean CAR for solo defendants that had more than 500 employees was -.56% (.18%), just slightly smaller than the return listed in the fourth row of Table 3.
+
+---
+
+16 - Troll - 10/11
+
+provide credible threats because of the multiple defendants. This explanation is supported by Allison et al. (2011) who find that NPE suits with multiple defendants are more likely to settle and, when they do go to trial, the plaintiffs are much more likely to lose (but see Shrestha 2010). This explanation is thus plausible, however, our evidence for it is not conclusive.
+
+### 3.4 Loss of wealth
+
+Nevertheless, just because the percentage loss of defendant firms is smaller in NPE lawsuits, this does not imply that the loss of wealth is small. Using the CAR estimates, we can calculate the loss of wealth that occurs upon a lawsuit filing. Columns 4 and 5 of Table 3 show the mean and median loss of wealth calculated by multiplying the mean CAR by each firm's capitalization. $^15$ The mean wealth lost per lawsuit is $ 122 million in 2010 dollars and the median loss is $ 20.4 million. These figures are substantially higher than the previous estimates for patent lawsuits of all types found by Bessen and Meurer (2007), shown in row 3. These estimates are, of course, much larger than the direct costs of legal fees. They also include the costs of lost business, management distraction and diversion of productive resources that might result from the lawsuit, possible payments needed to settle the suit, and the reduction in expectations of profits from future opportunities that are forestalled or foreclosed because of the suit.
+
+Investors' expectation of future profits are notoriously volatile. To the extent that one might want to gauge the effect of the lawsuits on current profits while excluding expectations about future profits, it is possible to make some crude adjustments to the above figures. One method is to divide the estimated loss of wealth by the ratio of the market capitalization of the
+
+15 We could, alternatively, calculate the average by summing the estimated loss from each suit, however, that
+
+$$\frac{1}{N}\sum_{i=1}^N(r+e_i)x_i$$ where $N$ is the procedure would provide a less efficient estimate. This alternative estimator is number of firms, $r$ is the true CAR, $e$ is the error in measuring the $i$ th firm's CAR, and $x$ is the $i$ th firm's market $$\frac{1}{N}\left(r+\frac{\sum_{i=1}^N e_i}{N}\right)\sum_{i=1}^Nx_i$$. It is straightforward to show that both are unbiased but that the latter has smaller variance assuming that $e$ and $x$ are uncorrelated.
+
+---
+
+17 - Troll - 10/11
+
+firm's common stock divided by the value of the firm's capital assets. $^16$ This reduces the mean wealth lost to $ 112 million in 2010 dollars. Alternatively, the loss can be divided by the ratio of the total market value of the firm to the value of the firm's capital assets, reducing the mean loss to $ 64 million in 2010 dollars. These figures are also quite substantial and, although investors' expectations of future profits might occasionally be “exuberant,” our basic estimate nevertheless captures the actual loss of wealth related to the lawsuit.
+
+Thus although the NPE CARs are lower than the CARs for other lawsuits, the mean loss per lawsuit is larger because the market capitalization of the NPE defendants is that much larger. This, combined with the tendency of NPE lawsuits to involve multiple defendants means that these suits have an outsized impact on firm wealth. Aggregating over the sample (column 6), shows that NPE lawsuits from 1990 through October 2010 are responsible for over half a trillion dollars in lost wealth (in 2010 dollars). From 2007 through October 2010, the losses average over $ 83 billion per year in 2010 dollars, over a quarter of US industrial R & D spending per annum. Moreover, because this total is only for publicly listed firms, it likely understates the true loss of wealth resulting from NPE lawsuits.
+
+Whatever the theoretical and historical role of NPEs might be in facilitating markets for technology, it is clear that the current crop of NPE litigation is responsible for an unprecedented loss of wealth. The next section looks at whether this private loss of wealth to the defendants is also a loss to society or not.
+
+### 3.5 Transfers
+
+As discussed in the Introduction, these private losses might or might not correspond to social losses. Litigation incurs static social losses when it involves socially wasteful activity. Aside
+
+16 For the capital assets, we use the inflation-adjusted value of the aggregate sum of accounting assets and R&D. For details on the computation of these quantities, see Bessen (2009). This adjustment implicitly gives the amount of investment that would be needed to restore the firm to its value before the lawsuit. The alternative calculation assumes that the lawsuit does not reduce the market value of the firm aside from the firm's common stock.
+
+---
+
+18 - Troll - 10/11
+
+from direct legal fees, litigation often involves a diversion of management resources away from productive activity. It may also involve a loss of consumer welfare. For example, preliminary injunctions can shut down production and sales while the litigation pends. Even without a preliminary injunction, customers may stop buying a product. And the threat of final injunction might require the defendant to drastically rework its product or even abandon it. Frequently, products require customers to make complementary investments; they may not be willing to make these investments if a lawsuit poses some risk that the product will be withdrawn from the market. Furthermore, patent owners can threaten customers and suppliers with patent lawsuits because patent infringement extends to every party who makes, uses, or sells a patented technology without permission, and sometimes to those who participate indirectly in the infringement.
+
+These social losses might be offset if NPE litigation acts like an investment in a reputation for toughness that deters future piracy. We doubt this is the case. There is simply no evidence that a significant number of defendants in NPE suits are pirates; later we discuss evidence showing that they are mostly inadvertent infringers. Furthermore, NPE litigation is rising over time, not declining as it should if the reputational story were true.
+
+A more important consideration is the extent to which private losses arise from transfers of wealth to other parties that do not incur a static loss of social welfare. When defendants make payments to NPEs to settle lawsuits or subsequent to legal judgments, the private loss to the defendant is not socially wasteful. To what extent do the half trillion dollars in private losses correspond to such expected transfers?
+
+To explore transfers to NPEs and, in turn, transfers from NPEs to independent inventors, we assembled a list of NPE firms in our database that are publicly listed. We identified 14 firms (see list in the Appendix). These firms account for 574 litigation events in our data, about 14 % of the total. The aggregate losses to the defendants in these lawsuits from 2000 through October
+
+---
+
+19 - Troll - 10/11
+
+2010 total $87.6 billion in 2010 dollars, about 17% of the total in our database.
+
+How much of this loss represents a transfer to the NPEs? Table 4 shows the cumulative flow of several financial variables over this same time period. Total revenues over these years come to $ 7.6 billion, about 9 % of the total loss to defendants. Revenues necessarily overstate any transfers from the defendants to the NPEs because they also include revenues from firms that are not involved in litigation and from private firms. Nevertheless, it is quite clear that most of the defendants' private loss is not a transfer to NPEs.
+
+Another possible transfer occurs to the defendant's competitors. To the extent that patent litigation causes customers to select a rival product or service, some of the lost business captured in the above calculations represents a transfer to rival firms. Of course, because the NPEs sue multiple parties, it happens frequently that a firm and its rivals are sued at the same time, so that no such transfer would occur. This provides us a simple test of the magnitude of potential transfers to rivals: if such transfers are substantial, we should see smaller CARs when a firm and its rival are sued than in cases where rivals are not sued. We identified 1,914 events (47 % of the events) where a firm was sued along with another firm in the same SIC 3-digit industry. However, the CARs for these events were slightly higher than in those cases where a rival firm was not also sued. $^17$ Thus this test is inconsistent with substantial transfers to rivals.
+
+Another transfer occurs to the lawyers, expert witnesses, etc. involved in lawsuits. Estimates of legal costs from Bessen and Meurer (2007) suggest that these transfers cannot be more than a few percent of the loss.
+
+We also conducted event studies of the NPE stocks around the lawsuit filings. The NPE stocks also lost wealth around the lawsuit filings. 18 Although other factors might cause a drop in the plaintiffs' market capitalizations (Bessen and Meurer 2007), this evidence is not consistent
+
+17 The difference was 0.02% with a standard error of 0.17%.
+
+18 The mean CAR was -5.2% with a standard error of 3.1%, significantly different from zero at P = .064. A loss to plaintiffs is frequently observed in the litigation literature.
+
+---
+
+20 - Troll - 10/11
+
+with large transfers of wealth to the NPEs.
+
+In summary, while there are some limited transfers to NPEs and to rivals and lawyers, most of the private losses incurred by defendants in NPE litigation do not appear to be transfers to other parties; presumably, most of the losses correspond to static losses of social welfare.
+
+Of course, NPE litigation might also produce dynamic gains in social welfare if transfers to independent inventors increase innovation incentives. How much of the transfer to NPEs is subsequently transferred to inventors outside of the NPEs? The investment that NPEs make in acquiring patents is included in the accounting category “net cash flow to investing activities.” This figure less capital expenditures is shown in Table 4. Although this figure includes other investments in addition to payments to outside inventors, it is small compared to the defendants' losses: $ 1.7 billion, or about 2 % of the defendants' losses. The investments made in patents are also included in the NPE's intangible assets, although these quantities are amortized. The table also reports intangible assets for fiscal 2010. It is less than $ 600 million, about 1 % of the defendants' losses. Note again that both the intangible assets and the net cash flow to investing activities generate revenues from sources other than our defendants, so these figures might overstate transfers to independent inventors. In any case, we can state that less than 2 % of the defendants' losses could represent a transfer to independent inventors and quite possibly the true figure is much smaller than 2 % . $^19$
+
+Some of the NPEs also conduct their own R & D. Indeed, capitalized R & D investments are included in the intangible assets of the firm. The R & D expense flows are also not large, around 2 % of the loss.
+
+It is likely that the R & D investments and acquisitions from outside inventors will yield value to the NPE firms beyond 2010. To the extent that this is true, all of these figures overstate the extent to which these investments are tied to the defendant losses occurring through 2010.
+
+19 Risch (2012) also finds evidence that NPE litigation does not help inventors raise funds from venture capitalists either.
+
+---
+
+21 - Troll - 10/11
+
+That is, some portion of these investments is related to defendant losses that will be incurred after 2010, so only a portion of the investment can be attributed to a transfer of wealth from the pre-2011 defendants.
+
+Although the transfer to inventors are small, it is still positive. Does this mean that NPE litigation nevertheless increases innovation incentives? There are three reasons to conclude that it does not. First and foremost, the losses to technology firms who are defendants in this litigation are two orders of magnitude larger. These losses imply a very large disincentive to innovation for these firms, firms that spend heavily on R & D. Studies show that the more a firm spends on R & D, the more likely it is to be sued for patent infringement (Bessen and Meurer 2005). Moreover, very rarely are the defendants in these lawsuits found to have actually copied the patented technology (Bessen and Meurer 2008, p. 126, Cotropia and Lemley 2009). Instead, they are inadvertent infringers, if infringers at all. This means that they have to anticipate the risk of future lawsuit-related losses as part of their cost of developing new technology and products. This risk is a disincentive to invest in innovation, and our results find that it is a very large disincentive, much larger than any possible incentives provided by transfers to independent inventors via NPEs. Even if incentives to small inventors were much more fertile than incentives provided to large technology firms — producing two, three or even ten times as many innovations — the incentives flowing to small inventors would not offset the very much larger disincentives imposed on the technology firms.
+
+Second, to the extent that independent inventors benefit by licensing or selling their inventions to large firms, this risk of inadvertent infringement reduces their innovation incentives as well. Because their prospective licensees have to anticipate the risk of an NPE lawsuit, this risk decreases the amount licensees are willing to pay. Thus the very large losses incurred by defendants tend to reduce the market for technology for independent inventors.
+
+Finally, the incentives provided to patent holders by the current crop of NPEs may be the
+
+---
+
+22 - Troll - 10/11
+
+wrong kind of incentives. NPE activity may skew the research agenda of small firms away from disruptive technologies and toward mainstream technology and associated patents that can be asserted against big incumbents. Even worse, small firms are encouraged to divert investment from genuine invention toward simply obtaining broad and vague patents that might one day lead to a credible, if weak, lawsuit.
+
+To summarize, there are a lot of big losers from NPE litigation, while hardly anyone benefits much. The defendant firms and their customers lose while patent holders gain very little by comparison. Even the investors in NPE firms have gained little — these firms barely break even based on their cumulative net income in Table 4 . Apparently, the only real beneficiaries are the lawyers and perhaps the principals of the NPE firms.
+
+## 4 The New Business Model
+
+These findings should be interpreted cautiously. While there are large losses from NPE litigation, not all NPEs today are opportunistic litigators. Nor does this imply that NPEs have not played a more positive role in the past. It is important to understand what is uniquely different about the NPEs who are behind today's litigation surge.
+
+Indeed, today's NPEs tell us they are different. Proponents tell us they are a new breed of company, a new business model, that is misunderstood (McDonough 2006, Myhrvold 2010). They tell us that NPEs are, in fact, good for society because they are creating “a capital market for invention” by buying patents and selling licenses. This helps “turbocharge technological progress” “by realigning market participant incentives, making patents more liquid, and clearing the patent market.”
+
+What, exactly, is new about this business model and what does it mean for innovation? Markets for technology have been around at least since the nineteenth century and studies have documented some of the benefits of these markets (for example, Arora et al. 2004): they allow
+
+---
+
+23 - Troll - 10/11
+
+inventors a way of getting money for their inventions, thus providing them with stronger incentives to invent, and they help spread new technologies to the companies who can commercialize them the best. But most of this literature concerns markets for technology , not markets for patents . There is no evidence the transactions occurring around NPE litigation involve the transfer of technology — news reports and judicial opinions indicate the defendants are already using the technology. Instead, these transactions involve just the transfer of patent rights (and money).
+
+Even so, some advocates hold that NPEs are socially beneficial because they reduce the costs of patent transactions (McDonough 2006) . To the extent that NPEs facilitate the clearance of patent rights before firms invest in technology this is a clear benefit. The patent brokers and auctions facilitate transactions, but that is not obviously true for those NPEs who are primarily involved in asserting and litigating patents. Moreover, to the extent that these NPE transactions occur only after firms invest in technology, any savings in transaction costs has to be offset by the associated dispute costs. We have shown that the litigation losses amount to over half a trillion dollars, so these dispute costs are substantial. No reasonable estimate of the transaction costs of licensing these patents could approach the magnitude of these litigation losses.
+
+The pattern of NPE patent litigation casts further doubt on the view that NPE patent enforcement has any connection to technology transfer. Is it possible that large numbers of innovative firms in case after case are pirating the technology disclosed in NPE patents? Why are the numbers so large? Perhaps the firms have colluded to jointly pirate the technology, or perhaps all of these firms have independently decided to pirate the same technology. Not likely. We think the plausible explanation is that the many firms who end up as defendants in these cases have independently created the invention or derived the claimed technology from some source other than the NPE patent.
+
+Multiple inadvertent infringements are especially likely for a general purpose technology
+
+---
+
+24 - Troll - 10/11
+
+like software. As noted above, NPE lawsuits are concentrated in one technology area, namely, software and software-related patents including business methods. Consequently, this litigation has a disproportionately large effect on firms working with these technologies. A thumbnail calculation suggests that NPEs account for about 41 % of patent litigation involving software patents. 20 So NPE litigation is quite significant for this technology.
+
+Thus the new business model for NPEs is not about licensing patents in general; it is mainly about licensing software patents, including patents on business and financial processes. This is significant because we have argued elsewhere that software patent litigation has risen dramatically because of eroding patent notice and that software patents have been an important contributor to this trend (Bessen and Meurer 2008) . That is, software patents have “ fuzzy boundaries ” : they have unpredictable claim interpretation and unclear scope, lax enablement and obviousness standards make the validity of many of these patents questionable, and the huge number of software patents granted makes thorough search to clear rights infeasible, especially when the patent applicants hide claims for many years by filing continuations. This gives rise to many situations where technology firms inadvertently infringe. And this means that there is a business opportunity based on acquiring patents that can be arguably read to cover existing technologies and asserting those patents, litigating if necessary in order to obtain a licensing agreement. Models by Reitzig et al. (2007) and Turner (2011) show that the patent troll business model only makes economic sense when there is such inadvertent infringement. And the rise in NPE litigation has closely mirrored the rise in software patent litigation (Bessen 2011) . Moreover, fuzzy boundaries can explain why so many NPE lawsuits have multiple defendants: many firms may have reasonably concluded that they did not infringe or the patents were invalid or they may have been unable to find these patents while conducting a clearance search. Later,
+
+20 From Bessen (2011), about 26% of patent lawsuits involve software patents. If 17% of lawsuits involve NPEs and if 62% of NPE lawsuits involve software patents (Table 2) then .17 x .62 / .26 = 41% of software patent lawsuits are filed by NPEs.
+
+---
+
+25 - Troll - 10/11
+
+they encounter an NPE who sues over an aggressively broad interpretation of the patent's scope and validity.
+
+Thus “ fuzzy boundaries ” for software and business method patents enable the rise of this new business model. Large numbers of hidden patents or patents with unpredictable boundaries provide an opportunity to extract rents from technology firms. Further, because NPEs have no operating business, technology firms cannot retaliate with countersuits. Combine this with capital markets to fund the acquisition of patents and to conduct litigation and you get a viable business model. But this is a very different business from the business pursued by those patent brokers, consultants and auctioneers who facilitate markets for technology.
+
+## 5 Conclusion
+
+Firms that buy and license technologies can improve the market for technology and thus improve the innovation incentives for independent inventors. Patent agents and markets for technology have been an important part of the US innovation system since the nineteenth century.
+
+But the role of the current NPEs who assert and litigate patents is something altogether different: it is focused on software and related technologies, it targets firms that have already developed technology, and it is very much about litigation, especially litigation in the special circumstances where multiple large parties can be sued at once. Whatever the general benefits of technology markets, this does not obscure the fact that this particular manifestation involves large amounts of costly litigation. It is hard to believe that markets can be somehow improved by having thousands of lawsuits that incur hundreds of billions of dollars in losses.
+
+We have shown that defendants have lost over half a trillion dollars in wealth—over $ 83 billion per year during recent years—and this has not improved incentives to innovate. While the lawsuits might increase incentives to acquire vague, over-reaching patents, they do not increase incentives for real innovation. The defendants in these lawsuits are firms that already invest a lot
+
+---
+
+26 - Troll - 10/11
+
+in innovation. Their losses make it more expensive for them to continue to do so and it also makes them less willing to license new technologies from small inventors. Meanwhile, independent inventors benefit very little from what the large companies lose.
+
+---
+
+27 - Troll - 10/11
+
+## 6 References
+
+Allison, John R., Mark A. Lemley and Joshua Walker, 2009. "Extreme Value or Trolls on Top? Evidence From the Most-Litigated Patents," University of Pennsylvania Law Review, 158, pp. 101-37.
+
+Allison, John R., Mark A. Lemley and Joshua Walker. 2010. "Patent Quality and Settlement Among Repeat Patent Litigants," Georgetown Law Journal, 99, pp. 677-712.
+
+Arora, Ashish, 1997. "Patents, licensing, and market structure in the chemical industry," Research Policy 26, pp. 391-403.
+
+Arora, Ashish, Andrea Fosfuri, and Alfonso Gambardella, 2004. Markets for Technology: The Economics of Innovation and Corporate Strategy, Cambridge, Ma.: MIT Press.
+
+Ball, Gwendolyn G. & Jay P. Kesan, 2009. "Transaction Costs and Trolls: Strategic Behavior by Individual Inventors, Small Firms and Entrepreneurs in Patent Litigation," Ill. Law & Econ. Papers Series, Research Papers Series No. LE09-005.
+
+Bessen, James, 2008. "The value of U.S. patents by owner and patent characteristics," Research Policy, 37, pp. 932-45.
+
+Bessen, James, 2009. "Estimates of Patent Rents from Firm Market Value," Research Policy 38, pp. 1604-16.
+
+Bessen, James, 2011. "A Generation of Software Patents," Boston Univ. School of Law, Law and Economics Research Paper No. 11-31.
+
+Bessen, James and Michael J. Meurer. 2005. "The Patent Litigation Explosion," B.U.S. L. Law and Economics Working Paper Series, No. 05-18.
+
+Bessen, James and Michael J. Meurer (2006), "Patent Litigation with Endogenous Disputes," American Economic Review, 96, no. 2, pp. 77-81.
+
+Bessen, James and Michael J. Meurer (2007), "The Private Costs of Patent Litigation," Boston Univ. School of Law Working Paper No. 07-08.
+
+Bessen, James and Michael J. Meurer (2008), Patent Failure: How judges, bureaucrats and lawyers put innovators at risk, Princeton, NJ: Princeton University Press.
+
+Bhagat, Sanjai, John M. Bizjak, and Jeffrey L. Coles, 1998. "The shareholder wealth implications of corporate lawsuits," Financial Management 27, pp. 5-27.
+
+Bhagat, Sanjai, James A. Brickley, & Jeffrey L. Coles. 1994. "The Costs of Inefficient Bargaining and Financial Distress: Evidence from Corporate Lawsuits," Journal of Financial Economics, v. 35, p. 221.
+
+Bhagat, Sanjai and Roberta Romano, 2002. "Event Studies and the Law: Part I: Technique and Corporate Litigation," American Law and Economics Review, 4, pp. 141-167.
+
+---
+
+28 - Troll - 10/11
+
+Brown, Stephen J. and Jerold B. Warner. 1985. "Using Daily Stock Returns: The case of event studies," Journal of Financial Economics, v. 14, pp. 3-31.
+
+Chien, Colleen, 2009. "Of Trolls, Davids, Goliaths, and Kings: Narratives and Evidence in the Litigation of High – Tech Patents," North Carolina Law Review, 87, pp. 1571-1615
+
+Cotropia, Christopher Anthony and Mark A. Lemley, 2009, "Copying in Patent Law," North Carolina Law Review, 87, p. 1421.
+
+Cutler, David and Lawrence Summers. 1988. "The costs of conflict resolution and financial distress: Evidence from the Texaco-Pennzoil Litigation," RAND Journal of Economics, 19, pp. 157-72.
+
+Dodd, Peter and Jerold B. Warner. 1983. "On Corporate Governance: A study of proxy contests," Journal of Financial Economics, v. 11, pp. 401-38.
+
+Federal Trade Commission, 2011. "The Evolving IP Marketplace: Aligning patent notice and remedies with competition."
+
+Fischer, Timo and Joachim Henkel. 2011. “Patent Trolls on Markets for Technology: An Empirical Analysis of Trolls’ Patent Acquisitions” working paper.
+
+Hall, Bronwyn H., Adam B. Jaffe, Manuel Trajtenberg, 2001. "The NBER Patent Citation Data File: Lessons, Insights and Methodological Tools," NBER Working Paper No. 8498.
+
+Haslem, Bruce. 2005. "Managerial Opportunism during Corporate Litigation," Journal of Finance, 60, no. 4, pp. 2013-41.
+
+Hosie, Spencer, 2008. "Patent Trolls and the New Tort Reform: A Practitioner's Perspective," A Journal of Law & Policy for the Information Society, 4, pp. 75-87.
+
+Kesan, Jay P. and Gwendolyn G. Ball. 2005. "How Are Patent Cases Resolved? An Empirical Examination of the Adjudication and Settlement of Patent Disputes," U. Illinois Law & Economics Research Paper No. LE05-027.
+
+Kramer, Lisa A. 2001. "Alternative Methods for Robust Analysis in Event Study Applications," in Lee, C. F., ed., Advances in Investment Analysis and Portfolio Management, v. 8, pp. 109-32.
+
+Lamoreaux, Naomi R., and Kenneth L. Sokoloff, 1999. "Inventors, Firms, and the Market for Technology in the Late Nineteenth and Early Twentieth Centuries." In Naomi R. Lamroeaux, Daniel M. G. Raff, and Peter Temin, eds., Learning by Doing in Markets, Firms, and Countries. Chicago: University of Chicago Press, 19–60.
+
+Lerner, Josh. 1995. "Patenting in the Shadow of Competitors," Journal of Law and Economics, 38, no. 2, pp. 463-95.
+
+Love, Brian J., 2010. "An Empirical Study of Patent Litigation Timing: Could a Patent Term Reduction Decimate Trolls Without Harming Innovators?," Working paper.
+
+Lunney, Glynn S. Jr. 2004. "Patent Law, the Federal Circuit, and the Supreme Court: A Quiet
+
+---
+
+29 - Troll - 10/11
+
+Revolution," Supreme Court Economic Review, 11, pg. 1.
+
+Mackinlay, A. Craig. 1997. "Event Studies in Economics and Finance," Journal of Economic Literature, 35(1), pp. 13-39.
+
+Magliocca, Gerard N. 2007. "Blackberries and Barnyards: Patent Trolls and the Perils of Innovation," Notre Dame Law Review, 82(5), 1809-1838.
+
+McDonough, James F. III, 2006. "The Myth of the Patent Troll: An Alternative View of the Function of Patent Dealers in an Idea Economy," Emory Law Journal 56, pp189-211.
+
+Meurer, Michael J. 1989. "The Settlement of Patent Litigation," RAND Journal of Economics, Spring 1989, v. 20, iss. 1, pp. 77-91.
+
+Morgan, Marc, 2008. "Stop Looking Under the Bridge for Imaginary Creatures: A Comment Examining Who Really Deserves The Title Patent Troll," Federal Circuit Bar Journal, 17, pp. 165-80.
+
+Myhrvold, Nathan, 2010. "The Big Idea," Harvard Business Review, March 2010, pp. 40-50.
+
+Patent Freedom, 2011. “Litigations over Time,” https://www.patentfreedom.com/researchlot.html , accessed August 2011.
+
+Reitzig, Markus, Joachim Henkel and Christopher Heath, 2007. “On sharks, trolls, and their patent prey—Unrealistic damage awards and firms’ strategies of ‘being infringed’,” Research Policy 36, pp. 134–154.
+
+Risch, Michael, 2012 forthcoming. "Patent Troll Myths," Seton Hall Law Review, 42.
+
+Salinger, Michael. 1992. "Standard Errors in Event Studies," The Journal of Financial and Quantitative Analysis 27(1), pp. 39-53.
+
+Satterthwaite, F. E. 1946. "An approximate distribution of estimates of variance components," Biometrics Bulletin, 2. pp. 110-14.
+
+Shrestha, Sannu K. 2011. “Trolls or Market-makers? An empirical analysis of nonpracticing entities,” Columbia Law Review, 110, pp. 114-60.
+
+Temple, James, 2011. “Tech-patent lawsuits discourage innovation,” San Francisco Chronicle, 7/31/2011.
+
+Turner, John L., 2011. "Patent Thickets, Trolls and Unproductive Entrepreneurship," working paper.
+
+Yanagisawa, Tomoya and Dominique Guellec, 2009. “The Emerging Patent Marketplace,” OECD STI Working Paper.
+
+---
+
+30 - Troll - 10/11
+
+## Tables and Figures
+
+Table 1. Summary Statistics of Defendant Firms (Millions of $2010)
+
+<table><tr><td></td><td>Mean</td><td>Median</td></tr><tr><td>Revenue</td><td>34,487</td><td>13,232</td></tr><tr><td>R\&amp;D spending</td><td>1,779</td><td>531</td></tr><tr><td>Intangible assets</td><td>9,792</td><td>1,269</td></tr><tr><td>Employees (1000s)</td><td>9.4</td><td>3.6</td></tr><tr><td>NPE lawsuits per firm (1990-2010)</td><td>26.1</td><td>12</td></tr><tr><td>Industry (2 digit SIC code)</td><td></td><td></td></tr><tr><td>Electronics (36)</td><td>22\%</td><td></td></tr><tr><td>Machinery \</td><td>computer equipment (35)</td><td>15\%</td></tr><tr><td>Retail/wholesale (50-59)</td><td>15\%</td><td></td></tr><tr><td>Software (73)</td><td>14\%</td><td></td></tr><tr><td>Communications (48)</td><td>9\%</td><td></td></tr><tr><td>Financial services (60-67)</td><td>8\%</td><td></td></tr></table>
+
+
+Means of firm characteristics are over 3,821 firm-events. This is a sample of publicly listed firms.
+
+---
+
+31 - Troll - 10/11
+
+Table 2. Summary Characteristics of Lawsuits
+
+<table><tr><td></td><td>Mean</td><td>Median</td></tr><tr><td>No. of publicly listed defendants</td><td>15.3</td><td>5</td></tr><tr><td>Sole defendant</td><td>17\%</td><td></td></tr><tr><td>In litigation with 10 or more defendants</td><td>32\%</td><td></td></tr><tr><td>Software patent</td><td>62\%</td><td></td></tr><tr><td>Patent Technology Classes (NBER)</td><td></td><td></td></tr><tr><td>Chemical</td><td>1\%</td><td></td></tr><tr><td>Computers \&amp; communications</td><td>75\%</td><td></td></tr><tr><td>Drugs \&amp; medical</td><td>1\%</td><td></td></tr><tr><td>Electrical \&amp; electronics</td><td>12\%</td><td></td></tr><tr><td>Mechanical</td><td>4\%</td><td></td></tr><tr><td>Other</td><td>8\%</td><td></td></tr></table>
+
+
+Note: The number of defendants in the lawsuits are for all lawsuits filed by the same NPE on the same day. Patent characteristics are for a sub-sample matched to Derwent Litalert and are for the first patent listed in the suit. The categorization of software patents is described in Bessen (2011) . We have adapted the NBER technology classes (Hall et al. 2001) to the current technology class system adding classes 398, 715, 717, 725 and 726 to the computers and communications category. This sample consists of publicly listed firms.
+
+---
+
+32 - Troll - 10/11
+
+Table 3. Cumulative Abnormal Returns (CARs) of Defendants in NPE Litigation
+
+<table><tr><td></td><td colspan="3">Cumulative Abnormal Returns</td><td colspan="3">Loss of common stock value (millions of $2010)</td><td></td></tr><tr><td rowspan="2">Sample</td><td>Mean</td><td>Median</td><td>Robust Z statistic</td><td>Mean</td><td>Median</td><td>Aggregate</td><td>N</td></tr><tr><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td><td>(5)</td><td>(6)</td><td></td></tr><tr><td>5 day event window</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>All NPE suits</td><td>-0.32\% (0.08\%)**</td><td>-0.52\%</td><td>-4.01**</td><td>$122.0</td><td>$20.4</td><td>$501,775</td><td>4,114</td></tr><tr><td>25 day event window</td><td></td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>All NPE suits</td><td>-0.37\% (0.14\%)**</td><td>-0.71\%</td><td>-2.04*</td><td>$140.6</td><td>$23.6</td><td>$579,217</td><td>4,119</td></tr><tr><td>All patent litigation</td><td>1984-99 (Bessen and Meurer 2007)</td><td></td><td></td><td></td><td></td><td></td><td></td></tr><tr><td>All suits</td><td>-0.50\% (0.16\%)**</td><td>-0.51\%</td><td>-3.24**</td><td>$75.9</td><td>$6.5</td><td></td><td>2,887</td></tr><tr><td>Single defendants</td><td>-0.61\% (0.18\%)**</td><td>-0.54\%</td><td>-2.94**</td><td></td><td></td><td></td><td>2,460</td></tr></table>
+
+
+Note: Standard errors in parentheses. Single asterisk indicates statistical significance at the 5 % level; double asterisk indicates 1 % significance. Average cumulative abnormal returns (CARs) are weighted means, with weights proportional to the inverse of the estimated variance of each return. Event window is 5 days (T-1 to T+4) or 25 days (T-1 to T+24). Cumulative abnormal returns are estimated using OLS. The robust $Z$ statistic is a joint test of the individual firm $t$ statistics (Kramer 2001) .
+
+---
+
+33 - Troll - 10/11
+
+Table 4. Wealth Transfer for Publicly Listed NPEs
+
+<table><tr><td></td><td>Millions of $2010</td><td>As share of Defendants' Losses</td></tr><tr><td>Cumulative for 14 NPEs, 2000-2010</td><td></td><td></td></tr><tr><td>Revenues</td><td>$7,639</td><td>$9 \%$</td></tr><tr><td>Net cash flow to investing activities less capital expenditures</td><td>$1,697</td><td>$2 \%$</td></tr><tr><td>R\&amp;D expense</td><td>$2,039</td><td>$2 \%$</td></tr><tr><td>Net income</td><td>$258</td><td>$0 \%$</td></tr><tr><td>Combined stock for 14 NPEs, 2010</td><td></td><td></td></tr><tr><td>Intangible assets</td><td>$562</td><td>$1 \%$</td></tr><tr><td>Defendant firms</td><td></td><td></td></tr><tr><td>Loss of wealth</td><td>$87,574</td><td>100%</td></tr></table>
+
+
+---
+
+34- Troll - 10/11
+
+Appendix. Public NPE firms
+
+Acacia Technologies Asure Software Burst.com Inc Decisioning.com Inc Interdigital Intertrust Technologies Corp LecTec Corp Mosaid Technologies Inc Network-1 Security Solutions Inc OPTi Inc Rambus Tessera Technologies Inc VirnetX Inc Wi-Lan
+

processed/00025_W3121881716/document.md

@@ -0,0 +1,2 @@
+![Figure](figures/figure_000.png)
+

processed/00025_W3121881716/metadata.json

@@ -0,0 +1,24 @@
+{
+  "source_file": "/tmp/tmpksht834i/00025_W3121881716_Regime_Shifting_The_TRIPs_Agreement_and_New_Dynamics_of_International_Intellectual_Property_Lawmaking.pdf",
+  "total_pages": 1,
+  "pages": [
+    {
+      "page_number": 1,
+      "elements": [
+        {
+          "bbox": [
+            39,
+            0,
+            570,
+            896
+          ],
+          "label": "fig",
+          "reading_order": 0,
+          "text": "![Figure](figures/figure_000.png)",
+          "figure_path": "/tmp/pdf_output/markdown/figures/figure_000.png"
+        }
+      ]
+    }
+  ],
+  "metadata": {}
+}

processed/00059_W1987143486/document.md

@@ -0,0 +1,1292 @@
+Skip to main content Accessibility help
+
+# Login Alert
+
+Cancel Log in ×
+
+![Figure](figures/figure_003.png)
+
+Discover Content
+
+Products and Services Register Log In (0) Cart
+
+Search
+
+- • Browse
+• Services
+• Open research
+Institution Login Search
+
+Menu links
+
+- 1. Browse
+---
+
+## 1. Subjects
+
+### 1. Subjects (A-D)
+
+- 1. Anthropology
+2. Archaeology
+3. Area Studies
+4. Art
+5. Chemistry
+6. Classical Studies
+7. Computer Science
+8. Drama, Theatre, Performance Studies
+### 2. Subjects (E-K)
+
+- 1. Earth and Environmental Science
+2. Economics
+3. Education
+4. Engineering
+5. English Language Teaching – Resources for
+Teachers
+6. Film, Media, Mass Communication
+7. General Science
+8. Geography
+9. History
+### 3. Subjects (L-O)
+
+- 1. Language and Linguistics
+2. Law
+3. Life Sciences
+4. Literature
+5. Management
+6. Materials Science
+7. Mathematics
+8. Medicine
+9. Music
+
+10. Nutrition
+### 4. Subjects (P-Z)
+
+---
+
+- 1. Philosophy
+2. Physics and Astronomy
+3. Politics and International Relations
+4. Psychiatry
+5. Psychology
+6. Religion
+7. Social Science Research Methods
+8. Sociology
+9. Statistics and Probability
+## 2. Open access
+
+### 1. All open access publishing
+
+- 1. Open access
+2. Open access journals
+3. Research open journals
+4. Journals containing open access
+5. Open access articles
+6. Open access books
+7. Open access Elements
+## 3. Journals
+
+### 1. Explore
+
+- 1. All journal subjects
+2. Search journals
+### 2. Open access
+
+- 1. Open access journals
+2. Research open journals
+3. Journals containing open access
+4. Open access articles
+## 3. Collections
+
+- 1. Cambridge Forum
+2. Cambridge Law Reports Collection
+3. Cambridge Prisms
+4. Research Directions
+---
+
+## 4. Books
+
+### 1. Explore
+
+- 1. Books
+2. Open access books
+3. New books
+4. Flip it Open
+### 2. Collections
+
+- 1. Cambridge Companions
+2. Cambridge Editions
+3. Cambridge Histories
+4. Cambridge Library Collection
+5. Cambridge Shakespeare
+6. Cambridge Handbooks
+### 3. Collections (cont.)
+
+- 1. Dispute Settlement Reports Online
+2. Flip it Open
+3. Hemingway Letters
+4. Shakespeare Survey
+5. Stahl Online
+6. The Correspondence of Isaac Newton
+### 5. Elements
+
+### 1. Explore
+
+- 1. About Elements
+2. Elements series
+3. Open access Elements
+4. New Elements
+### 2. Subjects (A-E)
+
+- 1. Anthropology
+2. Archaeology
+3. Classical Studies
+4. Computer Science
+---
+
+- 5. Drama, Theatre, Performance Studies
+6. Earth and Environmental Sciences
+7. Economics
+8. Education
+9. Engineering
+## 3. Subjects (F-O)
+
+- 1. Film, Media, Mass Communication
+2. History
+3. Language and Linguistics
+4. Law
+5. Life Sciences
+6. Literature
+7. Management
+8. Mathematics
+9. Medicine
+
+10. Music
+## 4. Subjects (P-Z)
+
+- 1. Philosophy
+2. Physics and Astronomy
+3. Politics and International Relations
+4. Psychology
+5. Religion
+6. Sociology
+7. Statistics and Probability
+## 6. Textbooks
+
+### 1. Explore
+
+- 1. Cambridge Higher Education
+2. Title list
+3. New titles
+## 7. Collections
+
+### 1. Book collections
+
+### 1. Cambridge Companions
+
+---
+
+- 2. Cambridge Editions
+3. Cambridge Histories
+4. Cambridge Library Collection
+5. Cambridge Shakespeare
+6. Cambridge Handbooks
+2. Book collections (cont.)
+
+- 1. Dispute Settlement Reports Online
+2. Flip it Open
+3. Hemingway Letters
+4. Shakespeare Survey
+5. Stahl Online
+6. The Correspondence of Isaac Newton
+3. Journal collections
+
+- 1. Cambridge Forum
+2. Cambridge Law Reports Collection
+3. Cambridge Materials
+4. Cambridge Prisms
+4. Series
+
+1. All series
+
+8. Partners
+
+1. Partners
+
+- 1. Agenda Publishing
+2. Amsterdam University Press
+3. Anthem Press
+4. Boydell & Brewer
+5. Bristol University Press
+6. Edinburgh University Press
+7. Emirates Center for Strategic Studies and
+Research
+8. Facet Publishing
+2. Partners (cont.)
+
+---
+
+- 1. Foundation Books
+2. Intersentia
+3. ISEAS-Yusof Ishak Institute
+4. Jagiellonian University Press
+5. Royal Economic Society
+6. Unisa Press
+7. The University of Adelaide Press
+8. Wits University Press
+## 2. Services
+
+### 1. About
+
+### 1. About Cambridge Core
+
+- 1. About
+2. Accessibility
+3. CrossMark policy
+4. Ethical Standards
+### 2. Environment and sustainability
+
+- 1. Environment and sustainability
+2. Reducing print
+3. Journals moving to online only
+### 3. Guides
+
+- 1. User guides
+2. User Guides and Videos
+3. Support Videos
+4. Training
+### 4. Help
+
+- 1. Cambridge Core help
+2. Contact us
+3. Technical support
+### 2. Agents
+
+- 1. Services for agents
+---
+
+- 1. Services for agents
+2. Journals for agents
+3. Books for agents
+4. Price list
+## 3. Authors
+
+### 1. Journals
+
+- 1. Journals
+2. Journal publishing statistics
+3. Corresponding author
+4. Seeking permission to use copyrighted
+material
+
+5. Publishing supplementary material
+6. Writing an effective abstract
+
+7. Journal production - FAQs
+### 2. Journals (cont.)
+
+- 1. Author affiliations
+2. Co-reviewing policy
+3. Anonymising your manuscript
+4. Publishing open access
+5. Convert your article to Gold Open Access
+6. Publishing Open Access - webinars
+### 3. Journals (cont.)
+
+- 1. Preparing and submitting your paper
+2. Publication journey
+3. Publishing agreement FAQs for journal
+authors
+4. Author Information Form FAQs
+5. Promoting your published paper
+6. Measuring impact
+7. Journals artwork guide
+8. Using ORCID
+## 4. Books
+
+---
+
+- 1. Books
+2. Marketing your book
+3. Author guides for Cambridge Elements
+## 4. Corporates
+
+### 1. Corporates
+
+- 1. Commercial reprints
+2. Advertising
+3. Sponsorship
+4. Book special sales
+5. Contact us
+### 5. Editors
+
+### 1. Information
+
+- 1. Journal development
+2. Peer review for editors
+3. Open access for editors
+4. Policies and guidelines
+### 2. Resources
+
+- 1. The editor's role
+2. Open research for editors
+3. Engagement and promotion
+4. Blogging
+5. Social media
+### 6. Librarians
+
+### 1. Information
+
+- 1. Open Access for Librarians
+2. Transformative agreements
+3. Transformative Agreements - FAQs
+4. Evidence based acquisition
+5. Cambridge libraries of the world podcast
+6. Purchasing models
+7. Journals Publishing Updates
+---
+
+## 2. Products
+
+- 1. Cambridge frontlist
+2. Cambridge journals digital archive
+3. Hot topics
+4. Other digital products
+5. Perpetual access products
+6. Price list
+7. Developing country programme
+8. New content
+## 3. Tools
+
+- 1. Eligibility checker
+2. Transformative agreements
+3. KBART
+4. MARC records
+5. Using MARCEdit for MARC records
+6. Inbound OpenURL specifications
+7. COUNTER usage reporting
+## 4. Resources
+
+- 1. Catalogues and resources
+2. Making the most of your EBA
+3. Posters
+4. Leaflets and brochures
+5. Additional resources
+6. Find my sales contact
+7. Training
+8. Read and publish resources
+## 7. Peer review
+
+### 1. Peer review
+
+- 1. How to peer review journal articles
+2. How to peer review book proposals
+3. How to peer review Registered Reports
+4. Peer review FAQs
+5. Ethics in peer review
+---
+
+- 6. Online peer review systems
+7. A guide to Publons
+## 8. Publishing ethics
+
+### 1. Journals
+
+- 1. Publishing ethics guidelines for journals
+2. Core editorial policies for journals
+3. Authorship and contributorship for journals
+4. Affiliations for journals
+5. Research ethics for journals
+6. Competing interests and funding for
+journals
+### 2. Journals (cont.)
+
+- 1. Data and supporting evidence for journals
+2. Misconduct for journals
+3. Corrections, retractions and removals for
+journals
+4. Versions and adaptations for journals
+5. Libel, defamation and freedom of
+expression
+6. Business ethics journals
+### 3. Books
+
+- 1. Publishing ethics guidelines for books
+2. Core editorial policies for books
+3. Authorship and contributorship for books
+4. Affiliations for books
+5. Research ethics for books
+6. Competing interests and funding for books
+### 4. Books (cont.)
+
+- 1. Data and supporting evidence for books
+2. Misconduct for books
+3. Corrections, retractions and removals for
+books
+---
+
+- 4. Versions and adaptations for books
+5. Libel, defamation and freedom of
+expression
+6. Business ethics books
+## 9. Publishing partners
+
+### 1. Publishing partners
+
+- 1. Publishing partnerships
+2. Partner books
+3. eBook publishing partnerships
+4. Journal publishing partnerships
+### 2. Publishing partners (cont.)
+
+- 1. Journals publishing
+2. Customer support
+3. Membership Services
+4. Our Team
+## 3. Open research
+
+### 1. Open access policies
+
+### 1. Open access policies
+
+- 1. Open research
+2. Open access policies
+3. Cambridge University Press and Plan S
+4. Text and data mining
+5. Preprint policy
+6. Social sharing
+### 2. Journals
+
+- 1. Open access journals
+2. Gold Open Access journals
+3. Transformative journals
+4. Green Open Access policy for journals
+5. Transparent pricing policy for journals
+---
+
+## 3. Books and Elements
+
+- 1. Open access books
+2. Gold open access books
+3. Green Open Access policy for books
+4. Open access Elements
+## 2. Open access publishing
+
+### 1. About open access
+
+- 1. Open research
+2. Open Access Week
+3. What is open access?
+4. Open access glossary
+5. Open access myths
+6. Hybrid Open Access FAQs
+7. Eligibility checker
+### 2. Open access resources
+
+- 1. Open access resources
+2. Benefits of open access
+3. Creative commons licences
+4. Funder policies and mandates
+5. Article type definitions
+6. Convert your article to Gold Open Access
+7. Open access video resources
+### 3. Open research initiatives
+
+### 1. Research transparency
+
+- 1. Transparency and openness
+2. Open Practice Badges
+3. OA organisations, initiatives & directories
+4. Registered Reports
+5. Annotation for Transparent Inquiry (ATI)
+### 2. Journal flips
+
+- 1. Open access journal flips
+---
+
+## 2. OA Journal Flip FAQs
+
+### 3. Flip it Open
+
+- 1. Flip it Open
+2. Flip it Open FAQs
+### 4. Open access funding
+
+### 1. Open access funding
+
+- 1. Funding open access publication
+2. Cambridge Open Equity Initiative
+3. Completing a RightsLink (open access)
+transaction
+### 5. Cambridge Open Engage
+
+### 1. Cambridge Open Engage
+
+- 1. Cambridge Open Engage
+2. Partner With Us
+3. Branded Hubs
+4. Event Workspaces
+5. Partner Resources
+6. APSA Preprints
+7. APSA Preprints FAQs
+Unfortunately you do not have access to this content, please use the Get access link below for information on how to access this content.
+
+𠄍
+
+- • Home
+• > Journals
+• > Journal of Law, Medicine & Ethics
+• > Volume 37 Issue 2
+• > Comment: Compulsory Licensing of Patented
+Pharmaceutical...
+• English
+• Français
+---
+
+[image] Journal of Law, Medicine & Ethics
+
+# Article contents
+
+- • Extract
+• References
+## Comment: Compulsory  Licensing of Patented  Pharmaceutical Inventions: Evaluating the Options
+
+Published online by Cambridge University Press: 01 January 2021
+
+Jerome H. Reichman
+
+- • Article
+• Metrics
+Article contents
+
+- • Extract
+• References
+Get access
+
+[image] Share [image] Cite [image]Rights & Permissions [Opens in a new window]
+
+## Extract
+
+Few topics in international intellectual property law have been as
+
+---
+
+controversial in recent years as the one we are about to examine. In the 1980s and early 1990s, a Diplomatic Conference attempted to revise the oldest international convention providing some protection for patented inventions outside of the domestic laws. Those efforts broke down, largely because developed and developing countries could not agree on the powers that governments should retain to issue compulsory licenses or on the grounds for which these powers could be exercised. The failure of this Conference, held under the auspices of the World Intellectual Property Organization (WIPO), persuaded the technology-exporting countries to link future negotiations concerning international intellectual property protection to the Multilateral Trade Negotiations, known as the Uruguay Round, which got underway in 1986. The end result was Annex IC of the Agreement Establishing the World Trade Organization of 1994, which incorporated a new, comprehensive and relatively elevated set of international minimum standards of patent protection into the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS Agreement).
+
+## Information
+
+Type Symposium Information Journal of Law, Medicine & Ethics, Volume 37, Issue 2, Summer 2009, pp. 247 - 263 DOI: https://doi.org/10.1111/j.1748-720X.2009.00369.x [Opens in a new window] Copyright Copyright © American Society of Law, Medicine and Ethics 2009
+
+## Access options
+
+Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or
+
+---
+
+personal access. Content may require purchase if you do not have access.)
+
+## Article purchase
+
+Temporarily unavailable
+
+## References
+
+1
+
+Paris Convention for the Protection of Industrial Property, March 20, 1883, as revised at Stockholm (1967), 21 UST 1583, 828 UNTS 305 [hereinafter cited as Paris Convention].Google Scholar
+
+2
+
+See, e.g., Reichman, J. H. with Hasenzahl, C., Non-Voluntary Licensing of Patented Inventions: Historical Perspective, Legal Framework under TRIPS, and an Overview of the Practice in Canada and the U.S.A., U.N. Conference on Trade & Development [UNCTAD] & International Centre for Trade & Sustainable Development [ICTSD], Project on IPRs and Sustainable Development, Issue Paper No. 5, June 2003, available at < http://
+
+ictsd.net/i/publications/11764/> (last visited April 17, 2009)
+
+[hereinafter cited as Reichman with Hasenzahl].
+
++(last + visited + April + 17,
+
++ 2009) + [hereinafter + cited + as + Reichman + with + Hasenzahl]. > Google Scholar
+
+3
+
+For comprehensive accounts of this process, see Drahos, P. and Braithwaite, J., Information Feudalism: Who Owns the Knowledge Economy? (London: Earthscan Publications, 2002); Sell, S. K., Power and Ideas: North-South Politics of Intellectual Property and Antitrust (Albany, New York: State University of New York Press, 1998); Sell, S. K., Private Power, Public Law: The Globalization of Intellectual Property Rights (Cambridge, U.K./New York: Cambridge University Press, 2003); Helfer, L. R., “Regime Shifting: The TRIPS Agreement and New Dynamics of International Intellectual Property Lawmaking, ” Yale Journal of International Law 29, no. 1 (2004): 1-83.Google Scholar
+
+---
+
+4
+
+See Agreement on Trade-Related Aspects of Intellectual Property Rights, April 15, 1994, 33 I.L.M. 81, art. 27–34 (1994), available at <http://www.wto.org/english/docs_e/legal_e/27-trips.pdf> [hereinafter cited as TRIPS]; Marrakesh Agreement Establishing the World Trade Organization, April 15, 1994, 1867 UNTS 154; 33 ILM 1144 (1994), available at <http://www.wto.org/english/docs_e/ legal_e/legal_e.htm> (last visited April 17, 2009). The United States implemented the WTO Agreements in the Uruguay Round Agreements Act, Public Law 103–465, 108 Stat. 4809 (1994).
+
++ [hereinafter + cited + as + TRIPS]; + Marrakesh + Agreement + Establishing + the + World + Trade + Organization + April + 15, + 1994, + 1867 + UNTS + 154; + 33 + ILM + 1144 + (1994), + available + at + + (last + visited + April + 17, + 2009). + The + United + States + implemented + the + WTO + Agreements + in + the + + Public + Law + 103–465, + 108 + Stat. + 4809 + (1994). > Google Scholar
+
+## 5
+
+Id. (TRIPS), at art. 65–66, 70; World Trade Organization, Declaration on the TRIPS Agreement and Public Health, November 20, 2001, WT/MIN(01)/DEC/2, at para. 4, available at <http:// www.wto.org/english/thewto e/minist e/min01 e/ mindecl_trips e.htm> (last visited April 17, 2009) [hereinafter cited as Declaration on TRIPS]. From this date on, developing countries were required to provide at least 20 years of patent protection to a broad range of products, including pharmaceutical products, and mailboxes with pending patent applications were opened and began being processed. A few least-developed countries (LDCs) remain exempt from protecting patents until 2013 and patents on pharmaceuticals until 2016. See Extension of the Transition Period Under Article 66.1 for Least-Developed Country Members, Decision of the Council for TRIPS of 19 November 2005, WTO doc. IP/C/40, November 30, 2005; Extension of the Transition Period under Article 66.1 of the TRIPS Agreement for Least-Developed Country Members for Certain Obligations with Respect to Pharmaceutical Products, Decision of the Council for TRIPS of 27 June 2002, WTO doc. IP/C/25, July 1, 2002. + (last + visited + April + 17, + 2009) + [hereinafter + cited + as + Declaration + on + TRIPS]. + From + this + date + on,
+
+---
+
++ developing + countries + were + required + to + provide + at + least + 20 + y + including + pharmaceutical + products, + and + mailboxes + with + pending + patent + applications + were + opened + A + few + leastdeveloped + countries + (LDCs) + remain + exempt + from + protecting + pate + See + Extension + of + the + Transition + Period + Under + Article + 66.1 + f . Developed + Country + Members, + Decision + of + the + Council + for + TRIPS + of + 19 + November + 2005, + WTO + doc. + IP/C/40, + November + 30, + 2005; + Extension + of + the + Transition + Period + under + Article + 66.1 + of + the . Developed + Country + Members + for + Certain + Obligations + with + Respect + Decision + of + the + Council + for + TRIPS + of + 27 + June + 2002, + WTO + doc. + IP/C/25, + July + 1, + 2002. > Google Scholar
+
+6
+
+Understanding on Rules and Procedures Governing the Settlement of Disputes, April 15, 1994, 33 ILM 1226, art. 23 (1994), available at <http://www.wto.org/english/tratop_e/dispu_e/dsu_e.htm> (last visited April 17, 2009) [hereinafter cited as DSU]; see infra note 166 and accompanying text. + (last + visited + April + 17, + 2009) + [hereinafter + cited + as + DSU]; + see + infra + note + 166 + and + accompanying + text. > Google Scholar
+
+7
+
+See TRIPS, supra note 4, at art. 31.Google Scholar 8
+
+See Watal, J., Intellectual Property Rights in the WTO and Developing Countries (Boston: Kluwer Law International, 2002): at 33–44.Google Scholar
+
+9
+
+Compare Paris Convention, supra note 1, at art. 5A, with TRIPS, supra note 4, art. 31; see Reichman, and Hasenzahl, , supra note 2.Google Scholar
+
+10
+
+See TRIPS, supra note 4, at art. 31(b).Google Scholar
+
+11
+
+Indeed, it was the United States' inability to distinguish its routine exercise of government use licenses from other compulsory licenses that led to the breadth of article 31 in the first place. See, e.g., Watal,, supra note 8.Google Scholar
+
+12
+
+---
+
+See Paris Convention, supra note 1, at art. 1.Google Scholar 13
+
+See Reichman, with Hasenzahl, , supra note 2.Google Scholar 14
+
+Debrulle, J., De Cort, L. and Petit, M., “La license obligatoire belge pour raison de santé publique,” in van Overwalle, G., ed., Gene Patents and Public Health (Brussels: Bruylant, 2007): at 159 (note: English translation and summary available at 199); van Zimmeren, E. and Requena, G., “Exofficio Licensing in the Medical Sector: The French Model,” in van Overwalle, G., ed., Gene Patents and Public Health (Brussels: Bruylant, 2007): at 123.Google Scholar 15
+
+For the reasons, see Flynn, S., Hollis, A. and Palmedo, M., “An Economic Justification for Open Access to Essential Medicine Patents in Developing Countries,” Journal of Law, Medicine & Ethics 37, no. 2 (2009): 184–209.CrossRefGoogle Scholar 16
+
+See Declaration on TRIPS, supra note 5, at para. 4 (emphasis added).Google Scholar
+
+17 Id., at para. 4–5.Google Scholar
+
+18
+
+Id., at para. 5.b.Google Scholar 19
+
+Abbott, F. M. and Reichman, J. H., “The Doha Round's Public Health Legacy: Strategies for the Production and Diffusion of Patented Medicines Under the Amended TRIPS Provisions, ” Journal of International Economic Law 10, no. 4 (2007): 921–987, at 929; see Declaration on TRIPS, supra note 5, at para. 6; TRIPS, supra note 4, at art. 31(f).CrossRefGoogle Scholar 20
+
+See id. (Abbott & Reichman), at 937.Google Scholar 21
+
+For a detailed article analyzing Article 31's disease scope, see Outterson, K., “Disease-Based Limitations on Compulsory Licenses under Articles 31 and 31bis,” in Correa, C., ed., Research Handbook on Intellectual Property Law and the WTO (forthcoming 2009). An earlier version of this paper can be found at: Outterson, K., “Should Access to Medicines and TRIPS Flexibilities Be Limited to Specific Diseases?” American Journal of Law & Medicine 34, nos. 2–3
+
+---
+
+(2008): 279–301 [hereinafter Access to Medicine and TRIPS].CrossRefGoogle Scholar
+
+22
+
+World Trade Organization, Implementation of Paragraph 6 of the Doha Declaration on the TRIPS Agreement and Public Health, Decision of the General Council of 30 August 2003, August 30, 2003, WT/ L/540, available at <http://www.wto.org/english/tratop e/trips e/ implem para6 e.htm> (last visited April 17, 2009) [hereinafter cited as Paragraph 6 Decision]; World Trade Organization, Amendment of the TRIPS Agreement, December 6, 2005, WT/L/641, available at <http://www.wto.org/english/tratop e/trips e/ wtl641 e.htm> (last visited April 17, 2009) [hereinafter cited as Amendment of TRIPS]; see Abbott, and Reichman,, supra note 19. + (last + visited + April + 17, + 2009) + [hereinafter + cited + as + Paragraph + 6 + Decision]; + World + Trade + Organization, + Amendment + of + the + TRIPS + Agreement, + December + 6, + 2005, + WT/L/641, + available + at + + (last + visited + April + 17, + 2009) + [hereinafter + cited + as + Amendment + of + TRIPS]; + see + Abbott, + and + Reichman, +, + supra + note + 19. > Google
+
+Scholar
+
+23
+
+See TRIPS, supra note 4, at art. 31(h). See generally Taubman, A., “Rethinking TRIPS: ‘Adequate Remuneration’ for Non-Voluntary Patent Licensing, ” Journal of International Economic Law 11, no. 4 (2008): 927–970.CrossRefGoogle Scholar
+
+24
+
+See Abbott, and Reichman,, supra note 19, at 944.Google Scholar 25
+
+See Paragraph 6 Decision, supra note 22. See also, e.g., Matthews, D., “WTO Decision on Implementation of Paragraph 6 of the Doha Declaration on the TRIPS Agreement and Public Health: A Solution to the Access to Essential Medicines Problem?” Journal of International Economic Law 7, no. 1 (2004): 73– 107.CrossRefGoogle Scholar
+
+26
+
+See Amendment of TRIPS, supra note 22.Google Scholar
+
+27
+
+Id.; see Abbott, and Reichman, supra note 19, at 929.Google Scholar 28
+
+---
+
+See, e.g., Mercurio, B. C., “TRIPS, Patents, and Access to Life-Saving Drugs in the Developing World,” Marquette Intellectual Property Law Review 8, no. 2 (2004): 211; see also Abbott, and Reichman, , supra note 19, at 947.Google Scholar
+
+29
+
+Goodwin, P. E., “Right Idea, Wrong Result – Canada's Access to Medicines Regime, ” American Journal of Law & Medicine 34, no. 4 (2008): 567–584.CrossRefGoogle Scholar 30
+
+Mercurio, B., “TRIPS-Plus Provisions in FTAs: Recent Trends, ” in Bartels, L., Ortino, F., eds., Regional Trade Agreements and the WTO Legal System (Oxford: Oxford University Press, 2006): at 215– 238; see also Abbott, and Reichman,,, supra note 19, at 958–66; Dwyer, L., “Patent Protection and Access to Medicine: The Colombian and Peruvian Trade Promotion Agreements, ” Law & Business Review of the Americas 13, no. 4 (2007): 825-857. For an example from a developed country, see Outterson, K., “Agony in the Antipodes: The Generic Drug Provisions in the Australia – US Free Trade Agreement, ” Journal of Generic Medicine 2, no. 4 (2005): 316-326.Google Scholar
+
+31
+
+See Flynn, et al., supra note 15.Google Scholar
+
+32
+
+Downward pressures on AIDS drugs were further exerted by private foundations (especially the Clinton Foundation) and both national (e.g., PEPFAR in the U.S.) and international aid programs.Google Scholar
+
+33
+
+Lybecker, K. M. and Fowler, E., “Compulsory Licensing in Canada and Thailand: Comparing Regimes to Ensure Legitimate Use of the WTO Rules, ” Journal of Law, Medicine & Ethics 37, no. 2 (2009): 222–240.CrossRefGoogle Scholar 34
+
+See, e.g., Cohen, J., “Brazil, Thailand Override Big Pharma Patents, ” Science, May 11, 2007, at 816; Abbott and Reichman, supra note 19, at 950–952 (noting that, until 2007, Brazil had reached negotiated settlements with foreign suppliers without formally issuing a compulsory license).Google Scholar
+
+35
+
+See Outterson (Access to Medicines and TRIPS), supra note 21, at
+
+---
+
+320–321; Outterson, K. (Disease-Based Limitations), supra note 21. But see Lybecker, and Fowler,, supra note 33 (who disagree on this point).Google Scholar
+
+36
+
+See Goodwin,, supra note 29, at 569; Lybecker, and Fowler,, supra note 33, at 2.Google Scholar 37
+
+Tunsarawuth, S., “Indonesia Mulls Compulsory Licenses on Three More HIV/AIDS Drugs, ” Intellectual Property Watch, November 26, 2007, available at <http://www.ip-watch.org/ weblog/2007/11/26/indonesia-mulls-compulsory-licences-on-threemore-hivaids-drugs/> (last visited April 21, 2009). + (last + visited + April + 21, + 2009). > Google Scholar 38
+
+See van Zimmeren, and Requena, , supra note 14, at 123 and 139 n. 38; Love, J. Packard, "Recent Examples of Compulsory Licensing of Patents," KEI, May 6, 2007, available at <http://
+
+www.keionline.org/index.php?
+
+option = com_content&task = view&id = 41 > (last visited April 21, 2009). + (last + visited + April + 21, + 2009). > Google Scholar 39
+
+See van Zimmeren, & Requena,, supra note 14, at 140.Google Scholar
+
+40
+
+See Love,, supra note 38, at 3; Abbott, and Reichman,, supra note 19, at 939; Carroll, J. and Winslow, R., “Bayer Agrees to Slash Prices for Cipro Drug,” Wall Street Journal, October 25, 2001, at A3 (“The agreement comes after a high-stakes threat by Tommy Thompson, HHS secretary, to break Bayer's patent for Cipro if he didn't get the price he wanted.”) The relevant U.S. compulsory license statutes are 7 U.S.C. § 2404 (2000) (plant variety protection certificates necessary for the nation's fiber, food, or feed supply); 28 U.S.C. § 1498 (2000) (government use of patents); 35 U.S.C. § 203 (2000) (patents developed through the use of government research funding under the Bayh-Dole Act); and 42 U.S.C. § 2183 (2000) (atomic energy). The U.S. compulsory license statutes do not meet all of the restrictions required by Article 31 of TRIPS. See TRIPS, supra note 4, at art. 31.Google Scholar
+
+41
+
+See van Zimmeren, and Requena, , supra note 14, at 133-
+
+---
+
+134.Google Scholar
+
+42
+
+See Debrulle, et al., supra note 14, at 159 and 163.Google Scholar 43
+
+Id., at 171; but see infra note 90.Google Scholar
+
+44
+
+See van Zimmeren, and Requena,, supra note 14, at 137.Google Scholar
+
+45
+
+See Love,, supra note 38, at 10–11. For details see Coco, R. and Nebbia, P., “Compulsory Licensing and Interim Measures in Merck: A Case for Italy or for Antitrust?” Journal of Intellectual Property Law & Practice 2, no. 7 (2007): 452–462.Google Scholar 46
+
+Castle, S. and Kanter, J., “European Antitrust Regulators Raid Drug Companies, ” International Herald Tribune, January 16, 2008, available at <http://www.iht.com/articles/2008/01/16/business/ drug.php> (last visited March 7, 2009); DG Competition, Pharmaceutical Sector Inquiry: Preliminary Report, November 28, 2008, available at <http://ec.europa.eu/competition/sectors/ pharmaceuticals/inquiry/preliminary report.pdf> (last visited March 7, 2009). + (last + visited + March + 7, + 2009); + DG + Competition, + Pharmaceutical + Sector + Inquiry: + Preliminary + Report, + November + 28, + 2008, + available + at + + (last + visited + March + 7, + 2009). > Google Scholar
+
+47
+
+For a view consistent with that of the patent-based drug industry, see Danzon, P. M. and Towse, A., “Differential Pricing for Pharmaceuticals; Reconciling Access, R&D and Patents, ” AEIBrookings Joint Center for Regulatory Studies, Working Paper No. 03–7, 2003, at 28–29, available at <http://papers.ssrn.com/sol3/ papers.cfm?abstract id=422821> (last visited April 21, 2009). For a synthesis more attuned to the problems of access, see Outterson, K., “Pharmaceutical Arbitrage: Balancing Access and Innovation in International Prescription Drug Markets, ” Yale Journal Health Policy, Law & Ethics 5, no. 1 (2005): 193-291 [hereinafter cited as Pharmaceutical Arbitrage]. + (last + visited + April + 21, + 2009). + For + a + synthesis + more + attuned + to + the + problems + of + access, + see + Outterson, + K., + “Pharmaceutical + Arbitrage: + Balancing + Access + and + Innovation + in + International + Prescription +
+
+---
+
++ Law + & + Ethics + 5, + no. + 1 + (2005): + 193-291 + [hereinafter + cited + as + Pharmaceutical + Arbitrage]. > Googl Scholar
+
+48
+
+Outterson, K. and Kesselheim, A., "Market-Based Licenses for HPV Vaccines in Developing Countries," Health Affairs 27, no. 1 (2008): 130–139, at 136–137.CrossRefGoogle Scholar 49
+
+Scherer, F. M., “A Note on Global Welfare in Pharmaceutical Patenting, ” World Economy 27, no. 27 (2004): 1127–1142, at 1141.CrossRefGoogle Scholar
+
+50
+
+Gilead has announced a policy of price discrimination for major AIDS products that has elicited complaints from other companies.
+
+See Baron, D. P., Krehbiel, K. and Tayan, B., The Gilead Access Program for HIV Drugs, 2007.Google Scholar
+
+51
+
+See TRIPS, supra note 4, at art. 6; Declaration on TRIPS, supra note 5, at par. 5(d).Google Scholar 52
+
+Danzon, P. M. and Towse, A., "Theory and Implementation of Differential Pricing for Pharmaceuticals," in Maskus, K. E., Reichman, J. H., eds., International Public Goods and Transfer of Technology under a Globalized Intellectual Property Regime (Cambridge: Cambridge University Press, 2005): at 425. See also Dutfield, G., "Delivering Drugs to the Poor: Will the TRIPS Amendment Help?" American Journal of Law & Medicine 34, nos. 2-3 (2008): 107-124, at 114 (citing authorities). The term “reference pricing” is somewhat confusing as it also applies to pricing mechanisms within wealthy countries. An alternative term is “virtual arbitrage.” See Outterson, , supra note 47, at 283.Google
+
+Scholar
+
+53
+
+DiMasi, J. A., Hansen, R. W. and Grabowski, H. G., "The Price of Innovation: New Estimates of Drug Development Costs," Journal of Health Economics 22, no. 2 (2003): 151–185; see also Reichman, J. H., "Rethinking the Role of Clinical Trial Data in International Intellectual Property Law: The Case for a Public Goods Approach," Marquette Intellectual Property Law Review 13, no. 1 (2009): 1-68, at 9-11 (citing authorities with different estimates).CrossRefGoogle
+
+---
+
+Scholar 54 See Danzon, and Towse,, supra note 52, at 445.Google Scholar 55 McElligott, S., “Addressing Supply Side Barriers to Introduction of New Vaccines to the Developing World, ” American Journal of Law & Medicine 35 (forthcoming 2009).CrossRefGoogle Scholar 56 See Flynn, et al., supra note 15.Google Scholar 57 Id.Google Scholar 58 Id.Google Scholar 59 Id. While relatively few countries may fit the 10–90 matrix of South Africa, the principle could apply equally to other countries where the relevant matrix was 20–80 or 30–70, rather than 10–90.Google Scholar 60 Id.Google Scholar 61 The Ministry of Public Health & The National Health Security Office Thailand, Facts and Evidences on the 10 Burning Issues Related to the Government Use of Patents on Three Patented Essential Drugs in Thailand, at 14–15, available at <www.moph.go.th/hot/ WhitePaperCL-EN.pdf> (last visited April 23, 2009); see Abbott, and Reichman,, supra note 19, at 953. + (last + visited + April + 23, + 2009); + see + Abbott, + and + Reichman, +, + supra + note + 19, + at + 953. > Google Scholar 62 See Outterson,,, supra note 47 at 203–216, 232–235.Google Scholar 63 Personal communication with Professor Kevin Outterson, March 10, 2009.Google Scholar 64 See Flynn, et al., supra note 15.Google Scholar 65 For details, see Reichman, J. H. with Hasenzahl, C., Non-Voluntary Licensing of Patented Inventions: The Canadian Experience, UNCTAD/ICTSD Project on IPRs and Sustainable Development,
+
+---
+
+October 2002, at 37–38; see also McFetridge, D. G., Intellectual Property, Technology, Diffusion and Growth in the Canadian Economy, in Anderson, R. D. and Gallini, N. T., eds., Competition Policy and Intellectual Property Rights in the Knowledge Economy (Calgary: University of Calgary Press, 1998): at 65.Google Scholar 66
+
+Id. (Reichman with Hasenzahl), at 38–44. Industry Canada has, at various times, expressed satisfaction with the results of this arrangement.Google Scholar
+
+67
+
+Id., at 42–43; see North American Free Trade Agreement, CanadaU.S. -Mexico, at ch. 17, December 17, 1992, 32 I.L.M. 605 (1993) [hereinafter NAFTA].Google Scholar
+
+68
+
+See TRIPS, supra note 4, at arts. 31(a), (c).Google Scholar
+
+69
+
+Id., at art. 31(b). This provision is waived in the case of national emergency, other circumstances of extreme urgency or in cases of governmental use. Id.Google Scholar
+
+70
+
+Id., at art. 31(g). For example, an abusive use of a patent could be purged, allowing a claim to void such a license.Google Scholar
+
+71
+
+Id., at art. 31(j).Google Scholar
+
+72
+
+Id., at art. 31(h). The remuneration must take “into account the economic value of the authorization. ” Id. See also Love, J., Remunerations Guidelines for Non-Voluntary Use of a Patent on Medical Technologies, WHO/TCM/2005.1, 2005, available at <http://www.who.int/medicines/areas/technical cooperation/ WHOTCM2005.1 OMS.pdf> (last visited April 23, 2009); see also Taubman,, supra note 23. + (last + visited + April + 23, + 2009); + see + also + Taubman, +, + supra + note + 23. > Google Scholar
+
+73
+
+See Flynn, et al., supra note 15.Google Scholar
+
+74
+
+See TRIPS, supra note 4, at art. 31(k).Google Scholar
+
+75
+
+Id.Google Scholar
+
+76
+
+---
+
+Id.Google Scholar 77
+
+See, e.g., Hovenkamp, H. J., “The Intellectual Property-Anti-trust Interface, ” University of Iowa College of Law, Legal Studies Research Paper No. 08–46, 2008, at 1979–2007, available at <http://ssrn.com/abstract=1287628> (last visited April 23, 2009); Fox, E. M., “Can Antitrust Policy Protect the Global Commons from the Excesses of IPRs?” in Maskus, K. E. and Reichman, J. H., eds., International Public Goods and Transfer of Technology under a Globalized Intellectual Property Regime (Cambridge: Cambridge University Press, 2005): at 758,758– 769.Google Scholar
+
+78
+
+See, e.g., Arezzo, E., “Intellectual Property Rights at the Cross-road between Monopolization and Abuse of a Dominant Position: American and European Approaches Compared, ” John Marshall Journal of Computer & Information Law 24, no. 3 (2006): 456– 494.Google Scholar 79
+
+See, e.g., Fox, , supra note 77, at 759–64; Ullrich, H., “Expansionist Intellectual Property Protection and Reductionist Competition Rules: A TRIPS Perspective, ” in Maskus, K. E. and Reichman, J. H., eds., International Public Goods and Transfer of Technology Under a Globalized Intellectual Property Regime (Cambridge: Cambridge University Press 2005): at 726, 726–757.Google Scholar 80
+
+See Hovenkamp,, supra note 77, at 1991; Illinois Tool Works v. Indep. Ink, 126 Sup. Ct. 1281 (2006); see also Cotter, T. F., “Misuse, ” Houston Law Review 44, no. 4 (2007): 901–964; Janis, M. D., “‘Minimal’” Standards for Patent-Related Antitrust Law under TRIPS, ” in Maskus, K. E. and Reichman, J. H., eds., International Public Goods and Transfer of Technology under a Globalized Intellectual Property Regime (Cambridge: Cambridge University Press, 2005): 774, 790-792.Google Scholar 81
+
+See, e.g., Fox,, supra note 77, at 758; Ullrich,, supra note 79, at 726.Google Scholar
+
+82
+
+See, e.g., Ghidini, G., Intellectual Property and Competition Law: The Innovation Nexus (Cheltenham: Elgar, 2006): 103–
+
+---
+
+111.CrossRefGoogle Scholar 83
+
+See supra note 77; Drexl, J., “The Critical Role of Competition Law in Preserving Public Goods in Conflict with Intellectual Property Rights, ” in Maskus, K. E. and Reichman, J. H., eds., International Public Goods and Transfer of Technology under a Globalized Intellectual Property Regime (Cambridge: Cambridge University Press, 2005): at 709–725.Google Scholar 84
+
+See Flynn, et al., supra note 15.Google Scholar 85
+
+See, e.g., Coco, R., “Antitrust Liability for Refusal to License Intellectual Property: A Comparative Analysis and the International Setting, ” Marquette Law Review 12, no. 1 (2008): 10–21 (citing authorities).Google Scholar
+
+86 See Fox, , supra note 77, at 768.Google Scholar 87
+
+See Paris Convention, supra note 1, at art. 5A; Bodenhausen, G. H. C., Guide to the Application of the Paris Convention for the Protection of Industrial Property, as Revised at Stockholm in 1967, 1968, at 71. However, current precedents governing misuse in the U.S. may no longer support such outcomes. See Hovenkamp,,, supra note 77, at 1091–1095; Reichman, with Hasenzahl,, supra note 2; Cotter,, supra note 80.Google Scholar 88
+
+See TRIPS, supra note 4, at art. 3, 8.2, 40; Paris, Convention, supra note 1, at art. 2(1).Google Scholar 89
+
+However, for the disappointing story of the first attempt to use the Waiver, see Goodwin,, supra note 29.Google Scholar 90
+
+For an existing WTO panel decision that casts doubt on efforts to use art. 8 of the TRIPS Agreement to expand the limited exceptions to a patentee's exclusive rights under art. 30 of that Agreement, see Canada-Patent Protection of Pharmaceutical Products, WT/DS114/ R, March 17, 2000.Google Scholar 91
+
+Bird, R. C., "Developing Nations and the Compulsory License: Maximizing Access to Essential Medicines while Minimizing
+
+---
+
+Investment Side Effects, ” Journal of Law, Medicine & Ethics 37, no. 2 (2009): 209–222.CrossRefGoogle Scholar 92 Id., at 1 (citing World Health Organization, The World Medicines Situation at 61 (2004)).Google Scholar 93 Id.Google Scholar 94 Id.Google Scholar 95 Id.Google Scholar 96 Id.Google Scholar 97 Id.Google Scholar 98 Id.Google Scholar 99 Id.Google Scholar 100 Id.Google Scholar 101 Id.Google Scholar 102 See infra text accompanying notes 166–172.Google Scholar 103 See Abbott, and Reichman,, supra note 19, at 973–980; see infra text accompanying notes 159–160.Google Scholar 104 See Outterson,, (Access to Medicine and TRIPS), supra note 21, at 289–91.Google Scholar 105 Id., at 296.Google Scholar 106 See Abbott, and Reichman,, supra note 19, at 952, 954–956.Google Scholar 107 See Flynn, et al., supra note 15; Bird,, supra note 91.Google Scholar 108 See infra note 135 and accompanying text.Google Scholar
+
+---
+
+109 See Bird,, supra note 91.Google Scholar 110 Id.Google Scholar 111 See Lybecker, and Fowler,, supra note 33.Google Scholar 112 Id.Google Scholar 113 See Goodwin,, supra note 29, at 2.Google Scholar 114 Compare Abbott, and Reichman,, supra note 19, at 957.Google Scholar 115 See Lybecker, and Fowler,, supra note 33.Google Scholar 116 Id.Google Scholar 117 See Abbott, and Reichman,, supra note 19, at 953.Google Scholar 118 Id.Google Scholar 119 Id., at 952.Google Scholar 120 See Lybecker, and Fowler,, supra note 33.Google Scholar 121 See Outterson, (Access to Medicine and TRIPS), supra note 21, at 282.Google Scholar 122 Id.Google Scholar 123 Id. (citing Declaration on TRIPS, supra note 5, at para. 4, 5[b], 5[c], 5[d]; U.S. Gen. Accounting Office, GAO Report 07–1198, U.S. Trade Policy Guidance on WtO Declaration on Access to Medicines May Need Clarification 15, 19, 23, September 2007); accord Abbott, and Reichman,, supra note 19, at 936–37 (citing Paragraph 6 Decision supra note 22; Amendment of TRIPS, supra note 22).Google Scholar 124 Regulation (EC) No. 816/2006 of the European Parliament and of the Council of 17 May 2006 on compulsory licensing of patents
+
+---
+
+relating to the manufacture of pharmaceutical products for export to countries with public health problems.Google Scholar 125 See Abbott, and Reichman,, supra note 19, at 947.Google Scholar 126 See Lybecker, and Fowler,, supra note 33.Google Scholar 127 Id.Google Scholar 128 Id.Google Scholar 129 Id.Google Scholar 130 See supra note 123 and accompanying text; see also Outterson,, (Access to Medicine and TRIPS), supra note 21.Google Scholar 131 See TRIPS, supra note 4, at art. 31(b).Google Scholar 132 Id.Google Scholar 133 See Abbott, and Reichman,, supra note 19, at 952. The extent to which negotiations actually occurred is disputed by both sides.Google Scholar 134 See id., at 953.Google Scholar 135 Id., at 952–953 (noting compensation at 0.5%).Google Scholar 136 See infra notes 166–172 and accompanying text.Google Scholar 137 See, e.g., Bird,, supra note 91 (case of Pfizer with respect to Viagra in Egypt).Google Scholar 138 See, e.g., Lybecker, and Fowler,, supra note 33 (case of Abbott Laboratories in Thailand).Google Scholar 139 See Bird,, supra note 91.Google Scholar 140 See Lybecker, and Fowler,, supra note 33.Google Scholar 141
+
+---
+
+See, e.g., Yu, P., "Intellectual Property, Economic Development, and the China Puzzle," in Gervais, D., ed., Intellectual Property, Trade and Development (Oxford: Oxford University Press, 2007): 173, 176–183.Google Scholar
+
+142
+
+Maskus, K. E., “The Role of Intellectual Property Rights in Encouraging Foreign Direct Investment and Technology Transfer, ” Duke Journal of Comparative & International Law 9, no. 1 (1998): 109–161, at 129–130.Google Scholar 143
+
+See TRIPS, supra note 4, at arts. 2.1 (incorporating Paris Convention arts. 1–12), 27 (novelty); Paris Convention, supra note 1, at arts. 2(1) (national treatment), 4bis(1) (independence of patents).Google Scholar
+
+144
+
+The task of reverse-engineering key active ingredients (APIs) nonetheless remains difficult and costly for most developing countries. See, e.g., Abbott, F. M. and Dukes, G., Global Pharmaceutical Policy: Ensuring Medicines for Tomorrow's World (Cheltenham: Elgar, forthcoming 2009): at 123–24.CrossRefGoogle Scholar
+
+145
+
+See TRIPS, supra note 4, at art. 6; Paragraph 6 Decision, supra note 22; Abbott, and Reichman,, supra note 19, at 975-976.Google
+
+Scholar 146
+
+See e.g., Lybecker, and Fowler,, supra note 33; Bird,, supra note 91.Google Scholar
+
+147
+
+See supra note 22.Google Scholar
+
+148
+
+Compare Flynn, et al., supra note 15.Google Scholar 149
+
+See infra text accompanying notes 159–162; compare Reichman, J. H., “The TRIPS Agreement Comes of Age: Conflict or Cooperation with the Developing Countries?” Case Western Reserve Journal of International Law 32, no. 3 (2000): 441–470.Google Scholar 150
+
+See Lybecker, and Fowler, , supra note 33; Bird, , supra note 91; Sykes, A. O., "TRIPs, Pharmaceuticals, Developing Countries, and
+
+---
+
+the Doha ‘Solution, ’” Chicago Journal of International Law 3, no. 1 (2002): 47–68. But see Bird,, supra note 91, noting contrary empirical studies.Google Scholar 151
+
+See, e.g., Outterson, K., “Patent Buy-Outs for Global Disease Innovations for Low- and Middle-Income Countries, ” American Journal of Law & Medicine 32, nos. 2–3 (2006): 159–173, at 161; see also Lanjouw, J. O., Beyond TRIPS: A New Global Patent Regime, Brief No. 3, The Center for Global Development, July 2002 (market segmentation); Love, J., Proposal for Patent Pool for Essential Medicines (PPEM), Addis Ababa, March 13, 2005.Google Scholar 152
+
+See, e.g., Outterson, and Kesselheim,, supra note 48, at 136–137; Outterson,, supra note 151.Google Scholar 153
+
+Sykes, Compare, supra note 150.Google Scholar 154
+
+See Lybecker, and Fowler,, supra note 33.Google Scholar 155
+
+See Outterson, and Kesselheim, supra note 48, at 136–137; Outterson,, supra note 151, at 159–173.Google Scholar 156
+
+See Moran, M., Ropars, A. -L., Guzman, J., Diaz, J. and Garrison, C., The New Landscape of Neglected Diseases Drug Development, 2005, available at <www.thegeorgeinstitute.org/shadomx/apps/fms/ fmsdownload.cfm?file_uuid=F2B06396-EEA0–851E-3049-
+
+C9A030AEDE0F&siteName = iih > (last visited April 23, 2009); Drugs for Neglected Diseases Initiative (DNDi), New, Once-a-Day Fixed-Dose Combination Against Malaria Now Available, March 1, 2007; DNDi, MSF and DNDi Call For Scale-Up of R&D For Neglected Diseases, Press Release, February 23, 2009, available at < http:// www.msfaccess.org/media-room/press-releases/press-releasedetail/?tx_ttnews%5Btt_news%5D = 1535&cHash = 86e19504b2 > (last visited April 23, 2009); Kiechel, J. -R. and Pecoul, B., “Innovative Partnership Brings New ACT Free of Patents, ” Guest Blog, PLOS, February 26, 2007, available at < http://www.plos.org/ cms/node/208 > (last visited April 23, 2009). + (last + visited + April + 23, + 2009); + Drugs + for + Neglected + Diseases + Initiative + (DNDi), + New, + Once-a-Day + Fixed-
+
+---
+
+Dose + Combination + Against + Malaria + Now + Available, + March + 1, + 2007; + DNDi, + MSF + and + DNDi + Call + For + ScaleUp + of + R&D + For + Neglected + Diseases, + Press + Release, + February + 23, + 2009, + available + at + + (last + visited + April + 23, + 2009); + Kiechel, + J. -R, + and + Pecoul, + B., + “Innovative + Partnership + Brings + New + ACT + Free + of + Patents, ” + G + PLOS, + February + 26, + 2007, + available + at + + (last + visited + April + 23, + 2009). > Google Scholar 157 See, e.g., So, A. D., “Enabling Conditions for the Scientific Commons, ” paper presented at the Conference on Technology Development in the Life Sciences, Program on Science, Technology and Global Development, The Earth Institute at Columbia University, May 20–21, 2004.Google Scholar 158 See, e.g., Abbott, and Reichman,, supra note 19, at 981–982.Google Scholar 159 Id., at 969–981 (“Making the Amendment System Work”).Google Scholar 160 Id., at 973–974.Google Scholar 161 Id., at 977–978. See also Reichman, J. H. and Lewis, T., “Using Liability Rules to Stimulate Local Innovation in Developing Countries, ” in Maskus, K. E. and Reichman, J. H., eds., International Public Goods and Transfer of Technology under a Globalized Intellectual Property Regime (Cambridge: Cambridge University Press, 2005): at 337, 337–366.Google Scholar 162 See Flynn, et al., supra note 15; Abbott,, Reichman,, supra note 19, at 982 (“Changing the Marketing Model”).Google Scholar 163 See Bird,, supra note 91; Lybecker, and Fowler,, supra note 33.Google Scholar 164 See Abbott, and Reichman,, supra note 19, at 954.Google Scholar 165
+
+---
+
+Food & Drug Letter, Brazil, Thailand Lose Trading Privileges in Wake of IP Disputes, August 17, 2007. In 2006, Thailand had $4.2 billion in GSP qualified exports to the U.S. Jones, V. C., CRS Report for Congress, Generalized System of Preferences: Background and Renewal Debate, at Table A2, January 25, 2008, available at <http:// www.nationalaglawcenter.org/assets/crs/RL33663.pdf> (last visited April 23, 2009). + (last + visited + April + 23, + 2009). > Google Scholar
+
+166
+
+The DSU, supra note 6, at art. 23 provides as follows: 1. When Members seek the redress of a violation of obligations or other nullification or impairment of benefits under the covered agreements or an impediment to the attainment of any objective of the covered agreements, they shall have recourse to, and abide by, the rules and procedures of this Understanding. 2. In such cases, Members shall: (a) not make a determination to the effect that a violation has occurred, that benefits have been nullified or impaired or that the attainment of any objective of the covered agreements has been impeded, exceptGoogle Scholar
+
+167
+
+See Abbott, and Reichman,, supra note 19, at 980–81. Freedom from unilateral action is, indeed, a major reason developing countries signed onto the Agreement Establishing the WTO of 1994 in the first place.Google Scholar
+
+168
+
+See Panel Report, United States – Sections 301–310 of the Trade Act of 1974, WT/DS152/R, December 22, 1999; Abbott,, Reichman,, supra note 19, at 980–981. While GSP privileges are not bound by GATT, and therefore remain revocable, one may doubt that this revocation can be used as a unilateral sanction for some alleged violation of TRIPS.Google Scholar
+
+169
+
+See Marrakesh Agreement Establishing the WTO, supra note 4, at art. II (2) (DSU is “binding on all Members”); DSU, supra note 6, at art. 23 (quoted supra note 166).Google Scholar 170
+
+Vienna Convention on the Law of Treaties, UN Doc. A/Conf.39/27, 1155 UNTS 331, 8 I.L.M. 679, at art. 60 (1969), available at <http://untreaty.un.org/ilc/texts/instruments/english/ conventions/1 1 1969.pdf> (last visited April 23, 2009).
+
+---
+
++ (last + visited + April + 23, + 2009). > Google Scholar 171 Id., at art 60.2.Google Scholar 172 See TRIPS, supra note 4, at art. 64.Google Scholar
+
+56
+
+Cited by
+
+## Cited by
+
+Loading...
+
+[image] Cited by
+
+- • [image] 56
+• [image]
+Crossref Citations [image]
+
+This article has been cited by the following publications. This list is generated based on data provided by Crossref.
+
+Outterson, Kevin 2009. Pharmaceutical Innovation: Law & the Public's Health. Journal of Law, Medicine & Ethics, Vol. 37, Issue. 2, p. 173.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Byeon, Jin-Ok and Chung, Jung-Hoon 2010. Compulsory Licensing as a price control and supply policy of patented drugs : Is it a possible alternative in South Korea?. Korean Journal of Health Policy and
+
+---
+
+Administration, Vol. 20, Issue. 1, p. 64.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Orsi, Fabienne and d'Almeida, Cristina 2010. Soaring antiretroviral prices, TRIPS and TRIPS flexibilities: a burning issue for antiretroviral treatment scale-up in developing countries. Current Opinion in HIV and AIDS, Vol. 5, Issue. 3, p. 237.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Barnighausen, T. Kyle, M. Salomon, J. A. and Waning, B. 2012. Assessing the population health impact of market interventions to improve access to antiretroviral treatment. Health Policy and Planning, Vol. 27, Issue. 6, p. 467.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Löfgren, Hans and Williams, Owain David 2013. The New Political Economy of Pharmaceuticals. p. 1.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+ZAMAN, Khorsed 2013. The TRIPS Patent Protection Provisions and Their Effects on Transferring Climate Change Technologies to LDCs and Poor Developing Countries: A Critical Appraisal. Asian Journal of International Law, Vol. 3, Issue. 1, p. 137.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+---
+
+Shah, Aakash Kaushik Warsh, Jonathan and Kesselheim, Aaron S. 2013. The Ethics of Intellectual Property Rights in an Era of Globalization. Journal of Law, Medicine & Ethics, Vol. 41, Issue. 4, p. 841.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Timmermann, Cristian and van den Belt, Henk 2013. Intellectual Property and Global Health: From Corporate Social Responsibility to the Access to Knowledge Movement. Liverpool Law Review, Vol. 34, Issue. 1, p. 47.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Nicol, Dianne and Owoeye, Olasupo 2013. Using TRIPS flexibilities to facilitate access to medicines. Bulletin of the World Health Organization, Vol. 91, Issue. 7, p. 533.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Mishra, Amrita and Bubela, Tania 2014. Legal Agreements and the Governance of Research Commons: Lessons from Materials Sharing in Mouse Genomics. OMICS: A Journal of Integrative Biology, Vol. 18, Issue. 4, p. 254.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Stavropoulou, Charitini and Valletti, Tommaso 2015. Compulsory licensing and access to drugs. The European Journal of Health Economics, Vol. 16, Issue. 1, p. 83.
+
+---
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Guennif, Samira 2015. La licence obligatoire: outil emblématique de la protection de la santé publique au Sud. Revue de la régulation, Vol. 17, Issue. ,
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Goyal, Yugank 2015. Compulsory Licensing. Vol. 22, Issue. , p. 437.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Rana, Proteesh and Roy, Vandana 2015. Generic medicines: issues and relevance for global health. Fundamental & Clinical Pharmacology, Vol. 29, Issue. 6, p. 529.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Zapatero Miguel, Pablo 2015. Legal and policy foundations for global generic competition: Promoting affordable drug pricing in developing societies. Global Public Health, Vol. 10, Issue. 8, p. 901.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Okpechi, Ikechi G. Swanepoel, Charles R. and Venter, Francois 2015. Access to medications and conducting clinical trials in LMICs. Nature Reviews Nephrology, Vol. 11, Issue. 3, p. 189.
+
+- • CrossRef
+---
+
+- • OpenURL query
+• Google Scholar
+Rai, Arti K and Sherkow, Jacob S 2016. The changing life science patent landscape. Nature Biotechnology, Vol. 34, Issue. 3, p. 292.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Lee, Kah Seng Khan, Tahir M. and Ming, Long Chiau 2016. Review of data exclusivity provisions in the Trans-Pacific Partnership agreement. Research in Social and Administrative Pharmacy, Vol. 12, Issue. 4, p. 545.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+2016. Trade, Investment, Innovation and their Impact on Access to Medicines. p. 219.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+de Morais, Rafael Pinho 2016. Competition Law Enforcement in the BRICS and in Developing Countries. p. 149.
+
+- • CrossRef
+• OpenURL query
+• Google Scholar
+Download full list
+
+![Figure](figures/figure_010.png)
+
+---
+
+## Our Site
+
+- • Accessibility
+• Contact & Help
+• Legal Notices
+• Cookie Settings
+## Quick Links
+
+- • Cambridge Core
+• Cambridge Open Engage
+• Cambridge Aspire website
+## Our Products
+
+- • Journals
+• Books
+• Elements
+• Textbooks
+• Courseware
+## Join us online
+
+Location
+
+Please choose a valid location.
+
+Update Legal Information
+
+---
+
+- • Rights & Permissions
+• Copyright
+• Privacy Notice
+• Terms of Use
+• Cookies Policy
+Cambridge University Press 2025
+
+Cancel
+
+Confirm
+
+𠄍
+
+## Save article to Kindle
+
+To send this article to your Kindle, first ensure noreply@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about saving to your Kindle.
+
+Note you can select to save to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be saved to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.
+
+Find out more about the Kindle Personal Document Service.
+
+Comment: Compulsory Licensing of Patented Pharmaceutical Inventions: Evaluating the Options
+
+- • Volume 37, Issue 2
+• Jerome H. Reichman
+• DOI: https://doi.org/10.1111/j.1748-720X.2009.00369.x
+Your Kindle email address
+
+@free.kindle.com @kindle.com (service fees apply)
+
+---
+
+Available formats PDF
+
+By using this service, you agree that you will only keep content for personal use, and will not openly distribute them via Dropbox, Google Drive or other file sharing services
+
+Cancel
+
+Save
+
+![Figure](figures/figure_004.png)
+
+## Save article to Dropbox
+
+To save this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Dropbox account. Find out more about saving content to Dropbox.
+
+Comment: Compulsory Licensing of Patented Pharmaceutical Inventions: Evaluating the Options
+
+- • Volume 37, Issue 2
+• Jerome H. Reichman
+• DOI: https://doi.org/10.1111/j.1748-720X.2009.00369.x
+Available formats PDF
+
+By using this service, you agree that you will only keep content for personal use, and will not openly distribute them via Dropbox, Google Drive or other file sharing services
+
+Cancel
+
+Save
+
+![Figure](figures/figure_013.png)
+
+## Save article to Google Drive
+
+To save this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Google
+
+---
+
+Drive account. Find out more about saving content to Google Drive.
+
+Comment: Compulsory Licensing of Patented Pharmaceutical Inventions: Evaluating the Options
+
+- • Volume 37, Issue 2
+• Jerome H. Reichman
+• DOI: https://doi.org/10.1111/j.1748-720X.2009.00369.x
+Available formats PDF
+
+By using this service, you agree that you will only keep content for personal use, and will not openly distribute them via Dropbox, Google Drive or other file sharing services
+
+Cancel
+
+Save
+
+![Figure](figures/figure_007.png)
+
+Reply to: Submit a response
+
+Title * Please enter a title for your response.
+
+![Figure](figures/figure_010.png)
+
+Close Contents help
+
+- - No HTML tags allowed
+- Web page URLs will display as text only
+- Lines and paragraphs break automatically
+- Attachments, images or tables are not permitted
+Please enter your response.
+
+Your details
+
+First name * Please enter your first name. Last name * Please enter your last name.
+
+①
+
+---
+
+Email * Email help Close Email help
+
+Your email address will be used in order to notify you when your comment has been reviewed by the moderator and in case the author(s) of the article or the moderator need to contact you directly.
+
+Please enter a valid email address. Occupation Please enter your occupation. Affiliation Please enter any affiliation.
+
+You have entered the maximum number of contributors
+
+## Conflicting interests
+
+Do you have any conflicting interests? * Conflicting interests help
+
+![Figure](figures/figure_007.png)
+
+### Close Conflicting interests help
+
+Please list any fees and grants from, employment by, consultancy for, shared ownership in or any close relationship with, at any time over the preceding 36 months, any organisation whose interests may be affected by the publication of the response. Please also list any non-financial associations or interests (personal, professional, political, institutional, religious or other) that a reasonable reader would want to know about in relation to the submitted work. This pertains to all the authors of the piece, their spouses or partners.
+
+Yes No More information * Please enter details of the conflict of interest or select
+
+---
+
+'No'.
+
+Please tick the box to confirm you agree to our Terms of use. *
+
+Please accept terms of use.
+
+Please tick the box to confirm you agree that your name, comment and conflicts of interest (if accepted) will be visible on the website and your comment may be printed in the journal at the Editor's discretion. *
+
+Please confirm you agree that your details will be displayed.
+

processed/00067_W1981194769/document.md

@@ -0,0 +1,529 @@
+![Figure](figures/figure_000.png)
+
+Research Policy 42 (2013) 1780-1792
+
+Contents lists available at ScienceDirect
+
+Research Policy
+
+journal homepage: www.elsevier.com/locate/respol
+
+![Figure](figures/figure_005.png)
+
+# Reassessing patent propensity: Evidence from a dataset of R&D awards, 1977–2004
+
+![Figure](figures/figure_007.png)
+
+Roberto Fontana$^{a,*}$, Alessandro Nuvolari$^{b}$, Hiroshi Shimizu$^{c}$, Andrea Vezzulli$^{d}$
+
+a Department of Economics ad Management, University of Pavia, Via San Felice 5, 27100, Pavia & CRIOS – Bocconi University, Via Sarfatti 25, 20139 Milano, Italy
+
+$^{b}$ LEM - Sant' Arana School of Advanced Studies, Piazza Martini della Libertà 33, 56127 Pisa, Italy
+
+c Institute of Innovation Research – Hitotsubashi University, Tokyo, Japan
+
+d UECE-ISEG, Universitade Técnica de Lisboa, Rua Miguel Lupi, 20, 1249-078 Lisboa, Portugal
+
+## A R T I C L E I N F O
+
+### Article history:
+
+Received 28 November 2011 Accepted 25 May 2012 Available online 16 September 2013
+
+### Keywords:
+
+Innovation Patent propensity R&D awards
+
+## A B S T R A C T
+
+It is well known that not all innovations are patented, but the exact volume of innovative activities undertaken outside the coverage of patent protection and, relatedly, the actual propensity to patent an innovation in different contexts remain, to a major degree, a matter of speculation. This paper presents an exploratory study comparing systematically patented and unpatented innovations over the period 1977–2004 across industrial sectors. The main data source is the 'R&D 100 Awards' competition organized by the journal Research and Development. Since 1963, the magazine has been awarding this prize to the 100 most technologically significant new products available for sale or licensing in the year preceding the judgments. We match the products winners of the R&D 100 awards competition with USPTO patents and we examine the variation of patent propensity across different contexts (industries, geographical areas and organizations). Finally we compare our findings with previous assessments of patent propensity based on several sources of data.
+
+© 2013 Elsevier B.V. All rights reserved.
+
+## 1. Introduction
+
+Within the Economics of Technical change and Innovation Studies (ETIS) literature it is today widely acknowledged that many innovations are not patented. In principle, there may be three types of explanation accounting for the inventor's decision of not taking a patent ( Basberg, 1987 ). The first explanation is that the innovation is simply not patentable. In this case, the inventor believes that the innovation in question does not represent suitable patent matter (e.g. the patentability of ‘pure’ software program was still a matter of contention in many jurisdictions not so long ago). Alternatively, the innovation is in principle patentable but the inventor may anticipate that the inventive step embodied in her innovation is not 'high' enough to be deemed worthy of patent protection by patent examiners. In both these two examples the decision of not patenting is determined by the fact that this is not actually possible (or believed possible). The third possibility is that the inventor, even when conceiving taking a patent as a fully feasible course of action,
+
+decides not to patent the innovation because she actually prefers to do so. In this case, even though the innovation is patentable and worth patenting, the inventor prefers industrial secrecy or other alternative strategies to extract some economic returns from her innovation. This third case is the most interesting one from the viewpoint of innovation scholars.
+
+The existence of `appropriability strategies' that are alternative to patenting was initially documented by early economists of innovation ( Kuznets , 1962 ; Schmookler , 1966 ; Taylor and Silberston , 1973 ) . Later on, the survey studies by Mansfield ( 1986 ) and Levin et al. ( 1987 ) during the 1980s highlighted that, in most industries, patent protection was not the typical tool adopted by firms for the extraction of economic returns from innovations, a finding further corroborated by subsequent research both in US ( Cohen et al. , 2000 ) and Europe ( Arundel and Kabla , 1998 ) . All these research results are frequently cited and surely represent important pieces of evidence discussed in the innovation literature. However, as aptly noted by De Rasenfosse ( 2010 ) , on closer inspection, it is difficult to avoid the impression that the major implication of these findings (i.e. that a sizeable share of innovations is never patented) has gone completely neglected. To be sure, many empirical investigations acknowledge the limitations of patents as innovation indicators. However, once these limitations are gauged against their advantages (i.e. availability and richness of information they provide)
+
+* Corresponding author. Tel.: +39 0258363037; fax: +39 0258363399. E-mail addresses: roberto.fontana@unibocconi.it (R. Fontana), alessandro.nuvolari@sssup.it (A. Nuvolari), shimizu@iir.hit-u.ac.jp (H. Shimizu), andreav@iseg.utl.pt (A. Vezzulli).
+
+0048-7333/$ – see front matter c 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.respol.2012.05.014
+
+---
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+1781
+
+the final choice is to rely on patents if anything because of the sheer difficulty of constructing suitable indicators using alternative sources.1
+
+This state of affairs is deeply unsatisfactory as we incur the risk that even the most carefully designed empirical studies will provide us with a partial, and sometimes even distorted, representation of innovative activities. Furthermore, since our understanding of patent propensity in different contexts is still rudimentary, in many cases, it is also difficult to formulate a sound assessment of the margin of error and of the biases involved in the adoption of patents as innovation indicators. Consider, for instance, the concept of ‘propensity to patent’ usually defined in the literature as the ratio between patents and R&D expenditures ( Scherer, 1983; Hall and Ziedonis, 2001). Though surely legitimate, we should note that this definition of patent propensity is simply describing the overall relationship between patents and innovative efforts and it is only indirectly linked to the actual decision to patent or not a specific innovation (see De Rasenfosse, 2010 for a more extensive discussion).
+
+Interestingly enough, economic historians and historians of technology have instead adopted a more 'straightforward' definition of patent propensity, namely the share of patented innovations in the total number of innovations occurring in a given time period ( Sullivan , 1989 ; Moser , 2005 , 2012 ) . 2 This conceptualization of patent propensity, although intuitively appealing, is not of immediate empirical operationalization because it requires some form of direct assessment of the total amount of innovations occurring in a given time period. Still, historians have displayed considerable ingenuity both in the identification of sources (alternative to patents) that could be used for formulating quantitative assessments of overall innovative output in different contexts and periods, and in connecting these sources with the patent evidence for constructing estimates of patent propensity. In this respect, the recent contributions of Moser ( 2005 , 2012 ) can be regarded as among one of the most successful examples of this approach.
+
+This paper argues that these historical investigations suggest a framework of inquiry that can, and should, be fruitfully extended within the field of ETIS. In the paper we present an application of this method using a database of ‘important’ industrial innovations occurred between 1977 and 2004. Our source of data is the ‘R&D 100 Awards’ competition organized by the journal Research and Development. Since 1963, this journal (which at that time was called Industrial Research ) has been awarding a prize to 100 most technologically significant new products available for sale or licensing in the year preceding the judgement. The potential of this source was already reckoned by Carpenter et al. (1981) and Scherer (1989) . This source has also been more recently used by Block and Keller (2009) to document the increasing role of public institutions and public funding in the generation of innovations in the US economy in the period 1971–2006.
+
+Though recent and therefore not 'historical' in a strict sense, the database covers 30 years of innovations, several manufacturing industries, and different types of economic actors, both corporations and Universities and Public Research Organizations (PROs). These data seem particularly appropriate for studying the propensity to patent for the following reasons: (i) the data consider innovations that have been recognized by a jury of experts
+
+as significant and they should be commercially feasible at the time of the awards; (ii) most of the awards have been granted to large corporations accounting for a sizeable amount of total R&D investments; (iii) the data cover a relatively long time period allowing us to take into account changes in the determinants of the propensity to patent over time. Using these data we are able to assess systematically the relative influence of sector, organization, and inventor specific characteristics on the actual decision of taking or not a patent.
+
+Our study is based on a sample of about 3000 innovations that have received an award. For each innovation in our dataset we have retrieved information concerning: years of the award, description of the innovation, type and name of applicant organization(s), application domain of the innovation, country, and name of inventor(s). The first step of our analysis is to match awarded innovations with patents using the search engine of the USPTO website. Then, on the basis of the invention description contained in the journal, we classify all the awarded innovations in thirty different sectors of activity. In this way, the data allow a thorough comparison between patented and not patented innovations across different industrial sectors, countries, types of organization and types of innovation. Our results highlight the following patterns. First, a large share of innovations is not protected by means of patents. Second, we are able to point out the existence of systematic significant differences in patenting propensity across sectors, geographical areas, types of organization and types of innovation.
+
+The structure of the paper is as follows. Section 2 reviews the empirical literature on the effectiveness of patents as appropriability tools. Section 3 describes in detail our data source, our matching procedure and the limitations of the dataset. Section 4 presents our analysis of patent propensity across different dimensions. Section 5 compares our findings with those of previous assessments of patent propensity carried out using different types of data. Section 6 concludes and draws some methodological implications with particular reference to the possibility of extending the framework of inquiry adopted by economic historians and historians of technology to contemporary studies of patent propensity.
+
+## 2. Patents as indicators of innovation and their limitations
+
+Scholars within the ETIS tradition have relied intensively on patents to investigate the sources, nature, and the effects of innovative activities. Innovative activities are inherently elusive phenomena which almost by definition are bound to defy systematic attempts of (quantitative) measurement. It is not surprising then that the existence of patent records has been regarded for a long time, mostly by economists, but also by other scholars of innovation with different disciplinary backgrounds, as an almost unique source of insights into the nature of inventive activities. The main merits of patent records as a source for measuring innovation are easy to summarize: (i) they are by definition related to innovative activities; $^3$ (ii) they are readily available (allowing to economize efforts of data collection); $^4$ (iii) they are available for relatively long periods of time; (iv) they contain a significant depth of information (inventors' names and addresses, ownership of the
+
+1 In this respect, we would argue that a large bulk of the most recent research on innovation seems implicitly to apply to patents Winston Churchill's famous quip on democracy: "Democracy is the worst form of government, except all the others". In the case of patents, the jibe would probably sound like: "patents are the worst innovation indicator, except all the others".
+
+2 Moser (2005, 2012) uses the term 'patenting rate' to define the share of patented inventions in the total number of inventions.
+
+3 In the words of Griliches (1990, p. 1169): "[A] patent represents a minimal quantum of invention that has passed both the scrutiny of the patent office as to its novelty and the test of the investment of effort and resources by the inventor and his organization into the development of this product or idea, indicating thereby the presence of a non-negligible expectation as to its ultimate utility and marketability".
+
+4 The `accessibility' of patent as a source has greatly increased over the last 20 years or so thanks to the creation of on line search engine such as ESPACENET and the efforts of construction of data-bases containing information gathered by patent records such as the NBER-US patent data-set ( Hall et al. , 2001 ) and the OECD-PATSTAT dataset.
+
+---
+
+1782
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+innovation, description of the innovation and its relation with previous ones, as represented by patent citations). These factors have made patents the most adopted indicator for scholars interested especially in measuring the output of innovation activities.
+
+Although much progress has been achieved in this way, it is well known that indicators of innovation based on patents suffer from several limitations. These limitations can be summarized citing again from Griliches (1990, p. 1169): “Not all inventions are patentable. Not all inventions are patented and the inventions that are patented differ greatly in their 'quality', in the magnitude of inventive output associated with them”.
+
+The first limitation is clearly the most obvious one and probably the easiest to tackle. Some domains of inventive activity do not constitute patentable subject matter. The solution is to resort to alternative indicators for assessing inventive output in these areas. The second limitation is that not all patentable innovations are actually patented. This means that in contexts characterized by low patent propensity, i.e. in environments in which firms prefer to adopt alternative appropriability strategies, the use of patents as innovation indicator may result in a biased assessment of the volume of innovative activities. $^5$ The existence of a different propensity to patent across industries has indeed been the most important finding of studies based on surveys of the attitudes of R&D personnel towards the use of patents. Mansfield (1986) examined how many patentable innovations were actually patented in a random sample of large US firms in different industries. His results highlighted that in sectors where patents are not regarded as particularly effective ‘appropriability’ mechanisms (i.e. primary metals, electrical equipment, instruments, office equipment, motor vehicles, rubber and textiles) around 34% of potentially patentable inventions were not patented. This percentage is around 16% in those sectors where patents are considered to be more important (i.e. pharmaceuticals, chemicals, petroleum, machinery, and fabricated metal products). Results from subsequent surveys corroborate these early findings. In some industries, secrecy and lead times seem to be more important than patenting for appropriating the returns from innovation ( Levin et al. , 1987 ) . Moreover, patent propensity varies depending on the type of innovation, with firms more likely to apply for a patent for product innovation than for process innovation ( Cohen et al. , 2000 ) . $^6$ Arundel and Kabla ( 1998 ) also find that the (sales weighted) propensity to patent differs across innovation type with relatively lower rates for process innovation (24.8%) than for product innovations (35.9%).
+
+Among the available alternatives, secrecy seems to play an important role in protecting innovation. Looking at a sample of European innovative firms, Arundel (2001) finds that secrecy is generally rated as more valuable than patenting. This is particularly true in the case of product innovation, though the probability of being considered more valuable declines with firm size. Hussinger (2007) carries out a similar analysis using sales figures to assess the importance of alternative means of appropriation rather than
+
+individual evaluations. Her findings suggest that secrecy is relatively more important for innovations that are not commercialized.
+
+The third limitation is that patents and innovations differ greatly in their technological and economic significance. 7 In particular, several studies have shown that the `size' distribution of innovations is sharply skewed with the majority of innovations of little or limited technical and economic significance and a restricted number of highly significant innovations ( Silverberg and Verspagen, 2007 ) . Innovation scholars have attempted to deal with this problem by weighting patents using citations or other information such as claims, and family size ( Trajtenberg, 1990 ) . Still, it is acknowledged that these methods represent only imperfect proxies of the quality of the innovation underlying the patent in question. In fact, the most sensible use of these proxies of patent value is to use them as `probabilistic markers' of the underlying economic value of the patents and employ them for the identification of groups of potentially valuable patents ( van Zeebroeck, 2011 ) . Clearly not taking properly into account these variations in the underlying value of patents may again lead to biased assessments of inventive output.
+
+Economic historians seem to have been more sensitive to the limitations of patents as innovation indicators and have explored the potentialities of alternative sources. Moser (2005, 2012) has constructed a dataset of innovations on the basis of the catalogues of nineteenth century industrial exhibitions (in her case she has used the Crystal Palace exhibition of 1851 and the Centennial exhibition in Philadelphia in 1876). According to Moser (2002, pp. 1–2), this type of data covers “economically useful innovations” (i.e. the commercial introduction of new products or processes), rather than “inventions” (i.e. the additions to the stock of technological knowledge). Furthermore, exhibition data measure innovations regardless of whether they are patented or not. Moser’s findings have produced novel insights on the sources of innovation across countries and sectors during the second half of the nineteenth century. Her findings show that in 1851, 89% of British innovations on display at the Crystal palace exhibition were not patented. Even among prize winning innovations, 84% were not patented. Moreover, she finds that patent propensity is affected by the characteristics of the sectors where the innovation occurs, the location of the invention (urban vs. rural), and the quality of the invention (incremental vs. radical).
+
+Brunt et al. (2012) have instead relied upon information on prizes and awards. Using a dataset of awards for inventions promoted by the Royal Agricultural Society of England from 1839 to 1939, they have studied, among other aspects, whether and how prizes affect innovation and patenting. Their findings point to the presence of a positive relationship between prizes and patenting. In particular, the propensity to patent in the technology category targeted by the award increased by 42% for those inventions awarded a gold medal. Moreover, patents are more likely to be renewed when they were taken out of awarded inventions thus suggesting a positive relationship between awards and the quality of the patent.
+
+These recent contributions by historians clearly illustrate the potentialities of assessments of innovative output using data sources that are alternative to patents.8 In particular, we would like to suggest that innovation scholars should consider with particular interest prize and exhibition data, because some of their intrinsic characteristics are likely to render them less prone to the
+
+5 The fact that firms in some contexts prefer alternative appropriability strategies to patenting does not imply that patents are completely irrelevant as a measure of innovation. For example, in semiconductors, appropriability strategies are based more on secrecy and lead times than on patents, but firms are increasingly resorting to patent protection in order to use patents as "bargaining chips" in negotiation with other firms ( Hall and Ziedonis , 2001 ) . As a result, even though in this industry patents are taken for strategic motives rather than for reaping economic returns from a specific innovation, they still provide a useful measure of innovative activities.
+
+6 The survey questionnaire of Levin et al. (1987) did not contain a specific question asking what percentage of innovation a firm typically patented, but simply contained a question asking to assess the relative effectiveness of different appropriability strategies. The survey questionnaire of Cohen et al. (2000) instead contained a specific question asking respondents what percentage of their innovations was patented.
+
+7 In the words of Kuznets (1962, p. 37): "[T]he main difficulty with patent statistics is, of course, the enormous range in the magnitude of the inventions covered […] patented inventions do differ widely in their potential economic magnitude".
+
+8 Interestingly, Schmookler, one of the pioneer of the use of patent statistic in the field of ETIS, was also one of the early scholars to argue in favour of cross-checking the assessments of inventive output based on patent with data on 'important innovations' (Schmookler, 1966).
+
+---
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+1783
+
+pitfalls that typically affect patent data. First, both prize and exhibition data refer to valuable, or to use Moser's words “economically useful”, innovations. In the case of prizes this is almost a tautology given that they have been recognized by experts in the field as superior to alternative available solutions and probably also to existing practices. For the case of innovations displayed at industrial exhibitions, their economic and technological significance will depend on the exact criteria that an artefact must satisfy for being included in the exhibition. Second, and most importantly, both types of data typically comprise innovations with and without patents ( Moser, 2012 ) . In other words, using this type of data allows an assessment of the share of innovations occurring outside of the patent system as well as the construction of a more 'direct' indicator of propensity to patent. This is a crucial advantage with respect to other sources of data.
+
+## 3. The 'R&D 100 Award' database
+
+This paper presents an extension of the method recently employed by historians to the field of ETIS. Using a source of data that so far has received little attention we provide new estimates of patent propensity across industries and over time. Our source of data is the 'R&D 100 Award' competition organized by the magazine Research and Development (previously called Industrial Research). The magazine was founded in 1959 and it represents probably one of the most authoritative regular publications for R&D practitioners. Currently it has an estimated monthly readership of over 80,000. It is estimated that about 75% of the readers works in high-tech industries, whereas the remaining 25% works for government laboratories, universities, and similar organizations. Over 60% of the readers have managerial or executive type of jobs. The 'R&D 100 Award' competition has been running since 1963. Each year the magazine awards with a prize the 100 most technologically significant products available for sale or licensing in the year preceding the judgement. Throughout the years, key breakthroughs inventions such as Polacolor film (1963), the flashcube (1965), the automated teller machine (1973), the halogen lamp (1974), the fax machine (1975), the liquid crystal display (1980), the printer (1986), the Kodak Photo CD (1991), the Nicoderm antismoking patch (1992), Taxol anticancer drug (1993), lab on a chip (1996), and HDTV (1998) have received the prize. In order to apply for the prize inventors, or their employees, must fill an application form providing a detailed description of the product in question. The prize consists of a plaque which is presented in a special ceremony. There is no sum of money involved. The prize is awarded by a jury composed of university professors, industrial researchers and consultants with a certified level of competence in the specific areas they are called to assess. The members of the jury are selected by the editor of the magazine. The main criteria for assessment are two: (i) technological significance (i.e., whether the product can be considered a major breakthrough from a technical point of view); (ii) competitive significance (i.e., how the performance of the product compares to rival solutions available on the market). R&D 100 awards are accolades comparable to the Oscars for the motion picture industry as “they carry considerable prestige within the community of R&D professionals” (Block and Keller, 2009, p. 464).
+
+The technological significance requirement is to be understood in fairly broad terms:9
+
+"[...] products and processes that can change people's lives for the better, improve the standard of living for large numbers of
+
+people, save lives, promote good health, clean up the environment, etc. [...] A cure for cancer or AIDS. An engine that runs on water. A safe, cheap method for cleaning up toxic waste. A vehicle that can fly 800 passengers from New York to Tokyo in two hours. A device that would cut automotive accidents or one that would reduce workplace injuries. A pollution-free herbicide that would increase crop production in Third World countries”.
+
+Accordingly products with a wide potential of application are preferred to those catering to very specific sets of user needs:
+
+“Products or processes that solve very specialized or circumscribed problems could be judged less significant than those that meet larger, more broad-based needs. For example, a new scientific instrument that only benefits a few scientists in a narrow field of interest would have difficulty competing against a device with much broader application. It would depend on how significant the two fields of interest were and how much the technical improvements contributed to the success of each device.”
+
+Furthermore, for attaining the prize there should be a proven link between the effect of the innovation and an improvement in technology:
+
+“[...] these improvements must be attributed to significant breakthroughs in technology. In general, this means your product should exhibit multiple levels of improvement - 53 times faster, 103 greater throughput, 503 times more accurate - or, preferably, orders of magnitude improvement over existing technology. Again, we're looking for ‘leapfrog’ gains in performance, not expected, incremental improvements.”
+
+Additionally the product should also represent a major improvement in comparison with alternative solutions already existing on the market. For this reason, the applicant is requested to provide a 'competitive matrix' illustrating how the product compares with rival solutions already available on the market:
+
+“The competitive matrix should show how your product compares to existing products in terms of the crucial factors involved in the technology. This is your opportunity to give the judges a quick overview of how your product beats the competition. [ . . . ] Include only factors crucial to the technology. Don't waste space (and the judges’ time) throwing in every conceivable factor, just to pad your entry. However, you must list all factors that are indeed crucial to the technology, even if you don't 'win' that particular point. For example, if you fail to include 'hardness' in an entry involving a new alloy, your entry may be looked upon with suspicion by the judges. Some typical factors you might want to include: signal-to-noise ratio, weight, speed, reliability, resolution, cost, accuracy, life expectancy, mean time between failures, sensitivity, reproducibility, strength, power consumption, production yield, environmental operating, intensity, efficiency, size, output rate, bandwidth, number of materials tested, stability”.
+
+The product must exist in marketable form, i.e. it "must have been first available for sale or licensing during the calendar year preceding the judging". Applicants are not restricted to firms, but also governmental laboratories, universities, public research centres are allowed to compete. In case of products resulting from research collaborations, the application form requires to include all the organizations that have provided a "significant contribution to [the] creation of the product" and to provide a description of their precise role in the project. Hence, the rules of the competition make sure that all the parties that have participated to the innovation are
+
+9 All the following quotes concerning the rules and organization of the R&D 100 competition have been retrieved from the magazine website, www.rdmag.com, accessed 23 July, 2010 and on 7 April, 2012.
+
+---
+
+1784
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+properly acknowledged. $^10$ Finally an organization may submit as many products as it wishes at each yearly competition.
+
+There are a number of characteristics of the R&D 100 awards competition that, at least prima facie, appear particularly promis- ing for using this data source to measure innovative output. First, the R&D 100 awards competition seems to represent a good oppor- tunity for companies, government laboratories, etc. to showcase the outcome of their innovative activities. Thus, we can expect that the awards will provide us with a fairly reliable sample of inno- vations attained by R&D performers. Second, R&D 100 awards are granted to innovations that, at least in principle, should embody a significant improvement over the existing state-of-the-art that is clearly documented. In other words awarded innovations should represent, at least in principle, a technological breakthrough. Third, the selection of the awards is made by what appears a competent, authoritative jury of experts. Fourth, R&D awards may be assigned both to patented and not-patented innovations. Finally, there seems to be limited space for strategic behaviour and attempts to conditioning the jury, because the nature of the prize is sim- ply honorific. Alongside these advantages, some biases exist which prevent us from considering these awarded innovations as a fully representative sample of innovative activities. For example, Scherer was struck by the fact that awards covered a very limited number of new weapon systems and a relatively few pharmaceutical products, both sectors notoriously characterized by high R&D investments. Nevertheless, he still regarded the source as capable of providing useful insights on the nature of innovative processes (Scherer, 1999, pp. 67–68).
+
+Retrieving the information from different issues of the magazine, we have constructed a data set of all the R&D 100 awards granted in the period 1977–2004. Our dataset contains 2802 inventions. The total is not equal to 2800, because the requirement of awarding 100 inventions was apparently interpreted with some degree of flexibility. Thus, the amount of awards given in each year in the period we are considering ranges from 97 to 109. A major limitation of the data set is that we do not have information on applying innovations that were not awarded the prize. For this reason, we cannot control whether specific factors, besides the specific technical and economic merits of the innovation, affected the selection of the awards.
+
+### 3.1. Matching awards with patent data
+
+In order to assess the propensity to patent for the awarded innovations included in our sample, we had to look for a possible match between each awarded innovations and one or more USPTO patents. 11 We do not expect to find an exact match between each awarded innovation and one patent. As noted above, awarded innovations represents ‘products’ available for commercialization or license, so it is possible that, in certain areas, individual components of a specific product may be protected by different patents. We regarded the awarded innovation as ‘patented’ also in cases in which one or more components of the innovation in question were actually patented. To carry out the matching exercise we relied upon the following information contained in the R&D 100 database: (i) name of award winning organization(s), (ii) the name of the innovation, (iii) the year of award, (iv) the name of developer(s) and (v) the description of the innovation. We have searched USPTO
+
+patents granted in a time interval ranging from 3 years before to 3 years after the award. The criteria for ascertaining a 'positive' match were the name of the inventors, the name of the organization and the consistency between the description of the ‘R&D 100’ innovation and the title and abstract of the patent (also taking into account the possibility that one or more components of the awarded innovation could have been patented as a separate item). In particular, the patent search procedure entailed the following steps. First, the name of developer as Inventor and the name of organization as Assignee were used to search patents in the USPTO online database. If any patents were found, the patent title and abstract were checked by looking at the information provided by the R&D 100 to see if the patent was corresponding to the award winning invention. Second, if the name of developer was not available, abstract key word search with the name of organization was carried out. The key words were selected from the technological information of the innovations contained in the R&D 100 list. If a match was found at this stage, a further check was carried out to see whether the patent was related with the award winning innovation by cross-checking the information of the patent and the R&D 100 innovation.
+
+We should note that the matching process may be subjected to errors. More specifically, there may be two limitations in the patent searching procedures we have adopted. The first is the time span of the searching. The search considers as relevant to the innovation a patent obtained in plus/minus 3 years from the year of the award. It means that this procedure can overlook the relevant patent(s) that were granted more than 3 years before or after the year of the award. Second, the product name and description contained in the R&D 100 database may not always provide enough information for the identification of one or more possible underlying patents. These limitations notwithstanding, we are confident that our matching procedure provided reliable results in most of the cases. Nevertheless, in uncertain cases, the ‘benefit of the doubt’ was given to a positive match in the sense that we considered the awarded invention as covered by a patent. $^12$ For this reason, if anything, the adopted matching procedure does not contain any in-built bias leading to a systematic underestimation of patent propensity.
+
+### 3.2. The 'quality' of R&D 100 awards
+
+As highlighted by Moser (2012) , one of the chief advantages of employing data on awards and prizes as indicators of innovative output is that, with respect to patents, this type of data should in principle contain only relatively `important' innovations, namely those inventions deemed worthy of receiving the prize or of being put on display. Accordingly, the first exercise we carried out was an attempt of checking whether our R&D 100 dataset contains inventions that are above a certain quality threshold. This is done by replicating an exercise originally performed by Carpenter et al. (1981) . Carpenter et al. (1981) used the 1969 and 1970 R&D awards list and matched these inventions with the corresponding US patents. In this way, they obtained a set of 100 patents whose technological significance had been `certified' by the granting of the award. They then compared the citations received by this group of patents with the citations received by a random sample of patents distributed within the same time cohort. Their results showed that the patents covering the R&D 100 awards received a significantly higher number of citations than the control group. This obviously suggests that R&D 100 innovations are on average of better quality than the `average' patent.
+
+10 In this respect it is also important to note that the rules of the competition state explicitly that: "existing technologies purchased by third parties who then conduct sales, [marketing and other commercialization] efforts" are considered eligible for the award only if the original developer is included in the application.
+
+11 Given the nature of our data source, the most obvious choice was to match R&D 100 innovations with patents taken in US.
+
+12 In this respect our matching procedure was also robust to changes in the 'time range' before and after the award was received.
+
+---
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+1785
+
+Table 1 Patent citations received by awarded innovations and by a random sample of patents (matched by granted year and technology class).
+
+<table><tr><td></td><td>Number</td><td>Mean</td><td>Median</td><td>Standard deviation</td><td>Min</td><td>Max</td></tr><tr><td>R&amp;D 100 patents</td><td>535</td><td>12.88037</td><td>7</td><td>16.17822</td><td>0</td><td>137</td></tr><tr><td>Random sample</td><td>5331</td><td>8.483024</td><td>4</td><td>14.11133</td><td>0</td><td>329</td></tr></table>
+
+
+Note: Mann-Whitney test rejects the Null Hypothesis of equal populations.
+
+Table 2
+
+Total awarded innovations and patents.
+
+<table><tr><td></td><td>Awarded innovations</td><td>Patented innovations</td><td>Share not patented (%)</td></tr><tr><td>All the sample (1977–2004)</td><td>2802</td><td>255</td><td>90.9</td></tr><tr><td>Non-corporate</td><td>886</td><td>25</td><td>97.16</td></tr><tr><td>Corporate only</td><td>1751</td><td>220</td><td>87.44</td></tr></table>
+
+
+Our results for the period 1977–2004 confirm the early findings of Carpenter et al. (1981). For each R&D 100 innovation with one or more USPTO patents we constructed a 'matched random' sample of ten granted patents of the same granted year and of the same International Patent Classification (IPC) class and then compared the number of citations received by patents in this random sample with the citations received by the patents covering an invention within the R&D100 award. 13 The results of this exercise are reported in Table 1. 14
+
+The non-parametric Mann–Whitney test confirms that the median number of citations of patents associated with a R&D 100 invention is significantly not lower than the median of the random matched sample. Overall, this exercise confirms that the innovations which received a R&D 100 award are more significant from a technological or economic viewpoint than the ‘average’ patent in their technological class.
+
+## 4. R&D 100 awards and patent propensity
+
+This section presents our estimates of patent propensity defined as the share of patented innovations in the total number of inventions that have received an R&D 100 award. We compute our estimates across different dimensions of innovative activities (industry, geographical area, organization and type of invention).
+
+### 4.1. Most awarded innovations are not patented
+
+Table 2 reports the number of awarded innovations and the percentage of not patented ones.
+
+As highlighted above, the awarded innovations contained in the R&D 100 list refer to products that are available on the market or for licence when the application is submitted. Hence, it is possible that the data will contain a bias against organizations such as universities and PROs that lack 'downstream' assets for the commercialization of a product. However, it is interesting to note that results in Table 2 are consistent with the results of Block and Keller (2009) showing that a significant share of awarded innovations (more than 30 % ) are generated by non-corporate type of organizations. So it would seem that, the presence of bias notwithstanding, our data cover also a significant segment of the population of non-corporate organization involved in R&D
+
+activities.15 Overall, we found that 269 awarded innovations (slightly less than 10%) were patented according to our matching criteria suggesting that the great majority of innovations were not patented. This percentage is slightly higher (12.56%) when we consider only innovations that have been made by firms.
+
+This estimated patent propensity is in line with the findings of Moser who reports total patenting rates between 11% and 14% for the inventions displayed at the Crystal Palace exhibition of 1851 ( Moser , 2005 , p. 1221) . Of course, this finding should be interpreted keeping in mind the inherent limitations the data set discussed in Section 3 . However, even if we consider possible errors that may have led us to underestimate patenting rates, the result that such a sizeable share of major innovations is not patented is remarkable. In particular, if we consider that `The R&D 100 Award' is a competition aimed at acknowledging the output of formalized R&D efforts, which is notoriously one of the contexts with the highest propensity to patent and that, we are in principle dealing with breakthrough innovations , our findings reveal that patent protection, even in this context, is actually a much less used appropriability strategy than it is generally believed. In this respect, our findings are actually a powerful corroboration of the findings of Moser (2005) . Additionally, they are not inconsistent with the results of both the Yale ( Levin et al. , 1987 ) and the Carnegie Mellon survey ( Cohen et al. , 2000 ) indicating that only in a very restricted number of contexts patents are considered as effective tools for protecting innovation. The obvious policy implication is that the recent developments towards the strengthening of IPR regimes may actually represent a step going in the wrong direction, as it would appear considering the predominant share of innovative activities which is actually carried out without resorting to patent protection. $^16$
+
+Fig. 1 displays the evolution over time of the propensity to patent for all our sample of innovations.
+
+Our estimated propensity to patent is never higher than 20%. Moreover, contrary to what has been suggested by other studies ( Kortum and Lerner, 1999), our evidence does not seem to indicate the existence of significant structural breaks in the time period
+
+13 Our control group is made of granted patents and not of patent applications. At the USPTO a sizeable percentage of patent applications are not granted. In addition to this, many patent applications are discontinued. For these reasons it would have been more rigorous to compare the sample of patented awarded innovations with patent applications. However, these data are not fully available for US, thus the need to rely upon granted patents.
+
+14 The random matched sample includes 5331 patents and not 5350 because for some specific years in some technology classes it was not possible to collect enough patents to create the match.
+
+15 The requirement of availability for sale or licensing implies that the awarded product must be available either for purchase or licensing during the year preceding the award and not necessarily already launched in the market on an extensive scale. Clearly, this allows also organizations with limited capabilities in the commercialization and in the “downstream” development of new products to compete, as also prototypes ready to be licensed are eligible for the award. Some notable examples of awarded products that were developed only by academic institutions are: the “Cost Thin Film Solar Cell” (University of Delaware, 1979); the “Low-Palmitic Soybean Oil” (Iowa State University, 1991); the “SC-54 Oral Vaccine” (Iowa State University, 1996); the “Nanoruler” (M.I.T., 2004); the “Netsolve 1.2” (University of Tennessee & U.C.S.D, 1999) and the “Chromium” (University of Virginia & Stanford University, 2004) software.
+
+16 For a more elaborate discussion of this point see Boldrin and Levine (2008).
+
+---
+
+1786
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+![Figure](figures/figure_002.png)
+
+Fig. 1. Pattern of change in the propensity to patent. All the samples.
+
+![Figure](figures/figure_004.png)
+
+Fig. 2. Pattern of change in the propensity to patent by type of inventor.
+
+considered.17 In our case, it is suggested that the propensity to patent has been remarkably stable and possibly characterized by a fluctuating behaviour around what seems to be a constant level of 10%.
+
+Fig. 2 shows the evolution of the propensity to patent broken down by type of inventors, distinguishing between corporate (i.e. firms) and non-corporate (i.e. PRO and universities) organizations.
+
+As expected Fig. 2 suggests that the propensity to patent is higher for firms than for PROs and universities. In several years the propensity to patent of PROs and universities is equal to zero indicating that no innovation has been patented. From early 1990s onwards the propensity to patent of universities seems to increase. It would be tempting to interpret this evidence as a consequence of the strengthening of IPRs following the introduction of the Bayh-Dole Act in US ( Mowery and Ziedonis , 2002 ; Mowery et al. , 2001 ) though an analysis of the mechanisms underlying this trend falls beyond the scope of the present paper. 18
+
+### 4.2. Patenting rates vary across sectors
+
+Awarded innovations are classified by the magazine in several categories on the basis of their technological content. The classification is not consistent over time and in some cases the innovations were not even assigned to a specific category. Thus, in order to examine the distribution of awarded innovations across different technological fields, we have proceeded as follows. First we reclassified each awarded innovation according to a technology-oriented classification of 30 different sectors based on the co-occurrence of the IPC codes proposed by the Observatoire des Sciences et des Techniques (OST). 19 In a few doubtful cases, we have relied both upon the classification in product categories of the R&D100 awards and on the innovation description. It is important to note that we have assigned each awarded innovation to only one of the 30 OST sectors. These sectors have been further aggregated into 5 ‘macro’ technological classes (called ‘OST5’ henceforth) defined according to the ISI-INIPI-OST patent classification based on the EPO IPC technological classes.
+
+Table 3 displays the shares of awarded innovations that have been patented (patenting rates) classified by both 5 and 30 OST sectors.
+
+The share of patented innovation varies considerably across sectors. In terms of macro-sectors, the sector with the highest propensity to patent is chemical/pharmaceuticals, a result which is also in line with the results of the Yale and Carnegie Mellon surveys on the effectiveness of patents for protecting innovations in these fields. Finally, the macro-sector with the lowest patenting rate is instruments. 20 In this case we should remember that many organizations active in this sector are non-corporate institutions such as universities and public research centres which traditionally display a very low patenting attitude.
+
+### 4.3. Patenting rates vary across geographical areas
+
+Table 4 compares patent propensities across different world regions.
+
+The awarded innovations with at least one applicant from US show a lower patent propensity with respect to the average level of the whole sample (−1.38%, statistically significant at 10% significance level), whereas applicants from Asia tend to patent more their innovations than the average level (+15.53%, statistically significant at 1% significance level). We should note that the large bulk of Asian innovations are awarded to Japanese companies. Hence our results pointing to a significantly higher propensity to patent of Asian (and especially Japanese) firms are consistent with previous research on the aggressive patterns of foreign patenting of Japanese firms in a comparative international perspective (Granstrand, 1999, pp. 134–175).
+
+It may be argued that this finding is the consequence of differences in the institutional mix of US and Asian prize winners. In order to understand whether this is the case with our data we have compared the propensities across subsets of similar organizations. Table 5 confirms that these overall differences in patent propensity rates across countries are not driven by different institutional mix of US prize-winners with respect to other geographical areas. In fact the higher propensity to patent of Asian prize-winners is
+
+17 Hall (2005) finds several structural breaks in patent application series at USPTO during the 1967–1997 time period. In our case, both the Phillips–Perron (PP) and the Augmented Dickey–Fuller (ADF) tests for unit root reject the null hypothesis that our patent propensity series contains a unit root at the 1% significance level, both after including a time trend and lagged terms (up to the 4th order) in the associated regression. The Andrews (1983) test for structural breaks at unknown points (sup. F=8.6538, T=28) does not reject the null hypothesis of 'no structural change' at the 5% significance level.
+
+18 A preliminary analysis along these lines actually suggests that both the number of patents and the number of awarded inventions attributable to University and PROs in our sample have increased.
+
+19 See Hinze et al. (1997).
+
+20 A statistical test for binary variables was carried out to check whether the difference in proportion between each industry and the total sample is statistically significant. Results rejected the null hypothesis of equality in two cases: Chemical and Pharmaceuticals (+4.6% with respect to the average patent propensity of the whole sample, statistically significant at 5% significance level); Instruments (−2.8% with respect to the average patent propensity of the whole sample, statistically significant at 1% significance level).
+
+---
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+1787
+
+Table 3 Patenting rates of 'R&D 100' innovations.
+
+<table><tr><td colspan="2">OST5</td><td rowspan="2">OST30 Electrical engineering &amp; devices Audiovisual technology Telecommunications Information Technologies Semiconductors 728</td><td colspan="2">All applicants No. of innovations</td><td colspan="2">Only corporate No. of innovations</td></tr><tr><td>Electrical engineering</td><td>1 2 3 4 5</td><td>No. of innovations 274 19 32 255 148 728</td><td>Share patented 0.1350 0.1053 0.0824 0.1149 0.1126</td><td>177 12 20 157 90 728</td><td>Share patented 0.1751 0.1667 0.25 0.1210 0.1333</td></tr><tr><td rowspan="5">Instruments</td><td>6</td><td>Optics</td><td>198</td><td>0.1111</td><td>123</td><td>0.1545</td></tr><tr><td>7</td><td>Control technology</td><td>629</td><td>0.0493</td><td>384</td><td>0.0729</td></tr><tr><td>8</td><td>Medical technology</td><td>125</td><td>0.1120</td><td>91</td><td>0.1209</td></tr><tr><td>27</td><td>Nuclear engineering</td><td>75</td><td>0.0400</td><td>41</td><td>0.0732</td></tr><tr><td></td><td></td><td>1027</td><td>0.0682</td><td>639</td><td>0.0954</td></tr><tr><td rowspan="6">Chemistry, Pharma</td><td>9</td><td>Organic chemistry</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td>10</td><td>Polymers</td><td>47</td><td>0.1489</td><td>41</td><td>0.1463</td></tr><tr><td>11</td><td>Pharmaceutics</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td>12</td><td>Biotechnology</td><td>87</td><td>0.0690</td><td>51</td><td>0.0588</td></tr><tr><td>14</td><td>Food chemistry</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td>15</td><td>Basic materials chemistry</td><td>42</td><td>0.2857</td><td>31</td><td>0.3870</td></tr><tr><td></td><td></td><td>176</td><td>0.1420</td><td>123</td><td>0.1703</td></tr><tr><td rowspan="8">Process engineering</td><td>13</td><td>Materials metallurgy</td><td>240</td><td>0.1458</td><td>167</td><td>0. 2036</td></tr><tr><td>16</td><td>Chemical engineering</td><td>220</td><td>0.1091</td><td>118</td><td>0. 1525</td></tr><tr><td>17</td><td>Surface technology</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td>18</td><td>Materials processing</td><td>8</td><td>0</td><td>0</td><td>–</td></tr><tr><td>20</td><td>Environmental technology</td><td>154</td><td>0.0714</td><td>81</td><td>0. 1235</td></tr><tr><td>24</td><td>Handling &amp; printing</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td>25</td><td>Food processing</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td></td><td></td><td>622</td><td>0.1125</td><td>366</td><td>0.1694</td></tr><tr><td rowspan="8">Mechanical engineering</td><td>19</td><td>Thermal processes</td><td>34</td><td>0.0882</td><td>23</td><td>0.1304</td></tr><tr><td>21</td><td>Machine tools</td><td>77</td><td>0.1169</td><td>47</td><td>0.1702</td></tr><tr><td>22</td><td>Engines</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td>23</td><td>Mechanical elements</td><td>43</td><td>0.0465</td><td>30</td><td>0.0333</td></tr><tr><td>26</td><td>Transport</td><td>27</td><td>0.0370</td><td>0</td><td>–</td></tr><tr><td>28</td><td>Space technology</td><td>9</td><td>0.1111</td><td>0</td><td>–</td></tr><tr><td>29</td><td>Consumer goods</td><td>59</td><td>0.1017</td><td>44</td><td>0.1364</td></tr><tr><td>30</td><td>Civil engineering</td><td>0</td><td>–</td><td>0</td><td>–</td></tr><tr><td></td><td></td><td>249</td><td>0.0884</td><td>144</td><td>0.125</td></tr><tr><td colspan="3">Total</td><td>2802</td><td>0.0960</td><td>1728</td><td>0.1336</td></tr></table>
+
+
+confirmed even when considering awarded innovations with at least one corporate applicant (columns 3–4) or only corporate applicants (columns 5–6).
+
+### 4.4. Multivariate regression analysis
+
+Though interesting, the previous results only account for the effect of a single characteristic at a time (i.e. technological sector,
+
+country of origin, applicant type) on the propensity to patent. In order to study the joint effect of all these variables in an unified framework we perform a probit multivariate regression analysis using the probability to patent a given innovation Pr(PAT=1) as dependent variable and a set of applicant and innovation specific characteristics as independent and control variables respectively. The list of additional variables includes: dummies for the technological sector (both at OST 5 and OST 30 aggregation level),
+
+Table 4
+
+Patenting rates by industry across countries.
+
+<table><tr><td rowspan="2">Sector (OST 5)</td><td colspan="2">Full Sample</td><td colspan="3">USA</td><td colspan="3">Europe</td><td colspan="3">Asia</td><td colspan="3">Other</td></tr><tr><td>(1) Total</td><td>(2)  $\overline{x} P at$ </td><td>(3) Total</td><td>(4)  $\overline{x} P at$ </td><td>(5) Diff. (4)–(2)</td><td>(6) Total</td><td>(7)  $\overline{x} P at$ </td><td>(8) Diff. (7)–(2)</td><td>(9) Total</td><td>(10)  $\overline{x} P at$ </td><td>(11) Diff. (10)–(2)</td><td>(12) Total</td><td>(13)  $\overline{x} P at$ </td><td>(14) Diff. (13)–(2)</td></tr><tr><td>Elec. Eng.</td><td>728</td><td>11.2</td><td>597</td><td>8.21</td><td>−2.99%*</td><td>16</td><td>0</td><td>−11.2%</td><td>86</td><td>31.40</td><td>+20.2%**</td><td>29</td><td>3.45</td><td>−7.75%</td></tr><tr><td>Instruments</td><td>1027</td><td>6.8</td><td>876</td><td>5.58</td><td>−1.22%</td><td>56</td><td>8.92</td><td>+2.12%</td><td>50</td><td>22</td><td>+15.2%**</td><td>45</td><td>4.44</td><td>−2.36%</td></tr><tr><td>Chemistry</td><td>176</td><td>14.2</td><td>158</td><td>13.92</td><td>−0.28%</td><td>8</td><td>0</td><td>−14.2%</td><td>5</td><td>20</td><td>+5.8%</td><td>5</td><td>0</td><td>−14.2%</td></tr><tr><td>Proc. Eng.</td><td>622</td><td>11.2</td><td>559</td><td>11.09</td><td>−0.11%</td><td>17</td><td>11.76</td><td>+0.56%</td><td>25</td><td>12</td><td>+0.8%</td><td>21</td><td>0</td><td>−11.2%</td></tr><tr><td>Mech. Eng.</td><td>249</td><td>8.8</td><td>207</td><td>7.73</td><td>−1.07%*</td><td>9</td><td>11.11</td><td>+2.31%</td><td>21</td><td>23.81</td><td>+15.01%**</td><td>12</td><td>0</td><td>−8.8%</td></tr><tr><td>All</td><td>2802</td><td>9.6</td><td>2397</td><td>8.22</td><td>−1.38%*</td><td>106</td><td>7.55</td><td>−2.05%</td><td>187</td><td>25.13</td><td>+15.5%***</td><td>112</td><td>2.68</td><td>−6.92%**</td></tr></table>
+
+
+Innovations with multiple applicants from different industries are double counted in the table.
+
+* Difference is statistically significant at 10% significance level. ** Difference is statistically significant at 5% significance level. *** Difference is statistically significant at 1% significance level.
+
+---
+
+1788
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+Table 5
+
+Patenting rates by geographical area and organization type.
+
+<table><tr><td rowspan="2">Type of applicant</td><td colspan="2">Full sample</td><td colspan="3">USA</td><td colspan="3">Europe</td><td colspan="3">Asia</td><td colspan="3">Other</td></tr><tr><td>(1) Total</td><td>(2)  $\%$  Pat</td><td>(3) Total</td><td>(4)  $\%$  Pat</td><td>(5) Diff. (4)–(2)</td><td>(6) Total</td><td>(7)  $\%$  Pat</td><td>(8) Diff. (7)–(2)</td><td>(9) Total</td><td>(10)  $\%$  Pat</td><td>(11) Diff. (10)–(2)</td><td>(12) Total</td><td>(13)  $\%$  Pat</td><td>(14) Diff. (13)–(2)</td></tr><tr><td>At least 1 corporate</td><td>1916</td><td>12</td><td>1557</td><td>11.11</td><td>−0.89%</td><td>102</td><td>7.84</td><td>−4.16%</td><td>183</td><td>25.68</td><td>+13.68%***</td><td>74</td><td>2.70</td><td>−9.3%**</td></tr><tr><td>Only corporate</td><td>1728</td><td>13.36</td><td>1389</td><td>11.74</td><td>−1.62%</td><td>94</td><td>7.45</td><td>−5.91%*</td><td>182</td><td>25.82</td><td>+12.46%***</td><td>63</td><td>3.28</td><td>−10.08%**</td></tr><tr><td>ALL</td><td>2802</td><td>9.6</td><td>2397</td><td>8.22</td><td>−1.38%*</td><td>106</td><td>7.55</td><td>−2.05%</td><td>187</td><td>25.13</td><td>+15.5%***</td><td>112</td><td>2.68</td><td>−6.92%**</td></tr></table>
+
+
+* Difference is statistically significant at 10% significance level. ** Difference is statistically significant at 5% significance level. *** Difference is statistically significant at 1% significance level.
+
+Table 6 Variable description.
+
+<table><tr><td></td><td>Description</td><td>Type</td></tr><tr><td colspan="3">Dependent variable</td></tr><tr><td>PAT</td><td>– 1 if the innovation has been patented (see Section 3.1)</td><td>Dummy</td></tr><tr><td colspan="3">Independent variables</td></tr><tr><td>MAPPL</td><td>– 1 for multiple applicant organizations, = 0 otherwise</td><td>Dummy</td></tr><tr><td>NINV</td><td>= number of inventors</td><td>Count</td></tr><tr><td>USA</td><td>= 1 if at least one applicant is a U.S. organization, = 0 otherwise</td><td>Dummy</td></tr><tr><td>EUROPE</td><td>– 1 if all the applicants are from continental Europe, = 0 otherwise</td><td>Dummy</td></tr><tr><td>ASIA</td><td>= 1 if all the applicants are from Asia, = 0 otherwise</td><td>Dummy</td></tr><tr><td>PRO</td><td>= 1 if at least one applicant is a public research organization, = 0 otherwise</td><td>Dummy</td></tr><tr><td colspan="3">Controls</td></tr><tr><td>dum1986.1995</td><td>= 1 the innovation has been awarded in the 1986–1995 decade, = 0 otherwise</td><td>Dummy</td></tr><tr><td>dum1996.2005</td><td>– 1 the innovation has been awarded in the 1996–2005 decade, = 0 otherwise</td><td>Dummy</td></tr><tr><td>Electrical.Eng</td><td>= 1 if the innovation belongs to the Electrical Engineering OSTS macro sector, = 0 otherwise</td><td>Dummy</td></tr><tr><td>Instruments</td><td>= 1 if the innovation belongs to the Instruments OSTS macro sector, = 0 otherwise</td><td>Dummy</td></tr><tr><td>Chemistry.Pharma</td><td>– 1 if the innovation belongs to the Chemistry &amp; Pharma OSTS macro sector, = 0 otherwise</td><td>Dummy</td></tr><tr><td>Process.Eng</td><td>= 1 if the innovation belongs to the Process Engineering OSTS macro sector, = 0 otherwise</td><td>Dummy</td></tr></table>
+
+
+dummies for the geographical macro areas (i.e. USA, EUROPE, ASIA), and for the time period decades (1976–1985, 1986–1995, 1996–2005) to capture the year of award. Another set of regressors includes: a dummy (PRO) to account for innovations with at least one public research organization applicant (i.e. either an academic or a governmental organization), a dummy (MAPPL) for collaborative innovations to indicate the presence of more than one applicant, and a count variable (NINV) that reports the number of applicants. All the variables used in the regressions analysis are summarized in Table 6. Table 7 provides instead their main descriptive statistics.
+
+Results from the multivariate probit regression are reported in Table 8.
+
+Column (1) reports the estimated coefficients for the most parsimonious model which does not include sector and time-period dummies. Column (2) includes both time-period and OST 5 macrosector dummies. Columns (3–4) report the estimated coefficients and average marginal effects (AMEs) for the full model with both
+
+time-period and OST 30 macro-sector dummies (not reported for the sake of clarity).
+
+Table 7 Descriptive statistics.
+
+<table><tr><td>Variable</td><td>Obs</td><td>Mean</td><td>Std. Dev.</td><td>Min</td><td>Max</td></tr><tr><td>PAT</td><td>2802</td><td>0.096</td><td>0.295</td><td>0</td><td>1</td></tr><tr><td>MAPPL</td><td>2802</td><td>0.256</td><td>0.437</td><td>0</td><td>1</td></tr><tr><td>NINV</td><td>2802</td><td>1.665</td><td>0.902</td><td>1</td><td>5</td></tr><tr><td>USA</td><td>2802</td><td>0.877</td><td>0.329</td><td>0</td><td>1</td></tr><tr><td>EUROPE</td><td>2802</td><td>0.038</td><td>0.191</td><td>0</td><td>1</td></tr><tr><td>ASIA</td><td>2802</td><td>0.067</td><td>0.250</td><td>0</td><td>1</td></tr><tr><td>PRO</td><td>2802</td><td>0.375</td><td>0.484</td><td>0</td><td>1</td></tr><tr><td>dum1986,1995</td><td>2802</td><td>0.356</td><td>0.479</td><td>0</td><td>1</td></tr><tr><td>dum1996,2005</td><td>2802</td><td>0.322</td><td>0.467</td><td>0</td><td>1</td></tr><tr><td>Electrical.Eng</td><td>2802</td><td>0.260</td><td>0.439</td><td>0</td><td>1</td></tr><tr><td>Instruments</td><td>2802</td><td>0.366</td><td>0.482</td><td>0</td><td>1</td></tr><tr><td>Chemistry.Pharma</td><td>2802</td><td>0.063</td><td>0.243</td><td>0</td><td>1</td></tr><tr><td>Process.Eng</td><td>2802</td><td>0.222</td><td>0.416</td><td>0</td><td>1</td></tr></table>
+
+
+In terms of geographical and sectoral effect, our regression analysis confirms the previous findings from the univariate analysis. Ceteris paribus, awarded innovations with all applicants from Asia show a larger probability to be patented with respect to the excluded category (which includes non-US, non-European and non-Asian countries), followed by awarded innovations with at least one US applicant. No significant difference in the average patent propensity is detected between the OST5 macro sectors represented by the four dummy variables included in the multivariate regression and the excluded one (Mechanical Engineering).
+
+In terms of differences in the propensity across applicants and/or between collaborative and non-collaborative innovations, we find that innovations with at least one PRO as applicant show a lower patent propensity with respect to the excluded category (only corporate applicants). Finally patent propensity increases with the number of inventors (NINV) whereas having multiple applicants (MAPPL) does not seem to exert a significant effect on the propensity to patent an awarded innovation.
+
+The former finding confirms our previous point about the lower propensity to patent for non-corporate inventors as a consequence of their relative lack of downstream capabilities when compared to firms. The latter finding seems to be in contrast with previous studies, such as Brouwer and Kleinknecht (1999) and Kleinknecht and van der Panne (2011), who instead found a positive correlation between R&D collaborations and propensity to patent. This discrepancy can be explained in term of different research design. These two studies employed firm level data (the Community Innovation Survey) whereas our analysis is carried out at the innovation level. Moreover these studies adopted a completely different dependent variable (i.e. whether the firm applied or not for a patent at the European Patent Office or the number of firm's EPO patents) and a different proxy for collaborative innovations (i.e. the firm's
+
+---
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+1789
+
+Table 8
+
+Estimation results for the propensity to patent.
+
+<table><tr><td>Dependent variable: Pr(PAT = 1)</td><td>(1) Coefficients</td><td>(2) Coefficients</td><td>(3) Coefficients</td><td>(4) Marginal effects</td></tr><tr><td>USA</td><td>0.406 **</td><td>0.387 **</td><td>0.407 **</td><td>0.045 ***</td></tr><tr><td></td><td>(0.165)</td><td>(0.169)</td><td>(0.169)</td><td>(0.014)</td></tr><tr><td>EUROPE</td><td>0.165</td><td>0.209</td><td>0.247</td><td>0.040</td></tr><tr><td></td><td>(0.216)</td><td>(0.221)</td><td>(0.222)</td><td>(0.042)</td></tr><tr><td>ASIA</td><td>0.971 ***</td><td>0.965 ***</td><td>1.006 ***</td><td>0.240 ***</td></tr><tr><td></td><td>(0.196)</td><td>(0.201)</td><td>(0.201)</td><td>(0.067)</td></tr><tr><td>PRO</td><td>−0.635 ***</td><td>−0.617 ***</td><td>−0.605 ***</td><td>−0.077 ***</td></tr><tr><td></td><td>(0.083)</td><td>(0.084)</td><td>(0.085)</td><td>(0.009)</td></tr><tr><td>MAPPL</td><td>−0.109</td><td>−0.087</td><td>−0.094</td><td>−0.013</td></tr><tr><td></td><td>(0.091)</td><td>(0.091)</td><td>(0.092)</td><td>(0.012)</td></tr><tr><td>NINV</td><td>0.153 **</td><td>0.144 ***</td><td>0.138 **</td><td>0.019 ***</td></tr><tr><td></td><td>(0.037)</td><td>(0.038)</td><td>(0.038)</td><td>(0.005)</td></tr><tr><td>dum1986.1995</td><td></td><td>−0.157 **</td><td>−0.128</td><td>−0.017</td></tr><tr><td></td><td></td><td>(0.079)</td><td>(0.082)</td><td>(0.011)</td></tr><tr><td>dum1996.2005</td><td></td><td>−0.144</td><td>−0.128</td><td>−0.017</td></tr><tr><td></td><td></td><td>(0.093)</td><td>(0.099)</td><td>(0.013)</td></tr><tr><td>Electrical.Eng</td><td></td><td>0.115</td><td></td><td></td></tr><tr><td></td><td></td><td>(0.129)</td><td></td><td></td></tr><tr><td>Instruments</td><td></td><td>−0.120</td><td></td><td></td></tr><tr><td></td><td></td><td>(0.128)</td><td></td><td></td></tr><tr><td>Chemistry.Pharma</td><td></td><td>0.259</td><td></td><td></td></tr><tr><td></td><td></td><td>(0.169)</td><td></td><td></td></tr><tr><td>Process.Eng</td><td></td><td>0.189</td><td></td><td></td></tr><tr><td>OST30 Sector Dummies</td><td></td><td>(0.137)</td><td></td><td></td></tr><tr><td>Constant</td><td>−1.816 ***</td><td>−1.755 ***</td><td>−5.281 ***</td><td></td></tr><tr><td></td><td>(0.177)</td><td>(0.206)</td><td>(0.275)</td><td></td></tr><tr><td>Observations</td><td>2802</td><td>2802</td><td>2802</td><td>2802</td></tr><tr><td>Log likelihood</td><td>−811.436</td><td>−796.360</td><td>−792.076</td><td></td></tr></table>
+
+
+Probit Maximum Likelihood estimates with clustered robust standard errors (in parenthesis).
+
+** Statistically significant at 5% level. *** Statistically significant at 1% level.
+
+past collaborations with other R&D partners). Results from other innovation-level studies such as Mäkinen (2007) find instead a nonsignificant effect of collaborations on the propensity to patent a given innovation. Moreover, our result can be understood on the basis of the two effects suggested by Peeters and Van Pottelsberghe de la Potterie (2006) to explain the patenting of collaborative innovations. On the one hand there is a 'need effect' which refers to "[...] a higher need for patent protection resulting from the mutual access to the partners' knowledge bases" (Peeters and Van Pottelsberghe de la Potterie, 2006, p. 127). On the other hand, there is a 'novelty effect' which refers to a "[...] potentially more fundamental and breakthrough knowledge generated by R&D collaborations compared to in-house R&D alone, which would result in more patents" (Peeters and Van Pottelsberghe de la Potterie, 2006, p. 127). In our case the ‘novelty effect’ is probably better proxied by the number of inventors (NINV), which has a positive and statistically significant coefficient, and dominates the ‘need effect’ which instead is better proxied by the dummy variable MAPPL.
+
+## 5. Reassessing patent propensity: a reappraisal of the empirical evidence
+
+At this point, it is instructive to compare our findings with prior estimates of patent propensity. This is done in Table 9.
+
+Table 9 is based on an extensive recognition of the existing literature and summarizes the findings of all the studies we have been able to identify, that provide direct or indirect empirical estimates of the importance of patent protection for product innovations in different industries using different types of methods. The last row of the table reports the findings of this paper. It has to be stressed that this exercise has important limitations, because the approach used to assess patent propensity is not consistent across studies.
+
+For instance, in some cases the estimation of patent propensity was not the main goal of the research (this is, in particular, the case for Acs and Audretsch, 1990). In other cases ( Mansfield, 1986), the survey asked how many patentable innovations were actually patented which is different from our focus on the ratio between the number of patented innovations and the total number of innovations. Levin et al. (1987) asked respondents to assess the relative effectiveness of patents in comparison to other alternative appropriability strategies. Cohen et al. (2000) asked firms to report the share of innovations for which they have applied for a patent which, again, is somewhat different from the variable we have constructed in this paper. These limitations notwithstanding, we believe that it is useful to compare our results with the main findings concerning patent propensity emerging from the literature.
+
+It is immediately interesting to note that, consistently with what we have highlighted in the introduction, four of the studies listed in the table ( Thomson , 2009 ; Moser , 2012 ; Meisenzahl and Mokyr , 2011 ; Nicholas , 2011 ) are contributions of economic historians or historians of technology.
+
+In terms of research strategies, we can draw a distinction between studies using a survey approach ( Mansfield , 1986 ; Levin et al. , 1987 ; Arundel and Kabla , 1998 ; Brouwer and Kleinknecht , 1999 ; Cohen et al. , 2000 ) and other contributions that instead estimate patent propensity using indicators of overall innovative output not based on patents. Although providing very detailed snapshots on innovative activities, it is well known that also the data collected by means of innovation surveys suffer from several shortcomings. The main one is that this type of data, unavoidably, reflects the personal judgement of actors that are required to monitor and self-assess their own innovative activities and performance ( Mairesse and Mohnen , 2010 ) .
+
+Turning our attention to the research strategies and sources employed by the studies that are not based on surveys, Acs and
+
+---
+
+1790
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+Table 9
+
+Estimates of patent propensity in different studies.
+
+<table><tr><td></td><td>Average patent propensity (%)</td><td>1st highest propensity to patent</td><td>2nd highest propensity to patent</td><td>1st lowest propensity to patent</td><td>2nd lowest propensity to patent</td><td>Indicator of patent propensity</td><td>Data &amp; methods</td><td>Period</td><td>Geographical scope</td></tr><tr><td>Mansfield (1986)</td><td>77</td><td>Petroleum (86%)</td><td>Machinery (86%)</td><td>Primary metals (50%)</td><td>Motor vehicles (65%)</td><td>Propensity to patent (sales weighted)</td><td>Survey of a random sample of large firms (over 1 million $8 \cdot 8 D$ in 1981)</td><td>1981-1983</td><td>USA</td></tr><tr><td>Levin et al. (1987)</td><td></td><td>Drugs</td><td>Organic chemicals</td><td>Pulp and paper</td><td>Computers</td><td>Perceived effectiveness of patents as appropriation tool</td><td>Yale Survey (650 R&amp;D executives from 130 business lines)</td><td>1981</td><td>USA</td></tr><tr><td>Acs and Audretsch (1990)</td><td></td><td>Lumber and furniture</td><td>Food and tobacco</td><td>Petroleum</td><td>Rubber and plastics</td><td>Patents/Innovations</td><td>Small Business Administration Innovation Dataset</td><td>1982</td><td>USA</td></tr><tr><td>Arundel and Kabla (1998)</td><td>35.90</td><td>Pharmaceuticals (79.2%)</td><td>Chemicals (57.3%)</td><td>Textiles (8.1%)</td><td>Basic metals (14.6%)</td><td>Propensity to patent product innovations (sales weighted)</td><td>PACE Survey (604 large firms)</td><td>1993</td><td>Europe</td></tr><tr><td>Browser and Kleinkenecht (1999)</td><td>25.40</td><td>Rubber and plastic products (36.4%)</td><td>Pharmaceuticals, Chemicals, Petroleum (36.3%)</td><td>Basic metals (9.9%)</td><td>Class, clay and ceramics (11.8%)</td><td>Percentage of firms for which patent are “very important” or “crucial” for product innovations</td><td>CIS Survey</td><td>1992</td><td>Netherlands</td></tr><tr><td>Cohen et al. (2000)</td><td>49.12</td><td>Drugs (95.5%)</td><td>Mineral products (79.2%)</td><td>Metals (2.97%)</td><td>Steel (4.46%)</td><td>Propensity to patent product innovations</td><td>CMS Survey, 1165 large firms</td><td>1991-1993</td><td>USA</td></tr><tr><td>Thomson (2009)</td><td>60.2</td><td>Electricity (87.5%)</td><td>Agriculture: harvesting (85.7%)</td><td>Clocks (11.1%)</td><td>Metal working (40%)</td><td>Share of exhibitors displaying patented innovations</td><td>Catalogue of New York Great Exhibition</td><td>1853</td><td>USA</td></tr><tr><td>Moser (2012)</td><td>11.10</td><td>Manufacturing machinery (29.8%)</td><td>Engines (24.6%)</td><td>Mining and metallurgy (5%)</td><td>Chemicals (5.1%)</td><td>Number of patented innovations in the total number of innovations exhibited</td><td>Catalogue of Crystal Palace Great Exhibition</td><td>1851</td><td>UK</td></tr><tr><td>Meisenzahl and Mokyr (2011)</td><td>60.00</td><td>Textiles (81%)</td><td>Ships (70%)</td><td>Instruments, Construction (35%)</td><td>Mining (41%)</td><td>Number of “great inventors” with 1 patent or more</td><td>Analysis of biographical dictionaries</td><td>1660-1830</td><td>UK</td></tr><tr><td>Nicholas (2011)</td><td>44.6</td><td>Machinery and machine tools (70.8%)</td><td>Auto (67.7%)</td><td>Food (33.1%)</td><td>Chemicals (34.9%)</td><td>Number of R&amp;D performing firms with 1 patent or more</td><td>Matching patents to firms in NRC surveys R&amp;D 100 competition</td><td>1921-1938</td><td>USA</td></tr><tr><td>This paper</td><td>9.6</td><td>Chemistry, pharma (14.2%)</td><td>Process engineering (11.2%)</td><td>Instruments (6.8%)</td><td>Mechanical engineering (8.8%)</td><td>Number of patented innovations in the total number of innovations awarded the prize</td><td>R&amp;D 100 competition</td><td>1977-2004</td><td>World</td></tr></table>
+
+
+---
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+1791
+
+Audretsch (1990) is a study of innovation in small firms in the US in 1982 based on the Small Business Administration Innovation dataset. This dataset was constructed by retrieving innovations in a large number of technical and trade journals. The results obtained may seem somewhat perplexing, because ‘petroleum’ and ‘rubber’, sectors that in innovation surveys usually appear characterized by a relative high propensity to patent, exhibit a very low propensity. In this respect, two considerations are in order. First, the Acs and Audretsch (1990) study is focused on small firms. Second, the indicator of patent propensity is constructed by comparing innovations with patents at industry level, but these quantities do not refer to the same groups of firms.
+
+The studies by Moser (2005, 2012) estimate patent propensity as the share of patented inventions in the total number of inventions presented at the great nineteenth century world fairs. 21 The most important result emerging from these studies is the relatively low levels of patent propensity. In this respect, as we have already noted, the findings of our paper are intriguingly in line with this type of evidence. The study of Thomson (2009) is an application of the research strategy of Moser (2005, 2012) to the case of the New York exhibition of 1853. Interestingly enough, Thomson finds levels of patent propensity that are much higher than those estimated by Moser (2012).
+
+Meisenzahl and Mokyr (2011) is a study based on the ‘prosopographical’ analysis of a sample of 759 British inventors, engineers, mechanics and skilled craftsmen active in the period 1660–1830 constructed using biographical dictionaries, an approach originally pioneered by Khan and Sokoloff (1993). In this case the propensity to patent was defined as the share of inventors that have been granted at least one patent in the total number of inventors. The study of Nicholas (2011) adopts a similar approach, but in this case the actors are not individuals, but firms. Here propensity to patent is defined as the share of firms with at least 1 patent in the total number of firms with R&D establishment that were included in the surveys of the National Research Council in the US in the period 1921–1938.
+
+Overall, the main result emerging from Table 9 , holding in different periods and locations, is that a sizeable share of inventive activities is not covered by patents. Hence, the obvious recommendation is that it is crucial, whenever possible, to assess inventive output combining evidence from patents with evidence from other sources. However, the wide range of the estimated levels of patent propensity (even in studies such as Moser (2012) and Thomson (2009) that adopt similar research designs), also suggest that we are still very far from a robust understanding of patent propensity in different contexts and that more research on this theme is needed.
+
+## 6. Concluding remarks
+
+This paper has reassessed the propensity to patent using a novel data source on R&D awards that so far had received little attention. Our analysis has provided two types of evidence. First, a relative low number of important innovations are patented. Second, patent propensity tends to vary across industrial sectors and types of organizations. While these results mirror, to a certain extent, some of the earlier findings in the literature, an important difference exists concerning the method. In particular, inspired by the recent works of economic historians ( Moser, 2005, 2010), we have adopted a straightforward definition of patent propensity in terms of the share of innovations patented over the total innovative output, a choice that marks an important departure with respect to
+
+investigations carried out in the field of ETIS which typically assess innovative output using indicators based on patents or innovation surveys.
+
+In this respect, we believe that recent contributions in the field of economic history and history of technology have probably some important methodological lessons to teach to the ETIS field. In our judgement, the main lesson is that much more effort ought to be devoted to the exploration of the potentialities of new sources of data for constructing indicators of innovative output. More specifically, the contributions of economic historians suggest that, with some ingenuity the literature produced by technologists and engineering practitioners (e.g. technical journals and magazines, exhibition and fair catalogues, engineering prizes and awards) can be successfully employed for constructing indicators of innovative output that can usefully integrate the patent evidence.
+
+We would like to argue that this is precisely one of the methodological lessons emerging from some of the early contributions of Nick von Tunzelmann. Although still not fully appreciated, important insights of his famous analysis of the development and impact of steam power technology in the British economy were based on the evidence gathered from a contemporary engineering journal published in Cornwall called Lean's Engine Reporter. 22 Exploiting the information contained in the journal, von Tunzelmann was able to provide new estimates for the rates of technical progress as well as for the rate of diffusion of new technical practices in the field of steam power technology ( von Tunzelmann, 1970, 1978, pp. 252–264). It would be wrong to think that similar endeavours pertain exclusively in the domain of historical research. One of the aims of this paper was precisely to show that similar sources are probably available also for modern studies of innovation and technical change. At the same time, our plea for the use of new sources should not be read as invoking the complete abandonment of patents as indicator of inventive output. Rather, we believe that real progress can be only attained by systematically combining different types of innovation indicators constructed using different types of sources and measurement techniques.
+
+## Acknowledgements
+
+Comments from participants to the conference “Technical change: History, Economics and Policy” (SPRU, University of Sussex, 29–30 March 2010) are gratefully acknowledged. We are also particularly grateful to Fred Block, Maria Savona, Jonathan Sapsed and three anonymous referees for very helpful suggestions on subsequent drafts. Finally, we would like to thank Nick von Tunzelmann who, throughout the years, has been a constant source of inspiration and encouragement.
+
+## References
+
+Acs, Z.J., Audretsch, D.B., 1990. Innovation and Small Firms. MIT Press, Cambridge, MA.
+
+Arundel, A., 2001. The relative effectiveness of patents and secrecy for appropriation. Research Policy 30, 611-624.
+
+Arundel, A., Kabla, I., 1998. What percentage of innovations are patented? Empirical estimates for European firms. Research Policy 27, 127-141.
+
+Basberg, B.L., 1987. Patents and the measurement of technological change: a survey of the literature. Research Policy 16, 131-141.
+
+Block, F., Keller, M.R., 2009. Where do innovations come from? Transformations in the US economy 1970-2006. Socio-Economic Review 7, 459-483.
+
+Boldrin, M., Levine, D., 2008. Against Intellectual Monopoly. Cambridge University Press, Cambridge, UK.
+
+Brouwer, E., Kleinknecht, A., 1999. Innovative output and a firm's propensity to patent. An exploration of CIS micro data. Research Policy 28, 615-624.
+
+21 In Table 9 we have reported only the results of Moser (2012) for the UK which may be taken as representative of the overall findings.
+
+22 For a recent discussion of the historical significance of this engineering journal see Nuvolari and Verspagen (2007, 2009).
+
+---
+
+1792
+
+R. Fontana et al. / Research Policy 42 (2013) 1780-1792
+
+Brunt, L., Lerner, J., Nicholas, T., 2012. Inducement prizes and innovation. Journal of Industrial Economics 60, 657-696.
+
+Carpenter, M.P., Narin, F., Woolf, P., 1981. Citation rates to technologically important patents. World Patent Information 3, 160-163.
+
+Cohen, W.M., Nelson, R.R., Walsh, J.P., 2000. Protecting their intellectual assets: appropriability conditions and why US manufacturing firms patent (or not). In: NBER working paper no. 7552.
+
+De Rasenfosse, G., 2010. How much do we know about firms' propensity to patent and should we worry about it? In: Paper presented at the 2010 Druid Summer Conference.
+
+Granstrand, O., 1999. The Economics and Management of Intellectual Property: Towards Intellectual Capitalism. Edward Elgar, Aldershot.
+
+Griliches, Z., 1990. Patents statistics as economic indicators. A survey. Journal of Economic Literature 28, 1661-1707.
+
+Hall, B.H., 2005. Exploring the patent explosion. Journal of Technology Transfer 30, 35–48.
+
+Hall, B.H., Jaffe, A., Trajtenberg, M., 2001. The NBER-Citations datafile: lessons, insights and methodological tools. In: NBER working paper no. 8498.
+
+Hall, B.H., Ziedonis, R., 2001. The patent paradox revisited. An empirical study of patenting in the US semiconductor industry, 1979-1995. Rand Journal of Economics 32, 101-128.
+
+Hinze, S., Reiss, T., Schmoch, U., 1997. Statistical analysis on the distance between fields of technology. In: Report for the European Commission, TSER project.
+
+Hussinger, K., 2007. Is silence golden? Patent versus secrecy at the firm level. Economics of Innovation and New Technology 15, 735-752.
+
+Khan, B.Z., Sokoloff, K.L., 1993. "Schemes of Practical Utility": entrepreneurship and innovation among 'Great Inventors' in the United States, 1790–1865. Journal of Economic History 53, 289–307.
+
+Kleinknecht, A., van der Panne, G., 2011. The propensity to patent an innovation: comparing entrepreneurial to routinized innovators. In: Audretsch, D.B., Falck, O., Heblich, S. (Eds.), Handbook of Research on Innovation and Entrepreneurship. Edward Elgar, Northampton, MA.
+
+Kortum, S., Lerner, J., 1999. What is behind the recent surge in patenting? Research Policy 28, 1–22.
+
+Kuznets, S., 1962. Inventive activity: problems of definition and measurement. In: Arrow, K.J, Nelson, R.R (Eds.), The Rate and Direction of Inventive Activity. Princeton University Press, Princeton.
+
+Levin, R.C., Klevorick, A.K., Nelson, R.R., Winter, S.G., 1987. Appropriating the returns from industrial research and development. Brookings Papers on Economic Activity 3, 783–831.
+
+Mairesse, J., Mohnen, P., 2010. Using innovation surveys for econometric analysis. In: Hall, B.H, Rosenberg, N (Eds.), Handbook of Economics of Innovation, vol. II. Elsevier, Amsterdam.
+
+Mäkinen, I., 2007. To Patent or not to patent? An innovation-level investigation of the propensity to patent. In: VTT Publication Series No. 646.
+
+Mansfield, E., 1986. Patents and innovation: an empirical study. Management Science 32, 173-181.
+
+Meisenzahl, R., Mokyr, J., 2011. The rate and direction of invention in the British Industrial Revolution. Incentives and institutions. In: NBER Working paper, no. 16993.
+
+Moser, P., 2002. The determinants of innovation: new evidence from nineteenth century world fairs. University of California at Berkeley (PhD thesis).
+
+Moser, P., 2005. How do patent laws influence innovation? Evidence from nineteenth century world's fairs. American Economic Review 95, 1214-1236.
+
+Moser, P., 2012. Innovation without patents. Evidence from world fairs. Journal of Law and Economics 55, 43–74.
+
+Mowery, D.C., Ziedonis, A.A., 2002. Academic patent quality and quantity before and after the Bayh-Dole act in the United States. Research Policy 31, 399-418.
+
+Mowery, D.C., Nelson, R.R., Sampat, B., Ziedonis, A., 2001. The growth of patenting and licensing by U.S. universities: an assessment of the effects of the Bayh-Dole Act of 1980. Research Policy 30, 99–119.
+
+Nicholas, T., 2011. Did R&D firms used to patent? Evidence from the first innovation surveys. Journal of Economic History 71, 1032–1059.
+
+Yuvolari, A., Verspagen, B., 2007. Lean's Engine Reporter and the development of the Cornish engine. A reappraisal. Transactions of the Newcomen Society 77, 167–189.
+
+Nuvolari, A., Verspagen, B., 2009. Technical choice, innovation and British steam engineering, 1800-1850. Economic History Review 62, 685-710.
+
+Peeters, C., Van Pottelsberghe de la Potterie, B., 2006. Innovation strategy and the patenting behavior of firms. Journal of Evolutionary Economics 16, 109–135.
+
+Scherer, F.M., 1983. The propensity to patent. International Journal of Industrial Organization 1, 107-128.
+
+Scherer, F.M., 1989. Comments on Z. Griliches patents: recent trends and puzzles. Brookings Papers on Economic Activity 9, 291-330.
+
+Scherer, F.M., 1999. New Perspectives on Economic Growth and Technological Innovation. Brookings Institution, New York.
+
+Schmookler, J., 1966. Invention and Economic Growth. Harvard University Press, Cambridge, MA.
+
+Silverberg, G., Verspagen, B., 2007. The size distribution of inventions revisited: an application of extreme value statistics to citation and value measure of patent significance. Journal of Econometrics 139, 318–339.
+
+Sullivan, R.J., 1989. England's "age of invention": the acceleration of patents and patentable invention during the Industrial Revolution. Explorations in Economic History 26, 429-443.
+
+Taylor, C.T., Silberston, Z.A., 1973. The Economic Impact of the Patent System. A Study of the British Experience. Cambridge University Press, Cambridge, UK.
+
+Thomson, R., 2009. Structures of Change in the Mechanical Age. Technological Innovation in the United States. John Hopkins University Press, Baltimore, pp. 1790–1865.
+
+Trajtenberg, M., 1990. A penny for your quotes: patent citations and the value of innovations. Rand Journal of Economics 21, 172-187.
+
+van Zeebroeck, N., 2011. The puzzle of patent value indicators. Economics of Innovation and New Technology 20, 33–62.
+
+von Tunzelmann, G.N., 1970. Technological diffusion during the Industrial Revolution. The case of the Cornish pumping engine. In: Hartwell, R (Ed.), The Industrial Revolution. Basil Blackwell, Abingdon.
+
+von Tunzelmann, G.N., 1978. Steam Power and British Industrialization to 1860. Clarendon, Oxford.
+

processed/00096_W2532229852/document.md

@@ -0,0 +1,706 @@
+![Figure](figures/figure_000.png)
+
+Asian Journal of Technology Innovation
+
+ISSN: 1976-1597 (Print) 2158-6721 (Online) Journal homepage: http://www.tandfonline.com/loi/rajt20
+
+![Figure](figures/figure_003.png)
+
+# Why do we need a theory and metrics of technology upgrading?
+
+Slavo Radosevic & Esin Yoruk
+
+To cite this article: Slavo Radosevic & Esin Yoruk (2016) Why do we need a theory and metrics of technology upgrading?, Asian Journal of Technology Innovation, 24:sup1, 8-32, DOI: 10.1080/19761597.2016.1207415
+
+To link to this article: http://dx.doi.org/10.1080/19761597.2016.1207415
+
+![Figure](figures/figure_008.png)
+
+© 2016 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
+
+![Figure](figures/figure_010.png)
+
+Published online: 21 Oct 2016.
+
+![Figure](figures/figure_012.png)
+
+Submit your article to this journal
+
+![Figure](figures/figure_014.png)
+
+Article views: 269
+
+![Figure](figures/figure_016.png)
+
+View related articles ☏
+
+![Figure](figures/figure_018.png)
+
+View Crossmark data
+
+Full Terms & Conditions of access and use can be found at
+
+
+http://www.tandfonline.com/action/journalInformation?journalCode=rajt20
+
+
+1
+
+Download by: [5.8.47.223]
+
+Date: 21 November 2016, At: 07:05
+
+---
+
+Asian Journal of Technology Innovation, 2016 Vol. 24, No. S1, 8–32, http://dx.doi.org/10.1080/19761597.2016.1207415
+
+![Figure](figures/figure_001.png)
+
+# Why do we need a theory and metrics of technology upgrading?
+
+Slavo Radosevic$^{a*}$\copyright and Esin Yoruk$^{b}$
+
+$^{a}$ School of Slavonic and East European Studies, University College London, London, UK
+
+$^{6}$ Coventry Business School, Coventry University, Coventry, UK
+
+This paper discusses why we need a theory and metrics of technology upgrading. It critically reviews existing approaches to technology upgrading, and proposes a theoretically relevant and empirically grounded intermediate conceptual and statistical framework to illustrate the types of challenges facing economies with different levels of income. It conceptualises technology upgrading as a three-dimensional process that considers the intensity and type of technology upgrading based on different types of innovation and technology activities; the broadening of technology upgrading through exploitation of technology and knowledge diversification; and interaction with the global economy via the import, adoption, and exchange of knowledge. We consider these to be necessary first steps towards a theory and metrics of technology upgrading and the generation of more relevant composite indicator of technology upgrading.
+
+Keywords: technology upgrading, growth, composite indicators, middle-income economies, R&D and innovation
+
+## 1. Introduction
+
+Why do we need a theory of technology upgrading? There are three major motivations for our study. First, the search for universal growth factors is futile (see Easterly 2001 ) . Theories of aggregate economic growth generally consider a single variable or a single factor when trying to explain economic growth. The search for a universal theory or a single set of factors able to explain growth at different levels of development and in different geographical areas faces huge empirical and methodological challenges. Similarly, technology as one of the major drivers of growth over the long term is not reducible to a single variable such as research and development (R&D) or total factor productivity (TFP) (Lee 2013 ) . Improvements in technological capability stem from increased investment across a number of drivers of technology upgrading (Furman and Hayes 2004 ) . It is important to identify which (if any) of these drivers are common, and which are country specific. In addition, the drivers of technology upgrading differ according to the stage of development and, thus, may be quite different for low-, middle-, and high-income economies.
+
+Second, current technological upgrading metrics are either not theoretical or not rooted in the stylised facts related to technology upgrading and, thus, are not relevant to middle-income economies. For example, the Global Innovation Index and the EU Innovation Union Scoreboard (IUS) are pragmatic, but atheoretical analytical frameworks. By this we mean that they are not underpinned by an understanding of how technology upgrading takes place at different income levels.
+
+*Corresponding author. Email: s.radosevic@ucl.ac.uk
+
+© 2016 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
+
+This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
+
+---
+
+Asian Journal of Technology Innovation
+
+9
+
+The frequently used Crépon, Duguet, and Mairesse ( 1998 ) or CDM model brings together within a simple framework, R & D, innovation, and productivity, giving the impression that it is theoretically grounded. However, we consider this framework to be of limited relevance to countries operating behind the technology frontier since it is based on R & D as an input into the innovation process which reduces its utility for middle-income economies. The World Economic Forum Global Competiveness Reports are considered theoretically and empirically grounded since they acknowledge differences in the drivers of growth, but this approach mixes technological and institutional variables which we consider to be problematic (see below).
+
+Third, theoretically grounded, but not measurable frameworks are of limited value, as are metrics that are not based on stylised facts or theory. Such metrics could lead to irrelevant policies through their focus on issues not directly relevant to the specific growth challenges. A paradigmatic example of this is the contradiction in the current EU approach between its dominant metrics (cf. IUS ), which assume identical technological paths and drivers of growth, and its aim to push countries along divergent `smart specialisation' paths. The EU is encouraging countries and regions to formulate their own smart specialisation strategies to avoid what have been described as `adding up' problems (Spence 2011 , pp. 94 – 96) or the situation where too many regions are focusing on similar technologies and markets and, thus, out competing one another. However, the EU's dominant IUS metrics, which countries and regions are using as policy targets, are reinforcing imitative policies aimed at R & D-based growth with the result that metrics are determining policy rather than policy determining the metrics.
+
+Based on these three factors, we argue that there is need to generate a theoretically relevant, but empirically grounded, middle-level conceptual and statistical framework which highlights the type of challenges that would seem to be relevant to a large number of countries classified by the World Bank as middle-income and `lower' high-income countries (from $ 1K $-$ 30 K per capita income) to escape the potential middle-income trap. Our notion of middle-income economies does not correspond strictly to the World Bank classification, which we consider out-dated. A more accurate definition of middle-income economies would include those countries currently classified as middle-income economies (from $ 1046 $-$ 12,745 per capita) and `lower' highincome economies ( $ 12,745 $-$ 30 K per capita) as opposed to `upper' high-income economies (above $ 30 K per capita). Our so-called broad middle-income group includes middle-income and lower high-income economies (see Appendix Tables A1 and A2 ). $^1$
+
+In order to avoid measurement without (some) theory or at least metrics grounded in the stylised facts of technology upgrading, we construct a composite indicator of technology upgrading which complements existing metrics, especially IUS, and which better reflects the drivers and patterns of technology upgrading in our broadly defined middle-income economies group.
+
+We do not aim to use technology upgrading as a substitute for economic growth, but rather as a major determinant of economic growth. We recognise that some countries can achieve high levels of income without much technological capability. For example, resource-based economies and entrepot economies can earn high incomes without necessarily being innovators. Various income levels can be achieved based on a variety of institutional systems. From the perspective of technology upgrading we are concerned primarily with technology accumulation. We assume that very different institutional systems can lead to technology upgrading, or that institutional forms are secondary to the process of technology accumulation. Similar to functional views on innovation systems we are concerned primarily with technology activities that increase firm, sector, and country-level capabilities. The institutional context is a quite important variable which ultimately is necessary to explain different growth performance, but our primary concern is with technology upgrading via the accumulation of technology capabilities at different levels.
+
+---
+
+10
+
+S. Radosevic and E. Yoruk
+
+In what follows, we discuss technology upgrading in the context of the broad literature on this issue (Section 2 ). Section 3 discusses the conceptual framework and its assumptions which we propose as a way forward to theorising and measuring technology upgrading. We provide a brief discussion of technology upgrading in the context of the distinction between catch-up and post-catch-up (Section 4 ), and conclude in Section 5 by summarising the major points.
+
+## 2. Past and current contributions to research on technology upgrading
+
+### 2.1. Past and recent contributions
+
+Considering growth from an upgrading perspective is not new. Marshall ( 1890 ) in his Principles of Economics , acknowledges the big variations in the dynamics among different economic sectors and their important aggregate effects on economic growth. Schumpeter's theory of `Business Cycles' is firmly rooted in analysis of the emergence and decline of leading industries, which, in aggregate, lead to macroeconomic cycles. Kuznets in `Secular, Movements in Prices and Production' ( 1930 ) recognised that shifts in the relative importance of leading industries follow sales and innovation patterns. Chenery and Syrquin ( 1975 ) in Patterns of Development 1950-70 analysed the structural characteristics of the economy by grouping industries into early, middle and late. The idea of development as an evolving process that goes through several stages was first formulated by Rostow ( 1960 ) in his stages of growth model. This was based on the idea of industry life cycles and `leading sectors', driving economic growth in specific stages. A common feature of these modes is the assumption that `all nations [go] through the same stages in the same order, though not necessarily at the same time' (von Tunzelmann 1995 , p. 69).
+
+A similar logic of structural change, but in an international context has been conceptualised, based on Japanese experience, as the `Flying-geese model' (Akamatsu 1962 ) . This model depicts changing patterns of industry specialisation based on an import – domestic production – export sequence, which induces structural change in both leader and follower countries. Countries exhibit similarities also in terms of sequences of structural change in industrial development, of capital goods following consumer goods and the progression from crude to simple goods to complex and refined goods . This pattern accompanied by a sequential positioning of the developing countries lined up behind the advanced nations, allowing the former group to emulate, learn from and capitalie on growth stimuli/externalities via economic interactions.
+
+A recent example of similar thinking about growth is Ozawa's ( 2009 ) structural growth stages model, which is a synthesis of stage theories of growth (i.e. Rostow and Akamatsu) and Schumpeter’s model. Ozawa's model is based on actual historical industry (hence technological) developments or ladders of economic development, which are driven by innovation. Ozawa synthesises sequential growth across five stages involving a leading sector in each stage. Ozawa's stylisation of Japan's industry upgrading follows the sequence: labour-driven $\rightarrow$ scale-driven $\rightarrow$ assembly-driven $\rightarrow$ R & D-driven $\rightarrow$ IT-driven industries. Ozawa and other's contributions are based largely on history and map upgrading based on the leading economies' historical experience. This is a strength but also a weakness of these models since catching up countries may not adopt the same pattern as Lee ( 2013 ) shows.
+
+The alignment of countries in the industrial upgrading process has been described as `countries gradually mov[ing] up in technological development by following the pattern of countries just ahead of them in the development process' (Radelet and Sachs 1997 , p. 52) . However, the new structural economics of Lin ( 2011 , p. 14) updates this to economic development as:
+
+---
+
+Asian Journal of Technology Innovation
+
+11
+
+a process of continuous industrial and technological upgrading in which any country, regardless of its level of development, can succeed if it develops industries that are consistent with its comparative advantage, determined by its endowment structure. The successful strategy for developing countries is to exploit the late-comer advantage by building up industries that are growing dynamically in more advanced countries that have endowment structures similar to theirs.
+
+So, implicit in notions of new structural economics is the process of technology upgrading, which is considered to be based on the country's `latent comparative advantages' ( Lin 2011 ) . The econometric evidence for this proposition is quite persuasive and has been tested in the context of the transition economies (Bruno, Douarin, Korosteleva, and Radosevic 2015 ) . However, this approach seems largely applicable in the transformation from low to middle-income levels but less so to the transition from middle-income to (upper) high income. As Lee ( 2013 ) demonstrates countries that have successfully moved to high-income group did so by entering newly emerging short-cycle technologies rather than mature industries. Lee shows clearly how the role of structural change, especially technological diversification rather than imitation, is one of the major factors in catching up to high-income levels. While new structural economics accounts ( Lin 2012a , b ; Lin and Rosenblatt 2012 ) show the path to technology upgrading as based on `copying industries' using latent comparative advantages in the transition from low to middleincome levels, Lee ( 2013 ) shows middle-income economies taking `detours' and establishing their own paths to transition from middle to high income. A very recent contribution to our thinking about technology upgrading in the context of technology specialisation and growth is Foray's ( 2015 ) contribution which conceptually frames and theorises issues based on EU smart specialisation strategies.
+
+A major addition to our understanding of technology upgrading came from the exploration of upgrading via global value chains (GVCs). In this literature, industrial upgrading is defined as substantial changes to a country's specialisation and knowledge base, which increase its capacity for value generation (Ernst 1998 ) . Gereffi ( 1999 , pp. 51 – 52) defines it as `a process of improving the ability of a firm or an economy to move to more profitable and/or technologically sophisticated capital and skill-intensive economic niches'. Upgrading is usually defined as a gradual shift from lower to higher value-added activities; for example, from cheap and simple products to complex and expensive ones; from mass production of standardised products to flexible production of differentiated products; and, from simple assembly to more integrated forms of production (such as original equipment manufacturing (OEM), original design manufacturing (ODM), and original brand manufacturing (OBM)) (see Table 1 ). Hierarchy is a common feature of upgrading taxonomies (see Table 1 , particularly Gereffi ( 1999 ) and Ernst ( 1998 ) ). Some taxonomies are theory driven (Gereffi 1999 ; Ernst 2001 ) and some are the outcome of firm-level empirical work (firm-based case studies) (Hobday 1995 ; Humphrey and Schmitz 2004 ) .
+
+It is not surprising that technology upgrading is most often discussed in an international context. Ernst ( 1997 ) and Gereffi ( 1999 ) pioneered the analysis of upgrading via global production networks (GPNs) and a GVC framework. However, Yoruk ( 2013 ) shows that upgrading is currently unjustifiably reduced to upgrading within value chains or GPN. In the first in-depth study of technology upgrading through value chains, in the context of Central and Eastern Europe, she shows the major importance of both production networks and knowledge networks. Yoruk's research shows that learning by doing and learning by exporting have no statistically significant effect on functional upgrading. She shows that the opportunities offered by GVCs are of little use unless firms have the ability to internalise this external knowledge through their human resources, and through in house training and research. She shows also that managerial upgrading is important for technology upgrading, but global buyers do not support it. This highlights the
+
+---
+
+12
+
+S. Radosevic and E. Yoruk
+
+Table 1: Taxonomies of firm-level upgrading in international (GVC) context.
+
+<table><tr><td>Authors</td><td>Taxonomy/trajectory</td><td>Locus of upgrading</td></tr><tr><td>Hobday (1995)</td><td>Original equipment manufacturing (OEM)Original design manufacturing (ODM)Original brand manufacturing (OBM) Within - factories, - inter-firm networks, - local or national economies, and - supranational macro-regions</td><td>International production networksGlobal value chains</td></tr><tr><td>Ernst (2001)</td><td>hierarchy of - industries, - factors of production, - consumption, - value chain stages, and - forward and backward linkages</td><td>Global production networks (2001, 2006),Global knowledge networks (2008),Global innovation networks (2009)</td></tr><tr><td>Humphrey and Schmitz (2004)</td><td>Process upgradingProduct upgradingFunctional upgradingInter-sectoral upgrading</td><td>Global value chains</td></tr></table>
+
+
+importance of organisational capabilities and suggests that the firm's structure is an important dimension of technology upgrading.
+
+Do these contributions have insights that are relevant to our understanding of the theory and metrics of technology upgrading? First, existing contributions provide most qualitative insights, which have not been converted into models or stylisations of countries' technology upgrading. Rostow and Chenery's contributions have not been extended or applied in a new context. Nevertheless, these existing contributions assume a similar or identical path to upgrading. In light of recent contributions, especially Lee ( 2013 ) and Foray ( 2015 ) , this assumption needs some qualification or needs to be relaxed, especially in relation to the transition from middle to high income. So, `copying industries' may help in the transition from low to middle income ( Lin 2012b ) , but diversity and variety of upgrading are more critical for subsequent upgrading. Second, technology upgrading is an interactive process between `leaders' and `followers' ( Akamatsu 1962 ) . The literature on GVC clearly shows the overwhelming importance of the international context of upgrading, but also its limits ( Yoruk 2013 ) . Third, several contributions on technology upgrading show that upgrading is a multi-level process that takes place at the firm, industry, inter-industry, and country levels. We discuss this important feature of technology upgrading in more depth below.
+
+### 2.2. Multi-level perspectives on technology upgrading
+
+The literature suggests that upgrading is multi-level phenomenon operating at the firm, industry, and country levels (see Table 2 for summary).
+
+---
+
+Asian Journal of Technology Innovation
+
+13
+
+Table 2: Different perspectives on technology upgrading.
+
+<table><tr><td>Types/levels</td><td>Conceptual framework</td></tr><tr><td>Intra-firm level</td><td>- Reverse product life cycle: a combination of the product life cycle model in advanced firms by Abernathy and Utterback (1978) and Kim's (1980) three-stage catch-up model of acquisition - assimilation - implementation - Importance of minor improvements during reverse learning trajectory (Hobday 1995; Hobday, Rush, and Bessant 2004)</td></tr><tr><td>Intra-industry and inter-industry level</td><td>- Industry life cycle and dominant design (Klepper 1997) - Upgrading towards high value-added industries (value chain upgrading)</td></tr><tr><td>Country level</td><td>Sequential upgrading of countries based on ‘leading-sector’ (Ozawa 2009) WEF rankings based on differing drivers of growth</td></tr></table>
+
+
+Innovation studies scholars were interested in technology upgrading at the firm level based on the so-called technology capabilities approach (Dahlman, Ross-Larson, and Westphal 1987 ; Lall 1992 ; Bell and Pavitt 1993 ; Dutrenit 2000 ) . Firm-level evidence shows that the paths to upgrading of firms in developing countries encompass through a variety of interrelated, sometimes similar and sometimes unique taxonomies. Based on Korean experience, Kim ( 1980 ) proposes a threestage catch-up model for developing country firms, going from the acquisition of foreign technology, to its assimilation and the implementation of new product lines. Hobday ( 1995 ) explores the path to technology accumulation among East Asian electronics firms during the 1960s $-$ 1990s. He shows that a best approximation of the process is the inverse product life cycle (PLC). In contrast to R & D and design led strategies typical of technology leaders and followers, East Asian latecomers began with minor improvements to the manufacturing process and moved to mastering elements of process technology through reverse engineering and finally mastering the elements of design capabilities.
+
+As some of these countries have moved to the group of high-income economies, the issue of firm upgrading has become an issue of transition from catching up to post-catch up stages. The move to a post-catch up stage or the level of high-income economies involves countries operating at the technology frontier by solving problems that have not been solved by others. Unlike the catch-up stage where firms enter largely through a reverse PLC pattern, in the post-catch up stage firms enter at various stages in the PLC (Choung, Hwang, and Song 2014 ) . They may enter the PLC via large firms in design and R & D stages; or via networks of new technology based firms immediately after dominant design has been established; or via cooperation between public R & D organisations and firms in early stages of new PLCs (Choung et al. 2014 ) . Other firms can follow a `strategic niche' strategy, `which involves large number of public, private stakeholders through small-scale transition experiments that expand the scope of changes to a wider scale when the experiments succeed' (Seong, Kim, and Cho 2014 ) .
+
+The literature highlights a trend towards expansion from firm-level upgrading to `industrylevel linkages' or industry-level upgrading. The rationale is the realisation that countries' advancement of their firm-level upgrading is increasingly dependent on `industry linkages' (Ernst 2008a ) . More broadly, Ernst ( 2008b ) refers to three forms of `industrial upgrading': (i) inter-industry upgrading from low value-added industries (e.g. light industries) to higher valueadded industries (e.g. heavy and higher tech industries); (ii) inter-factor upgrading from endowed assets (i.e. natural resources and unskilled labour) to created assets (physical capital,
+
+---
+
+14
+
+S. Radosevic and E. Yoruk
+
+skilled labour, and social capital); and (iii) upgrading of demand within a hierarchy of consumption, from necessities, to conveniences, to luxury goods.
+
+How industries evolve and upgrade is much less well understood. The most developed stylisation of industry dynamics is the industry life cycle, which is either inseparable from or quite reliant on the already mentioned PLC (Abernathy and Utterback 1978 ) . This view identifies three product evolution stages (Gort and Klepper 1982 ; Klepper and Graddy 1990 ; Suarez and Utterback 1995 ; Klepper 1997 ) . A radical innovation leads to product innovations. In this stage, the entry barriers are low, R & D and capital requirements are limited, and new entrants are most often small firms. This is followed by the emergence of a dominant design, which induces a stream of process innovations, but which has a positive effect on cost performance ratios. Economies of scale increase and equipment becomes standardised. This increases the barriers to entry and leads to an industry shakeout. Finally, as the technology matures, a few incremental innovations are introduced, the industry becomes more concentrated and the barriers to entry increase further. Malerba and Orsenigo ( 1994 ) summarise the weaknesses in these stylisations. First, industries are reduced to products, which is too reductive. Second, the sequence of product innovations followed by process innovations does not hold in capital intensive industries such as commodity chemicals, synthetic fibres, plastics, and petrochemicals where innovations are mainly of the process type. Third, in some industries the emergence of a dominant design can lead to new discontinuities or to several dominant designs. Fourth, the radical product innovation that triggered the industry life cycle may be accompanied by very different industry dynamics in terms of entrants, industry concentration and incumbents. Entrants may be existing firms active in related industries while the links among firms may change continuously, and new type of actors emerge so that the industry boundaries are continuously being redefined. Thus, there seem not to be useful stylisations or regularities related to the metrics of technology upgrading. The upgrading process seems not to be correlated either to firm size or the type of innovation.
+
+A critique of industry life cycle models shows that industry boundaries are ill-defined and changing, and need to include a variety of other non-business actors. This highlights the need to include infrastructure and infrastructural factor in the evolution and upgrading of industries and economies (see Ozawa 2009 ) (see below).
+
+The rise in composite indicators has resulted in a proliferation of different attempts to measure progress in innovation and competiveness at the country level (see Archibugi, Denni, and Filippetti 2009 for a discussion and overview). Adopting a long-term historical perspective, Ozawa ( 2009 ) proposes a sequential upgrading process based on countries' leading sectors. The World Economic Forum (WEF) Global Competitiveness Report (GCR) ankings provide country rankings based on various drivers of economic growth and classify countries into factor, efficiency, and innovation based. The EU Innovation Scoreboard provides rankings based on composite indicators of innovation activities ranging from moderate, to followers, to innovation leaders. Among this group of models of upgrading we should highlight the indicator of economic complexity as a dimension used to measure upgrading of different countries based on the complexity of their export products. See Hausmann and Klinger ( 2006 ) , Hausmann, Hwang, and Rodrik ( 2007 ) , Hidalgo, Klinger, Barabasi, and Hausmann ( 2007 ) , and Hidalgo and Hausmann ( 2009 ) .
+
+As already highlighted, technology upgrading is a multi-level process and the ultimate pattern of upgrading at country level is a complex interaction between the micro, mezzo, and macro levels. This is the most widespread perspective in the context of growth in East Asia and the integration of East Asian firms in GPNs. Aggregate explanations of growth do not provide satisfactory answers to the question of how these economies managed to achieve upper middle-income status in such short period. Hence, an industrial upgrading perspective is useful to shed light on the micro- and mezzo-level processes of technology accumulation, and provide a richer account of the drivers of growth than is possible based on variables such as TFP, or institutions.
+
+---
+
+Asian Journal of Technology Innovation
+
+15
+
+In successful cases of catching up or forging ahead we observe high complementarity among different levels and various sub-systems, which generate increasing returns (Freeman and Louçã 2001 ; Freeman 2002 ) . However, a favourable congruence between various sub-systems of society which have been positive for economic growth in one period of technological development, may be less favourable in the context of fundamental changes to technology (Freeman 2002 ) . In short, history shows that technology upgrading is also an institutional process, and that institutional requirements also change over time. However, for analytical convenience, we abstract from institutional set-ups and focus only on the outcomes of learning processes demonstrated in technology upgrading activities. The complexities of the multi-levels at which technology accumulation takes place, the diversity of its patterns as depicted through structural change, and the interactive nature of technology upgrading are sufficient to justify our abstraction. Our fundamental assumption is rooted firmly in Schumpeterian theorising both formal (Aghion, Akcigit, and Howitt 2013 ) as well as of evolutionary and structuralist (Freeman and Louçã 2001 , Perez 2010 ) . We assume a relationship between modes of technology upgrading, and countries' positions in terms of level of development or distance from the technology frontier (Acemoglu, Zilibotti, and Aghion 2006 ) . This is depicted in Figure 1 which suggests that technology upgrading paths vary as economies move from low to high incomes. Technology distant countries can grow based on imitative technology efforts similar to the logic in Lin ( 2012a , b ) . As they move from middle to high income, imitative technology is not sufficient and catching up countries need to find alternative paths involving technology diversification rather than imitation of technology leaders (Lee 2013 ) . When they reach the post-catch stage they need to be operating at the technology frontier.
+
+Figure 1 suggests that there is an association between income level groups and modes of technology upgrading. The modes of technology upgrading in right-hand column of Figure 1 are not the only possible modes, but are those that are considered to drive technology upgrading to achieve a higher income level. In reality, the technology activities of the technology leader countries include a mix of frontier and behind the technology frontier activities. Growth is driven by more effective diffusion of new technologies and management practices among the technology leaders compared to followers. Thus, Figure 1 should be considered a simplification, although a useful one because it conveys the idea that the dominant mode of technology differs as economies grow. However, this stylisation does not tell us anything about the transitions to different types of upgrading activities characterising different income levels. These are discussed in the form of the components of technology upgrading (depicted in Figure 3 and discussed in Sections 3.2–3.4 ).
+
+![Figure](figures/figure_004.png)
+
+Figure 1: Different patterns of technology upgrading at different income levels.
+
+Note: Based on our discussion in Section 1 we define high income as `upper high income' while middle income refers to `broad middle income' category, which includes lower, upper middle income, and lower high-income groups.
+
+---
+
+16
+
+S. Radosevic and E. Yoruk
+
+So far, we have depicted technology upgrading as a stage process, and leave discussion of the transitions or transformations involved until we have a better empirical understanding of the modes of technology upgrading to different income groups. However, we would agree with an anonymous reviewer that the modes of technology upgrading as well as the transitions between different modes always involve a mix of types of technology activities. For example, our on-going empirical work which applies this conceptual framework, shows clearly that different types of technology upgrading activities exist at all income levels, but that their intensity and weight differs across different income groups (see Radosevic and Yoruk 2015 ). For example, production, R & D and technology capabilities are important at all income levels, but their composition varies. Also, all dimensions of structural change play a role in technology upgrading, but have different weights at different income levels.
+
+Based on the different weight and intensity of the different components of technology upgrading, it follows the targets for countries operating behind technology frontier should not mimic those of high-income countries. Structural differences including differences in the levels and intensities of the different components of technology upgrading between countries, should be taken into account when considering appropriate policies (see Radosevic and Kaderabkova 2011 for an application of this thinking in the context of Eastern Europe).
+
+## 3. Technology upgrading: conceptual issues
+
+In this section, we outline some key criteria that must be satisfied in order to develop a theory of technology upgrading. We outline our conceptual approach, which includes the three dimensions of technology upgrading and its sub-components, and could be used as a basis for building a metrics of technology upgrading.
+
+### 3.1. Criteria for building a theory of technology upgrading
+
+We argue that a theory of technology upgrading, grounded in the stylised facts of economic growth, must comply with the following criteria:
+
+- (1) One of the keys to economic growth is improved technology capability, which is not
+reducible to a single variable (Lee 2013 ) . Hence, our theory must be based on several
+drivers of technology capability.
+(2) Technology upgrading is a multidimensional process that includes technology, structural
+change, and interaction with the global economy.
+(3) Technology upgrading is based on a broader understanding of innovation which goes
+beyond R&D.
+(4) Technology upgrading is a multi-level process with micro, mezzo, and macro foun-
+dations. Technological change is never entirely aggregate or entirely micro based, but pri-
+marily is a process of structural change which takes place at the micro, mezzo, and macro
+levels. Lee ( 2013 ) takes this into account by exploring the issue of catching up at all three
+levels.
+(5) At the core of technology upgrading is structural change which is also a multidimensional
+process involving technological, industrial, and organisational change.
+(6) Technology upgrading is an outcome of the interactions between global actors (embodied
+in international trade and investment flows) and local technology accumulation activities
+(pursued by host country firms and governments) (Berend and Ranki 1982 ; Ernst and
+Kim 2002 ; Lall 1992 ) . The key to catch-up and post-catch-up is leveraging of domestic
+innovation efforts through global industry and knowledge networks (Ernst 2008a ) .
+---
+
+Asian Journal of Technology Innovation
+
+17
+
+![Figure](figures/figure_002.png)
+
+Figure 2: Dimensions of technology upgrading.
+
+Hence, the magnitude of knowledge inflows, and their coupling to domestic innovation efforts, are key dimensions of technology upgrading (Yoruk 2013 ) .
+
+Technology upgrading is also described as `industrial upgrading', which we find problematic. Economic sectors are increasingly diverse conglomerates of technologies at different levels of complexity, and their boundaries are often arbitrary due in great part to the changing nature of industry and services (cf. tertiaryisation of industry and industrialisation of services). The notion of industry is always context specific since its technological level is not discernible from the statistical definition of industry at whatever level of aggregation. However, we acknowledge that that there is degree of overlap between industry and technology upgrading since some industries are based on more complex technologies than others. In that respect, technology upgrading is about
+
+![Figure](figures/figure_006.png)
+
+Figure 3: Dimensions and components of technology upgrading.
+
+---
+
+18
+
+S. Radosevic and E. Yoruk
+
+changes in technology intensity, but equally about structural change. In fact, as we argue below, the two are inextricably linked.
+
+Based on a literature review, and at very general level, we conceptualise technology upgrading as a three-dimensional process (see Figure 2 ). It consists of dimension 1 (vertical axis) related to the intensity of technology upgrading exemplified by different types of innovation activities, dimension 2 (horizontal axis) which refers to the spread or breadth of the technology such as the diversity of technological knowledge, types of supporting infrastructure, and the firms' structure as the carrier of technology upgrading, and of dimension 3 (diagonal axis) which refers to knowledge inflows into the economy via a variety of channels such as trade, foreign direct investment (FDI) and GVCs. All three dimensions are grounded in the respective literatures on firmlevel technology upgrading, structural change, and growth, and integration in the global economy ( Figure 3 ). $^2$
+
+### 3.2. Intensity and types of technology upgrading (scale)
+
+This dimension of upgrading is about the acquisition of different types of technology capabilities, which also are a reflection of countries' different technological levels. Economies that operate behind the technology frontier are more likely to grow based on production rather than technology capability, while high-income economies are more likely to grow based on technology frontier (R&D based) activities.
+
+In clearly differentiating industry upgrading and technology upgrading and focusing only on the latter we are in danger of focusing only on disembodied knowledge and technology. Given the diversity of the forms in which technology is embodied, especially physical inputs and machinery, this would be problematic. Also, innovation activities in latecomer economies are largely about adoption of and improvements to imported machinery. Although, technology as a stock of knowledge should be separated from production, technological capacity and production capacity are strictly interconnected (Bell and Pavitt 1993 ) . In this respect,
+
+![Figure](figures/figure_007.png)
+
+Figure 4: From production capability to technology capability.
+
+Source: Based on and adapted from Amsden and Tschang (2003).
+
+---
+
+Asian Journal of Technology Innovation
+
+19
+
+we differ from Archibugi and Coco (2005) who abstract production from technological capability.
+
+Three types of capabilities, production capability, technology capability , and R & D and knowledge intensity are present in all economies, to different degrees. Their importance as drivers of growth vary according to their dependence on achieved income and their technology levels and the structural features of their economies. Majcen, Radosevic, and Rojec ( 2009 ) and Kravtsova and Radosevic ( 2011 ) show that, in Eastern Europe, production capability was a significant determinant of productivity growth at both the micro and macro levels. This is not unique to this region, but is a general feature of middle-income economies more generally, as highlighted in Figure 4 and in the literature on technological capabilities cited earlier. The types of technology capabilities are depicted in Figure 4 based on Amsden and Tschang's ( 2003 ) seminal discussion of R & D indicators.
+
+Figure 4 differentiates between production capability, process and product engineering capability, advanced and exploratory development, applied research and basic research.
+
+Production capability is the capability to produce at a given technology level at world levels of efficiency and productivity. This requires good operational efficiency and a skilled technical and blue-collar workforce.
+
+A more complex capability is product and process engineering, which involves improvements to existing products and processes. This capability is largely dependent on skilled engineers. An extension of this capability is advanced development which Amsden and Tschang ( 2003 ) distinguish from exploratory development. Advanced development is about manufacturing prototypes while exploratory development is about system prototypes. There is an important threshold capability level between advanced development enabling manufacture, to own design manufacture. Production capability, process and product engineering, and advanced development are achievable by OEM enterprises, while exploratory development is a feature of ODM.
+
+These stages are not necessarily hierarchically structured – that is, moving from advanced to exploratory development or from exploratory development to applied research or from applied research to basic research does not necessarily involve higher technology complexity (although it may); it simply involves a qualitatively different set of technology or knowledge requirements. Equally, upgrading to `higher' stages is not automatically more rewarding in terms of value added – that is, upgrading may not necessarily lead to increased income, but might be necessary simply to maintain the existing income levels.
+
+The literature on technological capabilities explores several cases of upgrading from production capability to design capability which can be considered paradigmatic. The best examples are cases in East Asia – a region that can be considered a paragon of technology upgrading and catch-up. Taiwan's IT industry is a good example of progress from production to design capability in a period of some 20 – 30 years (Ernst 2013 ) . Hobday et al. ( 2004 ) explore the transformation of Korean firms from catch-up to post-catch up. Dutrenit ( 2000 ) provides an in-depth example of the transformation from production to OBM world-class capability among Mexican firms. In the case of Eastern Europe, Radosevic and Yoruk ( 2004 ) analyse the transformation of ex-socialist electronics conglomerates to contract manufacturers.
+
+These examples show that technology upgrading is not a linear and autonomous process, but is a non-linear process involving several threshold levels. The move from one to another stage is not guaranteed and requires a new set of technical, financial, and organisational preconditions. Also, past successes are usually a source of weakness since it is necessary to pursue a dual strategy to move between stages. For example, success in the technology stage is based on low cost manufacturing capabilities and the capacity to imitate the technology leaders when introducing new products. The network on which this capability relies is not sufficient for the next hierarchical stage, which requires more differentiated knowledge networks
+
+---
+
+20
+
+S. Radosevic and E. Yoruk
+
+and much larger investments in R & D-based developments and marketing. Finally, reliance on GVCs is the key to success in the initial production and technology capability stages. However, GVCs can become a source of vulnerability since technology leadership requires more autonomous development, strong local demand for technology and a variety of specialised services and knowledge providers.
+
+### 3.3. Breadth of technology upgrading: structural change, infrastructure, and firm structure (scope)
+
+Technology upgrading is more than intensity or scale of technological activities. It includes the breadth or scope of structural factors that affect the intensity of technology upgrading. There are three structural factors: the structural change itself, the infrastructure (human, physical, and organisational), and the firm structure.
+
+### 3.3.1. Structural change
+
+There is no general theory of structural change , although there is a variety of theoretical approaches that use different methodologies that claim to explain structural shifts between three broad sectors, and among the industries within these sectors (Krüger 2008 ) . There is a common understanding that technological change affects structural change in relation to how industries with relatively low rates of productivity growth tend to shrink in terms of market shares while those with higher rates of productivity growth expand. Thus, structural change promotes aggregate productivity growth even if within industries, productivity growth is stagnant. However, the empirical evidence on the role of structural change in aggregate productivity growth differs widely across time periods, countries, and regions. Structural change often is dominated by the inter or intra-sectoral effects of productivity growth. Peneder ( 2003 ) concludes that structural change generates positive as well as negative contributions to aggregate productivity growth. However, since many of these effects net out, structural change appears to have only a weak impact on average. However, Sandven, Smith, and Kaloudis ( 2005 ) show that structural change in the manufacturing industries of the OECD countries is not a key feature of the growth process. Growth is based primarily on internal transformations in the low and medium tech sectors, rather than the creation of new sectors. For instance, the incorporation of information and communication technology (ICT) in medium or low-tech products is only one element of innovation in these sectors.
+
+The issue is complicated by the level at which structural change is observed. Jorgenson and Timmer ( 2011 ) show that splitting the economy into agriculture, industry, and services is no longer relevant. There is substantial heterogeneity within the services sector, while the use of ICT and skilled labour is increasing in all sectors, and especially in services. The linkages among industries and innovation, including generic technologies such as ICT, are permeating structural change. The empirical results do not support the idea that growth is correlated with the share of the high tech sectors ( Sandven et al. 2005 ) . Instead, we are observing a change in the nature of industries and services, and their convergence. This is exemplified by knowledge intensive business services (KIBS), which are especially prominent in these developments. $^3$ The importance of KIBS as sources of innovation, technologies, and inputs has increased steadily over time resulting in the linkages between KIBS and the manufacturing industries in different countries strengthening over time.
+
+Another structural change is the increasing importance of knowledge in all economic activities, which is being captured by knowledge intensive activities (KIA) defined as economic sectors
+
+---
+
+Asian Journal of Technology Innovation
+
+21
+
+Table 3: Components of structural factors of technology upgrading.
+
+<table><tr><td>Structural changes which promote technology upgrading</td><td>ICT as generic technology: proxy of structural change Knowledge intensive business services (KIBS) Knowledge intensive activities (KIA) Technology diversification</td></tr><tr><td>Infrastructural upgrading</td><td>An important element or externality of technology upgrading. Inefficiencies in infrastructure can hinder otherwise competitive firms to upgrade Infrastructure upgrading (Ozawa 2009) Human capital</td></tr><tr><td>Firms' structure</td><td>Share of large firms Share of SMEs</td></tr></table>
+
+
+in which more than 33% of the employed labour force have completed an academic tertiary education (i.e. ISCED 5 and 6 levels; ECR 2011 ) .
+
+Imbs and Wacziarg ( 2003 ) find that sectoral concentration in relation to per capita income follows a U-shape. Thus, increased sectoral specialisation applies only to high-income economies. Countries diversify over most of their development path (Imbs and Wacziarg 2003 , p. 64), which is in line with Lee's ( 2013 ) findings on growth and structural change which show that technology diversification is an important and robust feature of the transition from middle to (upper) high incomes.
+
+In summary, despite recognition of the importance of structural change we can say little about the importance of different sectors and industries related to economic growth. The essence of structural change driven by technology is that it changes both the boundaries to and the nature of industries. Hence, using high tech to proxy for structural change is misleading since many high-tech elements permeate many low-tech sectors. Also, catching-up countries are involved increasingly in high-tech industries, but in low value-added segments. So, instead of focusing on structural change at the industry level, we investigate more reliable trends related to technological change. By this we refer primarily to the increasing importance of ICT in all economic sectors and activities; the increasing importance of the convergence between manufacturing and services captured by KIBS; the increasing knowledge intensity of all sectors of the economy captured by the KIA indicator; and increasing technology diversification as countries upgrade technologically (see Table 3 ).
+
+### 3.3.2. Infrastructural upgrading
+
+Technology upgrading takes place primarily in firms, but is not just a firm-level issue. The accumulation of technology capability in firms must be accompanied by an organisational and institutional infrastructure that supports the acquisition of such capabilities. Choung et al. ( 2014 ) show that the transition from adoption (catching-up) to creation (post-catch-up) depends on the range of infrastructure to support a country's innovation activity, and the strategies and resources of individual companies. We consider infrastructure to be an important dimension of structural change.
+
+Infrastructure is one component of an economy's endowments, which generates large externalities for other firms in relation to transaction costs. It is both a public good and an input to the production of other intermediate inputs. Access to infrastructure services is strongly correlated with a country's average income. Infrastructure matters, but the evidence does not provide an unequivocal argument in favour of more or less infrastructure investment (Prud’homme 2005 ) . However, despite the obvious problems related to estimating the direction of causation
+
+---
+
+22
+
+S. Radosevic and E. Yoruk
+
+between infrastructure investment and growth, the evidence suggests that infrastructure causes growth, which, in turn, causes greater demand for (and supply of) infrastructure. As countries reach a certain stage of economic development, the extent to which infrastructure represents a binding constraint on development, changes.
+
+Human capital is an important driver of growth (Glaeser, La Porta, Lopez-de-Silanes, and Shleifer 2004 ) . The technology embodied in new machinery and equipment will not by itself lead to increased productivity without the human capabilities to use it effectively. Human capital based on education is an infrastructural precondition for or input into technology upgrading. In addition to school attainment, and years of schooling, differences in learning achievements matter even more for explaining cross-country differences in productivity growth (Hanushek and Wößmann 2007 ) . Long run growth is closely related to the level of cognitive skills in the population (Hanushek 2013 ) . In short, evidence on the `quantity' and quality of human capital supports its inclusion as an important structural component.
+
+However, in order to be effective, human skills need to be part of a specific organisational and economic process that rewards dexterity, learning, and innovation. Human skills need to be converted into firm specific skills to achieve technology upgrading. Eastern Europe is good example of a region whose labour force post-1989 had relatively high education levels, but low firm specific skills. Post-1989, this high level of education was considered an advantage and, in a few cases led to domestic innovations, but with meagre economic results. A good example is Estonia's ICT industry (see Hogselius 2005 ) which generated several innovations in IT services. However, a careful examination of this case shows that individual competencies without firm specific organisational capabilities were not sufficient. Inherited competencies are strongest at the level of individuals which is not enough to develop a dynamic innovation system, which requires organisational capabilities to harness individual competencies. Estonia's ICT sector has developed customised innovations, which, by definition, are not directly transferable to other contexts. These product innovations are either easy to imitate or apply to services rather than a unique, firm-specific-accumulated technological competence. Estonia's weak organisational capabilities were further reinforced by barriers to exporting. Another example is inherited skills in electronics in 10 socialist era electronics conglomerates. In only one case were these skills preserved due largely to organisational capabilities and the strategies of top management (see Radosevic and Yoruk 2004 ) . In all other cases, these labour force skills were not employed in a productive context enabling the development of new organisational capabilities within which this human capital could be successfully deployed.
+
+Another, perhaps better known case, is the Indian software sector whose success is usually attributed to individuals. Indeed, competition from MNEs in labour markets (not product markets) has induced productive efficiency among domestic firms. However, the key to India's exploitation of human capital was the organisational capabilities of Indian software firms, which were based on imitation and deployed the productive skills of its labour force (Athreye 2005 ) . This leads to the next important structural factor: firm structure.
+
+### 3.3.3. Firm structure
+
+Schumpeterian economics emphasises firm heterogeneity as an essential feature of industry dynamics (Nelson and Winter 1985 ) . One aspect of heterogeneity is size differences or the role of large versus small firms in technology upgrading. Views on the role of large versus small firms are quite divided. Chandler ( 1994 ) is best known account of the importance of big business in economic growth. With the advent of `third industrial revolution' and globalisation, a view emerged that `the Chandlerian firm is under siege from a panoply of decentralised and marketlike forms that often resemble some of the “inferior” nineteenth-century structures the managerial
+
+---
+
+Asian Journal of Technology Innovation
+
+23
+
+enterprise replaced' (Langlois 2003 , p. 355) . On the hand, there is increasing evidence that globalisation has reinforced the importance of big business. Despite the new institutional framework, which gives prominence to markets and networks, global enterprises have remained the major players in global markets. It is true that the new paradigm has significantly influenced both the firm boundaries and the patterns of inter-organisational division of labour, but the `visible hand' of organisations, and the relative competitive advantage of size persist (Dosi, Gambardella, Grazzi, and Orsenigo 2008 ) . Lee, Kim, Park, and Sanidas ( 2013 ) provides the most grounded confirmation of a significant and robust relationship between the number or sales of large firms, such as Fortune 500 firms, and national economic growth, after controlling for country size and endogeneity. Lee et al. show that among latecomer economies, China and Korea, have more large firms than their country size would predict, whereas some middle-income countries have fewer large firms than their size would predict and many high-income countries have many more such companies. This is in line with IFC ( 2013 ) survey data that shows that the share of large firms in employment is larger in high income and upper middle-income economies compared to lower income countries. Evidence on the role of small and medium-sized enterprises (SMEs) is also weaker. When not controlling for endogeneity, Beck, Demirguc-Kunt, and Levine ( 2005 ) find a positive, but weak correlation between SME growth and per capita income, and no significance when controlling for endogeneity.
+
+Since, by definition, organisational capabilities are more complex and more developed in large firms, firm structure is an important factor in technology upgrading. Based on the above evidence we formulate our working hypothesis that the share of large firms is conducive to technology upgrading other things being equal.
+
+### 3.4. Interaction with global economy for technology upgrading
+
+Successful technology upgrading is never an entirely autonomous process, and is always linked to the inflow of foreign knowledge skills coupled with intensive domestic technology efforts (Radosevic 1999) . There is a large body of literature on this topic; a few examples will suffice to emphasise this robust, but often forgotten stylised fact (Mowery and Oxley 1997, Kim 1997, Amsden 2001) . The focus is usually on one of two elements of catching up – domestic technology accumulation or inflows of foreign knowledge through trade, FDI and a generally open economic regime.
+
+The FDI and technology upgrading or knowledge spillovers literature is quite extensive. A meta-review of this literature by Bruno and Campos ( 2013 ) shows that the effect of FDI on economic performance and growth is conditional. Firms, sectors, or countries that are below certain `thresholds' (in terms of human capital, financial development, or institutional quality) are less likely to benefit from FDI. Overall, the benefits are significantly greater in low income than in lower and upper middle-income countries (at both the micro and macro levels). The available data provide strong support for the differentiating effect of FDI on growth across levels of development rather than geographic regions.
+
+The effect at the macro level depends on whether the recipient countries have achieved a minimum level of human capital, financial, and institutional development. The effects of FDI based on firm-level data tend to show that the (micro-) effect is conditional upon the type of linkage (with backward linkages, that is, links between the firm and its suppliers, dominating horizontal, or forward linkages).
+
+FDI is a potential source of technology upgrading. Integration into the global economy, and FDI can be important catalysts for change, while on its own FDI does not drive technology upgrading. The literature suggests that its effects on upgrading are highly differentiated and dependent on indigenous technology efforts. The fact that countries are globally integrated in
+
+---
+
+24
+
+S. Radosevic and E. Yoruk
+
+R & D networks does not mean that they are linked to the domestic manufacturing value chain, which can lead to what Ernst ( 2014 ) describes in an example based on Indian electronics, as ‘truncation of FDI based learning’. He explains this as due to the fragmented Indian national innovation system, in which local electronics manufacturing remains disconnected from India’s chip-design capabilities which are integrated into global innovation and production networks.
+
+FDI indicators are of limited value for detecting the true knowledge acquired through international industry networks. Research on GVCs is useful in that respect, although difficult to generalise. Various contributions show the positive and significant effects of learning through value chains, related to process, product, and functional upgrading up to the ODM level. Yoruk ( 2013 ) shows the major importance of both knowledge and production networks for firms' upgrading, but shows also that it is misleading to reduce the learning opportunities for upgrading to interactions with global buyers within GVCs. Her research in the case of Eastern Europe shows that learning by doing and learning by exporting do not have a statistically significant effect on functional upgrading. She shows that opportunities offered by GVCs are of little use unless firms have the ability to internalise this external knowledge through their human resources, and through training and research within the firm. Yoruk shows also that managerial upgrading is important for technology upgrading, but global buyers do not support this. Again, the importance of organisational capabilities or firm structure is highlighted as discussed in relation to the structural dimensions of technology upgrading.
+
+Globalisation of technology exploitation and collaboration, and also technology generation through the globalisation of the R & D process, further increases the importance of international linkages for industrial upgrading (UNCTAD 2005 ) . An example of the importance of integration into GVCs and its growth benefits is the German $-$ Central European Supply Chain Cluster (GCESC) (IMF 2013a,b ) . The increase in foreign value added in four major country locations of the GCESC (Czech Republic, Slovakia, Poland, and Hungary/CE4) appear to have led to increases in domestic value added through productivity increases and creation of demand for ancillary products and services in the host economies. It seems that participation in the GCESC has led to valuable technology transfer to the CE4 countries, although there is no clear consensus on its magnitude due to the heterogeneity among firms in terms of skills.
+
+The weakness of many technology upgrading metrics is that they focus on explicit domestic technology efforts or the import of knowledge via licences, and neglect knowledge inflows in embodied forms via imported inputs and equipment. In countries behind the technology frontier growth is driven mainly by the diffusion and absorption of technologies that are new to the firm or the country, but not new to the world. Domestic knowledge generation in the business enterprise sector through R & D or non – R & D technical activities, and the accumulation of knowledge in the public R & D system is not yet the major driver of growth compared to indirect knowledge and R & D embodied in imported inputs and machinery. R & D activity in other industries has a significant effect on productivity growth. Own-R & D activity accounts for about one-half of total R & D in countries at the technological frontier, and for between one-quarter and one-half of total R & D in countries below the frontier (Knell 2008 ) .
+
+Finally, mobility of people is an effective channel of knowledge transfer and technology upgrading. It is crucial for transferring tacit knowledge and initiating learning.
+
+Of the three dimensions of technology upgrading, interaction with the global economy is probably the most difficult to capture since technology transfer happens through capital equipment imports, is embedded in FD, networks and subcontracting, or is disembodied (in the form of licences). However, modes of technology transfer cannot be used as proxies for real knowledge transfer (Radosevic 1999 ) . Thus, we consider the distinction between technology (embodied) imports, knowledge imports (licences), and knowledge cooperation (R&D
+
+---
+
+Asian Journal of Technology Innovation
+
+25
+
+cooperation) as components of the interaction with the global economy dimension, which are distinctive, but can be captured in data.
+
+## 4. Technology upgrading in the context of catch-up and post-catch-up
+
+As highlighted in introduction, our point of departure is the assumption that it is unrealistic to search for universal and simple metrics of technology upgrading. We argued also that current metrics are particularly inappropriate for countries behind the technology frontier. Hence, we focus most; on a `broad' middle-income group, which is the largest country grouping and one where technology-upgrading issues play a key role in the catching-up process. Accordingly, our conceptual framework is best suited to depicting the transition from the `broad' middle to the (upper) high-income group.
+
+The papers in this issue of AJTI address the issue of transition from catch-up to post-catch up. In terms of the different income groupings discussed in Section 1 , this is a transition from the lower high income to the upper high-income group, or shift from $ 12,745 $-$ 30 K per capita to over $ 30 K per capita. $^4$ A `broad' middle-income group perspective should include the upper middle-income group ( $ 4126 $-$ 12,745pc). This has been actually done not so much through the country level analyses (cf. paper on Brazil as upper middle-income economy) but much more through analytical focus on technology accumulation issues of `broad' middle-income group (east Asia and Latin America).
+
+In terms of technology upgrading, the transition from catch-up to post-catch up represents a shift from technology diversification activities to technology frontier activities. However, as we have pointed out, each income group practices a variety of modes of technology upgrading, that is, a mix of technology imitation, diversification, and world frontier activities. However, the composition of these activities differs especially regarding the drivers of this technology accumulation. So, achieving the post-catch-up stage does not imply that all activities are at the technology frontier, but rather that technology frontier activities gradually are becoming those that drive technology accumulation. For example, in the post-catch-up stage, production capabilities are less important than technology capabilities, but remain an important part of the technology spectrum as demonstrated by recent attempts by the US to return manufacturing from Asia (Berger 2013 ) . Also, as countries transform from the catch-up to the post-catchup stage, R & D and science in general becomes more important in relation to its world knowledge frontier generation function than in terms of absorption capabilities (see Radosevic and Yoruk 2014 ) . The importance of the knowledge infrastructure increases compared to the physical infrastructure. The role of large firms' organisational capabilities, but also their interaction with technology active small firms and extra-mural research organisations, becomes especially important. Also, the nature of the knowledge trade changes from decreasing importance of foreign acquisition of national patents, towards co-inventions and national acquisition of foreign patents (see Jindra, Lacasa, and Radosevic 2015 ) . Conversely, the catch-up stage does not mean that the country does not include pockets of world frontier technology activities, but these are not the main drivers of technology accumulation in the economy. Due to modernisation efforts, the infrastructure in middle-income economies may be better developed that might be expected given their income level. However, other dimensions of technology upgrading may be lagging behind, thus, reducing potential complementarities and increasing returns. For example, a high level of technology openness may be accompanied by weak intensity of own technology activities in production, R & D or technology areas. Also, imbalances may be concentrated in specific dimensions of technology upgrading. For example, a welldeveloped R & D and innovation infrastructure may be neither firm nor technology specific
+
+---
+
+26
+
+S. Radosevic and E. Yoruk
+
+due to the absence of sophisticated large or medium sized firms to articulate demand for R&D and innovation services.
+
+In summary, the intensity and weight of different types of technology activities changes as countries move from catch-up to post-catch up. This process can be unbalanced or harmonious with different dimensions of technology upgrading reinforcing each other. We believe that our three pronged conceptual framework allows us to model the strengths and weaknesses of different dimensions of technology upgrading and would be useful to policy-makers.
+
+If catch-up countries adopt the metrics of the technology leaders they will lack important insights into their position in the transition to post-catch-up. We see major value in alternative metrics for understanding differences in the accumulation of innovation capabilities as reflected in the different dimensions of technology upgrading rather than summary country rankings. Approaches and metrics that go beyond the science and technology dimensions of technology upgrading, and recognise its multidimensional nature can provide new insights into innovation activities that allow latecomer countries to establish new technological trajectories for innovation (Choung et al. 2014 ) .
+
+## 5. Conclusions
+
+We argued in this paper that we need a theory and metrics of technology upgrading for several reasons. First, aggregate theories of growth are not useful and represent a kind of `holy grail' of growth theory. Second, current metrics of technology upgrading are either atheoretical or are not useful for countries behind technology frontier. Third, such metrics lead to confusing or irrelevant policies. Hence, there is a need to generate a theoretically relevant, but empirically grounded middle level conceptual and statistical framework to model the type of challenges relevant to a large number of economies at different income levels.
+
+We reviewed the relevant literature and derived a conceptual framework which we consider provides a useful approach to a theory and metrics of technology upgrading. In a nutshell, we conceptualise technology upgrading as a three-dimensional process that includes intensity and types of technology upgrading through various types of innovation and technology activities; a broadening of technology upgrading through different forms of technology and knowledge diversification; and interaction with the global economy through different forms of knowledge import, adoption, and exchange. We discussed each of these dimensions in some detail pointing to their major components and justifying their importance for an understanding of the three dimensions of technology upgrading. We consider this to be a necessary first step towards a theory and metrics of technology upgrading, and the generation of a relevant composite indicator of technology upgrading in countries at different levels of income. Finally, we consider this approach to be of relevance to understanding the transformation from catch-up to post catch-up. Although the aim is to reach the post-catch up stage, we argue that it would be misleading to apply the metrics of the technology leaders to measure this progress.
+
+Finally, we would highlight two limitations of this analysis. First, we focus on a new conceptual approach to the measurement of technology upgrading. For reasons of both space and complexity we do not discuss specific measures and indicators. However, our work in progress is aimed at combining a new conceptual framework with the existing data, which as might be suspected is far from satisfactory (see Radosevic and Yoruk 2015 for preliminary results). There is a need to generate new data related specifically to non-R & D activities, the technology content of trade and FDI, and the technological features of structural change.
+
+The existing metrics are excessively biased towards science and technology modes of learning and greatly underestimate `doing-using-interaction' modes of learning (see Jensen, Johnson, Lorenz, and Lundvall 2007 5 ). As our in depth case study suggests, `the dominant aggregated
+
+---
+
+Asian Journal of Technology Innovation
+
+27
+
+S&T indicators are of little, if any, relevance for understanding the process of innovativeness, knowledge formation and technical change in industry' (Laestadius 1998 , p. 393) except in `modern' academic-based industries such as IT and pharmaceuticals (Laestadius 1998 ) . However, diffusion of science based generic technologies into traditional and natural resource based activities will further increase these biases and, thus, provide an unrealistic basis for decision-making. The only solution is to generate new data which should provide a more realistic understanding of the innovation activities and different dimensions of technology upgrading.
+
+We also made it clear in the introduction to this paper that the institutional context is also important to explain different growth performance, but our primary concern is with technology upgrading as an accumulation of technology capabilities issue, at different levels. In this respect, we see great value in measurement approaches aimed at depicting the accumulation of capabilities, rather than their being confused with specific institutional variables such as types of labour markets, financial systems, etc. However, we are aware that different capabilities and institutional arrangements are required as catch-up economies transit from catch-up. Finally, the contributions in this Special Issue show the changing institutional requirements for the transformation from catch-up to post catch-up. We consider our analytical approach as complementing not competing with perspectives that explore and measure the interaction between technological characteristics and the institutional setting.
+
+## Acknowledgements
+
+We are grateful to an anonymous reviewer for comprehensive and grounded comments on an earlier version of this paper. All remaining errors are our responsibility.
+
+## Notes
+
+1. In what follows, the middle income group includes lower middle, upper middle income, and lower highincome countries, and constitutes a `broad' middle income group, which we describe simply as `middle income' economies. In the analysis we refer to three specific sub-groups within the broad middleincome category (lower middle, upper middle, and lower high income), and confine the high-income group to the `upper high income group'.
+
+2. We are aware that technology upgrading emerges through interactions between demand and supply factors. However, we consider that including demand upgrading would introduce another layer of complexity and would make our conceptual framework unmanageable.
+
+3. KIBS are defined according to the NACE classification, NACE REV 1.1. as including the categories computer and related activities (NACE 72), research and development (NACE 73), and other business activities (NACE 74).
+
+4. GNI per capita Atlas method, data for 2013.
+
+5. We are in full agreement with Jensen et al. ( 2007 , p. 685) who argue that the absence of DUI type indicators reflects `political priorities and decision-making rather than any inevitable state of affairs'.
+
+## Funding
+
+This work was supported by the European Commission (GRINCOH project Nr. 290657).
+
+## ORCiD
+
+Slavo Radosevic http://orcid.org/0000-0001-7905-2281
+
+---
+
+28
+
+S. Radosevic and E. Yoruk
+
+## References
+
+Abernathy, W.J., and Utterback, J.M. (1978), 'Patterns of industrial innovation', Technology Review, 8(7), 40-47.
+
+Acemoglu, D., Zilibotti, F., and Aghion, P. (2006), 'Distance to frontier, selection, and economic growth', Journal of the European Economic Association, 4(1), 37-74.
+
+Aghion, P., Akcigit, U., and Howitt, P. (2013), What Do We Learn From Schumpeterian Growth Theory? NBER Working Paper Series No 18824, February.
+
+Akamatsu, K. (1962), 'A historical pattern of economic growth in developing countries', The Developing Economies, 1(1), 3-25.
+
+Amsden, A.H. (2001), The Rise of “The Rest”’. Challenges to the West from Late-industrializing Economies, Oxford: Oxford University Press.
+
+Amsden, A.H., and Tschang, F. T. (2003), 'A new approach to assessing the technological complexity of different categories of R&D (with examples from Singapore)', Research Policy, 32, 553-572.
+
+Archibugi, D., and Coco, A. (2005), ‘Measuring technological capabilities at the country level: a survey and a menu for choice’, Research Policy, 34(2005), 175–194.
+
+Archibugi, D., Denni, M., and Filippetti, A. (2009), ‘The technological capabilities of nations: the state of the art of synthetic indicators’, Technological Forecasting and Social Change, 76(7), 917–931.
+
+Athreye, S. (2005), ‘The Indian software industry and its evolving service capability’, Industrial and Corporate Change, 14(3), 393–418.
+
+Beck, T., Demirguc-Kunt, A., and Levine, R. (2005), 'SMEs, growth and poverty: cross-country evidence', Journal of Economic Growth, 10(3), 199-229.
+
+Bell, M., and Pavitt, K. (1993), 'Technological accumulation and industrial growth: contrasts between developed and developing countries', Industrial and Corporate Change, 2(2), 157-210.
+
+Berend, I.T., and Ránki, G. (1982), The European Periphery and Industrialization, 1780–1914, New York: Cambridge University Press.
+
+Berger, S. (2013), Making in America: From Innovation to Market, (August 2013) with the MIT Task Force on Production in the Innovation Economy, Cambridge, MA: The MIT Press.
+
+Bruno, R.L., and Campos, N.F. (2013), Reexamining the Conditional Effect of Foreign Direct Investment , IZA Discussion Paper No. 7458, June 2013. http://papers.ssrn.com/sol3/papers.cfm?abstract_id= 2287068 .
+
+Bruno, R., Douarin, E., Korosteleva, J., and Radosevic, S. (2015), ‘Technology choices and growth: testing new structural economics in transition economies’, Journal of Economic Policy Reform, 18(2), 131– 152.
+
+Chandler, A.D. Jr (1994), Scale and Scope. The Dynamics of Industrial Capitalism, Cambridge, MA: Harvard University Press.
+
+Chenery, H., and Syrquin, M. (1975), Patterns of Development, 1950-1970, (World Bank), Oxford: Oxford University Press.
+
+Choung, J.-Y., Hwang, H. R., and Song, W. (2014), 'Transitions of innovation activities in latecomer countries: an exploratory case study of South Korea', World Development, 54, 156-167.
+
+Crépon, B., Duguet, E., and Mairesse, J. (1998), 'Research and development, innovation and productivity: an econometric analysis at the firm level', Economics of Innovation and New Technology, 7(2), 115158.
+
+Dahlman, C., Ross-Larson, B., and Westphal, L.E. (1987), 'Managing technological development: lessons from the newly industrializing countries', World Development, 15(6), 759-775.
+
+Dosi, G., Gambardella, A., Grazzi, M., and Orsenigo, L. (2008), ‘Technological revolutions and the evolution of industrial structures: assessing the impact of new technologies upon the size and boundaries of firms’, Capitalism and Society, 3(1), 1–47, Article 6.
+
+Dutrenit, G. (2000), Learning and Knowledge Management in the Firm: From Knowledge Accumulation to Strategic Capabilities, Aldershot: Edward Elgar.
+
+Easterly, W. (2001), The Elusive Quest for Growth. Economists' Adventures and Misadventures in the Tropics, Cambridge, MA: MIT Press.
+
+Ernst, D. (1997), From Partial to Systemic Globalization: International Production Networks in the Electronics Industry, BRIE Working Paper 98.
+
+Ernst, D. (1998), 'Catching-up crisis and industrial upgrading: evolutionary aspects of technological learning in Korea's electronics industry', Asia Pacific Journal of Management, 15(2), 247-283.
+
+Ernst, D. (2001), Global Production Networks and Industrial Upgrading – A Knowledge Centred Approach, East-West Center Working Papers, Economics Series, No. 25.
+
+---
+
+Asian Journal of Technology Innovation
+
+29
+
+Ernst, D. (2008a), ‘Asia’s “upgrading through innovation” strategies and global innovation networks: an extension of Sanjaya Lall’s research agenda’, Transnational Corporations, 17(3), 22–44. December 2008.
+
+Ernst, D. (2008b), Can Chinese IT Firms Develop Innovative Capabilities within Global Knowledge Networks? East-West Center Working Papers, Economics Series, 94.
+
+Ernst, D. (2013), Industrial Upgrading through Low-Cost and Fast Innovation—Taiwan’s Experience, EastWest Center working paper, Honolulu, Economics Series, No. 133, September 2013.
+
+Ernst, D. (2014), Upgrading India’s Electronics Manufacturing Industry: Regulatory Reform and Industrial Policy, Honolulu: East West Centre.
+
+Ernst, D., and Kim, L. (2002), ‘Global production networks, knowledge diffusion and local capability formation’, Research Policy, 31, 1417–1429.
+
+European Competitiveness Report (2011), Chapter 2 ‘Convergence of Knowledge Intensive Sectors and the EU’s External Competitiveness’, Brussels: EC.
+
+Foray, D. (2015), Smart Specialization. Opportunities and Challenges for Regional Innovation Policy, London: Routledge.
+
+Freeman, C. (2002), 'Continental, national and sub-national innovation systems—complementarity and economic growth', Research Policy, 31, 191-211.
+
+Freeman, C., and Louçã, F. (2001), As Time Goes by From the Industrial Revolutions to the Information Revolution, Oxford: Oxford University Press.
+
+Furman, J.L., and Hayes, R. (2004), 'Catching up or standing still? National innovative productivity among ‘follower’ countries, 1978–1999’, Research Policy, 33(2004), 1329–1354.
+
+Gereffi, G. (1999), 'International trade and industrial upgrading in the apparel commodity chain', Journal of International Economics, 48(1), 37-70.
+
+Glaeser, E.L., La Porta, R., Lopez-de-Silanes, F., and Shleifer, A. (2004), 'Do institutions cause growth?', Journal of Economic Growth, 9, 271-303.
+
+Gort, M., and Klepper, S. (1982), 'Time paths in the diffusion of product innovations', The Economic Journal, 92(367), 630-653.
+
+Hanushek, E. (2013), ‘Economic growth in developing countries: the role of human capital’, Economics of Education Review, 37, 204–212. doi:10.1016/j.econedurev.2013.04.005.
+
+Hanushek, E.A., and Wößmann, L. (2007), Education Quality and Economic Growth, Washington, DC: World Bank.
+
+Hausmann, R., Hwang, J., and Rodrik, D. (2007), 'What you export matters', Journal of Economic Growth, 12, 1-25.
+
+Hausmann, R., and Klinger, B. (2006), Structural Transformation and Patterns of Comparative Advantage, CID Working Paper No. 128. Cambridge, MA: Center for International Development, Harvard University.
+
+Hidalgo, C., and Hausmann, R. (2009), 'The building blocks of economic complexity', Proceedings of the National Academy of Sciences, 106(26), 10570-10575.
+
+Hidalgo, C., Klinger, B., Barabasi, A.L., and Hausmann, R. (2007), ‘The product space conditions the development of nations’, Science, 317, 482–487.
+
+Hobday, M. (1995), 'East Asian latecomer firms: learning the technology of electronics', World Development, 23(7), 1171-1193.
+
+Hobday, M., Rush, H., and Bessant, J. (2004), ‘Approaching the innovation frontier in Korea: the transition phase to leadership’, Research Policy, 33, 1433–1457.
+
+Hogselius, P.T. (2005), Dynamics of Innovation in Eastern Europe: Lessons from Estonia, Cheltenham: Edward Elgar.
+
+Humphrey, J., and Schmitz, H. (2004), ‘Governance in global value chains’, in Local Enterprises in the Global Economy. Issues of Governance and Upgrading, ed. H. Schmitz, Cheltenham: Edward Elgar, pp. 95–109.
+
+IFC. (2013), IFC Jobs Study. Assessing Private Sector Contributions to Job Creation and Poverty Reduction, Washington: International Financial Corporation.
+
+Imbs, J., and Wacziarg, R. (2003), 'Stages of diversification', American Economic Review, 93(1), 63-86.
+
+IMF. (2013a), German-central European Supply Chain-cluster Report, IMF Country Report 13/263, August 20th, 2013.
+
+IMF. (2013b), Trade Linkages, Balance Sheets, and Spillovers: The Germany-central European Supply Chain , Prepared by Selim Elekdag and Dirk Muir, IMF Working Paper, European Department, October 2013.
+
+---
+
+30
+
+S. Radosevic and E. Yoruk
+
+Jensen, M.B., Johnson, B., Lorenz, E., and Lundvall, B.A. (2007), ‘Forms of knowledge and modes of innovation’, Research Policy, 36, 680–693.
+
+Jindra, B., Lacasa, I.D., and Radosevic, S. (2015), Dynamics of Technology Upgrading of the Central and East European Countries in a Comparative Perspective: Analysis Based on Patent Data, Economics and Business Working Paper No.135, February, UCL Centre for Comparative Economics.
+
+Jorgenson, D.W., and Timmer, M.P. (2011), 'Structural change in advanced nations: a new set of stylised facts', Scandinavian Journal of Economics, 113(1), 1-29.
+
+Kim, L. (1980), 'Stages of development of industrial technology in a developing country: a model', Research Policy, 9(3), 254-277.
+
+Kim, L. (1997), Imitation to Innovation: The Dynamics of Korea’s Technological Learning, Boston, MA: Harvard Business School Press.
+
+Klepper, S. (1997), 'Industry life cycles', Industrial and Corporate Change, 6(1), 145-182.
+
+Klepper, S., and Graddy, E. (1990), 'The evolution of new industries and the determinants of market structure', The RAND Journal of Economics, 21, 27-44.
+
+Knell, M. (2008), Product-Embodied Technological Diffusion and Intersectoral Linkages in Europe, 28 April, Innovation Watch – SYSTEMATIC, NIFUSTEP.
+
+Kravtsova, V., and Radosevic, S. (2011), ‘Are systems of innovation in Eastern Europe efficient?’, Economic Systems, 36, 109–126. doi:10.1016/j.ecosys.2011.04.005.
+
+Krüger, J.J. (2008), 'Productivity and structural change: a review of the literature', Journal of Economic Surveys, 22(2), 330-363.
+
+Kuznets, S. (1930), Secular Movements in Production and Prices: Their Nature and Their Bearing upon Cyclical Fluctuations, Boston, MA: Houghton Mifflin, New York, repr. Kelley, New York, 1967.
+
+Laestadius, S. (1998), 'The relevance of science and technology indicators: the case of pulp and paper', Research Policy, 27, 385-395.
+
+Lall, S. (1992), 'Technological capabilities and industrialization', Research Policy, 20(2), 165-186.
+
+Langlois, R.N. (2003), 'Chandler in a larger frame: markets, transaction costs, and organizational form in history', Business and Economic History, 1(3), 355-375.
+
+Lee, K. (2013), Schumpeterian Analysis of Economic Catch-up, Knowledge, Path-creation and the MiddleIncome Trap, Cambridge: Cambridge University Press.
+
+Lee, K., Kim, B.-Y., Park, Y.Y., and Sanidas, E. (2013), ‘Big businesses and economic growth: identifying a binding constraint for growth with country panel analysis’, Journal of Comparative Economics , 41(2), 561–582.
+
+Lin, J.Y. (2011), From Flying Geese to Leading Dragons New Opportunities and Strategies for Structural Transformation in Developing Countries, Policy Research Working Paper 5702, World Bank.
+
+Lin, J.Y. (2012a), ‘From flying geese to leading dragons: new opportunities and strategies for structural transformation in developing countries’, Global Policy, 3(4), 397–409.
+
+Lin, J.Y. (2012b), New Structural Economics. A Framework for Rethinking Development and Policy , Washington, DC: The World Bank.
+
+Lin, J.Y., and Rosenblatt, D. (2012), 'Shifting patterns of economic growth and rethinking development', Journal of Economic Policy Reform, 15(3), 171-194.
+
+Majcen, B., Radosevic, S., and Rojec, M. (2009), 'Nature and determinants of productivity growth of foreign subsidiaries in central and East European countries', Economic Systems, 33, 168-184.
+
+Malerba, F., and Orsenigo, L. (1994), The Dynamics and Evolution of Industries, Working Paper WP-94120, November 1994, The International Institute for Applied Systems Analysis.
+
+Marshall, A. (1890), Principles of Economics (Vol. 1, 9th ed.), ed. C.W. Guillebaud, London: Macmillan/ Royal Economic Society, 1961.
+
+Mowery, D.C., and Oxley, J. (1997), 'Inward technology transfer and competitiveness: the role of national innovation systems', in Technology, Globalisation and Economic Performance, eds. D. Archibugi and J. Michie, Cambridge: Cambridge University Press, pp. 138-171.
+
+Nelson, R.R., and Winter, S.G. (1985), An Evolutionary Theory of Economic Change, Cambridge, MA: Belknap Press.
+
+Ozawa, T. (2009), The Rise of Asia, the ‘Flying Geese’ Theory of Tandem Growth and Regional Agglomeration, Cheltenham: Edward Elgar.
+
+Peneder, M. (2003), 'Industrial structure and aggregate growth', Structural Change and Economic Dynamics, 14(2003), 427-448.
+
+Perez, C. (2010), 'Technological revolutions and techno-economic paradigms', Cambridge Journal of Economics, 34, 185-202.
+
+---
+
+Asian Journal of Technology Innovation
+
+31
+
+Prud'homme, R. (2005), 'Infrastructure and development', in Lessons of Experience (Proceedings of the 2004 Annual Bank Conference on Development Economics), eds. F. Bourguignon and B. Pleskovic, Washington, DC: The World Bank and Oxford University Press, pp. 153-181.
+
+Radelet, S., and Sachs, J. (1997), Asia's Re-emergence, Foreign Affairs, November/December.
+
+Radosevic, S. (1999), International Technology Transfer and Catch-up in Economic Development, Cheltenham: Edward Elgar.
+
+Radosevic, S., and Kaderabkova, A. (eds.) (2011), Challenges for European Innovation Policy: Cohesion and Excellence from a Schumpeterian Perspective, Cheltenham: Edward Elgar.
+
+Radosevic, S., and Yoruk, D.E. (2004), ‘The growth of enterprise through entrepreneurship and network alignment’, in International Industrial Networks and Industrial Restructuring in Central Europe, Russia and Ukraine , eds. S. Radosevic and B. Sadowski, Dordrecht: Kluwer, pp. 109–130.
+
+Radosevic, S., and Yoruk, E. (2014), 'Are there global shifts in world science base? Analysis of catching up and falling behind of world regions', Scientometrics, 101, 1897-1924. doi:10.1007/s11192-014-1344-1.
+
+Radosevic, S., and Yoruk, E. (2015), A New Metrics of Technology Upgrading: The Central And East European Countries in a Comparative Perspective, Working Paper, No. 3.04, Available at http://www. grincoh.eu/ .
+
+Rostow, W.W. (1960), The Stages of Economic Growth: A Non-Communist Manifest (2nd enlarged edn., 1971), Cambridge: Cambridge University Press.
+
+Sandven, T., Smith, K., and Kaloudis, A. (2005), ‘Structural change, growth and innovation: the roles of medium and low tech industries, 1980-2000’, in Low-Tech Innovation in the Knowledge Economy, eds. H. Hirsch-Kreinsen, D. Jacobson, S. Laestadius, and K. Smith, Frankfurt: P. Lane, pp. 31–63.
+
+Seong, J., Kim, J., and Cho, Y. (2014), Korea's Post Catch-up Policy Experiment: Focusing on the Transition towards a Sustainable Socio-technical System, Paper prepared for thematic issue of AJTI.
+
+Spence, M. (2011), The Next Convergence: The Future of Economic Growth in a Multispeed World, New York, NY: FSG Books.
+
+Suarez, F., and Utterback, J. (1995), 'Dominant designs and the survival of firms', Strategic Management Journal, 16, 415-430.
+
+Yoruk, D.E. (2013), ‘Firm-level upgrading in low- and medium-technology industries in emerging markets: the role of learning networks’, Ph.D. thesis, University of Sussex, December, SPRU.
+
+UNCTAD. (2005), World Investment Report, Geneva: UN.
+
+Von Tunzelmann, G.N. (1995), Technology and Industrial Progress: The Foundations of Economic Growth, Aldershot: Edward Elgar.
+
+## Appendix
+
+Table A1: World Bank and our classification of countries based on income per capita, 2014.
+
+<table><tr><td></td><td>World Bank 2014 classification</td><td>Our classification</td><td></td></tr><tr><td>LOW INCOME</td><td>Low income (less than $1045)</td><td>–</td><td></td></tr><tr><td>MIDDLE INCOME</td><td>Lower middle income ($1046-$4125)</td><td>Lower middle income ($1046-$4125)</td><td>`BROAD' MIDDLE INCOME</td></tr><tr><td></td><td>Upper middle income ($4126-$12,745)</td><td>Upper middle income ($4126-$12,745)</td><td></td></tr><tr><td>HIGH INCOME</td><td>High income ($12,746+)</td><td>Lower high income ($12,746-$30,000)</td><td></td></tr><tr><td></td><td></td><td>Upper high income ($30,000+)</td><td>HIGH INCOME</td></tr></table>
+
+
+---
+
+32
+
+S. Radosevic and E. Yoruk
+
+Table A2: Grouping by GNI per capita 2013_atlas method, selected countries.
+
+<table><tr><td>Lower middle income (GNI pc atlas method $1046-$4125)</td><td>Upper middle income (GNI pc atlas method $4126-$12,745)</td><td>Lower high income (GNI pc atlas method $12,176-$30,000)</td><td>Upper high income (GNI pc atlas method $30,001-)</td></tr><tr><td>Ghana</td><td>Albania</td><td>Chile</td><td>Austria</td></tr><tr><td>India</td><td>Argentina</td><td>Czech Republic</td><td>Belgium</td></tr><tr><td>Indonesia</td><td>Belarus</td><td>Estonia</td><td>Germany</td></tr><tr><td>Moldova</td><td>Brazil</td><td>Greece</td><td>Ireland</td></tr><tr><td>Morocco</td><td>Bulgaria</td><td>Korea</td><td>Italy</td></tr><tr><td>Philippines</td><td>China</td><td>Poland</td><td>Japan</td></tr><tr><td>Ukraine</td><td>Hungary</td><td>Portugal</td><td>Norway</td></tr><tr><td>Vietnam</td><td>Jordan</td><td>Russia</td><td>Sweden</td></tr><tr><td></td><td>Kazakhstan</td><td>Slovenia</td><td>UK</td></tr><tr><td></td><td>Malaysia</td><td>Spain</td><td>USA</td></tr><tr><td></td><td>Mexico</td><td></td><td></td></tr><tr><td></td><td>Peru</td><td></td><td></td></tr><tr><td></td><td>Romania</td><td></td><td></td></tr><tr><td></td><td>South Africa</td><td></td><td></td></tr><tr><td></td><td>Thailand</td><td></td><td></td></tr><tr><td></td><td>Turkey</td><td></td><td></td></tr></table>
+
+

processed/00114_W2073380195/document.md

@@ -0,0 +1,469 @@
+Hum Rights Rev (2008) 9:213-232 DOI 10.1007/s12142-007-0042-2
+
+# Human Rights, Intellectual Property, and Struggles for Recognition
+
+Volker Heins
+
+Published online: 30 November 2007 © Springer Science + Business Media B.V. 2007
+
+Abstract This article examines recent controversies over the relationship between human rights and intellectual property rights (IPRs). Many activists have claimed that IPRs conflict with human rights. Others have argued that IPRs are themselves human rights. The article approaches the debate as an opportunity to clarify the nature of IPRs in relation to human rights, as well as the nature of contemporary struggles over these rights. After surveying the dual expansion of both human rights and IPRs and rejecting the view that IPRs are rooted in human rights, the author investigates the example of the HIV/AIDS crisis and the global Campaign for Access to Essential Medicines in order to illustrate attempts to represent IPRs as an outright threat to human rights. Highlighting the limitations of a human rights-based critique of IPRs, he concludes by proposing to study contemporary conflicts over IPRs and human rights as struggles for recognition and as struggles over the institutionalization of a transnational “recognition order.”
+
+## Introduction
+
+The “ age of rights ” $^1$ we live in has not only fostered human rights but also a host of new property rights, in particular rights protecting intellectual property. With the 1994 Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), which is overseen by the World Trade Organization (WTO) and binding for all its members, these rights were significantly expanded in scope and authority. TRIPS guarantees property rights in trademarks, copyrights, industrial designs, geographical indications, plant varieties, and patents – the two most important being copyrights and patents. $^2$ During and after the negotiations leading to this agreement, many scholars and activists have claimed that intellectual property rights (IPRs) conflict with human
+
+$^{1}$ Louis Henkin, The Age of Rights (New York: Columbia University Press, 1990).
+
+2On the context and history of this landmark treaty, see Christopher May and Susan K. Sell, Intellectual Property Rights: A Critical History (Boulder, Colo.: Lynne Rienner, 2006).
+
+V. Heins Department of Political Science, McGill University, 855 Sherbrooke Street West, Montreal, QC H3A 2T7, Canada e-mail: volker.heins@mcgill.ca
+
+![Figure](figures/figure_010.png)
+
+---
+
+214
+
+V. Heins
+
+rights. IPRs on medicines and plant varieties have been criticized for threatening the enjoyment of human rights like the right to health, food, or even self-determination. $^3$ The extended coverage of copyright laws is seen as jeopardizing freedom of speech. $^4$ On the other hand, there is a powerful discourse that claims that IPRs are themselves human rights. This discourse has been promoted not only by business representatives, but also by critics of capitalism who believe that, for example, indigenous peoples are being robbed of their “ intellectual property, ” as their cultures are increasingly exposed to global market forces. $^5$ Some argue that these two sets of rights are fruits from the same tree of Enlightenment ideas, whereas others view IPRs as pests infesting that tree.
+
+There can be no doubt that the entire controversy has compounded the already rampant confusion in the human rights debate, whether by ennobling intellectual property as a human right, or by rhetorically overstating the antagonism between human rights and IPRs. 6 Both views have shaped many of the exchanges between business lobbyists, nongovernmental organizations and others in recent years. This article approaches the debate as an opportunity to clarify the nature of IPRs in relation to human rights, as well as the nature of contemporary struggles over these rights.
+
+My conclusion will be that casting the harm possibly done by the expansion of IPRs in the language of human rights' infringements has not contributed to bolster the consensus on the very meaning of this language. In fact, it has unduly narrowed the range of interpretations that can be employed to make sense of the important conflicts over a number of current intellectual property issues. I will reach this conclusion in four steps. First, I will briefly sketch the dual expansion of both human rights and IPRs. Second, I will criticize the contention that IPRs are rooted in human rights. Third, looking at the example of the HIV/AIDS crisis and the global Campaign for Access to Essential Medicines, I will comment on the alternative contention that attempts to represent IPRs as an outright threat to human rights. Fourth, after having examined potentials and limitations of a human rights-based critique of IPRs, I will propose to study contemporary conflicts over IPRs and
+
+3 See, for example, the Resolutions 2000/7 and 2001/21 of the UN Sub-Commission for the Promotion and Protection of Human Rights; The Impact of the Agreement on Trade-Related Aspects of Intellectual Property Rights on Human Rights , Report of the High Commissioner for Human Rights, UNHCR, Geneva, E/CN.4/ Sub.2/2001/13, June 27, 2001; Human Rights and the Establishment of a WIPO Development Agenda , June 2006, by the Geneva-based group 3D, available at: http://www.3dthree.org/en/ . Among the numerous scholarly contributions to the debate, see Thomas Pogge, “Human Rights and Global Health: A Research Program,” Metaphilosophy 36 (2005), 182–209; Philippe Cullet, “Patents and medicines: the relationship between TRIPS and the human right to health,” International Affairs 79 (2003), 139–60. For more references, see also Jakob Cornides, “Human Rights and Intellectual Property: Conflict or Convergence?,” The Journal of World Intellectual Property 7(2) (2004), 135–67, as well as the internet portals IPRsonline.org , IP-watch.org and CPTech.org .
+
+4 See, for example, Mark A. Lemley and Eugene Volokh, "Freedom of Speech and Injunctions in Intellectual Property Cases," Duke Law Journal 48 (1998), 147-242.
+
+5 See Rosemary J. Coombe, "Intellectual Property, Human Rights, and Sovereignty: New Dilemmas in International Law Posed by the Recognition of Indigenous Knowledge and the Conservation of Biodiversity," Indiana Journal of Global Legal Studies 6 (1998), 59-115.
+
+6 See Gary B. Herbert, "Clarity and Confusion in the Human Rights Debate: An Editorial," Human Rights Review 5(1) (2003), 5-11.
+
+![Figure](figures/figure_009.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+215
+
+human rights from a broader perspective as struggles for recognition and as struggles over the institutionalization of a transnational “recognition order.”
+
+## The Expanding Universe of Rights
+
+Over the last decades, we have witnessed an enormous expansion of human rights norms and assertions. There are at least five indications of this development. First, human rights are no longer only for citizens but also for noncitizens such as undocumented immigrants. Rights have become to some extent, although not yet entirely, independent from membership in territorially exclusive nation-states. 7 Second, human rights have entered a growing range of diverse organizations, from the agendas of international aid agencies to the curricula of military academies. 8 Third, they have ceased to be the domain of liberal elites in the West and have taken root in countless non-Western regions and movements. 9 Fourth, they have begun to cut across the old separation between the law of war and the law of peace, which had limited the applicability of human rights to the latter. This separation has been gradually replaced by legal opinions and treaties containing clear stipulations regarding “nonderogable” human rights obligations that cannot be suspended even in times of war or public emergencies. 10 And fifth, the appeal of human rights norms is such that, even where their observance cannot be directly enforced, they are still conveying powerful messages that can lead to an effective transformation of the situation on the ground, often by legitimizing widespread public anger and protest. 11
+
+Unfortunately, this expansion into ever new fields and dimensions of social life is not accompanied by an increasingly robust consensus on the meaning of human rights. On the contrary, the more the universe of human rights expands, the less these rights seem to work as an agreed-upon frame of reference. This negatively affects the ability to ease conflicts between groups and nations whose common humanity has been invoked in numerous Declarations and Covenants. It has been rightly remarked that, instead of bringing people together, the universal belief in human rights, ironically, “ drives nations and peoples further apart. ” $^12$ This has partly to do with the belief that individual rights of personal empowerment are at loggerheads with collective rights of economic development or social well-being. Many Asian intellectuals and activists, who today often endorse a human rights vocabulary, still
+
+7 See David Jacobson, Rights Across Borders: Immigration and the Decline of Citizenship (Baltimore: Johns Hopkins University Press, 1996).
+
+8 See Volker Heins, "Democratic States, Aid Agencies, and World Society: What's the Name of the Game?," Global Society 9 (2005), 361-84; Russell W. Ramsey and Antonio Raimondo, "Human Rights Instruction at the U.S. Army School of the Americas," Human Rights Review 2(3) (2001), 92-116.
+
+9 See Thomas Risse, Stephen Ropp and Kathryn Sikkink, eds., The Power of Human Rights: International Norms and Domestic Change (Cambridge: Cambridge University Press, 1999).
+
+10 See Tereya Koji, "Emerging Hierarchy in International Human Rights and Beyond: From the Perspective of Non-derogable Rights," European Journal of International Law 12 (2001), 917-941.
+
+11 See Daniel C. Thomas, The Helsinki Effect: International Norms, Human Rights, and the Demise of Communism (Princeton, NJ: Princeton University Press, 2001); Amy L. Wax, "Expressive Law and Oppressive Norms," Virginia Law Review 86 (2000), 1731-79.
+
+$^{12}$ Herbert, fn. 6 above at 9.
+
+![Figure](figures/figure_012.png)
+
+---
+
+216
+
+V. Heins
+
+reject the "state-centredness" as well as the "individualistic ethic" of Western approaches to human rights. $^13$
+
+The situation is likely to become more complicated, as different categories of human rights will no longer to be pitted against each other along the traditional fault lines of East versus West or North versus South – fault lines that might soon be rendered obsolete by the “mega-trend” of globalization. 14 This was made clear when a worldwide controversy erupted over a Danish newspaper publishing cartoons of the prophet Muhammad in October 2005. The event revealed, among other things, a sense of uncertainty within Western societies about competing notions of respect as well as unprecedented global – local tensions. Note that, for the first time, UN human rights officials rebuked the government of the small Scandinavian democracy for its “intransigent defense of unlimited freedom of expression” at the expense of “religious freedom” and international “religious harmony.” 15
+
+Another recent trend is the trivialization of human rights in the affluent West. A growing number of cases judged by human rights watchdogs no longer bear the faintest resemblance to the conscience-shaking “ barbarous acts ” mentioned in the Preamble of the Universal Declaration of Human Rights. Take the example of the hockey mom in British Columbia, Canada, who recently filed a complaint with the province's Human Rights Tribunal because the local hockey association did not allow her 14-year-old daughter to change in the same locker room as the boys on her co-ed sports team. Bizarre as it sounds, the tribunal dealt with this case in all seriousness and decreed that separate change rooms are indeed unjust and in contradiction with human rights. $^16$ This and similar instances – like the elevation of the notorious Swiss bank secret to a human right $^17$ – show that apart from conflicts between competing human rights, we are facing a worrying disconnect between people who suffer real rights abuses and other people who are completely out of touch with the realities of oppression against which the idea of human rights was originally brought into play. From this perspective, the rhetorical overuse of human rights is only the flip side of a yawning vacuum at the heart of that rhetoric.
+
+This entire cauldron of conflict has been further stirred by the addition of property rights that have undergone an expansion in scope and authority quite similar to the expansion of human rights. The expansion of property rights cuts across the distinctions between mind and matter, nature and culture, North and South, and therefore raises a host of philosophical as well as political issues. Three aspects are worth highlighting.
+
+13 Rajni Kothari, "Human Rights: A Movement in Search of a Theory," in Smitu Kothari and Harsh Sethi, eds., Rethinking Human Rights: Challenges for Theory and Action (Delhi: Lokayan, 1991), 19-30.
+
+14 See National Intelligence Council (NIC), Mapping the Global Future: Report of the National Intelligence Council's 2020 Project (Washington DC: Government Printing Office, 2004), 10.
+
+15 Commission on Human Rights, "Situation of Muslims and Arab peoples in various parts of the world" (Special Rapporteur Doudou Diène), UN Economic and Social Council, E/CN.4/2006/17; February 13, 2006, 10-11.
+
+16 See Naomi Lakritz, "Girls-in-locker-room ruling makes a mockery of rights law," The Gazette, Montreal, September 23, 2005.
+
+17 See Georg F. Krayer, "Privacy: a Human Right at Risk," Paper given at a media conference, September 19, 2002. Available at: http://www.swissbanking.org/krayer_e.pdf (last accessed June 1, 2006).
+
+![Figure](figures/figure_011.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+217
+
+First, property rights made bold new strides into the vast area of intellectual activity. IPRs make sure that so-called non-rivalrous, copyable goods (like computer software, processes of manufacture, or genetic material) that can be used and possessed by an infinite number of individuals without losing their value, are nonetheless considered “ scarce ” so that they can be bought and sold like other commodities. Although patents and copyrights have been with us for centuries, more recent trends have led to the reframing of an increasing number of issues of authorship, originality, use, and access to ideas and expressions in terms of IPRs. Leaving older notions of copyright behind, the World Intellectual Property Organization, a specialized UN agency headquartered in Geneva, Switzerland, is now even pondering a special treaty that would protect new “ broadcaster ” or “ webcaster ” property rights in information distributed over radio, cable television, or through any wired or wireless computer network, including the Internet. All these trends have been spurred by the perception of a technological revolution in information and communication technologies and the concomitant perception of huge profits to be lost to “ pirates, ” or of equally huge rents to be gained through better intellectual property protection.
+
+Second, patents today can be granted for things that have not been “invented” in a strict sense, but are already there, like components of life processes that are considered new, useful and nonobvious (as soon as they have been isolated or tinkered with a little). Thus, patent-like protection has been accorded to sexually reproducing plants, laboratory mice, an oyster with an extra set of chromosomes, and numerous compounds of living organisms that have been purified away from their sources. Various courts have confirmed that, with advancements in technology, plants and living organisms can be patented. The whole of the animal kingdom is now targeted, and the system is slowly closing in on the human body.
+
+Third, with the establishment of a global regime for the enforcement of IPRs, a high and uniform level of intellectual property protection has expanded from developed to developing countries. Today, all WTO members have to provide patent protection for any invention, whether a product (such as a medicine) or a process (such as a method of producing the ingredients for a medicine), although certain exceptions are allowed.
+
+Similar to debates in the human rights field, intellectuals from non-Western backgrounds have contested the alleged cultural “ chauvinism ” of the emerging IPR regime and the pretense that the rights protected by this regime are universally applicable and beneficial. 18 Furthermore, like in the case of universal human rights, there is evidence that the globalization of property rights drives nations and peoples apart.
+
+## Intellectual Property Rights as Human Rights?
+
+The business lobbyists, who since the mid-1980s pushed for the establishment of a new global intellectual property regime, have framed the issue in a way that
+
+18 See, for example, Akalemwa Ngenda, "The Nature of the International Intellectual Property System: Universal Norms and Values or Western Chauvinism?," Information & Communications Technology Law 14(1) (2005), 59-79.
+
+![Figure](figures/figure_009.png)
+
+---
+
+218
+
+V. Heins
+
+resonated with the broader normative context of a liberal culture deeply imbued with “ rights talk. ” 19 IPRs were constructed as a subset of ordinary property rights that in turn were presented as a species of human rights. From early on, spokespersons of companies and business associations accused developing countries of encouraging or tolerating the imitation of pharmaceuticals or the infringement of copyrights and called for a treaty able to “ stop their pirates from boarding our ships. ” 20 Others denounced organized “ criminal gangs ” for stealing “ America's crown jewels. ” 21 By labeling infringers of intellectual property rules as “ pirates, ” business activists transformed the mundane economic issue of IPRs into a symbol redolent with powerful moral meanings. Since ancient times, the pirate was declared an enemy of mankind itself – hostis humani generis – and even relatively, recent court rulings in the USA placed the pirate under the same rubric as the “ torturer ” and the “ slave trader. ” 22 The prehistory of the TRIPS accord was strongly influenced by such kind of rhetoric that ultimately suggested that attacking intellectual properties is like attacking the property owners themselves. This rhetoric was further accentuated by the concomitant idea to apply “ zero tolerance ” to international trade policies – a notion that originated from the federal drug policy of the 1980s and then caught on throughout society.
+
+This entire endeavor was certainly successful to a degree. The appropriation of intellectual property by anybody who is not officially recognized as its owner was stigmatized as piracy, and pirates were symbolically expurgated as “the Other,” outside of the new international order. 23 Representing IPRs as akin to human rights helped mobilize power resources in favor of a new uniform international property regime that could coerce developing countries into compliance. However, the business discourse failed to remove the manifold meanings of legitimate property from public contestation. In this sense, things have not changed much since the nineteenth century in Europe, when people already fought bitterly over the question whether the gathering of fallen wood by poor people constituted “theft,” “pilfering,” or, as a would-be famous radical maintained, just the exercise of a “customary right.” 24
+
+Unsurprisingly, opponents of the globalization of high levels of intellectual property protection quickly joined the global rights talk. Instead of relying only on utilitarian arguments, the critique of TRIPS was couched in terms of fundamental rights. A high-level expert from India, for example, asked “whether we as human beings do not have a natural right to imitation , the right to imitate each other, to learn from each other and to elaborate and develop each other’s ideas and activities.”
+
+19 See Susan K. Sell, Private Power, Public Law: The Globalization of Intellectual Property Rights (Cambridge: Cambridge University Press, 2003), 50-1.
+
+20 C. L. Clemente, "A Pharmaceutical Industry Perspective," in C. Walker and M. Bloomfield, eds., Intellectual Property Rights and Capital Formation in the Next Decade (Lanham, MD: University Press of America, 1988), 133.
+
+$^{21}$ Jack Valenti of the Motion Picture Association of America, quoted in Cornides, fn. 3 above at 136n.
+
+22 Filartiga v. Pena-Irala, 630 F.2d 876 (2d Circuit, June 30, 1980); May and Sell, fn. 2 above at 164.
+
+23 This is what "ideologies" do. See John B. Thompson, Ideology and Modern Culture: Critical Social Theory in the Era of Mass Communication (Stanford, CA: Stanford University Press, 1990), 65.
+
+24 Karl Marx, "Debates on the law on theft of wood [1842]," in Marx-Engels Collected Works (New York: International Publishers, 1975), Vol. 1. Available at: http://www.marxists.org/archive/marx/works/1842/ 10/25.htm.
+
+![Figure](figures/figure_011.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+219
+
+She added: “ It is this activity that distinguished us from our hominid forebears. ” 25 There is nothing intrinsically anti-Western about this position, which is not too different from Thomas Jefferson's noted skepticism toward patents, and close to current A2K (Access to Knowledge) activists who sometimes romanticize the return of plebeian, anti-property traditions of “ social banditry. ” 26 Some of these arguments reject rigid IPR systems as just another instance of the exclusivist institution of private property. Others challenge the construction of IPRs as a subset of ordinary property rights and underscore the fundamental differences between these two sorts of property.
+
+Private property is controversial because it is synonymous with the power to exclude people from resources. Historically, an important side effect of the introduction of the right to private property has always been the effective dispossession of large groups of people who lost the few things they used to own and control. 27 For this and other reasons, and contrary to what many people believe, the “ right to property ” was included in the Universal Declaration, but not the right to private property. 28 Still, ordinary private property rights seem to be less controversial than IPRs. One reason may be that, again contrary to deeply held Western including Marxist beliefs, private property rights were always known beyond occidental capitalism. Non-Western property systems were (and still are) different, but the thesis of the nonexistence of private property in Asian social history has long been discredited. 29
+
+On the other hand, property rights in physical instantiations of the human mind are inherently controversial for at least two reasons. The main reason is that the scarcity of intellectual goods is imposed on them by law, unlike those things which are scarce in relation to given demands and desires. Like other private property rights, IPRs constitute “a relationship both to and through objects of social wealth.” 30 However, intellectual property is a relationship to objects whose very “thingness” has first to be created as a legal artifice before the relationship between rightholders and dutyholders can be filtered through these peculiar things. This particular origin of IPRs increases the burden of legitimation that falls on them. A second reason is that, from a historical perspective, the scope of private property has shrunk over the last 200 years, whereas the scope of IPRs seems to be ever-expanding : at least in Western societies, you cannot legally own public offices or workers, but you may own parts of
+
+25 Usha Menon, “Designing a regime of access to genetic resources: beyond the popular logic of Farmers’ Rights and Breeders’ Rights,“ in Ethics and Equity in Conservation and Use of Genetic Resources for Sustainable Food Security. Proceedings of a Workshop to Develop Guidelines for the CGIAR (Rome: International Plant Genetic Resources Institute, 1997), 101.
+
+26 Chris Rojek, "P2P Leisure Exchange: Net Banditry and the Policing of Intellectual Property," Leisure Studies 24 (2005), 357-69.
+
+27 See Stefan Andreasson, "Stand and Deliver: Private Property and the Politics of Global Dispossession," Political Studies 54 (2006), 3-22.
+
+28 See Johannes Morsink, The Universal Declaration of Human Rights: Origins, Drafting and Intent (Philadelphia: University of Pennsylvania Press, 1998), Chap. 4.
+
+29 See, for example, Dharma Kumar, "Private Property in Asia? The Case of Medieval South Asia," in Colonialism, Property and the State (Delhi: Oxford University Press, 1998), 135-70.
+
+30 David Lametti, "The concept of property: relations through objects of social wealth," University of Toronto Law Journal 53 (2003), 329.
+
+![Figure](figures/figure_011.png)
+
+---
+
+220
+
+V. Heins
+
+the genetic map of a living organism, stock phrases and songlines or the compound that gives lemons their particular taste. Thus, although there is no clash of civilizations about the acceptance of private property as such, conflicts over the relationship between “real” property rights, IPRs, and other forms of rights are clearly on the increase, both in international society and within countries and cultures.
+
+To elucidate differences between forms of proprietary and nonproprietary rights, I now draw on an analytical framework recently developed by Leif Wenar, who in turn leans heavily on the work of Wesley Hohfeld, an influential legal theorist in the early twentieth century. 31 Wenar first distinguishes privileges (exemptions) and claims as fundamental forms of rights-assertion. Privileges have the form “A has a right to phi,” where “phi” is an active verb. An example is the right to free speech, which is the right not to follow the majority’s ideas and opinions. Privilege rights have to be exercised to be effective. 32 Claims, by contrast, have the form “A has a right that B phi.” They are enjoyed rather than exercised. In addition to claims and privileges, Wenar identifies powers and immunities . These are second-order “rights to alter our privileges and claims, and rights that our privileges and claims not be altered.” 33 Most real rights are typically made up of several of these elementary assertions, although it is usually possible to isolate a dominant element in each right.
+
+Thus, both real property rights and IPRs are, among other elements, made up of claims that entitle their holders to protection against harm or to a specific behavior on the part of dutyholders. Claim rights imply that somebody other than the rightholder has a duty to do something (to pay for access, to refrain from interference, etc.). As a consequence, property rights constrain the ways in which titleholders can be harmed. Competing them out of business is legitimate; simply taking their property is not. If the right to property were enshrined in constitutional law, property owners would also enjoy far-reaching immunity against expropriation or any other changes of their normative situation. Immunity means that others do not have any second-order right to change the rules that assign first-order rights to physical and juridical persons. Although American constitutional law, like most of its European counterparts, does not explicitly recognize private property as a fundamental right, property is today considered “ a far more robust constitutional value than it was just twenty years ago. ” 34 It is thus fair to say that property rights confer at least a strong measure of immunity against demands and intrusions from non-rightholders and the general public. For some holders of property rights, in particular large corporations, these rights coincide with the (limited) power to create new rights of themselves or others, or to annul old rights. Again, this power right is not simply enjoyed, but actively exercised as the history of the TRIPS agreement illustrates, which is marked by the extraordinary mobilization of private companies trying to spin new rights out of their already existing ones. 35
+
+$^{31}$ See Leif Wenar, "The Nature of Rights," Philosophy & Public Affairs 33 (2005), 223-52.
+
+$^{32}$ Ibid, 233.
+
+$^{33}$ Ibid, 230; emphasis added.
+
+34 Gregory S. Alexander, "Property as a Fundamental Constitutional Right? The German Example," Cornell Law School Working Paper Series, No. 4 (2003), 56.
+
+35 See Sell, fn. 19 above, chap. 5. See also Rodney Bruce Hall and Thomas J. Biersteker, eds., The Emergence of Private Authority in Global Governance (Cambridge: Cambridge University Press, 2002).
+
+![Figure](figures/figure_010.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+221
+
+In themselves, IPRs are neither powers nor immunities, although they are increasingly well-entrenched and often treated on par with other property rights. In some cases, the rights of a patent holder were even seen as compromising the rights of owners of physical property. 36 Yet, the global IPR regime is still far from according an immunity right to intellectual property claimants. First, it is worth remembering that, unlike other property titles, patents (and copyrights) expire after some time. They entitle their holders only to temporary monopolies. All private property is, and has always been, conditioned by at least some rules limiting its use. There is no such thing as a nonderogable property right. 37 This is evidently even more so the case with regard to intellectual property. IPRs are premised on the idea that they ultimately do not restrict, but promote the dissemination of knowledge and wealth in society. Thus, IPRs contain a privilege granted by law, an “ exemption from a general duty ” 38 or normative expectation. Society expects the public disclosure of proprietary information and wants knowledge to flow freely; yet patent holders are temporarily shielded from this general expectation, because it is assumed that the granting of private rights in intellectual property is an incentive to innovate and to broaden the knowledge base of society.
+
+Once granted, IPRs also function as claims against others who have to pay fees or royalties and who are not allowed to copy protected materials. These claims, however, are in turn subjected to a different set of privileges on the part of those excluded from intellectual property. These privileges or property-limiting principles are internal as well as external to the IPR system. The fact that IPRs are granted on the basis of an agreed-upon social purpose to be served by these rights is reflected in Article 27(2) of the TRIPS agreement, which allows the exclusion of a subject matter from patenting in order “to protect ordre public or morality, including to protect human, animal or plant life or health or to avoid serious prejudice to the environment.” More importantly, Articles 30 and 31 permit the compulsory licensing of patents (their use without the patent owners consent) in cases of national emergencies, but also in a range of other situations. Developing countries have not made much use of these perfectly legal exemptions and flexibilities, which have recently been confirmed, largely because of real or anticipated extra-institutional bilateral pressures from European and US governments or powerful businesses. 39
+
+Two external property-limiting principles are relevant to our discussion: Farmers' Rights and Traditional Resource Rights . The idea of Farmers' Rights was developed in the late 1980s within the UN Food and Agricultural Organization in response to strengthened legal protection of plant varieties and breeders' rights, sometimes
+
+36 A landmark ruling by the Federal Court of Canada in 2002 established that a farmers' ownership of a canola plant does not supersede the claim-rights of the holder of a patent for a stray gene found in that plant (Monsanto Canada Inc. v. Schmeiser, 2002 FCA 309 at para. 51).
+
+37 See Lametti, fn. 30 above at 369-70. On the essential "nonabsoluteness" of private property rights, see also Alan Gewirth, The Community of Rights (Chicago: University of Chicago Press, 1996), 170-98.
+
+$^{38}$Wenar, fn. 31 above at 226.
+
+39 Kenneth C. Shadlen, "Patents and Pills, Power and Procedure: The North-South Politics of Public Health in the WTO," Studies in Comparative International Development 39(3) (2004), 76-108.
+
+![Figure](figures/figure_009.png)
+
+---
+
+222
+
+V. Heins
+
+explicitly with reference to the "right to food." The concept also entered the 2001 International Treaty on Plant Genetic Resources for Food and Agriculture. 40
+
+Farmers' rights is a socio-economic right invented to entitle farmers – particularly small farmers in developing countries – to measures in support of their, so far, largely underrated contributions to the conservation and cultivation of agricultural plant varieties. As the work of these farmers is continuously needed for breeding high-yielding food crops, farmers' rights provisions, incorporated in a number of current plant variety protection laws, seek to redress the imbalance in the reward structure that favors the finished products of scientific plant breeding that are increasingly subject to IPRs.
+
+Yet there is no consensual understanding of the content of these rights. Surveys among experts have found that farmers' rights are often conceived as privilege rights that allow farmers to save, exchange, or replant branded seeds of a protected variety. In addition, farmers' rights are seen as claim rights entitling farmers to improved participation in decision-making bodies as well as to some kind of compensation for their undervalued contribution to the production of new plants. 41 Unfortunately, given the complex pedigree of plants and the difficulties in accurately identifying the contributing farming communities, authorities in developing countries are faced with enormous, perhaps insurmountable problems in adjudicating on redistributive claims, provided that funds for redistribution can be mobilized at all. 42 As a consequence, farmers' rights are still largely aspirational and far from being firmly institutionalized.
+
+The concept of Traditional Resource Rights was developed by Western activists in collaboration with indigenous groups, again mostly in response to imbalances in the modern patent and plant variety protection systems that recognize certain forms of creative activity, but not others. 43 Resource rights, which are also still largely aspirational, have moved to center stage recently. This is partly because of cases that proved the usefulness of indigenous knowledge in identifying compounds in plants that were later used to develop potentially lucrative pharmaceutical drugs. 44 Like farmers' rights, traditional resource rights entail both privileges that entitle groups to use their own biological resources as they see fit without interference by the state or other actors ( “ self-determination ” ) and claims directed at these same actors who are expected to compensate indigenous groups for the misappropriation of resources.
+
+40 See Regine Andersen, The History of Farmers' Rights: A Guide to Central Documents and Literature. The Farmers' Rights Project Background Study No. 1 (Lysaker, Norway: Fridtjof Nansen Institute, 2005).
+
+41 See Regine Andersen, Results from an International Stakeholder Survey on Farmers' Rights. The Farmers' Rights Project Background Study No. 2 (Lysaker, Norway: Fridtjof Nansen Institute, 2005), 6-15.
+
+42 See C. S. Srinivasan, "Exploring the Feasibility of Farmers' Rights," Development Policy Review 21(4) (2003, 429-34.
+
+43 See Graham Dutfield, “Indigenous Peoples and Traditional Resource Rights,” in Sheldon Krinsky and Peter Shorett, eds., Rights and Liberties in the Biotech Age: Why We Need a Genetic Bill of Rights (Lanham, MD: Rowan & Littlefield, 2005), 107–13; Rosemary J. Coombe, “Protecting Traditional Environmental Knowledge and New Social Movements in the Americas: Intellectual Property, Human Right or Claims to an Alternative Form of Sustainable Development?,” Florida Journal of International Law 17(1) (2005), 115–35.
+
+$^{44}$ See, for example, Dutfield, ibid at 111; Coombe, fn. 5 above at $82-9$.
+
+![Figure](figures/figure_011.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+223
+
+## Intellectual Property Rights as a Threat to Human Rights?
+
+### Contrasting Rights
+
+In my view, human rights can be dissected so that they fit into the Hohfeld – Wenar scheme of rights assertions, although an additional dimension is needed to fully understand them. The first proposition of a primitive theory of human rights reads that certain things ought not to be done to a human being. Not only ought everybody be protected against persecution, torture, or censorship, but everybody should also enjoy a second-order right that her normative situation not be altered. In this sense, the idea of human rights is first and foremost about immunity rights for every single human being. Private property participated in the exalted status only until the seventeenth century when the term “ property ” still included our life, body, and spouses and was not yet narrowed down to “ things. ” $^45$ Human rights can be conceived as a power right able to generate new rights or new interpretations of old rights, as the foregoing discussion has demonstrated. In addition, human rights assertions have the form of claim rights against states and other entities. Finally, human rights are also about privilege rights . Where human rights are respected, individuals are free to opt out of the dominant religion or to act against prevailing opinions. Massive violations of human rights, on the other hand, have given rise to a different kind of privilege; they have conferred legitimacy to states that attack other states by military means to stop the atrocities. This is a privilege so long as the intervening state does not flatly deny the general duty to respect state sovereignty.
+
+Thus, human rights cannot be reduced to single privileges or claims, powers or immunities. Rather, they encompass and transcend all these incidences in a way that led Hannah Arendt to characterize them as the “ right to have rights. ” $^46$ Human rights protect human agency itself, and hence the dignity of persons. Oscillating between moral and legal standing, they are defined by the struggle for translating them from merely aspirational rights into justiciable welfare and liberty rights. These latter rights are enjoyed or exercised. The point I want to stress, however, is that no legal articulation and no court ruling taking its cues from the Declarations and Covenants ever fully exhausts the scope of human rights. It is this unredeemed normative surplus that is invoked whenever individuals stand up and claim their right; at the same time, this very act of standing up underwrites their human dignity. In asserting claims, individuals do not address official dutyholders but rather the world – or the moral public – upon which they obtrude their claim to have a right. $^47$ That is what, for example, Nelson Mandela did when he turned to the world outside the court that tried and convicted him in 1962, denouncing the “ lack of human dignity ” under the apartheid regime, and calling for “ equal political rights ” and “ a living wage. ” $^48$
+
+45 See C.B. Macpherson, "Human Rights as Property Rights," in The Rise and Fall of Economic Justice and Other Papers (Oxford: Oxford University Press, 1985), Chap. 6.
+
+$^{46}$ Hannah Arendt, The Origins of Totalitarianism, 2nd ed. (Cleveland: Meridian, 1958), 296-7.
+
+47 For these formulations, see Joel Feinberg, "The Nature and Value of Rights," in Rights, Justice, and the Bonds of Liberty: Essays in Social Philosophy (Princeton, NJ: Princeton University Press, 1980), 150-1.
+
+48 See James Boyd White, "Mandela's Speech from the Dock and Lincoln's Second Inaugural Address: Giving Meaning to Life in an Unjust World," in Acts of Hope: Creating Authority in Literature, Law, and Politics (Chicago: Chicago University Press, 1994), 290-1.
+
+![Figure](figures/figure_010.png)
+
+---
+
+224
+
+V. Heins
+
+Table 1 A comparison of rights
+
+<table><tr><td></td><td>IPRs</td><td>Property rights</td><td>Traditional resource rights/farmers' rights</td><td>Human rights</td></tr><tr><td>Form of right</td><td>Privileges claims ( real )</td><td>Claims Immunities Powers ( real )</td><td>Privileges claims ( largely aspirational )</td><td>All forms of right ( real and aspirational )</td></tr><tr><td>Subject matter</td><td>Inventions, expressions, insignia, some discoveries</td><td>Created and discovered things</td><td>Underappreciated contributions to the conservation and cultivation of plant varieties, etc.</td><td>Human agency dignity</td></tr><tr><td>Rightholders</td><td>Corporations individuals</td><td>Individuals Corporations</td><td>Groups</td><td>Individuals groups</td></tr><tr><td>Dutyholders</td><td>All non-rightholders</td><td>All non-rightholders</td><td>States corporations</td><td>States corporations individuals</td></tr><tr><td>Duration</td><td>Limited</td><td>Unlimited</td><td>Unlimited</td><td>Unlimited</td></tr><tr><td>Mode of assertion</td><td>Enjoyed exercised</td><td>Enjoyed exercised</td><td>nvoked</td><td>Exercised enjoyed invoked</td></tr></table>
+
+
+At this point, I wish to come back to a shortcoming of the typology of elementary rights assertions introduced by Hohfeld and modified by Wenar. This typology has been criticized for being too simple, as it assumes that property relations take place only between a rightholder and a corresponding dutyholder. In reality, however, a “ right-holder sees her right as good against the world ; it is not a simple bilateral relation. ” 49 Human rights introduce a similar element of asymmetry into legal oneon-one relations by being good not “ against, ” but “ for ” the world. Before they are enjoyed or exercised as mundane constitutional rights of citizens, human rights are invoked by speakers who address and appeal to the public. These speakers make aspirational assertions delivered from public platforms with the aim of activating moral commitments and solidary ties that often do not preexist but are created by discourse. Human rights assertions require listeners whose attention can today be captured worldwide.
+
+Property rights, by contrast, are good “ against the world ” and have therefore always drawn fire from the dispossessed. IPRs are even more likely to fuel political conflicts because their subject matter includes intangibles like instantiations of the human mind or gene sequences. The emerging global IPR system appears to exclude, quite literally, humankind itself from access to important knowledge assets; moreover, it appears to exclude humankind from what might be considered a part of its essence (Table 1 ).
+
+AIDS and the Human Rights-Based Critique of IPRs
+
+Over the past 10 years, this seemingly abstract idea about the potential antagonism between IPRs and human rights has been worked out and turned into a tool for
+
+$^{49}$ Lametti, fn. 30 above at 343; emphasis added.
+
+![Figure](figures/figure_009.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+225
+
+mobilization. Interestingly, this process was spurred by one particular crisis which had repercussions not foreseen by the drafters of the TRIPS agreement: the global AIDS epidemic. In 2005 alone, HIV/AIDS has killed more than three million people, most of them in the developing world, particularly in southern Africa. Yet the protest against possible negative consequences of the TRIPS agreement started first as a local concern in the West. In the mid-1990s, consumer activists in the USA began to protest the TRIPS agreement which they saw as leading to further price hikes for medicines, thereby making health care even less affordable for poor citizens. This initial campaign gained enormous momentum and became global when, shortly afterward, Third World groups and aid agencies began to draw a connection between the granting of product patents for pharmaceuticals, the ban of parallel imports of cheaper generic substitutes for patented AIDS drugs, and the avoidable death of people living with the virus. Overall, the Campaign for Access to Essential Medicines, as it soon became known, was remarkably successful in drawing attention to the human rights implications of stronger patent regimes for the treatment of HIV/AIDS, in naming and shaming powerful pharmaceutical lobbies, and in encouraging poor countries to seek amendments of the TRIPS agreement. 50 The details of this story have often been told, so I confine myself to the following question: How have human rights arguments been used against the IPR regime and pharmaceutical companies, and does this use make sense?
+
+The highly intuitive assumption that AIDS patients have a human right to whatever drug might help them (regardless of whether a corresponding dutyholder can be identified) has been expressed as part of a range of different moral narratives. Three of these narratives can be simplified as follows:
+
+First, it has been maintained that the right to the best drugs can be inferred from the way in which “the West” was allegedly always implicated in the global spread of HIV or may have even fabricated the virus to wipe out the African population. This claim, which I just want to mention here without discussing it any further, was put forward, in particular, by the Kenyan ecologist Wangari Maathai who won the Nobel Peace Prize in 2004. According to this view, the West has deliberately pushed the non-swimming (African) child into the deep end.
+
+Second, at the other end of the spectrum, the relationship between Western companies or governments and non-Western AIDS victims is modeled after the biblical tale of the good Samaritan – or the “bad Samaritans” who passed by the roadside victim without coming to his aid. 51 This is certainly the majority view held by the board members of the Global Fund to Fight Aids, Tuberculosis and Malaria, founded in 2001. They accept scientific evidence suggesting that the virus causing AIDS originated from wild chimps in Cameroon and spread in ways uncontrolled by Western interests, including those protecting intellectual property. Some would also agree that human rights considerations provide an important rationale for massive
+
+50 See Sell, fn. 19 above at 147–62; Shadlen, fn. 39 above; Steven Robins, "‘Long live Zackie, long live’: AIDS activism, science and citizenship after apartheid," Journal of Southern African Studies 30(3) (2004), 651–72. See also the homepage of the Access Campaign: http://www.accessmed-msf.org/
+
+51 See Joel Feinberg, Harm to Others (New York: Oxford University Press, 1984), 126.
+
+![Figure](figures/figure_008.png)
+
+---
+
+226
+
+V. Heins
+
+investments in treatment and prevention. 52 The rich countries are seen as being in the position of a good swimmer who can do a lot to rescue the drowning child who fell into the swimming pool.
+
+Third, activists from the Access Campaign and their intellectual supporters, like Thomas Pogge of Columbia University, have chosen a middle path between these two alternatives by linking crucial aspects of the AIDS epidemic to the new IPR regime. The West is accused of not just failing to benefit the drowning child, but of actively harming her, although not being responsible for the bad situation, she was already in. Instead of pulling the child out of the water, the good swimmer has (perhaps inadvertently) switched on the 100-hp pump for artificial wave generation, making it more difficult for the child to reach the rim of the pool.
+
+It is worth noting that even the second narrative, which is the least radical of the three, has strong human rights implications, if we take into account the growing consensus among legal philosophers and lawyers who have considerably narrowed the difference between “ harming ” and “ withholding benefit, ” “ acting ” and “ failures to act. ” 53 In Britain today, negligence claims against public authorities can be framed as breaches of human rights law. 54 Even if not strictly applicable to international relations, the duty of care is now taken much more seriously than in earlier times, when strangers had no right to be rescued.
+
+The more provocative hypothesis, however, is advanced by the third narrative that suggests that the new IPR rules are actively violating the physical integrity rights of HIV-positive persons in poor countries. The plausibility of this charge depends on whether one can show that people living with HIV are worse-off than they would have been without the consequences flowing from those rules. In this case, a wrongful harm done to strangers would be directly attributable to IPRs and the political forces backing the global IPR regime. In my view, however, the empirical basis required for reaching such a conclusion is incomplete. Consider the following undisputed points. Treatment of people suffering from HIV/AIDS is possible thanks to drugs, in particular, antiretrovirals. Product patents tend to increase the price of antiretrovirals, which results in fewer people being able to afford them. Some countries like India produce generic equivalents of patented antiretrovirals, which are much cheaper, yet TRIPS has made the “parallel import” of generic versions of patented drugs illegal or very cumbersome. In late 2005, however, member states of the WTO, including the USA, decided to make permanent a waiver enabling poor countries, in particular those with inadequate production facilities, to obtain such generics by setting aside the original TRIPS provision. Among other things, competition from producers of generics is likely to force brandname firms to lower their prices, which allows for more people to be treated. In fact, prices for antiretrovirals in poor countries have fallen , the number
+
+52 See Nicoli Nattrass and Nathan Geffen, "The impact of reduced drug prices on the cost-effectiveness of HAART in South Africa," African Journal of AIDS Research 4(1) (2005), 65-7.
+
+53 Feinberg, above fn. 51, chap. 4; Cherie Booth and Dan Squires, The Negligence Liability of Public Authorities (Oxford: Oxford University Press, 2006), 316.
+
+$^{54}$ Booth and Squires, ibid. at 28.
+
+![Figure](figures/figure_009.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+227
+
+of people on antiretroviral therapy has massively increased, and overall access to AIDS drugs is expanding. $^55$
+
+Besides these things we know, there are other questions we simply cannot answer. Here are a few examples. We do not know whether, in the longer run, manufacturers of generic antiretrovirals in India, which satisfy around half of the world's demand, are going to face incentives to give up this particular business activity, partly because the waiver issued by the WTO may be difficult to use in practice, partly because other activities might yield higher margins. 56 We also do not know how a recent US Supreme Court decision to exempt preclinical drug research from patent infringement liability will affect AIDS vaccine development. 57 Furthermore, at the most fundamental level, we do not know to what extent pharmaceutical companies need to recoup their R & D expenditures through TRIPS-style patent monopolies to develop new drugs. 58 If such a link could be proven, attending to those who are worst-off today by suspending intellectual property protections might have the effect of reducing the available resources of those who will suffer tomorrow .
+
+Given what we know for sure and what we do not know, we need to ask whether Thomas Pogge is right to claim that “ the rich countries' IPR initiative goes in the wrong direction, foreseeably causing many additional premature deaths among the global poor by cutting them off from life-saving patented medicines." 59 Although it is true that the recent bolstering of intellectual property claims has tended to make access to many medicines more expensive, various intervening variables seem to have offset the much-feared consequence of making infected people in poor countries worse-off than they were before the new intellectual property arrangements. One of these intervening variables is an enormous increase in international funding for AIDS treatment and preventive HIV vaccine research. Between 1996 and 2005, funds for fighting AIDS in low- and middle-income countries increased 28-fold, from US $ 300 million to US $ 8.3 billion. $^60$ The world's most generous program that buys AIDS drugs for patients in developing countries is the President's Emergency Plan for AIDS Relief (PEPFAR), financed by the US government. Another variable is, of course, the widespread col è re publique triggered by the
+
+55 See Joint United Nations Programme on HIV/AIDS (UNAIDS), 2006 Report on the global AIDS epidemic, 151-8.
+
+56 See Ken Shadlen, "Patents, India, and HIV/AIDS Treatment," LSE AIDS Update 4, April 2005. Available at: http://www.lse.ac.uk/collections/LSEAIDS/ (last accessed June 1, 2006).
+
+57 See Merck KGaA v. Integra Lifesciences I, Ltd. (03-1237), June 13, 2005; AIDS Vaccine Advocacy Coalition (AVAC), 2005 AVAC Report: AIDS Vaccines at the Crossroads , 34–5. For the argument that, perversely, going soft on patent infringements might make drug development even more costly, see Beverly W. Lubit, “The Economic Impact of the Supreme Court Decision in Merck v. Integra,” BioPharm International , January 1, 2006 (online edition).
+
+58 A high level of patent protection seems to stimulate innovation in some industries, but not in others. See William Fisher, "Theories of Intellectual Property," in Stephen R. Munzer, ed., New Essays in the Legal and Political Theory of Property (Cambridge: Cambridge University Press, 2001), 180-1.
+
+59 Thomas Pogge, "Montreal Statement on the Human Right to Essential Medicines," Equality and the New Global Order Conference, Harvard University, May 11–13, 2006, 9. Pogge claims that an alternative institutional arrangement of intellectual property protection is feasible, and that relative to such a possible alternative regime – not to other historical or subjunctive-historical baselines – the existing IPR regime violates basic human rights (personal communication, June 25, 2006).
+
+$^{60}$ See UNAIDS, fn. 55 above at 224.
+
+![Figure](figures/figure_011.png)
+
+---
+
+228
+
+V. Heins
+
+TRIPS agreement itself. This led, among other things, to the most recent amendments introducing more emergency-related exemptions for poor countries. Arguably, the narrative that has linked the emergence of a uniform global IPR regime to a scenario of exclusion of the most needy persons from life-saving drugs functioned as a self-defeating prophecy by causing many people to take precautions to avert that scenario.
+
+Thus, given the multiplicity of causal factors at work, it seems unwarranted to single out IPRs as the main or proximate cause for the current suffering of people in poor countries living with HIV or other fatal diseases. In the presence of countervailing forces, some of which were actually generated by the new intellectual property rules, there is certainly nothing “ foreseeable, ” as Pogge maintains, that will lead from IPRs to “ many additional deaths among the global poor. ” Pogge also ignores the well-documented fact that only 17 out of 319 drugs on the World Health Organization's Model List of Essential Medicines are patented at all. 61
+
+Summing up these brief remarks, I can now answer my initial question whether and in what ways the invocation of human rights against the new patent regime makes sense. First, in our capacity as observers , we should tread with great circumspection when we try to identify the causally relevant factors that lead to the harming of HIV-infected people in poor countries. Undoubtedly, some people living with life-threatening diseases have been harmed by the new IPR regime, specifically in the sense of being worse-off than they would have been without those new rules. Yet, it is far less obvious whether today or in the near future HIVinfected people – in spite of the partly negative impact of that particular set of global rules – are not overall actually better-off than they were a decade ago when far less funds were invested in treatment and research. The trickiest question here is to what extent IPRs have not only generated a worldwide protest movement bent on changing some of the new rules, but might also have helped to bolster the commitment of private companies to develop new drugs.
+
+Second, from the point of view of public health advocates , it is quite obvious that human rights talk has worked. Observers might insist that there is no unarguable hierarchy in rights claims and that human rights do not automatically “ trump ” other rights. 62 Yet, in reality, that is exactly what human rights talk often achieves. Due to their aura of sacredness, human rights trump lesser and more mundane rights. Moreover, we have seen that physical integrity rights, summarized under the supremely vague heading of a “ right to health, ” can still trump property rights, whose advocates failed to convince policy-makers and the public that IPRs are just another subset of human rights. Unlike human rights, IPRs are a tool invented by the legal system to promote socially desirable outcomes. If the costs needed to produce this outcome appear too high, the invocation of a “ right to health ” – a claim right paired with an immunity right – can be effective in wrenching concessions from governments and corporations.
+
+61 See Amir Attaran, "How Do Patents And Economic Policies Affect Access To Essential Medicines In Developing Countries?," Health Affairs 23 (2004), 155-66.
+
+62 Michael Ignatieff, Human Rights as Politics and Idolatry, ed. Amy Gutmann (Princeton, NJ: Princeton University Press, 2001), 20.
+
+![Figure](figures/figure_008.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+229
+
+Third, I believe that the exaggerated focus on IPRs as the cause of easily avoidable misery has a number of hidden and problematic consequences. For one, this approach is highly western-centric in suggesting that whether AIDS patients in poor countries continue to suffer “is up to us,” meaning the developed world. It leads one to believe that citizens in those countries themselves cannot affect the situation, and that they are wrong in stubbornly pointing to many other causal factors, apart from global IPRs, which appear to contribute to unnecessary suffering – from the social exclusion or persecution of AIDS victims to the irresponsible quackery favored even by some governments. 63 Furthermore, the attribution of a vast array of harmful consequences to IPRs fosters a worldview that represents distant others as needy and dependent on Western samaritanism and goodwill. In this sense, the discourse of IPR critics fits surprisingly well into the narcissistic post-Cold War search of Western elites for a meaningful self-image, to be realized by interventionist foreign policies inspired by higher aims beyond petty self-interest. Although this discourse is arguably more promising than a ruthless realpolitik, it may also drive and legitimize the extension of new mechanisms of external interference in weak countries. 64
+
+## The Intellectual Property Regime as a Recognition Order
+
+The language of human rights is caught in a dilemma. Either it is trivialized and used to cover the kind of minor injustices or setbacks that can be experienced even in Western societies (of which I gave an example earlier), or it is used in a “ selfconsciously minimalist ” 65 manner that is more likely to be accepted by larger sections of humankind. In the latter case, however, it restricts attention to a small segment of extreme instances of oppression and negligence, while not capturing less conspicuous experiences of humiliation and disrespect. For this reason, I want to conclude this article by proposing a broader approach that is better suited to capture the potential for misrecognition and injustice built into the intellectual property order. Drawing on Axel Honneth's recent innovation in critical social theory, the global IPR regime can be redescribed as an institutionalized “ recognition order. ” 66
+
+Without trying to summarize the complexities of the debate on the politics of recognition, I want to stress two important ideas. First, numerous authors concur that a just society is one that shows all its members due recognition. They differ with regard to the question of whether social struggles for recognition are separate from struggles over the redistribution of material resources or whether and how exactly
+
+63 See, for example, Adamson S. Muula and Joseph M. Mfutso-Bengo, "Important But Neglected Ethical and Cultural Considerations in the Fight Against HIV/AIDS in Malawi," Nursing Ethics 11(5) (2004), 479-88; John Moore and Nicoli Nattrass, "Deadly Quackery," New York Times, June 4, 2006.
+
+64 For an astute analysis of this irony, see David Chandler, "The other-regarding ethics of the 'empire in denial'," in David Chandler and Volker Heins, eds., Rethinking Ethical Foreign Policy: Pitfalls, Possibilities and Paradoxes (London and New York: Routledge, 2007), 161-83.
+
+$^{65}$ Ignatieff, fn. 62 above at 56 .
+
+66 See Axel Honneth, "Redistribution as Recognition: A Response to Nancy Fraser," in Nancy Fraser and Axel Honneth, Redistribution or Recognition? A Political-Philosophical Exchange (New York: Verso, 2003), 110-97. See also Simon Thompson, The Political Theory of Recognition: A Critical Introduction (Cambridge: Polity Press, 2006).
+
+![Figure](figures/figure_010.png)
+
+---
+
+230
+
+V. Heins
+
+these two types of conflict are intertwined. 67 Second, in Honneth's reading of the concept, recognition encompasses more than the rights of persons. In particular, it includes social esteem deserved by individuals and communities according to their achievements as “ productive citizens ” who contribute to the welfare of society. 68 How do these ideas shed light on current discussions about new intellectual property arrangements?
+
+Critics have characterized the global IPR regime as driven not so much by concerns about efficiency than by distributive norms and the desire to collect patent rents on a global scale. $^69$ Following Honneth, I believe that these distributive arrangements may not only often be unfair, but that they can be interpreted as an institutional outcome of underlying patterns of recognition and misrecognition. Every intellectual property order encodes deeply entrenched value preferences for certain kinds of labor at the expense of others, but as soon as this order becomes global, its potential for misrecognizing the achievements and qualification of others grows considerably. Beneath the obvious distributive consequences of the global IPR regime, we find imbalances in the way some achievements by some social groups are explicitly acknowledged, while others are taken for granted and go unrecognized. Moreover, observable distributive patterns and rent flows are, at least to a significant extent, the consequence of the underlying recognition order.
+
+Most legal commentators readily accept that IPRs are about remuneration as well as recognition and that these two aspects are closely related. Modern societies tend to value both socially useful innovations and expressive intellectual activities that are seen as quintessentially human. Again, this belief in the immeasurable value of the “ artistic and creative nature ” 70 of human beings is by no means confined to the West. The granting of a patent or a copyright implies that society considers products of human ingenuity – and hence their producers – as worthy of “ respect. ” 71 For this reason, few authors reject the idea of intellectual property in toto. On the contrary, even staunch critics of the current intellectual property order often advocate the creation of new IPRs to protect, for example, the indigenous knowledge of marginalized rural communities. Therefore, the real controversy is not about IPRs as such, but about how to protect physical instantiations of very different kinds of intellectual achievement in a way that adequately reflects a shared sense of value preferences. The TRIPS agreement has led to widespread criticism and protest because it is seen by many as the institutional expression of patterns of esteem that do not square with the self-perception of those who are affected by the agreement. Conflicts arise out of the growing impression that some achievements in modern society are overprotected by law, whereas others are underprotected and up for grabs. In this connection, farmers' rights and resource rights deserve a second look as symbols in a struggle for reevaluating the contributions of marginalized groups to global society.
+
+67 See Nancy Fraser and Axel Honneth, Redistribution or Recognition? A Political-Philosophical Exchange (New York: Verso, 2003).
+
+$^{68}$ See Honneth, fn. 66 above at 140-1.
+
+$^{69}$ See May and Sell, fn. 2 above at $159-60,187-8$.
+
+70 Gandhi, "For contraceptives," in Collected Works of Mahatma Gandhi Online, Vol. 68, 344. Available at: http://www.gandhiserve.org/cwmg/cwmg.html
+
+$^{71}$ Fisher, fin. 58 above at 193.
+
+![Figure](figures/figure_010.png)
+
+---
+
+Human rights, intellectual property and struggles for recognition
+
+231
+
+Honneth has referred to the example of the critical importance of childrearing and housework to the reproduction of society to show that feminist struggles for social services, tax benefits, or the remuneration of housework were in fact struggles for recognition or struggles against a dominant prestige order that had made the work of women invisible. 72 Similarly, activist social scientists, anthropologists, and others have pointed to the role of millions of small farmers all over the world in maintaining and developing plant varieties and their environment, thereby providing the “raw material” for more visible and rewarding activities like scientific plant breeding. These contributions by peasants and cultivators, many of them poor and women, are still vastly underappreciated despite the fact that they are as crucial for the reproduction of society as childrearing and housework. Like these latter activities, which over the last decades were at the center of feminist struggles, the unrecognized activity and qualification of caring for crop species or medicinal plants have given rise to transnational struggles for compensation and acknowledgment.
+
+The crux of my argument is that, to explain those struggles, the sense of disrespect conveyed by the IPR regime is more important than its distributive effects. $^73$ Mobilization against the new rules originates from the perception that Western standards of comparative evaluation of contributions to social reproduction have been imposed on the rest of the world. This impression of a curious evaluative imbalance built into the intellectual property order has been nicely summarized by James Boyle: “ Curare, batik, myths and the dance `lambda' flow out of developing countries, unprotected by IPRs, while Prozac, Levis, Grisham and the movie Lambada! flow in – protected by a suite of intellectual property laws, which in turn are backed by trade sanctions. ” $^74$ Regarding the origins of recent struggles over fair intellectual property rules, an educated guess would be that the value preferences of Western industrial and commercial elites were asserted at the same time when consultants of international organizations, Western-educated locals, and aid workers rediscovered and re-emphasized the stunning wealth of the ancient but still utilized agronomic, botanical, and medical knowledge of the rural indigenous peasantry in many developing countries. $^75$ Against this background of a renewed emphasis on the hidden contributions of marginal groups to global welfare, the emerging IPR regime looked unfair and biased against the achievements of groups made invisible.
+
+Seen in this light, the human rights-based critique of IPRs is unduly narrowing our attention to a segment of injustices possibly fostered by the global IPR regime. By their very nature, human rights discourses are emergency oriented. They are centered on situations that cannot be tolerated under any circumstances. The flip side is that, by focusing on exceptional situations, they tend to confirm the rule. Product patents are fine, we are prompted to think, as long as they do not lead to price hikes for medicines needed in Africa. The human rights-based critique correctly realizes the potential for
+
+$^{72}$ See Honneth, fn. 66 above at 153-5.
+
+$^{73}$ See ibid, 157.
+
+74 James Boyle, Shamans, Software and Spleens: Law and the Construction of the Information Society (Cambridge, MA: Harvard University Press, 1996), 125.
+
+75 See Kevin Healy, Llamas, Weavings, and Organic Chocolate: Multicultural Grassroots Development in the Andes and Amazon of Bolivia (Notre Dame, IN: University of Notre Dame Press, 2001), 89-94 and Chap. 5.
+
+![Figure](figures/figure_009.png)
+
+---
+
+232
+
+V. Heins
+
+disrespecting relatively powerless individuals and communities inside and outside the West, but fails to see the big picture. It ignores the potential of the IPR regime for infringing upon rights other than human rights, as well as the potential for misrecognizing the labor of marginalized and “ forgotten ” groups who have a hard time attracting the attention of the world unless they find themselves in a humanitarian emergency.
+
+## Conclusion
+
+Today, both human rights and IPRs are universal in scope and vision. Yet, the world order partly shaped by these rights is a “thin” one, which unlike a “thick” order, is not backed by a strong moral consensus on the universal applicability and fairness of those rights. Hence, instead of building a framework for managing conflicts, both sets of global rights often tend to exacerbate them. The foregoing discussion has sketched out a number of differences between ordinary property rights, IPRs, human rights, and other nonproprietary rights like farmers’ rights and traditional resource rights. Whereas human rights are based on moral intuitions that shape the legal discourse, IPRs started their career as a highly specialized legal doctrine that was only subsequently reinforced through a moralizing narrative.
+
+I have shown that the human rights-based critique of the emerging global intellectual property order galvanized around the HIV/AIDS crisis and, more specifically, around the role of product patents in making access to life-saving drugs more cumbersome or even impossible. This critique is simultaneously too radical and not radical enough. On one hand, the invocation of human rights against IPRs is often relying on questionable empirical claims about the causes of avoidable social suffering; too many consequences are attributed to a single cause, and local actors and institutions are regularly exonerated at the expense of global actors and institutions. As a result, the critique insinuates that “ it is up to us ” – enlightened Western elites – to fix the world. 76 On the other hand, the language of human rights is strongly emergency oriented. Its overuse limits our attention to exceptional situations like the HIV/AIDS crisis at the expense of less obvious long-term consequences of the normal operation of the new rules, particularly for developing countries. In this way, the AIDS epidemic as the privileged object of the human rights-based critique of IPRs has proved to be a red herring in the debate about intellectual property. Although of enormous human relevance, it has helped to deflect attention away from broader issues raised by expanding intellectual property claims. To address these issues, we need to learn from theories that situate human rights in the larger institutional context of multiple forms of recognition and misrecognition and their respective consequences.
+
+Acknowledgments I wish to thank Thomas Pogge, Axel Honneth, Niranjan Rao, Michael Flitner, Susan Sell, and, in particular, an anonymous referee for helping me to shape my ideas about rights and recognition. Thanks also to Mark Hill for his assistance and suggestions.
+
+76 Pogge concedes that much “severe suffering would be avoided if rulers of the developing countries behaved better,” but he believes that this behavior can in turn be explained by the workings of “global institutions upheld by the West” (personal communication, June 25, 2006).
+
+![Figure](figures/figure_008.png)
+

processed/00117_W2270641865/document.md

@@ -0,0 +1,474 @@
+International Studies Quarterly
+
+International Studies Quarterly (2014) 58, 781-792
+
+(WN)
+
+# An Integrated Model of Legal Transplantation: The Diffusion of Intellectual Property Law in Developing Countries*
+
+Jean-Frédéric Morin Laval University and Edward Richard Gold McGill University
+
+Why do some countries adopt exogenous rules into their domestic law when those rules contravene their specific interests? We draw on the policy-diffusion literature to identify four causal mechanisms that we hypothesize explain the adoption of such rules. While existing literature treats these mechanisms as independent, we argue that each works in combination with the others to facilitate legal transplantation. While one mechanism—coercion—tends to initiate the transplantation process, it fades over time and three others largely supplant it: contractualization, socialization, and regulatory competition. These mechanisms act in a mutually supportive manner. We test our claims via a quantitative analysis of legal transplants in the field of intellectual property (IP) that incorporates an original index of IP protection in 121 developing countries over more than 14 years. This article concludes with a plea for theoretical eclecticism, acknowledging multicausality and context-conditionality. Any comprehensive explanation of legal transplantation must include the identification of mutual reinforcement between causal mechanisms, rather than simply rank their relative contributions.
+
+“[Laws] should be so specific to the people for whom they are made, that it is a great coincidence if those of one nation can suit another” argued Montesquieu in The Spirit of the Laws (1961:295). Apparently, history is full of coincidence. Laws frequently travel across both time and space. Sections of the Code of Hammurabi, enforced in Babylonia four thousand years ago, were integrated into Persian law, made their way into Greek law, and were subsequently incorporated into Roman law (Watson 1974:22–23). The Roman legacy then inspired the European Civil Codes that include elements of the ancient text. More recently, the European Codes have served as models for legal reform in countries as diverse as Peru, Egypt, and Japan.
+
+Legal transplantation proves particularly puzzling in situations of asymmetric interests: Those in which the
+
+interests of the adopting state conflict with those of the state in which the rule originated. Why would a country adopt foreign rules that run counter to its own interests? Existing scholarship lacks adequate answers to this question.
+
+This article argues that the explanation for legal transplantation lies not in any single causal mechanism but in the succession and reinforcement of multiple mechanisms. In making this claim, we favor analytical eclecticism and answer the call of Sil and Katzenstein for “complex causal stories that forgo parsimony in order to capture the interactions among different types of causal mechanisms normally analyzed in isolation from each other within separate research traditions” (2010:412).
+
+The article proceeds in three main sections. The first draws upon legal and political scholarship to build a typology of causal mechanisms for legal transplantation. It introduces an integrated understanding for their interaction under the scope condition of asymmetric interests. While the literature typically presents the mechanisms as being mutually exclusive, this article explores the possibility that they may, in fact, act in concert with one another. In doing so, they facilitate the adoption and maintenance of the legal transplant. The second section presents the case of intellectual property (IP) as an example of the dynamics of legal transplantation under asymmetric interests. It also introduces a new index of IP rules in force in 121 developing countries more than 14 years. The third section of this article examines the integrated understanding in light of quantitative evidence relating to IP. It puts this evidence into context through the use of examples explored in the literature. The results of this examination lead us to conclude that multicausality and context-conditionality constitute critical factors in understanding complex phenomena such as legal transplantation.
+
+Jean-Frédéric Morin is associate professor at Laval University and Canada Research Chair in International Political Economy. Before joining Laval University in 2014, he was professor at Université libre de Bruxelles in Belgium. His most recent research projects look at regime complexes, transnational expert networks, and policy transfers in the fields of trade, intellectual property, and biodiversity.
+
+E. Richard Gold is a James McGill Professor at McGill University's Faculty of Law where he was the founding Director of the Centre for Intellectual Property Policy. He teaches in the area of international, domestic, and comparative intellectual property and innovation. His research centers on the nexus between innovation, development, and commerce, with an emphasis on the life sciences.
+
+*Author's notes: Value Addition through Genomics and GE3LS (ValGEN), a Government of Canada (through Genome Canada) sponsored project, Genome Prairie, and Genome Quebec, supported the authors' research. The authors would like to thank scholars of both political science and law including Susan K. Sell, Hong Pang, Ken Shadlen, Bona Muzaka, Erick Duchesne, Stephen J. Toope, Barbara Delcourt, Alexis Carles, Josué Mathieu, Amandine Orsini, and Nicholas Kasirer. The authors also gratefully acknowledge the research assistance of Kevin Daley, James Duffy, Michael Shortt, Erica Shadeed, Andrew Baker, Nafay Choudhury, Palmira Granados Moreno, Ola Dajani, Priscilla César, and Nikita Stepin.
+
+Morin, Jean-Frédéric and E. Richard Gold. (2014) An Integrated Model of Legal Transplantation: The Diffusion of Intellectual Property Law in Developing Countries. International Studies Quarterly, doi: 10.1111/isqa.12176 © 2014 International Studies Association
+
+---
+
+782
+
+An Integrated Model of Legal Transplantation
+
+## Theoretical Framework
+
+### Defining Legal Transplantation
+
+We define legal transplantation as the adoption into the national legal system by one state (the “adopter” country) of a rule originating in a foreign state (the “originator” country). Legal transplantation differs from both coordinated and uncoordinated legal convergence. Coordinated legal convergence occurs when two groups of lawmakers each agree to move their respective systems toward a third and common point, often defined by a treaty (Drezner 2001:53). Uncoordinated legal convergence arises when states adopt parallel yet independent legislative processes leading to identical legal rules in both countries, usually as a reaction to similar conditions (Holzinger and Knill 2005:792). In contrast, legal transplantation is a process through which the adopter implements a rule formulated in and for the originator country.
+
+Much of the debate among legal comparativists on transplantation centers on the suitability of the surgical metaphor. Watson, who coined the term “legal transplant,” conceptualizes law as a collection of codified rules that can be easily displaced since “there is no simple correlation between a society and its law” (1974:108). On the other hand, Legrand (2001), one of Watson's fiercest critics, argues that laws exist within cohesive legal systems that reject alien rules. Legrand views successful transplants as unlikely because they require transplanting an entire legal culture—including its doctrines, procedures, and institutions. Most authors fall between these formalist and culturalist perspectives, recognizing that the adopting legal system will likely interpret, apply, and enforce transplanted rules differently.
+
+We agree that the term “ transplant ” fails to adequately capture the subtle process of adaptation involved in taking a legal concept developed in one context and incorporating it into another. However, since legal scholars widely employ the term, we favor it over alternatives. When discussing the process of legal transplantation, we refer solely to the transplantation of the rules themselves and make no claim in respect of their interpretation or enforcement. $^1$
+
+While a rich and subtle legal scholarship examines the nature and types of legal transplants, jurists have yet to establish the mechanisms that explain legal transplantation in the first place. Most of the legal literature tends to rely on vague explanations for the cause of the transplant, such as economic liberalization or the prestige of foreign laws. Other studies suggest that transplantation is more likely between culturally, institutionally, or economically similar countries (Berkowitz, Pistor, and Richard 2003:167; Simmons and Elkins 2004:187) . While the latter may be true, it constitutes a facilitating condition rather than a causal mechanism.
+
+### A Typology of Causal Mechanisms
+
+In contrast to the juridical scholarship on legal transplants, the study of policy transfers and policy diffusion identifies a number of explanations for the adoption by one country of policies previously crafted in another (see
+
+Dobbin, Simmons, and Garrett 2007 for a review of this literature). While political scientists have yet to assess the scope, magnitude, and frequency of policy diffusion, they have proven particularly creative in conceptualizing causal processes. By combining the descriptive analysis of transplantation in the legal literature and the causal mechanisms studied in the policy-diffusion literature, this article provides an opportunity for interdisciplinary research integrating both legal and political studies and comparativist and internationalist perspectives. This creates the opportunity for each approach to palliate the weaknesses of the other.
+
+This article builds on the policy-diffusion literature to identify five ideal-type causal mechanisms of legal transplantation. We call the first mechanism emulation , a process that is sometimes referred to as “lesson-drawing” (Rose 1991:4) or cost-saving transplants (Miller 2003:845). It suggests that legal transplantation occurs when lawmakers, confronted with a problem, look across national borders for effective and transferable solutions. US-style pleabargaining could provide, for example, a solution to countries struggling with overburdened criminal justice systems. Using foreign experience becomes a rational strategy to save time or money associated with trial and error.
+
+Coercion , the second ideal-type mechanism, occurs when a state promotes its rules through the use of material power, whether military or economic. We distinguish between two types of coercive legal transplantation processes. Imperialistic transplantation, also called “ direct imposition ” (Dolowitz and Marsh 2000:9) , involves the imposition of foreign legal rules without the consent of the adopter country. Imperialistic transplantation occurred, for example, when Japan introduced USinspired civil and political rights into its constitution during the allied occupation. The more common “ indirect imposition ” occurs when the mere threat of negative sanctions provides the incentive for countries to voluntarily transplant exogenous rules.
+
+The third ideal-type mechanism, contractualization , occurs when states bargain with one another in relation to a legal rule. In order to acquire acceptance of transplanted legal rules, those negotiations usually include trade-offs linking two or more issue-areas. One state will typically promote its own legal rules as constituting the common standard governing a particular issue-area. At the same time, it offers compensation or side payments in another issue-area. This quid pro quo arrangement is then formalized by an international treaty. US law on investment protection, for example, spreads to several countries via free trade agreements, offering the prospect of greater access to the American market in exchange for adopting US-style investor protection rules (Morin and Gagné 2007:59). Contracting parties do not necessarily negotiate as equals, and their contractual agreement does not necessarily result in a balanced outcome. The boundary between coercion and contractualization blurs, particularly where a country's leaders feel compelled to enter into a treaty to avoid economic harm. Nevertheless, for the limited purposes of this article, we consider as a contractual arrangement (rather than as an occurrence of coercion) any treaty in which there exists a quid pro quo beyond the elimination of harm or the threat of harm.
+
+Under the fourth ideal-type mechanism, regulatory competition, lawmakers adopt foreign rules, whether or not effective in addressing domestic issues, in order to better position their country in a competitive world (Radaelli
+
+1 There is a recognized gap between the formal existence of a legal rule and its interpretation and enforcement. We limit our discussion to substantive legal rules but understand that a legal system is much broader than a collection of legal rules and also includes general principles, legal administrations, enforcement mechanisms, etc.
+
+---
+
+Jean-FrédéRIC Morin and E. Richard Gold
+
+783
+
+2004). Regulatory competition usually presents itself in one of two opposing versions. The “race to the bottom” version assumes that lawmakers adopt the lowest regulatory standards of competing countries, such as minimal levels of labor and tax regulations, to avoid capital flight (Drezner 2001:57). By contrast, the “race to the top” generally focuses on reputational, rather than economic, competition. Under this second version, countries seek the cutting edge of legal creativity in order to generate legitimacy or political rent, a trend especially prevalent in the areas of consumer rights and environmental protection (Vogel 1995).
+
+The fifth ideal-type mechanism is socialization , defined as a process directed toward the internalization of the principles, beliefs, and norms of a foreign community (Schimmelfennig 2000:112; Checkel 2005:804). In the context of legal transplantation, it suggests that ideational transfers can lead to legal transfers. A state more frequently adopts a foreign rule if persuaded of its appropriateness, that is, if the legal rule in question resonates with established social norms and fits with the collective identity of the adopter country. This process of legal transplantation relies on the flexibility of idea may actively transfer through a wide variety of entrepreneurs, including NGO activists, university professors, business organizations, and legal communities. The most powerful of those entrepreneurs can most successfully diffuse their ideas (Bach and Newman 2010:523).
+
+A typology is not a theory and, arguably, a causal mechanism examined in isolation cannot explain a complex phenomenon such as legal transplantation. Current studies on policy diffusion remain limited, however, either by focusing on a single mechanism or by overlooking the connections that may exist between mechanisms. This article addresses this lacuna by exploring complementarities and reinforcements between causal mechanisms under the scope condition of asymmetric interests.
+
+## The Scope Condition of Asymmetric Interests
+
+Most studies of legal transplantation focus on cases in which countries share an interest in harmonization. While one country might benefit more than another from harmonization, there exists an underlying assumption that harmonization of environmental or trade regulations contributes to the common good. In those cases, legal transplantation can provide an absolute gain for all jurisdictions even if benefits remain unevenly distributed among them. Explaining legal transplantation in situations of asymmetric interests between an adopter and an originator country continues, however, to puzzle.
+
+By asymmetric interests, we mean situations in which the originator country has a strong interest in exporting its rules while the prospective adopter perceives itself as having a strong interest in maintaining its dissimilar rules, at least with respect to the specific issue-area that the legal transplant covers. The diffusion of “ know your customer ” rules in domestic banking regulations provides a typical case of legal transplant under asymmetric interests (Sharman 2008:640). On the one hand, the United States and several European countries have an interest in promoting their own banking standards in an effort to prevent money laundering and tax evasion. On the other hand, some offshore jurisdictions have an interest in providing banking clients with greater privacy protection in order to attract capital to their financial institutions. Why then do those offshore jurisdictions increasingly duplicate
+
+US and European banking rules when doing so appears to run counter to their interests? This case, like other cases of legal transplant under asymmetric interests, can only be explained by looking to the power imbalance between originator and adopter countries.
+
+Of all the causal mechanisms identified in the previous section, emulation is the only one to ignore power imbalances. While emulation may be a prime mechanism behind many legal transplants, one would expect it to have a substantially diminished role in the context of perceived asymmetric interests. The reason for this is simple: the divergent interests in the asymmetric context undermine the cost savings logic of emulation.
+
+There are two limitations to this argument. First, while the interests of the adopter country may, as a whole, be opposed to those of the originator country, there will likely be actors (for example, transnational or domestic corporations) within the adopting country who see their interests aligned with those of the originator. Given our scope condition of asymmetric interests, however, the pro-adoption lobby will unlikely be the dominant interest group. Second, if policymaker perception of asymmetry actually becomes weakened whether through this internal lobbying or through the other mechanisms — for example, if a country becomes persuaded that its interests are aligned with those of the originator country — then the cost savings logic of emulation might come to dominate.
+
+The remaining four mechanisms — coercion, contractualization, regulatory competition, and socialization — should play a central role in the case of asymmetric interests. A state may use coercion to render the costs of not adopting a legal rule sufficiently high as to outweigh the rule's inherent costs. Through contractualization, a state can trade the negative costs of adopting an inconvenient legal rule against gains elsewhere. Under the regulatory competition model, a country may agree to transplant an unfavorable rule when it believes that it will suffer greater reputational or economic losses by not running with or even leading the crowd. And socialization may lead lawmakers to underestimate the costs or overestimate the advantages of an inappropriate rule. Under each of these mechanisms, more powerful states are likely to act as the originators of transplanted rules as they can more easily coerce other states, bargain from a position of strength, be an originating cause of regulatory competition, and diffuse their ideas globally. Thus, each of these mechanisms has the potential to explain legal transplantation, at least in part, in the context of asymmetric interests.
+
+## An Integrated Model of Legal Transplantation under Asymmetric Interests
+
+Most of the literature on policy diffusion suggests that each causal mechanism constitutes a competing and total explanation for policy transfers. Some studies suggest that the manipulation of material incentives is more effective than the alteration of substantive beliefs (Kelley 2004; Schimmelfennig and Sedelmeier 2004; Elkins, Guzman, and Simmons 2006; Cao 2009), while others claim the opposite: that socialization has a greater impact than does coercion (Berkowitz et al. 2003; Meseguer 2004). Contrary to these contentions, we argue that originator countries rely on a combination of coercion, contractualization, regulatory competition, and socialization—along with a smaller emulation effect—to disseminate and maintain their own set of rules in situations of
+
+---
+
+784
+
+An Integrated Model of Legal Transplantation
+
+asymmetric interests. $^2$ While some authors discount the potential of the mechanisms to interrelate, we lack a fully integrated understanding of this phenomenon (Sharman 2008:636).
+
+Given the conflicting nature of asymmetric interests, an originator country will likely initiate the process of legal transplantation through the use of coercion. The more that exogenous rules conflict with the material interests of the adopter, the more likely that active coercion initiated the transplantation. Yet coercion remains a politically limited mechanism and cannot explain, alone, the diffusion of the same set of rules in a large set of countries. Even when the most powerful global actor deploys it, unilateral coercion can only effectively target a few countries at a time — not the entire international community (Holzinger and Knill 2005:778) . Moreover, the extensive use of coercion risks provoking political backlash and feeding animosity against the coercing country and its legal standards.
+
+Despite these limitations, coercion can play an important role in the bargaining process leading to contractualization. At a certain point, the originator will rationally switch emphasis away from the mechanism of coercion to that of contractualization (as we have defined). Contractualization provides, as opposed to coercion alone, a stable institutional context and continuous incentive through which the originator can maintain and extend legal transplantation to additional states. This remains true even when the contractual agreement itself was negotiated under coercion. For adopter countries, formalizing the transplantation process through contractualization can also prove beneficial. This is because adopting foreign rules as part of a quid pro quo bargain renders political opposition less likely at the domestic level than accepting the same rules under threat or actual application of coercive pressures. Under this perspective, contractualization may provide stability and predictability to both the originator and the adopter of the transplanted rule.
+
+While contractualization may not completely displace coercion, it tends to supplant it. Once a contractual agreement is formalized, the further use of coercion to extend transplantation to rules beyond the agreed framework will likely prove politically and legally ineffective. Indeed, several contractual agreements provide safeguards against the use of direct coercion. Moreover, coercion violating internationally agreed upon norms is seen as illegitimate, and this perceived illegitimacy significantly decreases the effectiveness of the mechanism (Pelc 2010) .
+
+While contractualization may diminish the effectiveness of further unilateral coercion, contractualization can initiate a synergic dynamic with regulatory competition. Baldwin (1997:877) captures this dynamic through his “domino theory of regionalism,” under which one or several bilateral trade agreements produce trade diversion from countries that have not signed bilateral agreements. This generates, in turn, pressure on non-signatories to enter into their own bilateral agreements. For example, developing countries do not have an incentive to duplicate US rules on investment protection unless a competing investment-receiving country acquires a competitive advantage by signing a bilateral investment treaty with the United States (Elkins et al. 2006) . Owing to this reality,
+
+several developing countries have autonomously transplanted into their domestic legislation trade and investment rules from the United States; some even became demandeurs for bilateral agreements with the United States formalizing further transplants.
+
+Contractualization can also pave the way for socialization in three different ways (Ikenberry and Kupchan 1990:290). First, at the institutional level, contractualization can provide the context in which socialization takes place. Various agreements provide for the creation of joint committees or annual summits in which bureaucrats and policymakers interact, debate, and share ideas. Numerous studies show that contact at an intergovernmental level, especially within relatively confidential and informal settings, facilitates socialization processes (for example, Schimmelfennig 2000; Kelley 2004; Checkel 2005; Cao 2009; Greenhill 2010). Second, at the discoursive level, contractual agreements can become powerful rhetorical weapons. Law has the general characteristic of appearing as an intrinsically legitimate force (Kapczynski 2008; Brunnée and Toope 2010). Contractual agreements can thus favor what Pierre Bourdieu calls the “genesis amnesia,” (1995:3) that is, forgetting that the adopted rules resulted from a bargaining process in which coercion may have been used. Third, at the cognitive level, contractualization may generate an uncomfortable cognitive dissonance when the political elite in adopter countries do not believe in the value of agreed transplanted rules. According to cognitive psychologists, cognitive dissonance is usually solved by changing beliefs to fit actual behavior, rather than the other way around (Festinger 1957). Therefore, once countries reach an agreement, their political elites tend to convince themselves that its least favorable clauses are actually less harmful than they had previously feared.
+
+Regulatory competition can also favor socialization under a dynamic akin to what Finnemore and Sikkink (1998:895) call a “norm cascade.” Under this process, once a critical mass of countries adopts a given norm, the norm reaches a tipping point after which it begins to diffuse independently, without the need for the material incentive initially required to reach that point. As an increasing number of countries transplant a set of rules, those rules become normalized and create social and cognitive pressure for conformity. No longer viewed as exogenous, these rules become the standard for countries with a given collective identity, such as “democratic,” “developed,” or “liberal.” As Checkel argues, what starts as strategic and incentive-based behavior “often leads at later points to preference shifts” (2005:814).
+
+In turn, socialized decision makers believing in the virtue of transplanted rules, but facing domestic opposition, may seek to lock in the rules by contractually binding their country to them. The bargaining process would be further facilitated by the fact that those socialized decision makers no longer perceive the costs of the transplanted rule and do not therefore insist on receiving compensation for their adoption. Likewise, socialization can reduce barriers to regulatory competition as socialized decision makers desiring to extract material or reputational gains at the international level may seek to transplant an even greater number of foreign rules than their competitors (Likhovski 2009) .
+
+We derive three main hypotheses from this integrated model under asymmetric interests. Our first is that an originator state's coercion is likely to initiate legal transplantation in adopter countries and to favor contractual-
+
+2 We thus put aside, for the purposes of this article, whether and if the originator country modifies its own rules as a result of the process of legal transplantation to an adopter country.
+
+---
+
+Jean-FrédéRIC Morin and E. Richard Gold
+
+785
+
+ization, while contractualization is likely to lessen the effectiveness of coercion against actual and potential adopters (H1). The second hypothesis is that regional competition and socialization each also positively influence legal transplantation from originator to adopter countries (H2). The third hypothesis builds on the second and focuses on the interrelationships between the three mechanisms, stating that contractualization, socialization, and regional competition operate in synergy, facilitating and mutually reinforcing one another to promote further transplantation in adopter countries (H3). H1 examines the sequence of causal mechanisms, H2 their concomitance, and H3 their interrelationship.
+
+## Methods and Data
+
+### The Case of Intellectual Property Law
+
+We selected the empirical case of intellectual property (IP) to illustrate our integrated theoretical model. IP law provides a textbook example of asymmetric interests. Although economists may differ on the optimal level of IP protection for any given country, they do agree that developed and developing countries have very different optima (Chen and Puttitanun 2005; Kim, Lee, Park, and Choo 2012) . On the one hand, knowledge economies seek to maintain their economic advantage by providing and promoting strong IP standards. Not only does the United States (often at the behest of prominent multinational corporations) most actively promote higher IP protection globally, it benefits most from upward IP harmonization. In 2011 alone, it received more than net US $ 84 billion in international royalties and licensing fees. Developing countries, on the other hand, have an interest in providing lower levels of protection for IP, which foreign investors largely own. This strategy allows local businesses to acquire foreign technologies and ensures low retail prices for products ranging from medicines and agricultural plant varieties to word-processing software and auto parts. When judged solely on the basis of economic interests, IP rules should differ significantly between developed and developing countries.
+
+Despite these asymmetries of interest, IP has become increasingly harmonized globally. So strong is this tendency that legal comparativists often consider IP as a classic example of legal transplant emanating from the United States (Mattei 2003:19–23; Miller 2003:847; Shi 2010). The Agreement on Trade Related Aspects of Intellectual Property Rights (TRIPs), for example, notoriously compels WTO members to raise their legislated IP protection to a level closer to that of the United States.
+
+Less well known is the fact that many developing states have gone beyond the minimum obligations required of them under TRIPs to align their IP laws more substantively with higher US standards. For example, Honduras protects copyright for 75 rather than 50 years; Guatemala authorizes the patentability of plants; Cambodia has criminalized the dissemination of technologies intended to circumvent copy protection, and Vietnam accepts the patentability of new uses of previously known pharmaceutical products. TRIPs does not require these requirements but each aligns with US IP rules. Overall, a majority of lowand middle-income countries exceed the minimum requirements of the TRIPs agreement and have transplanted rules that are seen as largely benefiting foreign IP holders. As Deere observes, many of the poorest WTO members—countries that the economic literature would
+
+anticipate adopting the lowest levels of IP protection —“opted for some of the world’s highest IP standards and made limited use of TRIPs flexibilities that might have helped them address social challenges in the areas of public health, education, and agriculture” (2009:102). This broad dissemination of US IP rules in developing countries demands an explanation.
+
+### The Construction of the Transplant Index
+
+This article introduces a novel index to measure the extent to which developing countries have transplanted IP rules established in and specifically promulgated internationally by the United States. In doing so, we examine the processes through which the dominant global power, the United States, has caused developing countries to adopt IP rules that, by themselves, do not align with the economic interests of those countries. The new index is required as existing IP indexes are not suitable to uncovering the transplantation of post-TRIPs US IP rules. Those indexes include indicators, such as 20-year patent terms, that have become mandatory under TRIPs and thus are no longer relevant. The widely used Ginarte and Park index (1997), for example, captures little meaningful variation after 1994 when many of its indicators became mandatory for WTO members. Further, existing indexes often collect data for only a few years, or at 5-year intervals, whereas the mechanisms being examined require a more rigorous analysis based on yearly updates over longer periods. Last, some indexes are based on only a limited sample of countries (Sherwood 1997), or do not disclose their coding frames, making it impossible to update or expand on the data set (Ostergard 2000).
+
+US-style IP rules have set the international standard for IP protection. We developed therefore an index measuring the adoption of those IP rules that are not required under TRIPs and specific to US demands for increased IP protection. $^3$ This Transplant Index scores countries on a 0 – 9 scale. The higher a country scores, the more it has aligned its IP rules with those of the United States. Given the specificity of the rules modeled, we assume that it is unlikely that a particular country — even one favoring higher IP rights — would adopt those particular IP rules in the absence of US leadership. That is, we assume that in the absence of the United States having adopted and promoted those particular rules, the index would normally approach 0 for most developing countries.
+
+Data were collected for each year from 1995 to 2008, coinciding with the initial coming into force of the TRIPs agreement. The Transplant Index covers all developing countries with a population of more than 1 million and for which data were available. The result is 121 countries and a sample of 1694 country-years. These results, as well as the complete list of indicators, coding values, and data sources, are available from the Appendix of the online version of this article.
+
+### Variables that Operationalize Causal Mechanisms
+
+Unlike other studies on policy diffusion, we include all four mechanisms-coercion, contractualization, regional competition, and socialization-within our model. We
+
+3 We acknowledge the same limitations as all others—except Sherwood (1997) who uses an idiosyncratic and irreproducible approach—in looking only at the formal adoption of IP rules rather than at their enforcement by the executive and judiciary.
+
+---
+
+786
+
+An Integrated Model of Legal Transplantation
+
+also include GDP per Capita to capture innovation capacity that previous literature has linked to increases in IP protection, at least in part through the mechanism of emulation (Chen and Puttitanun 2005) . Although GDP per Capita imperfectly proxies emulation, other domestic indicators for endogenous interests in stronger IP protection, such as research and development expenditure, were not available for several developing countries. $^4$ Each causal mechanism finds representation in one or more variables (additional supporting information may be found in the Appendix of the online version of this article).
+
+Two variables represent coercion. The first, Priority Watchlist , indicates whether the country was listed on the US Priority Watch List because of an IP-related concern the previous year. Under so-called Special 301 legislation, the United States Trade Representative (USTR) has the obligation to publish annually a Priority Watch List of countries that "deny adequate and effective protection of intellectual property rights" (19USC § 2411). The coercive effect of this trade instrument stems from the fact that countries listed face the risk of trade sanctions unless they take appropriate measures to address the concerns raised. The second coercive variable, GSP Review, indicates whether a country was being reviewed under the generalized system of preferences (GSP) program due to concerns relating to IP protection the previous year. The GSP program offers unilateral duty-free treatment for thousands of products to designated developing countries. In determining whether a country may benefit from privileged trade access, the President has the statutory obligation to take into account "the extent to which such a country is providing adequate and effective protection of intellectual property rights" (19USC § 2462). If US authorities consider that the country under review has not provided adequate IP protection, the country can no longer benefit from preferential treatment. Importantly, since both Priority Watchlist and GSP Review are expressions of threats rather than actual trade sanctions, we avoid the common selection bias in the sanction literature of capturing only failed occurrences of coercion and ignoring occurrences in which threats were effective in inducing behavioral change before being implemented.
+
+Contractualization is represented by whether the country has signed and brought into force a bilateral free trade agreement (FTA) with the United States ( US Bilateral Agreement ). The Trade Promotion Authority of 2002 specifically required the USTR to ensure that US FTAs “reflect a standard of [IP] protection similar to that found in US law” (19USC § 2102 and 3801). Although several studies note that bilateral FTAs signed by the United States provide a level of IP protection that goes well beyond what is required under the TRIPs agreement (Deere 2009:114–118), the actual transposition of these FTAs into domestic law remains largely undocumented. A case study on China suggests that some developing countries may actually agree to international IP standards knowing very well that they do not intend to fully implement them in domestic legislation (Mertha and Pahre 2005). This remains only a hypothesis since there is very little systematic evidence on the actual implementation of FTAs. Including US FTAs in our model provides an opportunity to bridge this gap in the literature.
+
+Variables representing socialization are less straightforward. The transmission of ideas is notoriously difficult to capture empirically since socialization occurs through several pathways, not all of which are intentional, formal, or even conscious. Building on the existing socialization literature, we identify two pathways to socialization: university education and capacity-building training. While those pathways do not provide direct evidence that socialization occurred, they constitute recognized vehicles for socialization and can be assessed by quantitative measures.
+
+The first variable used to represent socialization, IP Training, is the stock of IP-related training activities labeled as “technical assistance” or “capacity building” offered to the country by the United States since 1995 to the year in question. $^5$ As the USTR notes, “perhaps the most important of the remaining tools [in promoting strong IP standards worldwide] is our ability to offer technical assistance” (USTR 2004). With this in mind, the US government established the Global Intellectual Property Academy (GIPA) to “present the US methods for protecting the IP rights of business owners” and expose foreign officials “to a US model of protecting and enforcing intellectual property rights” (USPTO 2010). In 2008 alone, GIPA provided training to more than 4,100 foreign officials (USPTO 2010). In addition, several business organizations sponsor training activities for developing country policymakers. The variable IP Training includes the number of training activities offered by either the US government or private organizations in partnership with the US government. In recent years, several studies noted that those programs could be major drivers of transplantation of IP rules in developing countries, but so far none has systematically investigated this hypothesis in controlling for alternative explanations (May 2004; Matthews and Munoz-Tellez 2006). Our article fills this gap.
+
+The second socialization variable, Population Studying in the US , measures the number of nationals of a country studying in the United States in the previous year. Foreign students who populate the business, engineering, and science programs of US universities provide an indirect channel through which US norms can make their way into developing countries. During their time abroad, students are fully immersed in US culture and are socialized to the specific ideas concerning the importance of high levels of IP protection. They also absorb more general normative principles underpinning the IP system, such as individualism, rationalism, liberalism, and modernism. Once they return home, those individuals often integrate into the local elite, bringing with them exogenous beliefs that they acquired abroad. Studies of the effects of socialization in other fields have shown that foreign education is a powerful driver of transnational socialization (for example, Atkinson 2010) . Scholars, however, have never fully examined its impact on a country's level of IP protection specifically.
+
+Our identification of an indicator of regulatory competition in the field of IP is a contribution to the literature that has, so far, lacked such a measure. Based on our unique Transplant Index , we have identified the highest value of that index within the region (other than the country in question), which we call Regional Top Score , as a strong indicator of regulatory competition. As the linear increase in the average value of the Transplant Index indi-
+
+4 Moreover, research and development expenditures would have been inappropriate here because of doubtful relation with IP other than patents.
+
+5 We follow UNCTAD in counting the number of events rather than the number of days. Since we are dealing with socialization, it is the networks created rather than time spent learning that we capture here.
+
+---
+
+Jean-FrédéRIC Morin and E. Richard Gold
+
+787
+
+cates, the form of regional competition involved in the field of IP is a race to the top. Countries compete against others in the region to attract FDI, especially in the much-coveted high-tech sector. One of the ways of doing so is to boost their credibility with foreign firms by strengthening IP laws. As investors will often compare countries within the same region before making their investments — and threaten to relocate their investments from countries with lower IP levels — some countries can be driven to adopt more protective IP rules. $^6$ The country with the highest level of IP protection in the region, captured through the variable Regional Top Score , sets the pattern for competition within that region. We thus expect that if states actually engage in regulatory competition, one should observe that countries move their respective levels of IP protection toward their region's Regional Top Score , if not, we should expect this variable to have little influence on a country's Transplant Index value. $^7$
+
+In this study, we limited our analysis of the relative contributions of the above mechanisms to what we could measure quantitatively. Variables supporting our models were obviously limited by data availability. We did not have access, for example, to quantitative data on foreign investor threats to locate investments elsewhere if lawmakers failed to increase IP protection. Similarly, we found no reliable data on efforts made by business organizations to socialize the general population on the idea that strong IP protection is in their country's best interests. Those shortcomings make our empirical analysis more state-centric than our theoretical model would actually require. We leave it to future studies to provide more fine-grained qualitative evidence on the involvement of nonstate actors in the mechanisms analyzed here.
+
+We selected a fixed effects model to effectively manage certain forms of unobserved heterogeneity and eliminate bias arising from omitted variables, provided that the omitted variables remain effectively fixed within each country. In order to assure the stability of our results, we only report those models whose patterns of significance maintained even when the standard errors were estimated using 1,000 bootstraps. The program StataSE 12 was used to perform all fixed effects analyses described in the article, and in all models, robust standard error estimates were used in order to correct for heteroscedasticity. All $F$ test results in respect of our models are statistically significant at 0.001.
+
+## Results and Discussion
+
+### H1: From Coercion to Contractualization
+
+Our empirical models are consistent with the first hypothesis that coercion sets the stage for legal transplantation but fades in its effect after contractualization takes hold. A number of scholars have already documented the importance of economic coercion in triggering the legal transplantation of US-style IP rules in the 1980s and early 1990s, both qualitatively (Braithwaite and Drahos 2000; May and Sell 2006) and quantitatively (Zeng 2002; Shadlen, Schrank, and Kurtz 2005) . Those studies show that
+
+coercion was instrumental during the Uruguay Round to conclude the TRIPs agreement, considered here as an instance of contractualization. In particular, the USTR effectively targeted Brazil and other developing countries opposed to that agreement with the use of its Special 301 and lifted privileged access to the US market under the GSP program to goods coming from Argentina, Honduras, India, Mexico, and Thailand. When countries finally yielded under this pressure, the US government noted that “the Special 301 annual review is one of the most effective instruments in our trade policy arsenal” (USTR 1997: 1) and that the GSP program was “an effective point of leverage with some of our trading partners” (USTR 2004). Scholars of IP politics largely concur with this assessment and recognize that the coercive strategies of the 1980s and early 1990s had a significant impact both on developing countries and on the negotiation process of the TRIPs agreement.
+
+Our regression analysis extends the above findings and provides strong evidence that coercion continued to operate in the immediate post-TRIPs period. Table 1 shows that US coercion remained effective between 1995 and 1999 as both Priority Watchlist and GSP Review are positive and significant in that period in raising a country’s Transplant Index value. Thus, the contractual arrangement of the TRIPs agreement does not appear to have immediately prevented further coercion.
+
+This finding does not, however, contradict our first hypothesis. In fact, it was only in 1999 that a WTO Panel interpreted article 23(2) (a) of the WTO Dispute Resolution Understanding, concluding that the United States cannot unilaterally suspend trade concessions against another WTO member. As Table 1 suggests, the ruling's apparent result was that the continued practice of US watch lists lost some of its credibility. In the 2000 – 2008 period, Priority Watchlist is no longer significantly associated with the Transplant Index , while GSP Review is actually significantly but negatively associated with the Transplant Index . Despite the continued deployment of US coercive instruments against countries such as India, Indonesia, and the Philippines, they have had a limited effectiveness since 2000. We can infer from this finding that, while US coercion may have lessened the opposition of some developing countries to the TRIPs agreement during the Uruguay Round, the resulting WTO agreements, and especially the WTO Dispute Resolution Understanding, provided opportunities for those same countries to seriously limit the effectiveness of direct coercive tactics.
+
+T ABLE 1. The Influence of Variables on the Transplant Index, Before and After 2000 (Fixed Effects)
+
+<table><tr><td></td><td>1995–1999 Coefficient (543 observations; 110 groups)</td><td>2000–2008 Coefficient (1023 observations; 116 groups)</td></tr><tr><td>GDP per Capita</td><td>−0.00003</td><td>0.0001***</td></tr><tr><td>Priority Watchlist</td><td>0.379*</td><td>0.192</td></tr><tr><td>GSP Review</td><td>0.281***</td><td>−0.624**</td></tr><tr><td>US Bilateral Agreement</td><td>–</td><td>1.772***</td></tr><tr><td>IP Training</td><td>0.009*</td><td>0.016**</td></tr><tr><td>Pop. Studying in US</td><td>0.212**</td><td>0.285*</td></tr><tr><td>Regional Top Score</td><td>0.153***</td><td>0.103*</td></tr><tr><td> $R^{2}$ </td><td>0.20</td><td>0.43</td></tr><tr><td> $F_{6,100}$ </td><td>13.48***</td><td>14.63***</td></tr></table>
+
+
+(Note. *Significant at ≤0.05; **Significant at ≤0.01; ***Significant at ≤0.001.)
+
+$^{6}$ See, for example, Special 301 Submissions of PhRMA.
+
+7 We assume that there are no independent correlations between a given country's Transpleat Index value and the value of any of the independent variables at play within neighboring countries. We further assume that regulatory competition is not playing out along other groupings that correlate with regional groupings.
+
+---
+
+788
+
+An Integrated Model of Legal Transplantation
+
+Beyond the WTO Panel ruling of 1999, we suggest at least two reasons why the Special 301 was less effective in the 2000s than it was in the early to mid-1990s. Both of these reasons are related, directly or indirectly, to the rise of contractualization in the form of the WTO agreements. The first is the growing perception among developing countries that coercion is a largely illegitimate means of pressuring governments to adopt TRIPs-Plus provisions. In public debate, the TRIPs agreement provides a baseline against which to assess the legitimacy of claims regarding the level of IP protection in developing countries. Although, from a legal point of view, TRIPs provides only a floor of obligation, it also provides a discursive ceiling on legitimate pressure. As a result, many countries and NGOs resist pressure to adopt higher standards of IP protection beyond that required by TRIPs. For example, when the Clinton administration sanctioned South Africa for its TRIPs-compliant measure favoring access to pharmaceutical patents, a network of transnational NGOs, including Oxfam and MSF, organized protests around the world and disrupted the lead up to the 2000 US presidential campaign (Sell and Prakash 2004:160–167). After several leading newspapers and members of Congress sided with South Africa, the US government backed off and reintroduced trade concessions. In a similar vein, coercion may also feed nationalist sentiments and provide developing country governments with an opportunity to increase their standing with domestic constituencies by resisting coercion perceived as illegitimate. This was the case in Argentina, Brazil, India, and Thailand where governments successfully opposed US pressure to increase their IP protection (Deere 2009). The negative effect of GSP Review on a country's Transplant Index in the 2000 – 2008 period may suggest that the use of coercion actually led to push back from targeted countries.
+
+Second, the increased trade interdependence that followed the Uruguay Round provided some developing countries with the capacity to resist and even deter coercion. In 2001, when the United States threatened Brazil with the withdrawal of its duty-free access under the GSP, the Brazilian government calculated that the cost to US industries of this withdrawal would dissuade its actual imposition. As a result, Brazil confidently replied to the US threat by filing a request for consultations at the WTO that the US Patents Code was not TRIPs compliant. One month later, the United States dropped its case against Brazil.
+
+The WTO agreements only represented the beginning of an era of contractualization as more and more countries agreed to sign bilateral FTAs endorsing US-style IP rules in exchange for preferential access to the US market. Arguably, the boundary between coercion and contractualization is blurred as all US FTAs were negotiated in the context of sharply asymmetric trade relationships. Until recently, the United States focused its FTA program on those countries that presented a combination of pronounced power asymmetries and the possibility of extracting tangible gains. It did not enter into a legally binding bilateral IP treaty with emerging economies such as Brazil, India, or China. Rather, it concluded bilateral agreements with countries that have some economic significance but that cannot realistically negotiate on an equal footing with the United States. Nicaragua, for example, whose FTA with the United States entered into force in 2006, exports more than 12 % of its GDP to the United States, its primary trading partner, while the Uni-
+
+ted States only exports 0.003% of its GDP to Nicaragua, its 78th trade partner. Some countries, such as Jordan and the Dominican Republic, were even placed on the Priority Watch List or on the Out-of-Cycle Review in the 3 years prior to the signature of their bilateral agreements with the United States.
+
+While active coercion likely gave rise directly to some contractual negotiations, our results suggest that the general rise of the contractualization wave did not depend on explicit coercion. Returning to Table 1 , our analysis supports the hypothesis that contractualization came to replace coercion in effectiveness in the 2000 – 2008 period. As the effectiveness of direct coercion was waning, contractualization became the principal mechanism driving the legal transplantation of IP rules. US Bilateral Agreement is not only significantly correlated with the Transplant Index in this period but, as Table 2 illustrates, outranks every mechanism other than socialization in terms of its explanatory strength over the entire 1995 – 2008 study. Most developing countries that signed a FTA with the United States are among those that demonstrate the largest increases in their levels of IP protection. These results suggest the existence of a strong link between the signing of bilateral agreements with the United States and higher levels of IP protection. The results further support the disputed claim that bilateral agreements are actually implemented in domestic legislation.
+
+Once a FTA is reached, coercive pressure becomes more relaxed. The WTO agreements led to a decline— although delayed—in the effectiveness of coercive instruments and the United States neither used the Special 301 nor the GSP review against any of its recent FTA partners in the years following the signature of an agreement. Thus, as hypothesized, contractualization takes over in effectiveness from coercion. As the next sections illustrate, however, contractualization does not operate in isolation.
+
+## H2: The Positive Effects of the Other Mechanisms on the Transplant Index
+
+Our quantitative analysis provides strong evidence that both regulatory competition and socialization contribute significantly to a country's Transplant Index score. First, the positive and significant effect of the Regional Top Score on the Transplant Index in both 1995-1999 and
+
+Table 2. The Influence, Strength, and Rank of Variables on the Transplant Index, Full Period (Fixed Effects)
+
+<table><tr><td>Variable</td><td>Coefficient (1566 observations; 116 groups)</td><td>Rank of Strength</td></tr><tr><td>GDP per Capita</td><td>0.00015***</td><td>2 (3.71%)</td></tr><tr><td>Priority Watchlist</td><td>0.313</td><td>7 (0.33%)</td></tr><tr><td>GSP Review</td><td>−0.393</td><td>6 (0.43%)</td></tr><tr><td>US Bilateral Agreement</td><td>1.948***</td><td>1 (4.19%)</td></tr><tr><td>IP Training</td><td>0.012***</td><td>4 (2.75%)</td></tr><tr><td>Pop. Studying in US</td><td>0.444***</td><td>3 (3.27%)</td></tr><tr><td>Regional Top Score</td><td>0.221***</td><td>5 (2.54%)</td></tr><tr><td> $R^{2}$ </td><td>0.53</td><td></td></tr><tr><td> $F_{7,115}$ </td><td>25.98***</td><td></td></tr></table>
+
+
+( Notes. *Significant at ≤0.05; **Significant at ≤0.01; ***Significant at ≤0.001.
+
+Ranking is based on the proportion (in parentheses) of within-country variation in the index uniquely explained by the predictor in question, measured using the Stata 12 semipartial $R^2$ statistic.)
+
+---
+
+Jean-FrédéRIC Morin and E. Richard Gold
+
+789
+
+2000–2008 in Table 1 and during the full period of 1995–2008 in Table 2 provides support for the role of regional competition in facilitating the adoption of USstyle IP rules. It suggests that when a country moves its IP protection closer to that of the United States, its neighbors try to catch up, most likely in the hope of avoiding the flight of capital or to remain attractive for further FDI. It is unlikely that they would have done so otherwise, considering their level of economic and technological development.
+
+Second, the regression analysis suggests that socialization mechanisms have a strong and positive effect on a country's Transplant Index score. Table 1 shows that both Population Studying in the US and IP Training are significantly associated with increases in the Transplant Index in both the 1995–1999 and 2000–2008 periods. Table 2 not only confirms this over the entire 1995–2008 period but shows that the combined socialization variables (that is, adding IP Training and Population Studying in the US ) outrank every other mechanism in terms of strength.
+
+This finding on the power of socialization to increase levels of IP protection supports existing case studies. According to an Oxfam study, the Guatemalan congress increased the level of IP protection beyond the minimal requirement of the TRIPs agreement under the guidance of US-funded legal advice (Oxfam 2002:9). Further case studies have documented a similar impact of IP capacitybuilding programs in other states such as Thailand, the Philippines, Nigeria, and West African countries (Deere 2009; Drahos 2010). Our quantitative analysis suggests that these cases are not deviant, but representative of IP socialization in developing countries.
+
+To the extent that GDP per Capita captures emulation — as this variable picks up other factors, we refrain from making stronger claims — the regression analysis is consistent with our assumption presented earlier that, while emulation is sometimes at work in facilitating the transplantation of US-style IP rules, it is far from the primary explanation of those transplantations. For example, GDP per Capita did not reach significance in the 1995 – 1999 period in Table 1 . On the other hand, GDP per Capita took on significance once socialization, regional competition, and contractualization took hold. This is illustrated not only in the significance of its effect in the 2000 – 2008 period in Table 1 , but in its overall strength in the full model in Table 2 , where it is third to contractualization and socialization (combining the strength values of IP Training and of Population Studying in the US ). One interpretation of these findings — subject to the same caveat above — is that by 2008, socialization had become so effective as to reduce developing countries' perceived asymmetries with the United States. Thus, those countries would consider emulating higher IP rules under the belief that those rules were as appropriate for them. This would account for both GDP per Capita's lack of significant effect in 1995 – 1999 — during which asymmetric interests were top of mind — and its significant effect in 2000 – 2008 when those asymmetries had lesser visibility. This result adds subtlety to previous studies in which GDP per Capita had been examined in isolation from the other variables studied here (Chen and Puttitanun 2005) .
+
+## H3: Interrelationship Between Contractualization, Socialization, and Regulatory Competition
+
+Our analysis provides partial support for the third hypothesis that each of contractualization, socialization,
+
+and regulatory competition are more prominent in the presence of the others, leading to a self-reinforcing cycle of transplantation of US-style IP rules.
+
+First, regression analyses — the results of which are illustrated in Table 3 below — suggest that the signing of a FTA leads to an increase in the volume of US-sponsored IP Training events directed at the country. It remains unclear whether it is the FTA itself that directly induces US training providers to increase their domestic presence, or whether the FTA acts to indirectly convince domestic authorities to welcome US offers of training in order to more aggressively attract FDI. In either case, a FTA seems to increase socialization opportunities.
+
+The regression analysis also illustrates the symbiotic relationship between contractualization and regional competition. Indeed, Table 4 below shows a peak in competitive activity in 1998 – 2000. This was the time that most countries were revising their laws to meet the 2000 deadline for developing countries to implement TRIPs within their domestic legal frameworks. Once those countries engaged in the process of revising their IP laws, they apparently took the opportunity to strategically position their IP rules so as to be competitive with others in the region, often surpassing the requirements of TRIPs. In the 1998 – 2000 period, Regional Top Score is one of only two variables (the other being Population Studying in the US ) that is significantly associated with the Transplant Index and its effect is markedly stronger than all other mechanisms put together. Thus, it seems that the contractual obligations set out in TRIPs provided a platform for jockeying for strategic advantage through further increases in the level of IP protection. Countries used the premise of contractualization to engage in competitive rule-making.
+
+In turn, developing countries seem to compete to secure a contractual agreement with the United States. Decisionmakers may believe that a FTA with the United States would make their countries more attractive for FDI and boost their exports. As a result, several developing countries now seek to enter into FTA negotiations with the United States, despite knowing that they will be required to comply with unfavorable IP provisions that they likely would have rejected in a different context. Tellingly, once the United States signs a FTA with a country, that country's neighbors often indicate an interest in signing a similar agreement. The Canada–US agreement of 1989 famously prompted Mexico to propose to the United States the negotiation of a similar FTA, which became trilateral at the request of Canada, providing the United States with an opportunity to diffuse its IP rules across all of North America. Likewise, in 1994, Trinidad
+
+T ABLE 3. The Influence of Variables on Intellectual Property (IP) Training (Fixed Effects)
+
+<table><tr><td colspan="2">IP Training (1694 Observations; 121 Groups)</td></tr><tr><td>Variable</td><td>Coefficient</td></tr><tr><td>GDP Per Capita</td><td>0.004***</td></tr><tr><td>Priority Watchlist</td><td>14.954***</td></tr><tr><td>GSP Review</td><td>0.107</td></tr><tr><td>US Bilateral Agreement</td><td>22.485**</td></tr><tr><td>Pop. Studying in US</td><td>5.935***</td></tr><tr><td> $R^{2}$ </td><td>0.27</td></tr><tr><td> $F_{5,120}$ </td><td>16.92***</td></tr></table>
+
+
+(Notes. *Significant at ≤0.05; **Significant at ≤0.01; ***Significant at ≤0.001.)
+
+---
+
+790
+
+An Integrated Model of Legal Transplantation
+
+T ABLE 4. The Influence and Strength of Variables on the Transplant Index for 1998-2000
+
+<table><tr><td>Variable</td><td>Coefficient</td><td>Rank of Strength</td></tr><tr><td>GDP Per Capita</td><td>-0.0001</td><td> $4(0.22 \%)$ </td></tr><tr><td>Priority Watchlist</td><td>0.068</td><td> $5(0.03 \%)$ </td></tr><tr><td>GSP Review</td><td>0.093</td><td> $5(0.03 \%)$ </td></tr><tr><td>US Bilateral Agreement</td><td>Dropped</td><td>Dropped</td></tr><tr><td>IP Training</td><td>0.016</td><td> $2(1.51 \%)$ </td></tr><tr><td>Pop. Studying in US</td><td>0.173*</td><td> $3(0.55 \%)$ </td></tr><tr><td>Regional Top Score</td><td>0.491***</td><td> $1(7.36 \%)$ </td></tr><tr><td> $R^{2}$ </td><td>0.16</td><td></td></tr><tr><td> $F_{6,111}$ </td><td>5.53***</td><td></td></tr></table>
+
+
+( Notes. *Significant at ≤ 0.05; **Significant at ≤ 0.01; ***Significant at ≤ 0.001. 2Ranking is based on the proportion (in parentheses) of within-country variation in the index uniquely explained by the predictor in question, measured using the Stata 12 semipartial $R^2$ statistic.)
+
+and Tobago concluded a bilateral IP agreement with the United States to avoid marginalization after a fellow Caribbean country, Jamaica, signed an investment treaty with the United States (Heron 2004). More recently, the Dominican Republic requested to join negotiations between the United States and central American countries, which included a chapter on IP, in order to maintain an access to the US market that was equal to that of its central American competitors (GAO 2004:12).
+
+As the United States can essentially select the countries with which to enter into a FTA, developing countries interested in acquiring a FTA must present themselves as reliable partners. In this context, offering a relatively high level of IP protection, even before FTA negotiations begin, could be a strategy for countries that aspire to a FTA, particularly if the IP changes would be required by the FTA in any case. The Dominican Republic increased, for example, its IP protection when it had difficulty convincing the United States to let it join ongoing negotiations with central American countries (GAO 2004:44–45). Similarly, Taiwan duplicated some US standards with the explicit hope that those reforms would facilitate negotiation of a full-fledged FTA (Smith and Rugaber 2002). Noticing those behaviors, the US government concluded that its “willingness to pursue bilateral FTAs has bolstered countries’ interest [in adopting US-style IP rules] and encouraged them to make the changes necessary to enter into FTA negotiations” (GAO 2004:11).
+
+Socialization can lead, in turn, to a re-evaluation of past trade concessions in light of newly acquired norms: Increases in IP protection may no longer be seen as compromises but as independently beneficial. Under this perspective, increased socialization, generated as a result of contractualization, can favor further rounds of contractualization. Tellingly, the United States Agency for International Development noted that capacity-building programs on IP can “ increase readiness to enter into free trade agreements ” (USAID 1998:4). Although a regression analysis cannot provide causal evidence to support this claim, it should be noted that several American FTA partners that were previously not known for advocating strong IP protection, including Jordan, Korea, Mexico, Morocco, Singapore, and the United Arab Emirates, have since joined the United States in the small group of countries that negotiated the Anti-Counterfeiting Trade Agreement (ACTA), a multilateral agreement providing
+
+high standards of IP enforcement. While those countries may also hope to receive contractual benefits in other fields, they already enjoy free access to the US market and one of the few plausible explanations for their participation in the ACTA negotiations is that a greater share of their socialized elite came to believe that IP protection is beneficial for their societies.
+
+Just as the boundary between coercion and contractualization is blurred, so too are those separating the activities of contractualization, socialization, and regulatory competition. Table 1 shows, for example, that socialization and regional competition operated in 1995–1999, during the period in which coercion was in decline, but prior to the post-TRIPs contractualization wave. Both continued once contractualization started. This complex interrelationship among the mechanisms illustrates the limits of any theory that does not address multiple mechanisms at once.
+
+## Conclusions
+
+This article contributes to the literature in a number of ways. It bridges the so-far unconnected literatures on legal transplants and policy diffusion by combining systematic legal comparisons with political causal explanations. It also articulates one of the first integrated understandings of the five discussed causal mechanisms, enabling a more complete and complex picture of the process of legal transplantation in the context of asymmetric interests.
+
+Furthermore, this article introduces an original and much needed index of IP protection, the Transplant Index , and a new indicator of regional competition, Regional Top Score . Our models, based on this index, provide strong empirical evidence—corroborating anecdotal examples in the literature—of (i) the declining effectiveness of coercion, (ii) the actual legal implementation of bilateral agreements, (iii) regional races to the top in IP standards setting, and (iv) the impact of US-sponsored capacity-building activities on domestic legislation.
+
+Three main conclusions follow from our findings. First, several causes can produce a similar effect. This amounts to a truism of social science, but scholars sometimes forget it in the context of quantitative analyses. As Dobbin et al. (2007:463) conclude in their review of the policy-diffusion literature, several researchers “ test only their own theory or simply show evidence of diffusion and impute that their favored mechanism is at work. ” This article demonstrates, in contrast, the significant contributions of combining several causal mechanisms.
+
+Second, acknowledging multicausality does not mean that the different causal mechanisms operate simultaneously, equally, or independently. Rather, this article suggests that the mechanisms of legal transplantation are historically conditioned. Power structures, the ideational environment, and the institutional context condition the availability of mechanisms and their ability to facilitate the process of legal transplantation. Coercion, for example, while an effective tool for the United States in the early 1990s, became less effective after the implementation of WTO agreements.
+
+Third, this article highlights how different causal mechanisms interrelate with one another. While each of the mechanisms studied may achieve a similar result, this does not imply that they are mere substitutes for one another, deployed according to the tastes of more powerful coun-
+
+---
+
+Jean-FrédéRIC Morin and E. Richard Gold
+
+791
+
+tries; rather, the mechanisms act in concert with one another and achieve maximum effect when combined in particular sequences so as to collectively drive continuous legal transplantation. For example, contractualization favors both regulatory competition—especially among FTA candidates—and socialization—through increased training programs. One possible effect of the sequencing of these mechanisms is that elites in developing countries may become so socialized to US norms that they cease to perceive the asymmetry of interest with the United States. Should this occur, the scope condition of this study—asymmetric interests—may lift, leading to a greater role for emulation. As the discussion at the end of hypothesis H2 suggests, this may actually be happening.
+
+Although a study of counter-hegemonic resistance exceeds the scope of this article, we note that both state and nonstate forces — in both developed and developing countries — resist the dynamic of legal transplantation (Morin 2014). In the past, those forces obliged the US government to change its strategy from coercion to contractualization. More recently, they may have forced Washington, along with some European governments, to pause the continuous expansion of IP protection within their own domestic systems. Nevertheless, these counter-hegemonic forces seem unable, so far, to reverse the current dynamic toward greater levels of legal transplantation of IP rules in developing countries.
+
+## References
+
+ATKINSON, CAROL. (2010) Does Soft Power Matter? A Comparative Analysis of Student Exchange Programs 1980-2006. Foreign Policy Analysis 6(1): 1-22.
+
+BACH, DAVID, AND ABRAHAM L. NEWMAN. (2010) Transgovernmental Networks and Domestic Policy Convergence. International Organization 64(3): 505–526.
+
+Baldwin, Richard. (1997) The Causes of Regionalism. World Economy 20 (7): 865-867.
+
+BERROWTTZ, DANIEL, KATHARINA PISTOR, AND JEAN-FRANCOIS RICHARD. (2003) The Transplant Effect. American Journal of Comparative Law 51(1): 163-203.
+
+BOURDIEU, PIERRE. (1995) La Cause de la Science. Actes de la Recherche en Sciences Sociales 106: 3-10.
+
+BRAITHWAITK, JOHN, AND PETER DRAHOS. (2000) Global Business Regulation. Cambridge: Cambridge University Press.
+
+BRUNNÉR, JUTTA, AND STEPHEN J. TOOPE. (2010) Legitimacy and Legality in International Law: An Interactional Account. Cambridge: Cambridge University Press.
+
+CAO, XUN. (2009) Network of Intergovernmental Organizations and Convergence in Domestic Economic Policies. International Studies Quarterly 53(4): 1095-1130.
+
+CHICKEL, JEFFLEY T. (2005) International Institutions and Socialization in Europe: Introduction and Framework. International Organization 59(4): 801-826.
+
+CHEN, YONGMIN, AND THITIMA PUTTITANUN. (2005) Intellectual Property Rights and Innovation in Developing Countries. Journal of Development Economics 78(2): 474-493.
+
+DEKRE, CAROLYN. (2009) The Implementation Game: The TRIPs Agreement and the Global Politics of Intellectual Property Reform in Developing Countries. Oxford: Oxford University Press.
+
+DOBBIN, FRANK, BETH SIMMONS, AND GEOFFREY GARRETT. (2007) The Global Diffusion of Public Policies: Social Construction, Coercion, Competition, or Learning? Annual Review of Sociology 33: 449-472.
+
+DOLOWITZ, DAVID P., AND DAVID MARSH. (2000) Learning From Abroad: The Role of Policy Transfer in Contemporary Policy-Making. Governance 13(1): 5-24.
+
+DRAHOS, PETER. (2010) The Global Governance of Knowledge: Patent Offices and Their Clients. Cambridge: Cambridge University Press.
+
+DREZNER, DANIEL W. (2001) Globalization and Policy Convergence. International Studies Review 3(1): 53-78.
+
+ELEINS, ZACHARY, ANDRAW T. GTZMAN, AND BETH A. SIMMONS. (2006) Competing for Capital: The Diffusion of Bilateral Investment Treaties, 1960-2000. International Organization 60(4): 811-846.
+
+FESTINGER, LEON. (1957) A Theory of Cognitive Dissonance. Stanford, CA: Stanford University Press.
+
+FINNEMORE, MARTHA, AND KATHRYN SIKKINK. (1998) International Norm Dynamics and Political Change. International Organization 52(4): 887-917.
+
+GAO. (2004) Intensifying Free Trade Negotiating Agenda Calls for Better Allocation of Staff and Resources. GAO-04-223. Washington, DC: General Accounting Office.
+
+GENARTE, JUAN C., AND WALTER G. PARK. (1997) Determinants of Patent Rights: A Cross-National Study. Research Policy 26(3): 283-301.
+
+GREENHILL, BRIAN. (2010) The Company You Keep: International Socialization and the Diffusion of Human Rights Norms. International Studies Quarterly 54(1): 127-145.
+
+HERON, TONY. (2004) The New Political Economy of United States-Caribbean Relations. Farnham, UK: Ashgate.
+
+HOLZINGER, KATHARINA, AND CHRISTOPH KNILL (2005) Causes and Conditions of Cross-National Policy Convergence. Journal of European Public Policy 12(5): 775–796.
+
+IRENBERRY, JOHN G., AND CHARLES A. KUPCHAN. (1990) Socialization and Hegemonic Poser. International Organization 44(3): 283-315.
+
+KAPCZYNSKI, AMY. (2008) The Access to Knowledge Mobilization and the New Politics of Intellectual Property. Yale Law Journal 117(5): 804885.
+
+KELLEY, JUDITH. (2004) International Actors on the Domestic Scene: Membership Conditionality and Socialization by International Institutions. International Organization 58(3): 425-457.
+
+Kim, Yin: Kyoung, Kun Lee, Walter G. Park, and Kinchung Catoo. (2012) Appropriate Intellectual Property Protection and Economic Growth in Countries at Different Levels of Development. Research Policy 41(2): 358-375.
+
+LEGRAND, PIERRE. (2001) What Legal Transplants? In Adapting Legal Cultures, edited by David Nelken and Johannes Feest. Oxford: Hart.
+
+LIKHOVSKI, ASSAF. (2009) Argonauts of the Eastern Mediterranean: Legal Transplants and Signaling. Theoretical Inquiries in Law 10(2): 619–651.
+
+MATTEL, UGO. (2003) A Theory of Imperial Law: A Study on U.S. Hegemony and the Latin Resistance. Global Jurist Frontiers 3(2): 161.
+
+MATTHEWS, DUNCAN, AND VIVIANA MUNOZ-TKILIZZ. (2006) Bilateral Technical Assistance and TRIPs: The United States, Japan and the European Communities in Comparative Perspective. Journal of World Intellectual Property 9(6): 629-653.
+
+MAY, CHRISTOPHER. (2004) Capacity Building and the (Re)Production of Intellectual Property Rights. Third World Quarterly 25(5): 821837.
+
+MAY, CHRISTOPHER, AND SUSAN K. SALL. (2006) Intellectual Property Rights: A Critical History. Boulder, CO: Lynne Rienner.
+
+MERTHIA, ANDREW, AND ROBERT PAIRR. (2005) Patently Misleading: Partial Implementation and Bargaining Leverage in Sino-American Negotiation on Intellectual Property Rights. International Organization 59(3): 695-729.
+
+MISSEGUER, COVADONGA. (2004) What Role for Learning? The Diffusion of Privatization in OECD and Latin American Countries. Journal of Public Policy 24(3): 299-325.
+
+MILLER, JONATHAN M. (2003) A Typology of Legal Transplants: Using Sociology, Legal History and Argentine Examples to Explain the Transplant Process. American Journal of Comparative Law 51(4): 839885.
+
+MONTHISQUIKU, CHARLES DE SUCONDAT. (1961) DE L'Ésprit des Lois. Paris: Garnier Frères.
+
+MORIN, JRAN-FRÉDÉRIC. (2014) Paradigm Shift in the Global IP Regime: The Agency of Academics. Review of International Studies Quarterly 21 (2): 275–309.
+
+MORIN, JEAN-FRÉDÉRIC, AND GILBERT GAGNE. (2007) What can Best Explain the Prevalence of Bilateralism in the Investment Regime. International Journal of Political Economy 36(1): 53-74.
+
+OSTERGARD, ROBERT L. (2000) The Measurement of Intellectual Property Rights Protection. Journal of International Business Studies 31 (2): 349-360.
+
+OXFAM. (2002) US Bullying on Drug Patents: One Year After Doka. Oxford: Oxfam.
+
+---
+
+792
+
+An Integrated Model of Legal Transplantation
+
+PELC, KRZVSZTOR. (2010) Constraining Coercion? Legitimacy and Its Role in US Trade Policy, 1975-2000. International Organization 64(1): 65-96.
+
+RADARLLI, CLAUDIO M. (2004) The Puzzle of Regulatory Competition. Journal of Public Policy 24(1): 1-23.
+
+ROSE, RICHARD. (1991) What Is Lesson-Drawing? Journal of Public Policy 11(1): 3-30.
+
+SCHIMMELFENNIG, FRANK. (2000) International Socialization in the New Europe: Rational Action in an Institutional Environment. European Journal of International Relations 6(1): 109-139.
+
+SCHIMMELFENNIG, FRANK, AND ULRICH SCHELMEER. (2004) Governance by Conditionality: EU Rule Transfer to the Candidate Countries of Central and Eastern Europe. Journal of European Public Policy 11(4): 661-679.
+
+SELL, SUSAN K., AND ASEEM PRAKASIL (2004) Using Ideas Strategically: The Contest Between Business and NGO Networks in Intellectual Property Rights. International Studies Quarterly 48(1): 143–175.
+
+SHADLEN, KENNETH C., ANDREW SCHIRANK, AND MARCUS J. KURTZ. (2005) The Political Economy of Intellectual Property Protection. International Studies Quarterly 49(1): 45-71.
+
+SHARMAN, JASON C. (2008) Poser and Discourse in Policy Diffusion: Anti-Money Laundering in Developing States. International Studies Quarterly 52(3): 635-656.
+
+SHERWOOD, ROBERT M. (1997) Intellectual Property Systems and Investment Stimulation: The Rating of Systems of Eighteen Developing Countries. IDEA 37: 261-370.
+
+Satt, Wici. (2010) Globalization and Indigenization: Legal Transplant of a Universal TRIPS Regime in a Multicultural World. American Business Law Journal 47(3): 455-507.
+
+SIL, RUDRA, AND PETER J. KATZENSTEIN. (2010) Analytic Electism in the Study of World Politics. Perspectives on Politics 8(2): 411-431.
+
+Simmons, Betti A., and Zachary Elkins. (2004) The Globalization of Liberalization: Policy Diffusion in the International Political Economy. American Political Science Review 98. (I): 171-189.
+
+SMITH, NOAH J., AND CHRISTOPHER S. RUGAHER. (2002) USTR Holds Intellectual Property Talks With Taiwan, Setting Off Talk of Possible FTA. International Trade Reporter Oct 17th:41.
+
+USAID. (1998) The USAID Fiscal Year 1998 Congressional Presentation— Central American Programs. Washington, DC: USAID.
+
+USPTO. (2010) Training and Education: Global Intellectual Property Academy. Washington, DC: USPTO. Available at www.uspto.gov/ip/training/ index.jsp (Accessed November 28, 2012).
+
+USTR. (1997) USTR Announces Results of Special 301 Annual Review , Press Release 97–37. Washington, DC: USTR.
+
+USTR. (2004) The Work of USTR—Intellectual Property. Washington, DC: USTR, www.bilaterals.org/spip.php?article709 (Accessed November 28, 2012).
+
+VOGEL, DAVID. (1995) Trading Up: Consumer and Environmental Regulation in a Global Economy. Cambridge: Harvard University Press.
+
+WATSON, ALAN. (1974) Legal Transplants: An Approach to Comparative Law. Edinburgh: Scottish Academic Press.
+
+ZENG, K. (2002) Trade Structure and the Effectiveness of America's Aggressively Unilateral Trade Policy. International Studies Quarterly 46 (1): 93-115.
+
+## Supporting Information
+
+Additional Supporting Information may be found in the online version of this article:
+
+Appendix S1. Full dataset.
+
+Appendix S2. Robustness and error distribution.
+
+Appendix S3. Sample.
+
+Appendix S4. Transplant index by country per year.
+
+Appendix S5. Variable definitions and data sources.
+
+Appendix S6. Variable test.
+
+---
+
+Copyright of International Studies Quarterly is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
+

processed/00120_W2164875280/document.md

@@ -0,0 +1,259 @@
+Genetics inMedicine
+
+ORIGINAL RESEARCH ARTICLE
+
+©American College of Medical Genetics and Genomics
+
+# “Trust is not something you can reclaim easily”: patenting in the field of direct-to-consumer genetic testing
+
+Sigrid Sterckx, PhD1,2, Julian Cockbain, DPhil3, Heidi Howard, PhD4,
+
+Isabelle Huys, PhD5,6 and Pascal Borry, PhD7
+
+Purpose: Recently, 23andMe announced that it had obtained its first patent, related to “polymorphisms associated with Parkinson's disease” (US-B-8187811). This announcement immediately sparked controversy in the community of 23andMe users and research participants, especially with regard to issues of transparency and trust. The purpose of this article was to analyze the patent portfolio of this prominent direct-to-consumer genetic testing company and discuss the potential ethical implications of patenting in this field for public participation in Web-based genetic research.
+
+Methods: We searched the publicly accessible patent database Espacenet as well as the commercially available database Micropatent for published patents and patent applications of 23andMe.
+
+Results: Six patent families were identified for 23andMe. These included patent applications related to: genetic comparisons between
+
+grandparents and grandchildren, family inheritance, genome sharing, processing data from genotyping chips, gamete donor selection based on genetic calculations, finding relatives in a database, and polymorphisms associated with Parkinson disease.
+
+Conclusion: An important lesson to be drawn from this ongoing controversy seems to be that any (private or public) organization involved in research that relies on human participation, whether by providing information, body material, or both, needs to be transparent, not only about its research goals but also about its strategies and policies regarding commercialization.
+
+Genet Med 2013:15(5):382-387
+
+Key Words: direct-to-consumer; ethics; genetic testing; patenting
+
+## INTRODUCTION
+
+In recent years, direct-to-consumer (DTC) genetic testing companies have been advertising and offering genetic tests directly to the public. 1 Some have sought to increase the value of their “biobanks” by asking customers to provide phenotypic information. 2 Various commentators have described participant-centric research initiatives, common features of which include voluntary and active participation, use of social media technology, active interaction between participants and researchers, and the appeal to promote public goods (e.g., scientific progress). 3–5
+
+23andMe is a DTC genetic testing company that provides its consumers with the opportunity to consent “to the use of their data for research.”6 Consumers are “given the option of contributing phenotype data via a series of Web-based surveys. The result is a single, continually expanding cohort, containing a self-selected set of individuals who participate in multiple studies in parallel.”6
+
+Within participant-centric research, 23andMe has focused on genetic and phenotypic correlations with Parkinson disease (PD), sarcoma, and myeloproliferative neoplasm. 2 23andMe states on its website that letting consumers participate in research in this way “can produce revolutionary findings that will benefit us all,” stimulating consumers to "direct research by participating in studies of conditions and traits you care about,” and to “join an effort to translate basic research into improved health care for everyone.” 7
+
+These encouragements to customers to advance research for the public good, however, stand in contrast with 23andMe's announcement on 28 May 2012 that it was to be granted a US patent—patent 8187811, for "polymorphisms associated with Parkinson's disease"'—the very next day. CEO Anne Wojcicki announced on the company website (“The Spittoon”) that the goal of the patent was to ensure that the underlying research could lead “towards successful translation of this discovery.” 8
+
+The announcement immediately sparked controversy on the Spittoon website among users and participants, with comments on many topics: the patentability of genes; the link between patents and medical advances; 23andMe's potential (ab)use of the patent, e.g., to charge royalties or block the performance of PD genetic tests; the lack of communication of 23andMe on its intention to patent discoveries; and the mismatch between applying for patents and the avowed mission of democratizing genomics. 8 These reactions suggest that the knowledge that 23andMe had sought a patent based on its participant-centric research could undermine trust in the company.
+
+A major criticism concerned the lack of transparency regarding any intention to patent discoveries related to PD. Therefore, we aimed to study which patent applications have been filed by 23andMe. This article presents the results of an analysis of the company's patent portfolio and discusses some of the potential
+
+1Bioethics Institute Ghent, Universiteit Gent, Gent, Belgium; 2Department of Philosophy and Moral Sciences, Vrije Universiteit Brussel, Brussel, Belgium; 3Consultant European Patent Attorney, Oxford, UK; 4Faculté de Médecine, Département d'Épidémologie et de Santé Publique, Université Toulouse III—Paul Sabatier, INSERM UMR U1027, Toulouse, France; 4Centre for Pharmaceutical Care and Pharmaco-Economics, Faculty of Pharmaceutical Sciences, Katholieke Universiteit Leuven, Leuven, Belgium; 5Centre for Intellectual Property Rights, Katholieke Universiteit Leuven, Leuven, Belgium; 6Centre for Biomedical Ethics and Law, Katholieke Universiteit Leuven, Leuven, Belgium. Correspondence: Pascal Borry (Pascal.Borry@med.kuleuven.be)
+
+Submitted 6 July 2012; accepted 1 October 2012; advance online publication 22 November 2012. doi:10.1038/gim.2012.143
+
+382
+
+Volume 15 | Number 5 | May 2013 | GENETICS in MEDICINE
+
+---
+
+Patenting in the field of DTC genetic testing | STERCKX et al
+
+ORIGINAL RESEARCH ARTICLE
+
+ethical implications of patenting in the context of public participation in Web-based genetic research.
+
+## MATERIALS AND METHODS
+
+We searched the European Patent Office's database Espacenet and the commercial database Micropatent for published patents and applications that showed 23andMe as the applicant or patentee. Espacenet provides access to more than 70 million patent documents. Our search revealed patents and applications accessible on 1 September 2012. Applications filed less than 18 months before, or filed independently by other companies working with 23andMe, would not have been revealed. We examined patent “families,” which contain patents and applications related to a single invention. For each family, we provide the following basic information: priority application filing date, patent or application publication numbers, and a summary of the broadest method claims.
+
+## RESULTS
+
+Six patent families were identified ( Table 1 ). Although our focus is on the PD patent family (family 6), this has not been 23andMe's only foray into patenting. Obviously, for a company involved in DTC genomic testing, it is reasonable that, without customer consent, they might seek to patent improvements in sample handling, sample testing, data analysis, and data presentation. Patent families 1, 2, 3, and 5 seem to fall in these categories. Family 4, which is concerned with a method of selecting a
+
+sperm or egg donor to maximize the chances of having a baby with desired phenotypic characteristics, raises many ethical questions but is not founded on participant-centric research and will not be commented on further here.
+
+The international application in patent family 6 (“polymorphisms associated with Parkinson's disease”) includes a claim directed to a method for screening a human subject for susceptibility to PD based on the determination of certain alleles. The nucleic acid primers or probes used in such analyses are claimed, as is a kit for assaying for PD susceptibility. In another claim (claim 10), however, we see 23andMe seeking to cover possible downstream inventions in PD therapy.
+
+## DISCUSSION
+
+Intellectual property (IP) protection is a well-established practice that aims to promote technological progress and investment. Patenting is common practice for any technology-based industry, but in the (bio)pharmaceutical sector, patents are considered vital, for example, to raise venture capital or justify further investment and as a way to transfer results of publicly funded research to commercially viable applications. 9–11 Although in the context of genetic diagnostics, patent offices have granted several patents of doubtful and disputed validity, 9–12 the mere fact that 23andMe has participated in IP-protecting activities since 2007, shortly after it was formed, seems quite normal.
+
+Although 23andMe has filed various patent applications, until now it has only drawn the attention of its consumers to
+
+Table 1 23andMe's patent families published by 1 September 2012
+
+<table><tr><td>Family no.</td><td>Earliest priority date</td><td>Patent/patent application publication nos.</td><td>Summary of the broadest method claim</td></tr><tr><td rowspan="2">1</td><td>15 October 2007</td><td>WO2009/051749; US-A-2009/118131</td><td>A method of comparing genetic information from a grandparent and a (presumed) grandchild involving calculating the similarity of the two sets of information and displaying this graphically and in color.</td></tr><tr><td></td><td>WO2009/051766; US-A-2009/119083</td><td>A method of comparing genetic information from two individuals involving comparing their genetic information and displaying this graphically using different symbols to represent genes that are identical and those that are “half”-identical.</td></tr><tr><td>2</td><td>15 October 2007</td><td>WO2009/051768; US-A-2009/112871</td><td>A method of sharing data between two data records that involves, on request from one data record holder, providing read-access to that requestor to selected data areas of non-public data from the other record.</td></tr><tr><td>3</td><td>26 August 2008</td><td>WO2010/024894; US-A-2010/057374; US-A-2010/057807</td><td>A method of providing a merged genetic information data set for an individual that consists of merging two or more genetic data information data sets for that individual and deciding, where there is a conflicting overlap, which input to use.</td></tr><tr><td>4</td><td>05 December 2008</td><td>WO2010/065139; US-A-2010/145981</td><td>A method of identifying a preferred gamete (sperm or ovum) donor to achieve a desired phenotype in the offspring of a particular recipient and a donor, by comparing the genetic data of the recipient and those of a set of possible donors and selecting the donor whose gametes are most likely to combine with those of the recipient to produce offspring with the desired phenotype.</td></tr><tr><td>5</td><td>31 December 2008</td><td>WO2010/077336; US-A-2010/223281; EP-A-2370929</td><td>A method of determining a relative relationship between two individuals that includes comparing genetic data for the two, calculating a predicted degree of genetic relationship between the two, and advising one if the other individual is likely to be a relation.</td></tr><tr><td>6</td><td>30 November 2009</td><td>WO2011/065982; US-A-2011/130337; US-B-8187811; CA-A-2782207</td><td>A method of screening an individual for susceptibility to Parkinson disease that includes determining whether the individual’s DNA has particular abberrations (single-nucleotide polymorphisms) at one or more of seven specific locations.</td></tr></table>
+
+
+GENETICS in MEDICINE | Volume 15 | Number 5 | May 2013
+
+383
+
+---
+
+ORIGINAL RESEARCH ARTICLE
+
+STERCKX et al | Patenting in the field of DTC genetic testing
+
+one patent case. $^8$ Moreover, the communication was made the day before the US patent was granted, even though the initial application was filed in 2009, and the research results were published in PLoS Genetics in 2011 6 and then rapidly drawn to the attention of its consumers. The delay in drawing attention to the patent application seems odd, given that 23andMe even recently underlined that “open dialogue about complicated issues like patents is important” and that it wanted to be “as open as possible about our intentions, including letting people know about our patent and why we have filed it.” $^8$
+
+How likely is it that these events might result in a loss of trust, and why is the issue of trust relevant in the first place? It seems that the strong position 23andMe has enjoyed so far in motivating customers to participate in research by providing phenotypic data may be weakened. Presumably motivated at least in part by altruism, paying customers have contributed to an increase of the informational value of the company's database by providing data. Altruism, confidence in scientific progress, and trust in research and researchers are quite common persuaders to induce participation in biobank research. 15 Studies 14–18 have shown that individuals donating biological samples and phenotypic information consider this to be an altruistic act. More relevant in this context is the importance of the donors' trust in the research and researchers, and the reputation of the entity planning the research. 14,15,18 23andMe appears to have been successful in meeting these expectations. The fact that it provides participants with their personal genetic information may also be a strong motivating factor.
+
+Now that it is clear that 23andMe is seeking patents, it is possible that various customers will withdraw their support because they do not consider such activities to be in line with their altruistic participation in the research projects. As one customer wrote on 23andMe's blog after the announcement of the patent: "this is simply crowd-sourced greed. As a longtime 23andMe customer, this patent is extremely disappointing and alarming. Our family is done with your service." $^8$
+
+In the context of biobanks, it has also been reported that donors are concerned with social fairness. Studies 19–23 have shown public distrust of for-profit companies in the context of biobanking, as many participants consider a profit motive to be at odds with their altruistic aim in participating. The point is not that there is anything inherently wrong in making a profit or using DTC genetic testing to create revenues, but rather that this may be perceived as conflicting with the open, altruistic, science-driven, and common-good image that 23andMe has clearly been trying to create. As one correspondent commented: “ I would not have talked my mother and others in my support group into participating if I had understood this was going to be a profit-driven enterprise. I believe 23andMe has been disingenuous in gathering a free database. ” 8
+
+This is not a case of 23andMe failing to meet the explicit expectations of participants; it did (by finding the biomarkers). Rather, this is a case of promising to build something with communal resources, building it, and then claiming ownership and (potentially) charging for access. The implied suggestion
+
+that the result would be a community good was misleading. An analogy might help: a company in a village next to a river says: “the village needs a bridge, give us the wood and we’ll build it”; the wood is given; the bridge is built; but the company charges a toll. In both cases, the contributors (the research participants/villagers) did not realize that contribution did not guarantee public ownership. The fault lies not in 23andMe/ the builder owning the result, but in the lack of transparency in the appeal for the necessary contributions. The contributors did not understand what was going on until after their contribution was made, and, had they understood, many might not have contributed. Having been misled, contributors may in the future be less likely to contribute to the attainment of public goods, fearing that they might not be public after all, which, in turn, might lead to a more morally impoverished community.
+
+The trust issue is not only related to the nature of the goals the company is pursuing (profit driven or not), but also to the extent of transparency surrounding the company's strategies. As far as the latter is concerned, the question of whether the participants had given truly informed (and thus valid) consent is clearly regarded as crucial by various correspondents. For example: “It would seem that the ethics of one company profiting from the knowledge of others because it patented a gene variant could do with some scrutiny, especially if it turns out that patients, who provided samples ..., were not aware that the results would be patented.” $^8$
+
+23andMe responded as follows to this comment: “We make reference to our intent to pursue intellectual property rights for discoveries made from our research in both [our] terms of service ... and in our research consent document ....”3 The relevant passages from the terms of service and consent documents mention that 23andMe might develop intellectual property and that participants have no right to share in any profits. The terms of service state: “By submitting ... user content, you give 23andMe ... a perpetual ... license to ... create derivative works from any user content you submit .... You acquire no rights in any research or commercial products that may be developed by 23andMe .... You specifically understand that you will not receive compensation for any research or commercial products ....”24 The consent document provides: “If 23andMe develops intellectual property and/or commercializes products or services, directly or indirectly, based on the results of this study, you will not receive any compensation.”25
+
+However, the word “patent” itself is only used in the context of information presented to the users. As stated in the terms of service: “You agree that 23andMe ... own all legal right, title, and interest in and to the services, including any intellectual property rights which subsist in the services .... You further acknowledge and agree that the services ... contain proprietary and confidential information that is protected by applicable intellectual property ... laws. You further acknowledge and agree that information presented to you through the services ... is protected by ... patents ....” 24
+
+The wording used by no means makes it clear that patents would be sought for the research results. Various users indicated
+
+384
+
+Volume 15 | Number 5 | May 2013 | GENETICS in MEDICINE
+
+---
+
+Patenting in the field of DTC genetic testing | STERCKX et al
+
+## ORIGINAL RESEARCH ARTICLE
+
+that they were unaware that 23andMe was planning to apply for patents, whereas, as noted by one of the bloggers: “everyone coming to [23andMe's] service, either by paying it or by funded invitation ... needs to know clearly what this is about and make their own informed decision to join or not.” 8
+
+Based on these reactions, it is clear that the consent procedure currently used is ethically inadequate, 26 especially in relation to patents. The reactions to the PD patent show the limitations of the use of a vague and unclear consent when there are potential commercial applications. We do not suggest that 23andMe has done the research without consent; rather the issue is whether the consent extended to cover the patenting of results. Participants may consent to donate biological materials and phenotypic data for the development of clinical applications. However, if they are not aware that this might be happening through commercialization involving patents, this might undermine the original trust and show the original consent to be invalid since participants were not told clearly “what it was about” and hence were not able to make “their own informed decisions to join or not.” These words, of one of the contributors to the blog reacting to the PD patent, illustrate the core ethical idea underlying consent. Although it may be impossible to inform people of all possible research uses of their material or data, the consent document should contain sufficient and adequately clear information to allow the individual to decide whether the project accords with her moral values and aspirations. Put more generally, consent serves to respect and promote the autonomy of people considering whether to participate in research. 26
+
+The reactions of various 23andMe users and participants suggest that this requirement was not met. This is problematic because it may result not only in a loss of trust, but also because it contravenes the principle of non-instrumentalization. As argued by bioethicist Julian Savulescu with regard to the use of leftover body material: “To ask a person’s permission to do something to that person is to involve her actively and to give her the opportunity to make the project a part of her plans. When we involve people in our projects without their consent, we use them as a means to our own ends.” 27 The reason why participants may perceive a research project as conflicting with their moral values may relate specifically to its commercial or IP aspects.
+
+Despite 23andMe's emphasis on the participant-driven nature of their research, it seems that its strategy regarding the PD project is not based on a true, well-informed involvement of the participants. As observed by one user: “Stating that “it is written in sections 13 and 22 and sections 3 and 5 that people signed” is not close to a decent answer to people you asked for partnering with you to advance research on PD. A company can be for profit or for social profit. You have the right to choose any form you like ..., but please make it clear. If you choose to be for profit only, I don't think you used the right messaging to call for participation of people ... And remember you can only play it once. Trust is not something you can reclaim easily.” 8
+
+Obviously, informed consent remains an imperfect tool to protect participants from being harmed. 28-30 For example,
+
+participants do not always read informed consent forms, and even those who do frequently do not understand.28 Moreover, many people make the decision to participate before the consent process is finalized.31
+
+Nonetheless, research shows that many participants have a desire to know about commercial aspects of research projects they might participate in. 31,32 Would this information make them change their minds about participating? Cook and Hoas conducted an interview study exploring the decision-making processes that participants use when deciding to participate in human subject research. They found, unsurprisingly, that trust plays an important part: "A trusting relationship with a healthcare provider or researcher seems to influence the decisions a prospective human subject makes." 31
+
+Cook and Hoas 31 were also interested in what was regarded as important information for the decision making. They found that most participants desired more information about the commercial purposes: “Prior to taking part in the interviews, most participants had not realized that some studies might be designed for commercial purposes, such as extending a patent .... Participants thought it was dishonest not to be transparent about ... the full purpose of a study.”
+
+Most participants wanted to know whether a study had a commercial purpose, and most reported that such information could influence their decisions about taking part in research in the future. To quote one participant: “I think the study participant should be told exactly what is going on. It’s coercion otherwise.” Another stated: “Patents. Sure. Absolutely, for sure. I absolutely want to know.” Even participants who stated that such information would not influence their decision still felt they should be informed about them. 31
+
+Because information regarding commercial and IP aspects of a study could make people change their mind about participating, withholding this information or not presenting it clearly and concisely, even if not legally required, may be ethically problematic as it may prevent informed decision making.
+
+As noted by Cook and Hoas in the context of another interview study (with institutional review board (IRB) members) “the purpose of a study including commercial purposes” is an issue “that may have a bearing on protection of human subjects but that (is) not well covered by ... regulatory guidance.” 33 They found that: “most IRBs were uncertain about how to handle disclosure of commercial purposes of research to either the IRB or the research participant .... The IRB members reported that the commercial purpose of a study was generally not “on the table” during the review process. 33 To quote one IRB member: “They say that the informed consent is already so long and ... there is no room to add another thing. But what is really important? It is like there is this fear that if you allow that kind of disclosure, the whole shebang will fall apart .... The trouble is how informed do people have to be? If the study was being conducted to extend a patent—I would not be willing to participate. And I would want to know that. Definitely. So there is intentional dishonesty in omitting information that could sway decision making.” 33
+
+GENETICS in MEDICINE | Volume 15 | Number 5 | May 2013
+
+385
+
+---
+
+ORIGINAL RESEARCH ARTICLE
+
+STERCKX et al | Patenting in the field of DTC genetic testing
+
+What patent-related information should be disclosed to gain adequate informed consent? In our view, the participant should be clearly advised that the researchers may seek to patent the results. They should also be informed of what licensing policy will be adopted, e.g., (non)exclusive licensing and (no) royaltyfree licensing of nonprofit entities. The participants' attention must be clearly drawn to this information, and its meaning must be understandable to a layperson. Not least, the word “patent,” rather than just “intellectual property”, should be used and explained.
+
+In addition to the question of what ought to be disclosed to participants, the question might also be raised whether the research setting (nonprofit vs. for-profit) makes a difference to the ethically required level/extent of disclosure. In this regard, we note that the kinds of patent claims filed by 23andMe could just as well be from a nonprofit applicant (cf. the patents of the University of Utah Research Foundation related to the BRCA1 and BRCA2 genes). Seeking these kinds of patents has become standard practice for universities.
+
+It could be argued that, in situations in which participants might assume there is no intent to commercialize, e.g., if research is done by academics or by companies suggesting an altruistic motive, the need to be transparent about the intent to commercialize is stronger. Perhaps participants who take part in research by for-profit organizations ought to expect that patenting will be involved. However, the reactions of several 23andMe users show that they had not realized or expected this. Moreover, studies suggest that, even if research is clearly intended for commercial purposes, participants recruited through people they trust do not appreciate the extent of the commercial dimension unless it is drawn to their attention, at which stage they show the belief that they should have been informed of this. 31 Therefore, in our view, the standards of disclosure should be the same for research conducted in profit and nonprofit settings.
+
+23andMe's PD patent is of concern also for reasons other than those discussed above. The diagnostic method claims are similar to the claims licensed to Myriad Genetics for assays for BRCA1/2 , i.e., the type of claim that might be used to prevent others screening for PD susceptibility. As noted by Cook-Deegan and Heaney, a "single blocking patent on a normal gene or any common disease-associated variant can be sufficient, if exclusively licensed to just one provider, to limit testing by other laboratories for that clinical condition.” 34 The validity of such claims however is in doubt following the US Supreme Court's decision in Mayo v. Prometheus 35 from March 2012, i.e., after 23andMe's patent application was accepted but before it was granted. In that case, the Supreme Court found that methods that are based on "laws of nature" are not patentable.
+
+Moreover, one claim of the international patent application covers the use of unproven (or even undiscovered) drugs in PD therapy. The intention with such a claim is to cover the activities of potential collaborators (e.g., licensees) in drug discovery and development, as well as those of potential
+
+competitors. Because this is the only thing 23andMe might have to offer to a pharmaceutical research and development company, other than the ability to mine their database, this claim, although speculative, is significant. Although 23andMe's granted PD patent was limited to the prognostic method, it should be noted that in April 2012 they filed a continuation patent application in the United States, USSN 13/452341, which may be used to seek patent coverage for the other aspects of their “invention.”
+
+## Conclusion
+
+The issues discussed in this article raise the question as to how an insufficiently informed participant can be regarded as an empowered participant or true partner in research. On the basis of the company communications, many customers clearly believed they were participating in an altruistic exercise to promote development of diagnostic tests and therapies. Yet, an analysis of 23andMe's patent portfolio has revealed that the exercise adds significant value to the company's database in ways that might hinder rather than promote the development and accessibility of diagnostic and treatment options. That there has been some loss of trust in 23andMe's consumer community is clear from reactions to the announcement of the patent. For some, what undermined trust was not so much the profit motive but rather the fact that the company did not provide any clear indication to consumers that it was seeking patents on its discoveries. Obviously, we are not suggesting that the quotes in this article are representative for all participants. It is unknown how many people will change their mind about participating and how many of those would change their mind again if a strategy of clear and forthright communication about patent policies were to be adopted.
+
+Patents are undoubtedly useful and important tools in smoothing the progression to market of novel diagnostic and therapeutic techniques, and no company need be reluctant to be open about the fact that it seeks to patent its inventions. We wish to emphasize that we see no inherent conflict between patenting and behaving responsibly. However, an important lesson to be drawn seems to be that any (private or public) organization involved in research that relies on human participation needs to be transparent, not only about its research goals but also about its strategies and policies regarding commercialization, including patenting and licensing policies. Such transparency is crucial to enable potential participants to make their own decisions as to whether those goals and policies are in line with their moral values, and, if so, whether they want to contribute to those goals by making information or body material available for those purposes. In the absence of such transparency, talk of “participant-centric research,” “empowered” research participants, and “democratized genomics” will continue to sound rather hollow.
+
+## DISCLOSURE
+
+The authors declare no conflict of interest.
+
+386
+
+Volume 15 | Number 5 | May 2013 | GENETICS in MEDICINE
+
+---
+
+Patenting in the field of DTC genetic testing | STERCKX et al
+
+## ORIGINAL RESEARCH ARTICLE
+
+### REFERENCES
+
+1. Caulfield T, McGuire AL. Direct-to-consumer genetic testing: perceptions, problems, and policy responses. Annu Rev Med 2012;63:23-33.
+
+2. Koch VG. PGTandMe: social networking-based genetic testing and the evolving research model. Health Matrix Cleve/ 2012;22:33-74.
+
+3. Kaye J, Curren L, Anderson N, et al. From patients to partners: participantcentric initiatives in biomedical research. Nat Rev Genet 2012;13:371-376.
+
+4. Tutton R, Prainsack B. Enterprising or altruistic selves? Making up research subjects in genetics research. Sociol Health Min 2011;33:1081-1095.
+
+5. Prainsack B. Voting with their mice: personal genome testing and the "participatory turn" in disease research. Account Res 2011;18:132-147.
+
+6. Do CB, Tung JY, Dorfman E, et al. Web-based genome-wide association study identifies two novel loci and a substantial genetic component for Parkinson's disease. PLoS Genet 2011;7:e1002141.
+
+7. 23andme. 23andWe Research. https://www.23andme.com/research Accessed 13 June 2012: 2012.
+
+8. Wojcicki A. Announcing 23andMe's first patent. http://spittoon.23andme. com/news/announcing-23andmes-first-patent/ Accessed 13 June 2012.
+
+9. Human Genetics Commission. Intellectual property and DNA diagnostics. http://www.hgc.gov.uk/Client/document.asp?DocId=318&CAtegoryld=10 Accessed 13 June 2012.
+
+10. Cook-Deegan R, Chandrasekharan S, Angrist M. The dangers of diagnostic monopolies. Nature 2009;458:405-406.
+
+11. Murdoch C, Caulfield T. Commercialization, patenting and genomics: researcher perspectives. Genome Med 2009;1:22.
+
+12. Sterckx S, Cockbain J. Exclusions from patentability. How far has the European Patent Office eroded boundaries? Cambridge University Press: Cambridge, 2012.
+
+13. Nobile H, Vermeulen E, Thys K, Bergmann MM, Borny P. Why do participants enroll in population biobank studies? A systematic literature review. Expert Rev Molecular Diagnostics 2012, in press.
+
+14. Haddow G. "We only did it because he asked us": gendered accounts of participation in a population genetic data collection. Soc Sci Med 2009;69:1010-1017.
+
+15. Allen J, McNamara B. Reconsidering the value of consent in biobank research. Bioethics 2011;25:155-166.
+
+16. Lemke AA, Wolf WA, Hebert-Beirne J, Smith ME. Public and biobank participant attitudes toward genetic research participation and data sharing. Public Health Genomics 2010;13:368-377.
+
+17. Streicher SA, Sanderson SC, Jabs EW, et al. Reasons for participating and genetic information needs among racially and ethnically diverse biobank participants: a focus group study. J Community Genet 2011;2:153-163.
+
+18. Dixon-Woods M, Ashcroft RE, Jackson CJ, et al. Beyond "misunderstanding": written information and decisions about taking part in a genetic epidemiology study. Soc Sci Med 2007;65:2212-2222.
+
+19. Skolbekken JA, Ursin L, Solberg B, Christensen E, Ytterhus B. Not worth the paper it's written on? Informed consent and biobank research in a Norwegian context. Critical Public Health 2005;15:335-347.
+
+20. Steinsbekk KS, Ursin LO, Skolbekken JA, Solberg B. We're not in it for the money-lay people's moral intuitions on commercial use of "their" biobank. Med Health Care Philos 2011; e-pub ahead of print.
+
+21. Steinsbekk KS, Solberg B, Myskja BK. From idealism to realism: commercial ventures in publicly funded biobanks. New Challenges for Biobanks: Ethics, Law and Governance (GeneBanC Project). Intersentia: Antwerp 2009:137-151.
+
+22. Hoeyer K. Trading in cold blood? In: Dabrock P, Taupitz J, Ried J (eds). Trust in Biobanking. Springer: Erlangen, 2012:21-41.
+
+23. Malinowski MJ. Technology transfer in biobanking: credits, debits, and population health futures. J Law Med Ethics 2005;33:54-69.
+
+24. 23andme. Terms of service. https://www.23andme.com/about/tos/ Accessed 13 June 2012.
+
+25. 23andme. Consent document. https://www.23andme.com/about/consent/ Accessed 13 June 2012.
+
+26. Beauchamp T, Childress J. Principles of Biomedical Ethics. 6th edn. Oxford University Press: New York, 2009.
+
+27. Savulescu J. No consent should be needed for using leftover body material for scientific purposes. Against BMJ 2002;325:648-651.
+
+28. Huntington I, Robinson W. The many ways of saying yes and no: reflections on the research coordinator's role in recruiting research participants and obtaining informed consent. IRB 2007;29:6-10.
+
+29. Sugarman J, McCrory DC, Powell D, et al. Empirical research on informed consent. An annotated bibliography. Hastings Center Report 1999;29: 51-42.
+
+30. Ormond KE, Cirino AL, Helenowski IB, Chisholm RL, Wolf WA. Assessing the understanding of biobank participants. Am J Med Genet A 2009;149A:188-198.
+
+31. Cook AF, Hoas H. Trading places: what the research participant can tell the investigator about informed consent. J Clinic Res Bioeth 2011; 2:1-7.
+
+32. Weinfurt KP, Hall MA, King NM, Friedman JY, Schulman KA, Sugarman J. Disclosure of financial relationships to participants in clinical research. N Engl J Med 2009;361:916-921.
+
+33. Cook AF, Hoas H. Exploring the obligation to inform: disclosing the purpose and benefits of research in an increasingly commercial research environment. AJOB Primary Research 2011;2:34-41.
+
+34. Cook-Deegan R, Heaney C. Patents in genomics and human genetics. Annu Rev Genomics Hum Genet 2010;11:383-425.
+
+35. Mayo Collaborative Services, DBA Mayo Medical Laboratories, et al. v. Prometheus Laboratories, Inc. 566 U.S. (2012), 132 S. Ct. 1289 (2012). 2012.
+
+GENETICS in MEDICINE | Volume 15 | Number 5 | May 2013
+
+387
+

processed/00120_W2164875280/metadata.json

@@ -0,0 +1,1390 @@
+{
+  "source_file": "/tmp/tmpi2y75sua/00120_W2164875280_“Trust_is_not_something_you_can_reclaim_easily”_patenting_in_the_field_of_direct-to-consumer_genetic_testing.pdf",
+  "total_pages": 6,
+  "pages": [
+    {
+      "page_number": 1,
+      "elements": [
+        {
+          "bbox": [
+            46,
+            30,
+            147,
+            69
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Genetics\ninMedicine"
+        },
+        {
+          "bbox": [
+            162,
+            46,
+            418,
+            70
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "ORIGINAL RESEARCH ARTICLE"
+        },
+        {
+          "bbox": [
+            453,
+            56,
+            633,
+            69
+          ],
+          "label": "header",
+          "reading_order": 2,
+          "text": "©American College of Medical Genetics and Genomics"
+        },
+        {
+          "bbox": [
+            51,
+            122,
+            627,
+            172
+          ],
+          "label": "sec_0",
+          "reading_order": 3,
+          "text": "“Trust is not something you can reclaim easily”: patenting in\nthe field of direct-to-consumer genetic testing"
+        },
+        {
+          "bbox": [
+            143,
+            177,
+            534,
+            216
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "Sigrid Sterckx, PhD1,2, Julian Cockbain, DPhil3, Heidi Howard, PhD4,\n\nIsabelle Huys, PhD5,6 and Pascal Borry, PhD7"
+        },
+        {
+          "bbox": [
+            51,
+            231,
+            334,
+            337
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "Purpose: Recently, 23andMe announced that it had obtained its\nfirst patent, related to “polymorphisms associated with Parkinson's\ndisease” (US-B-8187811). This announcement immediately sparked\ncontroversy in the community of 23andMe users and research par-\nticipants, especially with regard to issues of transparency and trust.\nThe purpose of this article was to analyze the patent portfolio of this\nprominent direct-to-consumer genetic testing company and discuss\nthe potential ethical implications of patenting in this field for public\nparticipation in Web-based genetic research."
+        },
+        {
+          "bbox": [
+            51,
+            340,
+            334,
+            377
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Methods: We searched the publicly accessible patent database\nEspacenet as well as the commercially available database Micropatent\nfor published patents and patent applications of 23andMe."
+        },
+        {
+          "bbox": [
+            51,
+            379,
+            334,
+            406
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "Results: Six patent families were identified for 23andMe. These\nincluded patent applications related to: genetic comparisons between"
+        },
+        {
+          "bbox": [
+            345,
+            231,
+            627,
+            281
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "grandparents and grandchildren, family inheritance, genome shar-\ning, processing data from genotyping chips, gamete donor selection\nbased on genetic calculations, finding relatives in a database, and\npolymorphisms associated with Parkinson disease."
+        },
+        {
+          "bbox": [
+            345,
+            289,
+            627,
+            361
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "Conclusion: An important lesson to be drawn from this ongoing\ncontroversy seems to be that any (private or public) organization\ninvolved in research that relies on human participation, whether by\nproviding information, body material, or both, needs to be transpar-\nent, not only about its research goals but also about its strategies and\npolicies regarding commercialization."
+        },
+        {
+          "bbox": [
+            345,
+            369,
+            477,
+            383
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Genet Med 2013:15(5):382-387"
+        },
+        {
+          "bbox": [
+            345,
+            392,
+            613,
+            407
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "Key Words: direct-to-consumer; ethics; genetic testing; patenting"
+        },
+        {
+          "bbox": [
+            141,
+            429,
+            238,
+            446
+          ],
+          "label": "sec_1",
+          "reading_order": 12,
+          "text": "INTRODUCTION"
+        },
+        {
+          "bbox": [
+            46,
+            446,
+            334,
+            570
+          ],
+          "label": "para",
+          "reading_order": 13,
+          "text": "In recent years, direct-to-consumer (DTC) genetic testing com-\npanies have been advertising and offering genetic tests directly\nto the public. 1 Some have sought to increase the value of their\n“biobanks” by asking customers to provide phenotypic informa-\ntion. 2 Various commentators have described participant-centric\nresearch initiatives, common features of which include volun-\ntary and active participation, use of social media technology,\nactive interaction between participants and researchers, and the\nappeal to promote public goods (e.g., scientific progress). 3–5"
+        },
+        {
+          "bbox": [
+            46,
+            570,
+            334,
+            652
+          ],
+          "label": "para",
+          "reading_order": 14,
+          "text": "23andMe is a DTC genetic testing company that provides its\nconsumers with the opportunity to consent “to the use of their\ndata for research.”6 Consumers are “given the option of contribut-\ning phenotype data via a series of Web-based surveys. The result is\na single, continually expanding cohort, containing a self-selected\nset of individuals who participate in multiple studies in parallel.”6"
+        },
+        {
+          "bbox": [
+            46,
+            652,
+            334,
+            763
+          ],
+          "label": "para",
+          "reading_order": 15,
+          "text": "Within participant-centric research, 23andMe has focused on\ngenetic and phenotypic correlations with Parkinson disease (PD),\nsarcoma, and myeloproliferative neoplasm. 2 23andMe states on its\nwebsite that letting consumers participate in research in this way\n“can produce revolutionary findings that will benefit us all,” stim-\nulating consumers to \"direct research by participating in studies\nof conditions and traits you care about,” and to “join an effort to\ntranslate basic research into improved health care for everyone.” 7"
+        },
+        {
+          "bbox": [
+            345,
+            431,
+            633,
+            542
+          ],
+          "label": "para",
+          "reading_order": 16,
+          "text": "These encouragements to customers to advance research for\nthe public good, however, stand in contrast with 23andMe's\nannouncement on 28 May 2012 that it was to be granted a US\npatent—patent 8187811, for \"polymorphisms associated with\nParkinson's disease\"'—the very next day. CEO Anne Wojcicki\nannounced on the company website (“The Spittoon”) that the\ngoal of the patent was to ensure that the underlying research\ncould lead “towards successful translation of this discovery.” 8"
+        },
+        {
+          "bbox": [
+            345,
+            542,
+            633,
+            694
+          ],
+          "label": "para",
+          "reading_order": 17,
+          "text": "The announcement immediately sparked controversy on the\nSpittoon website among users and participants, with comments\non many topics: the patentability of genes; the link between\npatents and medical advances; 23andMe's potential (ab)use of\nthe patent, e.g., to charge royalties or block the performance\nof PD genetic tests; the lack of communication of 23andMe on\nits intention to patent discoveries; and the mismatch between\napplying for patents and the avowed mission of democratiz-\ning genomics. 8 These reactions suggest that the knowledge that\n23andMe had sought a patent based on its participant-centric\nresearch could undermine trust in the company."
+        },
+        {
+          "bbox": [
+            345,
+            694,
+            633,
+            763
+          ],
+          "label": "para",
+          "reading_order": 18,
+          "text": "A major criticism concerned the lack of transparency regard-\ning any intention to patent discoveries related to PD. Therefore,\nwe aimed to study which patent applications have been filed by\n23andMe. This article presents the results of an analysis of the\ncompany's patent portfolio and discusses some of the potential"
+        },
+        {
+          "bbox": [
+            48,
+            780,
+            633,
+            831
+          ],
+          "label": "fnote",
+          "reading_order": 19,
+          "text": "1Bioethics Institute Ghent, Universiteit Gent, Gent, Belgium; 2Department of Philosophy and Moral Sciences, Vrije Universiteit Brussel, Brussel, Belgium; 3Consultant European\nPatent Attorney, Oxford, UK; 4Faculté de Médecine, Département d'Épidémologie et de Santé Publique, Université Toulouse III—Paul Sabatier, INSERM UMR U1027, Toulouse,\nFrance; 4Centre for Pharmaceutical Care and Pharmaco-Economics, Faculty of Pharmaceutical Sciences, Katholieke Universiteit Leuven, Leuven, Belgium; 5Centre for Intellectual\nProperty Rights, Katholieke Universiteit Leuven, Leuven, Belgium; 6Centre for Biomedical Ethics and Law, Katholieke Universiteit Leuven, Leuven, Belgium. Correspondence:\nPascal Borry (Pascal.Borry@med.kuleuven.be)"
+        },
+        {
+          "bbox": [
+            46,
+            834,
+            447,
+            847
+          ],
+          "label": "foot",
+          "reading_order": 20,
+          "text": "Submitted 6 July 2012; accepted 1 October 2012; advance online publication 22 November 2012. doi:10.1038/gim.2012.143"
+        },
+        {
+          "bbox": [
+            46,
+            858,
+            66,
+            871
+          ],
+          "label": "foot",
+          "reading_order": 21,
+          "text": "382"
+        },
+        {
+          "bbox": [
+            416,
+            858,
+            632,
+            870
+          ],
+          "label": "foot",
+          "reading_order": 22,
+          "text": "Volume 15 | Number 5 | May 2013 | GENETICS in MEDICINE"
+        }
+      ]
+    },
+    {
+      "page_number": 2,
+      "elements": [
+        {
+          "bbox": [
+            46,
+            56,
+            239,
+            69
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Patenting in the field of DTC genetic testing | STERCKX et al"
+        },
+        {
+          "bbox": [
+            394,
+            46,
+            632,
+            69
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "ORIGINAL RESEARCH ARTICLE"
+        },
+        {
+          "bbox": [
+            46,
+            88,
+            333,
+            117
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "ethical implications of patenting in the context of public par-\nticipation in Web-based genetic research."
+        },
+        {
+          "bbox": [
+            106,
+            127,
+            273,
+            143
+          ],
+          "label": "sec_1",
+          "reading_order": 3,
+          "text": "MATERIALS AND METHODS"
+        },
+        {
+          "bbox": [
+            46,
+            143,
+            334,
+            323
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "We searched the European Patent Office's database Espacenet\nand the commercial database Micropatent for published pat-\nents and applications that showed 23andMe as the applicant or\npatentee. Espacenet provides access to more than 70 million\npatent documents. Our search revealed patents and applica-\ntions accessible on 1 September 2012. Applications filed less\nthan 18 months before, or filed independently by other compa-\nnies working with 23andMe, would not have been revealed. We\nexamined patent “families,” which contain patents and applica-\ntions related to a single invention. For each family, we provide\nthe following basic information: priority application filing date,\npatent or application publication numbers, and a summary of\nthe broadest method claims."
+        },
+        {
+          "bbox": [
+            162,
+            333,
+            217,
+            350
+          ],
+          "label": "sec_1",
+          "reading_order": 5,
+          "text": "RESULTS"
+        },
+        {
+          "bbox": [
+            46,
+            350,
+            334,
+            461
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Six patent families were identified ( Table 1 ). Although our\nfocus is on the PD patent family (family 6), this has not been\n23andMe's only foray into patenting. Obviously, for a company\ninvolved in DTC genomic testing, it is reasonable that, without\ncustomer consent, they might seek to patent improvements in\nsample handling, sample testing, data analysis, and data presen-\ntation. Patent families 1, 2, 3, and 5 seem to fall in these catego-\nries. Family 4, which is concerned with a method of selecting a"
+        },
+        {
+          "bbox": [
+            345,
+            88,
+            632,
+            143
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "sperm or egg donor to maximize the chances of having a baby\nwith desired phenotypic characteristics, raises many ethical\nquestions but is not founded on participant-centric research\nand will not be commented on further here."
+        },
+        {
+          "bbox": [
+            345,
+            143,
+            633,
+            255
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "The international application in patent family 6 (“polymor-\nphisms associated with Parkinson's disease”) includes a claim\ndirected to a method for screening a human subject for sus-\nceptibility to PD based on the determination of certain alleles.\nThe nucleic acid primers or probes used in such analyses are\nclaimed, as is a kit for assaying for PD susceptibility. In another\nclaim (claim 10), however, we see 23andMe seeking to cover\npossible downstream inventions in PD therapy."
+        },
+        {
+          "bbox": [
+            450,
+            263,
+            527,
+            281
+          ],
+          "label": "sec_1",
+          "reading_order": 9,
+          "text": "DISCUSSION"
+        },
+        {
+          "bbox": [
+            345,
+            281,
+            633,
+            432
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Intellectual property (IP) protection is a well-established prac-\ntice that aims to promote technological progress and investment.\nPatenting is common practice for any technology-based indus-\ntry, but in the (bio)pharmaceutical sector, patents are consid-\nered vital, for example, to raise venture capital or justify further\ninvestment and as a way to transfer results of publicly funded\nresearch to commercially viable applications. 9–11 Although in\nthe context of genetic diagnostics, patent offices have granted\nseveral patents of doubtful and disputed validity, 9–12 the mere\nfact that 23andMe has participated in IP-protecting activities\nsince 2007, shortly after it was formed, seems quite normal."
+        },
+        {
+          "bbox": [
+            345,
+            432,
+            632,
+            461
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "Although 23andMe has filed various patent applications,\nuntil now it has only drawn the attention of its consumers to"
+        },
+        {
+          "bbox": [
+            50,
+            473,
+            372,
+            488
+          ],
+          "label": "cap",
+          "reading_order": 12,
+          "text": "Table 1 23andMe's patent families published by 1 September 2012"
+        },
+        {
+          "bbox": [
+            46,
+            488,
+            631,
+            843
+          ],
+          "label": "tab",
+          "reading_order": 13,
+          "text": "<table><tr><td>Family no.</td><td>Earliest priority date</td><td>Patent/patent application publication nos.</td><td>Summary of the broadest method claim</td></tr><tr><td rowspan=\"2\">1</td><td>15 October 2007</td><td>WO2009/051749; US-A-2009/118131</td><td>A method of comparing genetic information from a grandparent and a (presumed) grandchild involving calculating the similarity of the two sets of information and displaying this graphically and in color.</td></tr><tr><td></td><td>WO2009/051766; US-A-2009/119083</td><td>A method of comparing genetic information from two individuals involving comparing their genetic information and displaying this graphically using different symbols to represent genes that are identical and those that are “half”-identical.</td></tr><tr><td>2</td><td>15 October 2007</td><td>WO2009/051768; US-A-2009/112871</td><td>A method of sharing data between two data records that involves, on request from one data record holder, providing read-access to that requestor to selected data areas of non-public data from the other record.</td></tr><tr><td>3</td><td>26 August 2008</td><td>WO2010/024894; US-A-2010/057374; US-A-2010/057807</td><td>A method of providing a merged genetic information data set for an individual that consists of merging two or more genetic data information data sets for that individual and deciding, where there is a conflicting overlap, which input to use.</td></tr><tr><td>4</td><td>05 December 2008</td><td>WO2010/065139; US-A-2010/145981</td><td>A method of identifying a preferred gamete (sperm or ovum) donor to achieve a desired phenotype in the offspring of a particular recipient and a donor, by comparing the genetic data of the recipient and those of a set of possible donors and selecting the donor whose gametes are most likely to combine with those of the recipient to produce offspring with the desired phenotype.</td></tr><tr><td>5</td><td>31 December 2008</td><td>WO2010/077336; US-A-2010/223281; EP-A-2370929</td><td>A method of determining a relative relationship between two individuals that includes comparing genetic data for the two, calculating a predicted degree of genetic relationship between the two, and advising one if the other individual is likely to be a relation.</td></tr><tr><td>6</td><td>30 November 2009</td><td>WO2011/065982; US-A-2011/130337; US-B-8187811; CA-A-2782207</td><td>A method of screening an individual for susceptibility to Parkinson disease that includes determining whether the individual’s DNA has particular abberrations (single-nucleotide polymorphisms) at one or more of seven specific locations.</td></tr></table>"
+        },
+        {
+          "bbox": [
+            46,
+            858,
+            263,
+            870
+          ],
+          "label": "foot",
+          "reading_order": 14,
+          "text": "GENETICS in MEDICINE | Volume 15 | Number 5 | May 2013"
+        },
+        {
+          "bbox": [
+            612,
+            858,
+            632,
+            871
+          ],
+          "label": "foot",
+          "reading_order": 15,
+          "text": "383"
+        }
+      ]
+    },
+    {
+      "page_number": 3,
+      "elements": [
+        {
+          "bbox": [
+            46,
+            46,
+            285,
+            69
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "ORIGINAL RESEARCH ARTICLE"
+        },
+        {
+          "bbox": [
+            440,
+            56,
+            632,
+            69
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "STERCKX et al | Patenting in the field of DTC genetic testing"
+        },
+        {
+          "bbox": [
+            46,
+            88,
+            334,
+            226
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "one patent case. $^8$ Moreover, the communication was made the\nday before the US patent was granted, even though the initial\napplication was filed in 2009, and the research results were\npublished in PLoS Genetics in 2011 6 and then rapidly drawn to\nthe attention of its consumers. The delay in drawing attention\nto the patent application seems odd, given that 23andMe even\nrecently underlined that “open dialogue about complicated\nissues like patents is important” and that it wanted to be “as\nopen as possible about our intentions, including letting people\nknow about our patent and why we have filed it.” $^8$"
+        },
+        {
+          "bbox": [
+            46,
+            226,
+            334,
+            474
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "How likely is it that these events might result in a loss of trust,\nand why is the issue of trust relevant in the first place? It seems\nthat the strong position 23andMe has enjoyed so far in moti-\nvating customers to participate in research by providing phe-\nnotypic data may be weakened. Presumably motivated at least\nin part by altruism, paying customers have contributed to an\nincrease of the informational value of the company's database by\nproviding data. Altruism, confidence in scientific progress, and\ntrust in research and researchers are quite common persuaders\nto induce participation in biobank research. 15 Studies 14–18 have\nshown that individuals donating biological samples and phe-\nnotypic information consider this to be an altruistic act. More\nrelevant in this context is the importance of the donors' trust in\nthe research and researchers, and the reputation of the entity\nplanning the research. 14,15,18 23andMe appears to have been suc-\ncessful in meeting these expectations. The fact that it provides\nparticipants with their personal genetic information may also\nbe a strong motivating factor."
+        },
+        {
+          "bbox": [
+            46,
+            474,
+            334,
+            583
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "Now that it is clear that 23andMe is seeking patents, it is\npossible that various customers will withdraw their support\nbecause they do not consider such activities to be in line with\ntheir altruistic participation in the research projects. As one\ncustomer wrote on 23andMe's blog after the announcement of\nthe patent: \"this is simply crowd-sourced greed. As a longtime\n23andMe customer, this patent is extremely disappointing and\nalarming. Our family is done with your service.\" $^8$"
+        },
+        {
+          "bbox": [
+            46,
+            583,
+            334,
+            776
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "In the context of biobanks, it has also been reported that\ndonors are concerned with social fairness. Studies 19–23 have\nshown public distrust of for-profit companies in the context of\nbiobanking, as many participants consider a profit motive to be\nat odds with their altruistic aim in participating. The point is\nnot that there is anything inherently wrong in making a profit\nor using DTC genetic testing to create revenues, but rather that\nthis may be perceived as conflicting with the open, altruistic,\nscience-driven, and common-good image that 23andMe has\nclearly been trying to create. As one correspondent commented:\n“ I would not have talked my mother and others in my support\ngroup into participating if I had understood this was going to\nbe a profit-driven enterprise. I believe 23andMe has been disin-\ngenuous in gathering a free database. ” 8"
+        },
+        {
+          "bbox": [
+            46,
+            776,
+            334,
+            845
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "This is not a case of 23andMe failing to meet the explicit\nexpectations of participants; it did (by finding the biomark-\ners). Rather, this is a case of promising to build something with\ncommunal resources, building it, and then claiming ownership\nand (potentially) charging for access. The implied suggestion"
+        },
+        {
+          "bbox": [
+            345,
+            88,
+            633,
+            296
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "that the result would be a community good was misleading.\nAn analogy might help: a company in a village next to a river\nsays: “the village needs a bridge, give us the wood and we’ll\nbuild it”; the wood is given; the bridge is built; but the com-\npany charges a toll. In both cases, the contributors (the research\nparticipants/villagers) did not realize that contribution did not\nguarantee public ownership. The fault lies not in 23andMe/\nthe builder owning the result, but in the lack of transparency\nin the appeal for the necessary contributions. The contributors\ndid not understand what was going on until after their contri-\nbution was made, and, had they understood, many might not\nhave contributed. Having been misled, contributors may in the\nfuture be less likely to contribute to the attainment of public\ngoods, fearing that they might not be public after all, which, in\nturn, might lead to a more morally impoverished community."
+        },
+        {
+          "bbox": [
+            345,
+            296,
+            633,
+            446
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "The trust issue is not only related to the nature of the goals\nthe company is pursuing (profit driven or not), but also to the\nextent of transparency surrounding the company's strategies.\nAs far as the latter is concerned, the question of whether the\nparticipants had given truly informed (and thus valid) consent\nis clearly regarded as crucial by various correspondents. For\nexample: “It would seem that the ethics of one company profit-\ning from the knowledge of others because it patented a gene\nvariant could do with some scrutiny, especially if it turns out\nthat patients, who provided samples ..., were not aware that the\nresults would be patented.” $^8$"
+        },
+        {
+          "bbox": [
+            345,
+            446,
+            633,
+            679
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "23andMe responded as follows to this comment: “We make\nreference to our intent to pursue intellectual property rights for\ndiscoveries made from our research in both [our] terms of ser-\nvice ... and in our research consent document ....”3 The rele-\nvant passages from the terms of service and consent documents\nmention that 23andMe might develop intellectual property\nand that participants have no right to share in any profits. The\nterms of service state: “By submitting ... user content, you give\n23andMe ... a perpetual ... license to ... create derivative works\nfrom any user content you submit .... You acquire no rights in\nany research or commercial products that may be developed\nby 23andMe .... You specifically understand that you will not\nreceive compensation for any research or commercial products\n....”24 The consent document provides: “If 23andMe develops\nintellectual property and/or commercializes products or ser-\nvices, directly or indirectly, based on the results of this study,\nyou will not receive any compensation.”25"
+        },
+        {
+          "bbox": [
+            345,
+            679,
+            633,
+            817
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "However, the word “patent” itself is only used in the context\nof information presented to the users. As stated in the terms of\nservice: “You agree that 23andMe ... own all legal right, title,\nand interest in and to the services, including any intellectual\nproperty rights which subsist in the services .... You further\nacknowledge and agree that the services ... contain proprietary\nand confidential information that is protected by applicable\nintellectual property ... laws. You further acknowledge and\nagree that information presented to you through the services\n... is protected by ... patents ....” 24"
+        },
+        {
+          "bbox": [
+            345,
+            817,
+            632,
+            844
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "The wording used by no means makes it clear that patents\nwould be sought for the research results. Various users indicated"
+        },
+        {
+          "bbox": [
+            46,
+            858,
+            66,
+            871
+          ],
+          "label": "foot",
+          "reading_order": 12,
+          "text": "384"
+        },
+        {
+          "bbox": [
+            416,
+            858,
+            632,
+            870
+          ],
+          "label": "foot",
+          "reading_order": 13,
+          "text": "Volume 15 | Number 5 | May 2013 | GENETICS in MEDICINE"
+        }
+      ]
+    },
+    {
+      "page_number": 4,
+      "elements": [
+        {
+          "bbox": [
+            46,
+            56,
+            239,
+            69
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Patenting in the field of DTC genetic testing | STERCKX et al"
+        },
+        {
+          "bbox": [
+            394,
+            46,
+            632,
+            69
+          ],
+          "label": "sec_1",
+          "reading_order": 1,
+          "text": "ORIGINAL RESEARCH ARTICLE"
+        },
+        {
+          "bbox": [
+            46,
+            88,
+            334,
+            158
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "that they were unaware that 23andMe was planning to apply\nfor patents, whereas, as noted by one of the bloggers: “everyone\ncoming to [23andMe's] service, either by paying it or by funded\ninvitation ... needs to know clearly what this is about and make\ntheir own informed decision to join or not.” 8"
+        },
+        {
+          "bbox": [
+            46,
+            158,
+            334,
+            487
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "Based on these reactions, it is clear that the consent proce-\ndure currently used is ethically inadequate, 26 especially in rela-\ntion to patents. The reactions to the PD patent show the limi-\ntations of the use of a vague and unclear consent when there\nare potential commercial applications. We do not suggest that\n23andMe has done the research without consent; rather the\nissue is whether the consent extended to cover the patenting of\nresults. Participants may consent to donate biological materials\nand phenotypic data for the development of clinical applica-\ntions. However, if they are not aware that this might be happen-\ning through commercialization involving patents, this might\nundermine the original trust and show the original consent to\nbe invalid since participants were not told clearly “what it was\nabout” and hence were not able to make “their own informed\ndecisions to join or not.” These words, of one of the contribu-\ntors to the blog reacting to the PD patent, illustrate the core\nethical idea underlying consent. Although it may be impossible\nto inform people of all possible research uses of their material\nor data, the consent document should contain sufficient and\nadequately clear information to allow the individual to decide\nwhether the project accords with her moral values and aspira-\ntions. Put more generally, consent serves to respect and pro-\nmote the autonomy of people considering whether to partici-\npate in research. 26"
+        },
+        {
+          "bbox": [
+            46,
+            487,
+            334,
+            652
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "The reactions of various 23andMe users and participants\nsuggest that this requirement was not met. This is problematic\nbecause it may result not only in a loss of trust, but also because it\ncontravenes the principle of non-instrumentalization. As argued\nby bioethicist Julian Savulescu with regard to the use of leftover\nbody material: “To ask a person’s permission to do something to\nthat person is to involve her actively and to give her the oppor-\ntunity to make the project a part of her plans. When we involve\npeople in our projects without their consent, we use them as a\nmeans to our own ends.” 27 The reason why participants may per-\nceive a research project as conflicting with their moral values\nmay relate specifically to its commercial or IP aspects."
+        },
+        {
+          "bbox": [
+            46,
+            652,
+            334,
+            817
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "Despite 23andMe's emphasis on the participant-driven\nnature of their research, it seems that its strategy regarding the\nPD project is not based on a true, well-informed involvement\nof the participants. As observed by one user: “Stating that “it is\nwritten in sections 13 and 22 and sections 3 and 5 that people\nsigned” is not close to a decent answer to people you asked for\npartnering with you to advance research on PD. A company can\nbe for profit or for social profit. You have the right to choose any\nform you like ..., but please make it clear. If you choose to be for\nprofit only, I don't think you used the right messaging to call for\nparticipation of people ... And remember you can only play it\nonce. Trust is not something you can reclaim easily.” 8"
+        },
+        {
+          "bbox": [
+            46,
+            817,
+            334,
+            845
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Obviously, informed consent remains an imperfect tool\nto protect participants from being harmed. 28-30 For example,"
+        },
+        {
+          "bbox": [
+            345,
+            88,
+            632,
+            144
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "participants do not always read informed consent forms, and\neven those who do frequently do not understand.28 Moreover,\nmany people make the decision to participate before the con-\nsent process is finalized.31"
+        },
+        {
+          "bbox": [
+            345,
+            144,
+            632,
+            281
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "Nonetheless, research shows that many participants have a\ndesire to know about commercial aspects of research projects\nthey might participate in. 31,32 Would this information make\nthem change their minds about participating? Cook and Hoas\nconducted an interview study exploring the decision-making\nprocesses that participants use when deciding to participate in\nhuman subject research. They found, unsurprisingly, that trust\nplays an important part: \"A trusting relationship with a health-\ncare provider or researcher seems to influence the decisions a\nprospective human subject makes.\" 31"
+        },
+        {
+          "bbox": [
+            345,
+            281,
+            632,
+            391
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "Cook and Hoas 31 were also interested in what was regarded\nas important information for the decision making. They found\nthat most participants desired more information about the\ncommercial purposes: “Prior to taking part in the interviews,\nmost participants had not realized that some studies might be\ndesigned for commercial purposes, such as extending a patent\n.... Participants thought it was dishonest not to be transparent\nabout ... the full purpose of a study.”"
+        },
+        {
+          "bbox": [
+            345,
+            391,
+            632,
+            515
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Most participants wanted to know whether a study had a\ncommercial purpose, and most reported that such information\ncould influence their decisions about taking part in research\nin the future. To quote one participant: “I think the study par-\nticipant should be told exactly what is going on. It’s coercion\notherwise.” Another stated: “Patents. Sure. Absolutely, for sure.\nI absolutely want to know.” Even participants who stated that\nsuch information would not influence their decision still felt\nthey should be informed about them. 31"
+        },
+        {
+          "bbox": [
+            345,
+            515,
+            632,
+            583
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "Because information regarding commercial and IP aspects of\na study could make people change their mind about participat-\ning, withholding this information or not presenting it clearly\nand concisely, even if not legally required, may be ethically\nproblematic as it may prevent informed decision making."
+        },
+        {
+          "bbox": [
+            345,
+            583,
+            633,
+            845
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "As noted by Cook and Hoas in the context of another inter-\nview study (with institutional review board (IRB) members)\n“the purpose of a study including commercial purposes” is an\nissue “that may have a bearing on protection of human subjects\nbut that (is) not well covered by ... regulatory guidance.” 33 They\nfound that: “most IRBs were uncertain about how to handle\ndisclosure of commercial purposes of research to either the\nIRB or the research participant .... The IRB members reported\nthat the commercial purpose of a study was generally not “on\nthe table” during the review process. 33 To quote one IRB mem-\nber: “They say that the informed consent is already so long and\n... there is no room to add another thing. But what is really\nimportant? It is like there is this fear that if you allow that kind\nof disclosure, the whole shebang will fall apart .... The trouble\nis how informed do people have to be? If the study was being\nconducted to extend a patent—I would not be willing to par-\nticipate. And I would want to know that. Definitely. So there is\nintentional dishonesty in omitting information that could sway\ndecision making.” 33"
+        },
+        {
+          "bbox": [
+            46,
+            858,
+            263,
+            870
+          ],
+          "label": "foot",
+          "reading_order": 13,
+          "text": "GENETICS in MEDICINE | Volume 15 | Number 5 | May 2013"
+        },
+        {
+          "bbox": [
+            612,
+            858,
+            632,
+            871
+          ],
+          "label": "foot",
+          "reading_order": 14,
+          "text": "385"
+        }
+      ]
+    },
+    {
+      "page_number": 5,
+      "elements": [
+        {
+          "bbox": [
+            46,
+            46,
+            285,
+            69
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "ORIGINAL RESEARCH ARTICLE"
+        },
+        {
+          "bbox": [
+            440,
+            56,
+            632,
+            69
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "STERCKX et al | Patenting in the field of DTC genetic testing"
+        },
+        {
+          "bbox": [
+            46,
+            88,
+            334,
+            226
+          ],
+          "label": "para",
+          "reading_order": 2,
+          "text": "What patent-related information should be disclosed to gain\nadequate informed consent? In our view, the participant should\nbe clearly advised that the researchers may seek to patent the\nresults. They should also be informed of what licensing policy\nwill be adopted, e.g., (non)exclusive licensing and (no) royalty-\nfree licensing of nonprofit entities. The participants' attention\nmust be clearly drawn to this information, and its meaning\nmust be understandable to a layperson. Not least, the word\n“patent,” rather than just “intellectual property”, should be used\nand explained."
+        },
+        {
+          "bbox": [
+            46,
+            226,
+            334,
+            350
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "In addition to the question of what ought to be disclosed\nto participants, the question might also be raised whether the\nresearch setting (nonprofit vs. for-profit) makes a difference to\nthe ethically required level/extent of disclosure. In this regard,\nwe note that the kinds of patent claims filed by 23andMe could\njust as well be from a nonprofit applicant (cf. the patents of the\nUniversity of Utah Research Foundation related to the BRCA1\nand BRCA2 genes). Seeking these kinds of patents has become\nstandard practice for universities."
+        },
+        {
+          "bbox": [
+            46,
+            350,
+            334,
+            570
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "It could be argued that, in situations in which participants\nmight assume there is no intent to commercialize, e.g., if\nresearch is done by academics or by companies suggesting an\naltruistic motive, the need to be transparent about the intent\nto commercialize is stronger. Perhaps participants who take\npart in research by for-profit organizations ought to expect that\npatenting will be involved. However, the reactions of several\n23andMe users show that they had not realized or expected\nthis. Moreover, studies suggest that, even if research is clearly\nintended for commercial purposes, participants recruited\nthrough people they trust do not appreciate the extent of the\ncommercial dimension unless it is drawn to their attention, at\nwhich stage they show the belief that they should have been\ninformed of this. 31 Therefore, in our view, the standards of dis-\nclosure should be the same for research conducted in profit and\nnonprofit settings."
+        },
+        {
+          "bbox": [
+            46,
+            570,
+            334,
+            776
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "23andMe's PD patent is of concern also for reasons other\nthan those discussed above. The diagnostic method claims\nare similar to the claims licensed to Myriad Genetics for\nassays for BRCA1/2 , i.e., the type of claim that might be used\nto prevent others screening for PD susceptibility. As noted\nby Cook-Deegan and Heaney, a \"single blocking patent on\na normal gene or any common disease-associated variant\ncan be sufficient, if exclusively licensed to just one provider,\nto limit testing by other laboratories for that clinical condi-\ntion.” 34 The validity of such claims however is in doubt follow-\ning the US Supreme Court's decision in Mayo v. Prometheus 35\nfrom March 2012, i.e., after 23andMe's patent application was\naccepted but before it was granted. In that case, the Supreme\nCourt found that methods that are based on \"laws of nature\"\nare not patentable."
+        },
+        {
+          "bbox": [
+            46,
+            776,
+            334,
+            845
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Moreover, one claim of the international patent application\ncovers the use of unproven (or even undiscovered) drugs in\nPD therapy. The intention with such a claim is to cover the\nactivities of potential collaborators (e.g., licensees) in drug\ndiscovery and development, as well as those of potential"
+        },
+        {
+          "bbox": [
+            345,
+            88,
+            632,
+            213
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "competitors. Because this is the only thing 23andMe might\nhave to offer to a pharmaceutical research and development\ncompany, other than the ability to mine their database, this\nclaim, although speculative, is significant. Although 23and-\nMe's granted PD patent was limited to the prognostic method,\nit should be noted that in April 2012 they filed a continua-\ntion patent application in the United States, USSN 13/452341,\nwhich may be used to seek patent coverage for the other\naspects of their “invention.”"
+        },
+        {
+          "bbox": [
+            345,
+            225,
+            402,
+            239
+          ],
+          "label": "sec_1",
+          "reading_order": 8,
+          "text": "Conclusion"
+        },
+        {
+          "bbox": [
+            345,
+            239,
+            633,
+            542
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "The issues discussed in this article raise the question as to\nhow an insufficiently informed participant can be regarded\nas an empowered participant or true partner in research. On\nthe basis of the company communications, many custom-\ners clearly believed they were participating in an altruistic\nexercise to promote development of diagnostic tests and\ntherapies. Yet, an analysis of 23andMe's patent portfolio\nhas revealed that the exercise adds significant value to the\ncompany's database in ways that might hinder rather than\npromote the development and accessibility of diagnostic and\ntreatment options. That there has been some loss of trust\nin 23andMe's consumer community is clear from reactions\nto the announcement of the patent. For some, what under-\nmined trust was not so much the profit motive but rather\nthe fact that the company did not provide any clear indica-\ntion to consumers that it was seeking patents on its discover-\nies. Obviously, we are not suggesting that the quotes in this\narticle are representative for all participants. It is unknown\nhow many people will change their mind about participating\nand how many of those would change their mind again if a\nstrategy of clear and forthright communication about patent\npolicies were to be adopted."
+        },
+        {
+          "bbox": [
+            345,
+            542,
+            633,
+            803
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Patents are undoubtedly useful and important tools in\nsmoothing the progression to market of novel diagnostic and\ntherapeutic techniques, and no company need be reluctant to\nbe open about the fact that it seeks to patent its inventions. We\nwish to emphasize that we see no inherent conflict between pat-\nenting and behaving responsibly. However, an important lesson\nto be drawn seems to be that any (private or public) organi-\nzation involved in research that relies on human participation\nneeds to be transparent, not only about its research goals but\nalso about its strategies and policies regarding commercializa-\ntion, including patenting and licensing policies. Such transpar-\nency is crucial to enable potential participants to make their\nown decisions as to whether those goals and policies are in\nline with their moral values, and, if so, whether they want to\ncontribute to those goals by making information or body mate-\nrial available for those purposes. In the absence of such trans-\nparency, talk of “participant-centric research,” “empowered”\nresearch participants, and “democratized genomics” will con-\ntinue to sound rather hollow."
+        },
+        {
+          "bbox": [
+            345,
+            816,
+            415,
+            831
+          ],
+          "label": "sec_1",
+          "reading_order": 11,
+          "text": "DISCLOSURE"
+        },
+        {
+          "bbox": [
+            345,
+            831,
+            534,
+            844
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "The authors declare no conflict of interest."
+        },
+        {
+          "bbox": [
+            46,
+            858,
+            66,
+            871
+          ],
+          "label": "foot",
+          "reading_order": 13,
+          "text": "386"
+        },
+        {
+          "bbox": [
+            416,
+            858,
+            632,
+            870
+          ],
+          "label": "foot",
+          "reading_order": 14,
+          "text": "Volume 15 | Number 5 | May 2013 | GENETICS in MEDICINE"
+        }
+      ]
+    },
+    {
+      "page_number": 6,
+      "elements": [
+        {
+          "bbox": [
+            46,
+            56,
+            240,
+            68
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Patenting in the field of DTC genetic testing | STERCKX et al"
+        },
+        {
+          "bbox": [
+            394,
+            47,
+            632,
+            68
+          ],
+          "label": "sec_1",
+          "reading_order": 1,
+          "text": "ORIGINAL RESEARCH ARTICLE"
+        },
+        {
+          "bbox": [
+            46,
+            85,
+            116,
+            100
+          ],
+          "label": "sec_2",
+          "reading_order": 2,
+          "text": "REFERENCES"
+        },
+        {
+          "bbox": [
+            46,
+            100,
+            333,
+            121
+          ],
+          "label": "reference",
+          "reading_order": 3,
+          "text": "1. Caulfield T, McGuire AL. Direct-to-consumer genetic testing: perceptions,\nproblems, and policy responses. Annu Rev Med 2012;63:23-33."
+        },
+        {
+          "bbox": [
+            46,
+            121,
+            333,
+            142
+          ],
+          "label": "reference",
+          "reading_order": 4,
+          "text": "2. Koch VG. PGTandMe: social networking-based genetic testing and the\nevolving research model. Health Matrix Cleve/ 2012;22:33-74."
+        },
+        {
+          "bbox": [
+            46,
+            142,
+            333,
+            163
+          ],
+          "label": "reference",
+          "reading_order": 5,
+          "text": "3. Kaye J, Curren L, Anderson N, et al. From patients to partners: participant-\ncentric initiatives in biomedical research. Nat Rev Genet 2012;13:371-376."
+        },
+        {
+          "bbox": [
+            46,
+            163,
+            333,
+            185
+          ],
+          "label": "reference",
+          "reading_order": 6,
+          "text": "4. Tutton R, Prainsack B. Enterprising or altruistic selves? Making up research\nsubjects in genetics research. Sociol Health Min 2011;33:1081-1095."
+        },
+        {
+          "bbox": [
+            46,
+            185,
+            333,
+            205
+          ],
+          "label": "reference",
+          "reading_order": 7,
+          "text": "5. Prainsack B. Voting with their mice: personal genome testing and the\n\"participatory turn\" in disease research. Account Res 2011;18:132-147."
+        },
+        {
+          "bbox": [
+            46,
+            205,
+            333,
+            237
+          ],
+          "label": "reference",
+          "reading_order": 8,
+          "text": "6. Do CB, Tung JY, Dorfman E, et al. Web-based genome-wide association\nstudy identifies two novel loci and a substantial genetic component for\nParkinson's disease. PLoS Genet 2011;7:e1002141."
+        },
+        {
+          "bbox": [
+            46,
+            237,
+            333,
+            258
+          ],
+          "label": "reference",
+          "reading_order": 9,
+          "text": "7. 23andme. 23andWe Research. https://www.23andme.com/research\nAccessed 13 June 2012: 2012."
+        },
+        {
+          "bbox": [
+            46,
+            258,
+            333,
+            280
+          ],
+          "label": "reference",
+          "reading_order": 10,
+          "text": "8. Wojcicki A. Announcing 23andMe's first patent. http://spittoon.23andme.\ncom/news/announcing-23andmes-first-patent/ Accessed 13 June 2012."
+        },
+        {
+          "bbox": [
+            46,
+            280,
+            333,
+            310
+          ],
+          "label": "reference",
+          "reading_order": 11,
+          "text": "9. Human Genetics Commission. Intellectual property and DNA diagnostics.\nhttp://www.hgc.gov.uk/Client/document.asp?DocId=318&CAtegoryld=10\nAccessed 13 June 2012."
+        },
+        {
+          "bbox": [
+            46,
+            310,
+            333,
+            332
+          ],
+          "label": "reference",
+          "reading_order": 12,
+          "text": "10. Cook-Deegan R, Chandrasekharan S, Angrist M. The dangers of diagnostic\nmonopolies. Nature 2009;458:405-406."
+        },
+        {
+          "bbox": [
+            46,
+            332,
+            333,
+            353
+          ],
+          "label": "reference",
+          "reading_order": 13,
+          "text": "11. Murdoch C, Caulfield T. Commercialization, patenting and genomics:\nresearcher perspectives. Genome Med 2009;1:22."
+        },
+        {
+          "bbox": [
+            46,
+            353,
+            333,
+            384
+          ],
+          "label": "reference",
+          "reading_order": 14,
+          "text": "12. Sterckx S, Cockbain J. Exclusions from patentability. How far has the\nEuropean Patent Office eroded boundaries? Cambridge University Press:\nCambridge, 2012."
+        },
+        {
+          "bbox": [
+            46,
+            384,
+            334,
+            416
+          ],
+          "label": "reference",
+          "reading_order": 15,
+          "text": "13. Nobile H, Vermeulen E, Thys K, Bergmann MM, Borny P. Why do participants\nenroll in population biobank studies? A systematic literature review. Expert\nRev Molecular Diagnostics 2012, in press."
+        },
+        {
+          "bbox": [
+            46,
+            416,
+            334,
+            448
+          ],
+          "label": "reference",
+          "reading_order": 16,
+          "text": "14. Haddow G. \"We only did it because he asked us\": gendered accounts\nof participation in a population genetic data collection. Soc Sci Med\n2009;69:1010-1017."
+        },
+        {
+          "bbox": [
+            46,
+            448,
+            333,
+            469
+          ],
+          "label": "reference",
+          "reading_order": 17,
+          "text": "15. Allen J, McNamara B. Reconsidering the value of consent in biobank\nresearch. Bioethics 2011;25:155-166."
+        },
+        {
+          "bbox": [
+            46,
+            469,
+            333,
+            500
+          ],
+          "label": "reference",
+          "reading_order": 18,
+          "text": "16. Lemke AA, Wolf WA, Hebert-Beirne J, Smith ME. Public and biobank\nparticipant attitudes toward genetic research participation and data sharing.\nPublic Health Genomics 2010;13:368-377."
+        },
+        {
+          "bbox": [
+            46,
+            500,
+            334,
+            532
+          ],
+          "label": "reference",
+          "reading_order": 19,
+          "text": "17. Streicher SA, Sanderson SC, Jabs EW, et al. Reasons for participating and\ngenetic information needs among racially and ethnically diverse biobank\nparticipants: a focus group study. J Community Genet 2011;2:153-163."
+        },
+        {
+          "bbox": [
+            46,
+            532,
+            334,
+            564
+          ],
+          "label": "reference",
+          "reading_order": 20,
+          "text": "18. Dixon-Woods M, Ashcroft RE, Jackson CJ, et al. Beyond \"misunderstanding\":\nwritten information and decisions about taking part in a genetic\nepidemiology study. Soc Sci Med 2007;65:2212-2222."
+        },
+        {
+          "bbox": [
+            346,
+            86,
+            632,
+            118
+          ],
+          "label": "reference",
+          "reading_order": 21,
+          "text": "19. Skolbekken JA, Ursin L, Solberg B, Christensen E, Ytterhus B. Not worth\nthe paper it's written on? Informed consent and biobank research in a\nNorwegian context. Critical Public Health 2005;15:335-347."
+        },
+        {
+          "bbox": [
+            346,
+            118,
+            632,
+            151
+          ],
+          "label": "reference",
+          "reading_order": 22,
+          "text": "20. Steinsbekk KS, Ursin LO, Skolbekken JA, Solberg B. We're not in it for the\nmoney-lay people's moral intuitions on commercial use of \"their\" biobank.\nMed Health Care Philos 2011; e-pub ahead of print."
+        },
+        {
+          "bbox": [
+            346,
+            151,
+            632,
+            192
+          ],
+          "label": "reference",
+          "reading_order": 23,
+          "text": "21. Steinsbekk KS, Solberg B, Myskja BK. From idealism to realism: commercial\nventures in publicly funded biobanks. New Challenges for Biobanks:\nEthics, Law and Governance (GeneBanC Project). Intersentia: Antwerp\n2009:137-151."
+        },
+        {
+          "bbox": [
+            346,
+            192,
+            632,
+            214
+          ],
+          "label": "reference",
+          "reading_order": 24,
+          "text": "22. Hoeyer K. Trading in cold blood? In: Dabrock P, Taupitz J, Ried J (eds). Trust in\nBiobanking. Springer: Erlangen, 2012:21-41."
+        },
+        {
+          "bbox": [
+            346,
+            214,
+            632,
+            235
+          ],
+          "label": "reference",
+          "reading_order": 25,
+          "text": "23. Malinowski MJ. Technology transfer in biobanking: credits, debits, and\npopulation health futures. J Law Med Ethics 2005;33:54-69."
+        },
+        {
+          "bbox": [
+            346,
+            235,
+            632,
+            255
+          ],
+          "label": "reference",
+          "reading_order": 26,
+          "text": "24. 23andme. Terms of service. https://www.23andme.com/about/tos/\nAccessed 13 June 2012."
+        },
+        {
+          "bbox": [
+            346,
+            255,
+            632,
+            277
+          ],
+          "label": "reference",
+          "reading_order": 27,
+          "text": "25. 23andme. Consent document. https://www.23andme.com/about/consent/\nAccessed 13 June 2012."
+        },
+        {
+          "bbox": [
+            346,
+            277,
+            632,
+            298
+          ],
+          "label": "reference",
+          "reading_order": 28,
+          "text": "26. Beauchamp T, Childress J. Principles of Biomedical Ethics. 6th edn. Oxford\nUniversity Press: New York, 2009."
+        },
+        {
+          "bbox": [
+            346,
+            298,
+            632,
+            319
+          ],
+          "label": "reference",
+          "reading_order": 29,
+          "text": "27. Savulescu J. No consent should be needed for using leftover body material\nfor scientific purposes. Against BMJ 2002;325:648-651."
+        },
+        {
+          "bbox": [
+            346,
+            319,
+            632,
+            350
+          ],
+          "label": "reference",
+          "reading_order": 30,
+          "text": "28. Huntington I, Robinson W. The many ways of saying yes and no: reflections\non the research coordinator's role in recruiting research participants and\nobtaining informed consent. IRB 2007;29:6-10."
+        },
+        {
+          "bbox": [
+            346,
+            350,
+            632,
+            381
+          ],
+          "label": "reference",
+          "reading_order": 31,
+          "text": "29. Sugarman J, McCrory DC, Powell D, et al. Empirical research on informed\nconsent. An annotated bibliography. Hastings Center Report 1999;29:\n51-42."
+        },
+        {
+          "bbox": [
+            346,
+            381,
+            632,
+            413
+          ],
+          "label": "reference",
+          "reading_order": 32,
+          "text": "30. Ormond KE, Cirino AL, Helenowski IB, Chisholm RL, Wolf WA. Assessing\nthe understanding of biobank participants. Am J Med Genet A\n2009;149A:188-198."
+        },
+        {
+          "bbox": [
+            346,
+            413,
+            632,
+            445
+          ],
+          "label": "reference",
+          "reading_order": 33,
+          "text": "31. Cook AF, Hoas H. Trading places: what the research participant can\ntell the investigator about informed consent. J Clinic Res Bioeth 2011;\n2:1-7."
+        },
+        {
+          "bbox": [
+            346,
+            445,
+            632,
+            477
+          ],
+          "label": "reference",
+          "reading_order": 34,
+          "text": "32. Weinfurt KP, Hall MA, King NM, Friedman JY, Schulman KA, Sugarman J.\nDisclosure of financial relationships to participants in clinical research. N Engl\nJ Med 2009;361:916-921."
+        },
+        {
+          "bbox": [
+            346,
+            477,
+            632,
+            509
+          ],
+          "label": "reference",
+          "reading_order": 35,
+          "text": "33. Cook AF, Hoas H. Exploring the obligation to inform: disclosing the\npurpose and benefits of research in an increasingly commercial research\nenvironment. AJOB Primary Research 2011;2:34-41."
+        },
+        {
+          "bbox": [
+            346,
+            509,
+            632,
+            530
+          ],
+          "label": "reference",
+          "reading_order": 36,
+          "text": "34. Cook-Deegan R, Heaney C. Patents in genomics and human genetics. Annu\nRev Genomics Hum Genet 2010;11:383-425."
+        },
+        {
+          "bbox": [
+            346,
+            530,
+            632,
+            561
+          ],
+          "label": "reference",
+          "reading_order": 37,
+          "text": "35. Mayo Collaborative Services, DBA Mayo Medical Laboratories, et al. v.\nPrometheus Laboratories, Inc. 566 U.S. (2012), 132 S. Ct. 1289 (2012).\n2012."
+        },
+        {
+          "bbox": [
+            46,
+            858,
+            262,
+            870
+          ],
+          "label": "foot",
+          "reading_order": 38,
+          "text": "GENETICS in MEDICINE | Volume 15 | Number 5 | May 2013"
+        },
+        {
+          "bbox": [
+            613,
+            858,
+            632,
+            871
+          ],
+          "label": "foot",
+          "reading_order": 39,
+          "text": "387"
+        }
+      ]
+    }
+  ],
+  "metadata": {}
+}

processed/00126_W2605918107/document.md

@@ -0,0 +1,737 @@
+![Figure](figures/figure_000.png)
+
+# This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details.
+
+Author(s): Heikkilä, Jussi; Lorenz, Annika
+
+Title: Need for speed? : Exploring the relative importance of patents and utility models among German firms
+
+Year: 2018
+
+Version: Accepted version (Final draft)
+
+Copyright: © 2017 Informa UK Limited, trading as Taylor & Francis Group. This is a final draft
+
+Rights:
+
+Rights url:
+
+Please cite the original version:
+
+Heikkilä, J., & Lorenz, A. (2018). Need for speed? : Exploring the relative importance of patents
+
+and utility models among German firms. Economics of Innovation and New Technology, 27(1),
+
+80-105. https://doi.org/10.1080/10438599.2017.1310794
+
+---
+
+# Need for speed? Exploring the relative importance of patents and utility models among German firms
+
+Jussi Heikkilä School of Business and Economics, University of Jyväskylä, Jyväskylä, Finland Jyväskylä University School of Business and Economics, P.O. Box 35, FI-40014
+
+jussi.heikkila@jyu.fi
+
+Annika Lorenz* Department of Marketing and Strategy, Faculty of Business Economics, Hasselt University, Hasselt, Belgium Hasselt University, Martelarenlaan 42, B-3500 Hasselt, Belgium
+
+annika.lorenz@tu-berlin.de
+
+Chair for Innovation Economics, Faculty of Economics and Management, Technische Universität Berlin, Germany
+
+Technische Universität Berlin, Marchstr. 23, D-10587 Berlin
+
+*Corresponding author
+
+---
+
+## Need for speed? Exploring the relative importance of patents and utility models among German firms
+
+Despite the wide use of two-tiered patent systems (patents and utility models), there is little empirical evidence about how often utility models (UMs) are actually used, what kind of firms use them to protect their intellectual property, and how firms rank them relative to patents. We offer such an analysis using data from Germany. We find that larger firms are more likely to use both protection methods. Moreover, a short life cycle of products and services is associated with an increased likelihood to use UMs. The features and functioning of the German utility model system are of broader interest because it has been a benchmark for several second tier patent protection systems around the world.
+
+Keywords: patent; utility model; two-tiered patent system; intellectual property strategy; Germany
+
+JEL Classification: 031; 032; 034
+
+## 1 Introduction
+
+What kind of institutions and, in particular, systems of intellectual property rights (IPRs) should countries adopt in order to optimize the rate and direction of technological change and innovation? The patent system is arguably one of the most important innovation policy instruments, but its effect on the rate and direction of technological change, productivity growth, and, ultimately, citizens' welfare remains disputable (Mazzoleni and Nelson 1998; Kingston 2001; Moser 2005; Encaoua et al. 2006) . Moreover, the importance of the patent system for a country's productivity growth, as an example, may be conditional on the country's level of development (Kim et al. 2012) and its institutions. Thus, it is unclear which type of IPR system is desirable and how its design ought to vary across different institutional contexts. A necessary step, therefore, in the search for an optimal IPR system and combination of innovation
+
+---
+
+policy instruments is to understand the functioning of instruments in a given institutional context. This paper's analysis builds on the premise that understanding why firms in a certain country prefer one IPR over another is at the heart of understanding the functioning of its IPR system.
+
+Since its enactment in 1994, the Trade-Related Aspects of Intellectual Property Rights (TRIPS) agreement introduced minimum standards for many forms of intellectual property (IP), rendering it the most comprehensive international agreement on intellectual property to date. $^1$ Despite progressive global harmonization efforts, there are still significant differences across national IPR systems. Some countries, such as Germany, for instance, have a two-tiered patent system comprising a patent and a utility model (UM) system, whereas many other countries have a system consisting of patents only. Additionally, TRIPS defines the minimum standards of patent systems and leaves further room for discretion in designing national patent systems (Suthersanen 2006; Königer 2009; Grosse Ruse-Khan 2012, 2013) . However, since there are no international agreements on minimum standards of second tier patent systems, countries may design UM systems as they see fit. Although there are quite a few countries that possess dual-tier patent systems consisting of both patents and utility models, the average number of application is much lower for utility models in contrast to patents. This is mainly due to the distinct characteristics of patents and utility models. Table 1 illustrates some dimensions along which patent and UM protection typically differ.
+
+TABLE 1 Comparison of patent and utility model protection
+
+Notes: In some jurisdictions utility model applications are examined but typically there is just a registration process. Sources of information: WIPO; national patent offices; Janis 1999; Suthersanen 2006; Prud'homme 2014; Radauer et al. 2015.
+
+---
+
+Based on their distinct characteristics, UMs are usually described as a faster, cheaper, and simpler protection method than patents (e.g., European Commission 1995; Suthersanen 2006, Königer 2009, Radauer et al. 2015; Johnson et al. 2015) . $^2$ As a result, UM systems in advanced economies are often justified by emphasizing their importance to small- and medium-sized enterprises (SMEs) and individual inventors. UM and second tier patent systems have, therefore, been claimed to foster minor inventions, which do not satisfy patentability requirements (Johnson 2002; Beneito 2006; Encaoua et al. 2006; Radauer et al. 2015; Grosse Ruse-Khan 2013) . However, empirical evidence for these argued benefits is still scant, if not entirely lacking. For instance, Kim et al. (2012) did not find UM systems to be associated with economic growth in advanced economies.
+
+Although both patents and UMs can be used to protect technical inventions, we currently have a limited understanding of the factors that are associated with the use of these protection methods. Additionally, some studies have investigated patenting strategies of small firms (de Rassenfosse 2012; Holgersson 2013; Maurseth and Svensson 2013) but the strategies to file for utility models remain underexplored. What determines a firm's decision to opt for patents or UMs (or both) and, hence, to choose between these appropriation mechanisms? While insightful, prior theoretical (Anton and Yao 2004; Kultti et al. 2007; Hall et al. 2013) and empirical (Arundel 2001; Hussinger 2006; Hall et al. 2013, 2014; Heger and Zaby 2013) studies have mainly focused on the trade-off between patents and secrecy as IP protection methods. Our paper contributes to the rather limited literature on UMs (reviewed in the next section) by exploring the determinants of firms' decisions to use patents and/or UMs and by identifying firm characteristics that are associated with the use and relative importance of patents and UMs. Inspired by Cao et al. (2014) , we focus on the trade-off between UMs, which,
+
+---
+
+arguably, serve as appropriate protection methods for firms with a need to protect their inventions quickly (“need for speed”), and patents, which take more time to obtain, but are more reliable in case of infringement and provide longer-term protection.
+
+Our data come from Germany and allow exploring what kinds of firms use UMs to protect their intellectual property and how firms rank them relative to patents. We have two main findings. First, larger firms are more likely to use both protection methods. On the one hand, the UM system is an inclusive institution by promoting protection of IP among resource-constrained small firms; on the other hand, it concurrently extends the arsenal of IPR methods for large firms. Second, a short life cycle of products and services is associated with an increased likelihood to use UMs. This suggests that UMs do not serve only applicants with incremental inventions, but also applicants in need of quick protection. These findings bear on the debate of the harmonization of IPR systems worldwide, as decision-makers involved in the design of IPR institutions need to understand the effects that two-tiered patent systems have on potential innovators and innovation activity. Such understanding is called for if international harmonization proceeds and, at some point, forces countries to decide whether or not to implement (or maintain) a second tier patent protection system. $^3$
+
+The paper is structured as follows: Section 2 reviews the related literature. The German UM system is described in section 3. In section 4, we discuss the possible determinants of the use of patents and UMs and develop testable hypotheses. Data and methods are presented in section 5 and the empirical analysis in section 6. Section 7 concludes.
+
+## 2 Literature review
+
+Several studies and reports provide historical reviews and information on institutional details and justifications of national UM systems (European Commission 1995; Janis
+
+---
+
+1999; Suthersanen 2006; Königer 2009; Cummings 2010; Boztosun 2010; Grosse RuseKhan 2013; Prud’homme 2014; Radauer et al. 2015; Johnson et al. 2015) . These studies and reports highlight three features of UM systems. First, national UM systems vary significantly across countries, as they are not internationally regulated or harmonized. Second, the stated justification of UM systems often refers to creating innovation incentives for SMEs and incremental innovations. Third, a frequently mentioned disadvantage of the UM systems is the legal uncertainty that they create in many jurisdictions as unexamined IP rights (no examination, only a registration process).
+
+While UM systems have received some limited attention among legal scholars (e.g., Janis 1999; Suthersanen 2006; Björkwall 2009; Cummings 2010; Grosse RuseKhan 2012, 2013) , research on the economics of UMs has been scarcer. In quantitative patent data analyses, it is common to treat UMs as patents (e.g., Hu and Jefferson 2009) . Maskus and McDaniel (1999) found the Japanese UM system to have played an important role in the diffusion of technologies. They argue that the primary channels of diffusion were follow-up UM applications for incremental inventions, which built on prior technical knowledge embodied in patent applications. Johnson (2002) analyzes the interaction of “ technology acquisition forms ” (R&D and licensing) in the creation of new intellectual capital with Brazilian firm-level data. His results indicate that larger firms rely more on patents and less on utility models but otherwise the differences in determinants of patent and utility model applications are small. Beneito (2006) focuses on firm-level determinants of patent and UM use in Spanish research and development (R&D) intensive firms. She explicitly assumes patents and UMs to be measures of significant and incremental innovations, respectively, and found the number of patents to be associated with in-house R&D and UMs with contracted R&D. By using Korean data, Kim et al. (2012) show that UMs are positively associated with firm growth when
+
+---
+
+firms are technologically lagging. Kim et al. argue that such firms may use minor innovations protected by UMs as a learning device and as “a stepping stone for developing more patentable inventions later on” (2012, p. 358). Thomä and Bizer (2013) investigate combinations of IP rights utilized by SMEs in Germany using Community Innovation Survey (CIS) data. Although UMs are included in their analysis, Thomä and Bizer (2013) do not extensively discuss their role. It is, nevertheless, notable that, in their sample, SMEs in the “patent-oriented group” also assign a higher importance to UMs. This suggests that UMs are used as an auxiliary (i.e., complementary) protection method by patenting firms as highlighted by Radauer et al. (2015). This is particularly true for some jurisdictions, such as Germany, Finland, and Denmark, where patents and UMs can also be combined under certain conditions. This means that they are not necessarily mutually exclusive or substitutes (Guellec and van Pottelsberghe 2007; Prud’homme 2014; Radauer et al. 2015).
+
+Lemley et al. (2005) further suggest establishing a gold plate patent system to supplement the normal patent system (i.e., two-tiered patent system), and, thus, eliminate bad (low probability of validity) patents in the US. Based on Lemley et al. (2005), Atal and Bar (2014) construct a theoretical model highlighting the role of twotiered patent systems as signaling (or screening) devices. According to Atal and Bar (2014), introducing a second patent tier can reduce patent applications, decrease the number of bad patents, and increase social welfare: in a single-tiered system bad patents impose negative externalities to holders of “good” (high probability of validity) patents because they negatively affect the overall perception of patent quality. A two-tiered patent system enables screening and self-selection, which diminishes this negative externality. Importantly, Atal and Bar further claim that “since the two-tiered system
+
+---
+
+can be designed to mimic the best single-tiered system, welfare in an optimal two-tiered system is at least as high as welfare in the optimal single-tiered system ” (2014, p. 522).
+
+The disclosure function of the patent system aims to prevent duplication of R&D and allows rapid diffusion of innovations once the patent has expired (Mazzoleni and Nelson 1998; Johnson and Popp 2003; Denicolò and Franzoni 2004; Guellec and van Pottelsberghe 2007; Guellec et al. 2012; Graham and Hegde 2015) . Accessible patent information enables other agents, especially competitors, to obtain information on the state of the art and technical advances in their field of technology (Arundel and Steinmueller 1998; van Zeebroeck and van Pottelsberghe 2011) . UMs have the potential to further enhance welfare because technical information is published early on and normally much faster than in the case of patents. This means that knowledge spillovers may occur earlier.
+
+## 3 Institutional framework
+
+### 3.1 The German utility model system
+
+In 1891, Germany introduced a UM system, rendering it the oldest UM system in the world, which is why it is often used as a reference for other countries (Kingston 2001; Guellec and van Pottelsberghe 2007) . Initially, the aim of the German UM system was to fill the gap between patents and design rights, i.e., to offer protection for small technical inventions (Janis 1999; Grosse Ruse-Khan 2013) . Harhoff et al. (2003) , Cremers et al. (2013) , and Cremers et al. (2014) provide good descriptions of the German patent system.
+
+German UMs protect technical inventions, including chemical substances, food, and medical products, except processes (manufacturing and working processes, measuring processes, and others). The maximum duration of a patent is 20 years, while
+
+---
+
+a German UM can be extended for a maximum of 10 years. Although the UM is often referred to as a “petty patent” or “small patent”, in Germany the inventive step requirement has been the same for patents and UMs since 2006 (Björkwall 2009; Grosse Ruse-Khan 2013; Prud’homme 2014; Radauer et al. 2015) .
+
+Recent statistics on the UMs and patents in force in Germany, granted by the Deutsches Patent-und Markenamt (DPMA), show a slightly decreasing trend (as depicted in Figure 1). Moreover, Figure 2 indicates that the number of UM registrations is declining faster than the number of national patent applications. From 2000 to 2014, UM filings decreased by more than one-third (from 22,440 to 14,805), whereas patent applications increased by 1.7 % (from 64,862 to 65,958). In contrast, the number of European Patent Office (EPO) patents granted in Germany has steadily been increasing: from 255,303 in 2000 to 458,042 in 2014 (79.4 % ). Part of the decrease in UM filings could be explained by the decision of the German Supreme court in 2006 to apply the same inventive step requirement for UMs as for patents.
+
+Figure 1. Patents and utility models in force in Germany 2000-2014
+
+Source: DPMA annual reports 2006-2014, available at: http://www.dpma.de/english/service/publications/annualreports/index.html
+
+Figure 2. Patent applications and utility model filings at the DPMA
+
+Source: DPMA annual reports 2006-2014, available at: http://www.dpma.de/english/service/publications/annualreports/index.html
+
+The German UM system has been designed especially considering the needs of SMEs (Suthersanen 2006; Königer 2009; Grosse Ruse-Khan 2013) . The procedure to register a UM at the German patent office is simpler and somewhat cheaper than the
+
+---
+
+procedure to apply for a patent. This is reflected by statements made by the largest IP law firms and patent attorneys, as well as the Industrie-und Handwerkskammer (IHK) in Germany;
+
+“Due to a lack of examination it is significantly faster and more cost-effective to obtain a utility model, on the other hand, this makes it less legally secure.” (IHK München) 4
+
+“A utility model is due to the low costs, especially suitable for small and mediumsized businesses (SMEs) as well as for inventions, for which is not yet known whether or how they are economically exploited.” (Wittmann and Hernandez, patent attorneys) 5
+
+Additionally, Radauer et al. (2015, pp. 32-33) report six beneficial features of German UMs that stand out in the qualitative feedback from UM users: 1. speed of protection, which makes UMs suitable for products with short life cycles; 2. branchingoff a UM from a patent application; 3. protection for minor inventions; 4. grace period; 5. complying with German employee inventions law; and 6. using UMs as a cheap means for publication (create prior art and preserve freedom to operate). Hence, the UM adds flexibility to the German patent system.
+
+### 3.2 The process of registering a utility model in Germany6
+
+The examination and granting of a patent usually take several years. In Germany, the applicants also have the option for deferred examination, i.e., to postpone examination of the patent for a maximum of seven years (Harhoff et al. 2003; Jell et al. 2013; Harhoff et al. 2015) . In contrast, a German UM is registered on average within two to four months after filing the application (Radauer et al. 2015) , provided the documents filed comply with the provisions of the Utility Model Act (Gebrauchsmustergesetz). The DPMA itself advertises UMs as “the fast IP rights”. $^7$
+
+---
+
+Table 2 reports descriptive statistics for German patents and UMs concerning pendency times and publication lags. $^8$ The mean grant lag of patents and the registration lag of UMs are affected by right-skewed lag distributions as the median lags are considerably shorter than the respective means. $^9$ From 2000 to 2010, the median grant lag for German patents was 2.8 years, while for UMs, the median registration lag was approximately three months.
+
+Table 2. Publication, registration, and grant lags of patents and utility models
+
+Figure 3 shows the grant lag distribution of patents and Figure 4 the registration lag distribution of UMs. Interestingly, the distribution of UM publication lags is bimodal. As mentioned in the previous section, it is possible to postpone the publication of UMs by 15 months, and Figure 4 suggests that this option is exercised in a considerable number of UM filings.
+
+Figure 3. Grant lag distribution of German patents from 2000-2010
+
+Notes: Data source is the PATSTAT 2016 April edition. All priority patents were filed at the German patent office from 2000-2010. PCT filings are excluded.
+
+Figure 4. Registration/publication lag distribution of German utility models from 20002014
+
+Notes: The data are based on the PATSTAT 2016 April edition and cover all priority UMs filed at the German patent office from 2000-2010. Outliers, UMs with longer than 700 publication lags, are not reported, and PCT filings are excluded.
+
+A major strategic concern in patenting relates to the role of disclosure and the timing thereof. After having filed a patent, the application will be kept secret for 18
+
+---
+
+months and will then be published. The mean and median publication lags of patents reported in Table 2 are almost exactly 18 months (540 days). The published patent application will be available in the openly accessible DPMAregister database from the first publication date. $^10$ Figure 5 depicts the publication lags of patents. The 18-month secrecy period does not apply to UMs. Instead, if a UM application does not have any defects, or if the defects have been remedied, the UM is entered into the DPMAregister, and, hence, its contents become public knowledge. $^11$ For UMs, the publication date is the same as the registration date.
+
+Figure 5. Publication lag distribution of German patents from 2000-2014
+
+Note: Data source is the PATSTAT 2016 April edition. All priority patents filed at the German patent office from 2000-2014. Outliers, patents with longer than 700 publication lags, are not reported, and PCT filings are excluded.
+
+In Germany, a UM becomes effective upon registration and gives the same right to exclude others from using, producing, and marketing the protected invention as a patent. This makes it a potentially appropriate protection method for inventions whose owner needs quick enforcement against potential imitators. Patent applications, on the other hand, cannot be enforced when they are pending. Thus, while UMs can only be in force for 10 years, the difference in potential effective life (from grant to expiration) compared to patents is decreased by the fast grant. $^12$ Nonetheless, a UM is an unexamined IP right as it lacks substantive examination: patent examiners at the DPMA do not examine the novelty, inventive step, and industrial applicability of the invention prior to its registration. Therefore, inventors should conduct thorough searches to ensure that the application actually meets the requirements that apply to effective IP rights. Unexamined IP rights comprise inherent legal uncertainty (Suthersanen 2006; Königer 2009; Prud’homme 2014; Radauer et al. 2015) . Weak patents may be socially costly as
+
+---
+
+they can create the danger of patent hold-up, lead to costly litigations, and induce a vicious cycle of defensive patenting (Farrell and Shapiro 2009; Atal and Bar 2014).
+
+A peculiarity of the German UM system is the novelty grace period: If an inventor applies for a UM registration within six months from the publication of her invention, UM protection is still available (Radauer et al. 2015) . The patent systems of the US, Japan, and Canada apply grace periods, but European patent systems have not applied grace periods since the European patent convention was signed in 1973 (Franzoni and Scellato 2010) . This means that, in most European patent systems, publication of an invention destroys its novelty and a patent can no longer be granted (Franzoni and Scellato 2010) . In Germany the grace period of UMs enables inventors to protect their disclosed inventions with UMs.
+
+Moreover, in Germany, there is the possibility for double granting, i.e., an applicant can protect the same invention with both a patent and a UM (Guellec and van Pottelsberghe 2007; Prud’homme 2014; Radauer et al. 2015) . Splitting-off 13 a UM provides protection in the period between patent application and grant, when no or only limited protection is available. 14 By making a respective splitting-off declaration, an applicant may obtain an independent UM application, for which it is possible to claim the priority date from the patent application. Upon registration of the split-off UM, the invention enjoys full protection (a right to injunctive relief and claim for damages), irrespective of the outcome of the patent grant procedure. Hence, with complementary UM protection, it is possible to extend patent protection from the front end. Figure 3 provides an overview of patenting and UM processes in Germany and illustrates the gap in the time dimension of patent protection which the UM protection is filling.
+
+Figure 6. Patenting and UM registration processes in Germany
+
+---
+
+Although German UMs (and patents) are presumed valid until proven otherwise, the bifurcation principle (infringement and validity of a patent are decided independently by different courts) does not apply to UMs (Cremers et al. 2013; Cremers et al. 2014) . In UM disputes, the defendant is allowed to raise the invalidity defense (Cremers et al. 2013) , whereas this is not possible in patent suits. Then, cancellation proceedings will clarify whether the UM is actually valid, i.e., that the invention is new and involves an inventive step. $^15$ The UM infringement suit is then suspended until the resolution of cancellation proceedings. $^16$
+
+## 4 To patent or to register a utility model, or both?
+
+In this section, we discuss the possible determinants of the use of patents and UMs and develop testable hypotheses. The hypotheses are derived from the prior literature and economic theory, but they also mirror the particularities of the institutional context in which the firms in our data operated.
+
+Economic theory suggests that firms' IPR decisions are based on profit maximization and that firms take into account the prevailing IPR institutions when making those decisions. Surveys of patenting firms suggest that firms use patents for different strategic purposes (e.g., preventing imitation, securing freedom to operate (FTO), licensing, signaling capabilities to acquire funding, blocking, and pre-empting), and that these motives may differ by firm size and industry (Cohen et al. 2000; Blind et al. 2006; de Rassenfosse 2012; Holgersson 2013) . Presumably, the motives to use UMs should be similar to those of patents, as they are also a protection method for technical inventions. The two main differences between the German patents and UMs relate to the speed of protection and to the threshold of obtaining protection. However, the benefits and costs of using patents and UMs are expected to vary depending on firm
+
+---
+
+characteristics. Next, we develop hypotheses concerning the choices between UMs and patents.
+
+### 4.1 Economies of scale and scope in the use of IPRs
+
+Larger firms can benefit from economies of scale (Cohen et al., 2000) and learning when it comes to filing and using a particular type of IPR. They can spread the fixed costs of IPR activities over a larger number of innovations. In comparison to small firms, established firms have the financial capacity to develop internal R & D capabilities that generate more valuable innovations and, thus, patents with greater economic benefits (“valuable patents”) (Johnson 2002; Allison et al. 2004; Harhoff et al. 2003; Hussinger 2006) . They also often have an in-house R & D and/or IP department that is routinized in using the patent system (Hussinger 2006; Wagner 2006; Wang et al. 2009; Grimpe and Kaiser 2010) , whereas smaller firms possess fewer resources that they can invest in IP protection (Wernerfelt and Karnani 1987; Byma and Leiponen 2009) .
+
+There is no reason to expect that only small firms would use UMs, as it is unclear why large firms would not also want to register UMs to supplement their patent portfolios. Larger firms have the resources to invest in many kinds of R & D projects simultaneously, which increases their chances of discovering protectable inventions (Johnson 2002; Harhoff et al. 2003; Allison et al. 2004) . Such inventions can be incremental or more radical. Moreover, larger firms often have more product lines than small firms and, thus, a larger variety of potentially protectable technical inventions. Large firms have more resources and capabilities to utilize many kinds of IPR methods, which may lead to economies of scope. They can protect more valuable inventions with patents and register UMs for less valuable inventions (Beneito 2006) . In Germany, it is possible that inventions are protected with both a patent and a UM because the legislation allows double granting (Guellec and van Pottelsberghe 2007; Prud’homme
+
+---
+
+2014; Radauer et al. 2015) , as was explained in the previous section. Hence, the larger a firm is, the more likely it is to have several product inventions, some of which they protect with patents and some with UMs.
+
+Hypothesis 1: The likelihood to utilize patents and utility models simultaneously in protecting IP increases with firm size.
+
+### 4.2 Need for speed
+
+The DPMA states that UMs can be granted as quickly as four days after application and that they are typically granted in three to four weeks if a patent attorney is involved and, on average, in three months if no patent attorney is involved (Prud'homme 2014, p. 20). In contrast, Harhoff et al. (2015) report that the average duration of a patent examination at the DPMA was 2.6 years during the period from 1989 to 1996. Table 2 and Figures 2 and 3 in section 3.2 depict the difference in patent grant lag and UM registration lag distributions for the population of German priority filings from 20002010. The speed advantage of UMs compared to patents in obtaining protection for technical inventions is considerable and counted in years, rather than months (see Table 2).
+
+Radauer et al. (2015) have recently conducted a survey of UM users in Germany and find fast protection to be the most important motive to register UM among the 47 respondents. Furthermore, Radauer et al. (2015, p.32) report the most useful feature of UMs, which stood out in interviews with IP professionals and UM users, was speed: “Foremost to mention is speed. UMs are granted quickly and can be used particularly well for products that have a short product life cycle.” Hence, an important difference between patents and UMs lies in the time dimension of protection: patents offer slower but longer protection, whereas UMs offer fast but shorter protection (Cao 2014) .
+
+---
+
+Previous literature has defined the length of the product life cycle (i.e., product lifetime) as the time between a product's introduction and withdrawal from the marketplace (Bayus 1994, 1998).
+
+Particularly in dynamic industries, firms may rely more on informal protection methods, such as lead-time and secrecy, than on the slower patents (Cohen et al. 2000; Arundel 2001; Byma and Leiponen 2009; Thomä and Bizer 2013; Hall et al. 2014) . Therefore, we expect firms operating in markets with short life cycles of products and services to be more likely to use UMs and rate a higher importance to UMs relative to patents.
+
+Hypothesis 2a: Firms reporting to have short life cycles of products and services are more likely to use utility models than other firms are.
+
+Based on the arguments above, we developed a stated preference version of the previous hypothesis:
+
+Hypothesis 2b: Firms reporting to have short life cycles of products and services assign higher importance to utility models.
+
+It is possible that, in dynamic and fast-moving markets characterized by intense competition, the risk of overlapping invention is higher, irrespective of the life cycle of products. In this situation, firms require strong protection and may, therefore, prefer patents to uncertain UMs. Additionally, firms rarely possess “one patent per product”. Hence, even though product life cycles might be short, the underlying patented technology might still be applicable to new products; therefore, it would be rational to patent and protect inventions against competitors. If this reasoning were correct, then the association between short life cycles and the choice of UMs would be confounded.
+
+---
+
+## 5 Data and methods
+
+In this section, we present the data for analysis, the choice and construction of dependent and independent variables, and the estimation methods which we apply to test our hypotheses.
+
+### 5.1 Data sources
+
+We use the Mannheim Innovation Panel (MIP) 2005, MIP2005, which includes the core Eurostat Community Innovation Survey (CIS) questions and a few additional countryspecific questions. Previously, Thomä and Bizer (2013) utilized these data to study the innovation protection mechanisms of small German firms. They also provide a detailed description of how the MIP2005 data were collected. The questions in the MIP2005 refer to the three-year period from 2002 to 2004. Thus, the survey considers the period before the 2006 German Supreme Court's decision to abolish the inventive step difference between patents and UMs. As shown in Figure 1 , it seems that this legal change led to a slight, but not dramatic, decrease in the level of UM applications. The random sample of German firms is stratified by region, size, and sector. The survey contains information about the perceived business environment and firm characteristics, enabling us to test the developed hypotheses. Most importantly, the questionnaire includes questions about the protection methods that firms used to protect their intellectual property (IP).
+
+### 5.2 Variables
+
+### 5.2.1 Dependent variables
+
+We measure how firms use UMs and patents and their stated ranking (“stated preference”) of these two methods of IP protection. We, therefore, construct two
+
+---
+
+alternative dependent variables for our empirical analysis.
+
+The relevant questions in the MIP2005 ask whether the respondent firm had used the listed formal protection methods (patent, utility model, design right, trademark and copyright) to protect their IP during past three years. 17 We construct binary variables to indicate the use of patents (P) and UMs:
+
+$${  U s e  } _ { i } ^ { j } = \left\{ \begin{array} { c } { { 1 ,  i f ~ f i r m ~ i ~ u s e d ~ p r o t e c t i o n ~ m e t h o d ~ j ~ t o ~ p r o t e c t ~ I P ~  2 0 0 2 - 2 0 0 4 \, , } } \\ { { 0 ,  o t h e r w i s e  } } \end{array} \right.$$
+
+$$ where  j \in\{P,UM\} .$$
+
+Conditional on using a specific IP protection method, the respondents were also asked to rate the importance of the respective method on a three-point Likert scale: low importance, medium importance, and high importance. We assume that firms which did not use a specific protection method are considered to assign the lowest possible importance i.e. no importance to the protection method. Hence, we construct the following dependent variable when modelling the stated importance of protection methods in protecting IP:
+
+$$ I m p  _ { i } ^ { j } = \left\{ \begin{array} { c } { 0 ,  i f \, U s e  _ { i } ^ { j } = 0 } \\ { 1 ,  i f \, f i r m \, i \, {  r e p o r t s \, j \, t o \, b e \, o f \, l o w \, i m p o r t a n c e  } \\ { 2 ,  i f \, f i r m \, i \, {  r e p o r t s \, j \, t o \, b e \, o f \, m e d i u m \, i m p o r t a n c e  } \\ { 3 ,  i f \, f i r m \, i \, {  r e p o r t s \, j \, t o \, b e \, o f \, h i g h \, i m p o r t a n c e  } , } \end{array} \right.$$
+
+$$ where  j \in\{P,UM\} .$$
+
+### 5.2.2 Independent variables of key interest
+
+We measure firm size by logarithm of reported full-time equivalent employees in 2004, log (Employees). Short life cycle of products and services is proxied with a binary variable D (Short life cycle), which obtains value 1, if a firm totally agrees or agrees $^18$
+
+---
+
+that in its main market “products/services mature rapidly,” and 0 otherwise (cf. Thomä and Bizer 2013). We acknowledge that product and service life cycles differ across industries and that, in the survey, the term “rapidly” is subjectively interpreted. If D (Short life cycle) contains a high measurement error due to subjective interpretations (and, therefore, does not systematically measure how rapidly the products of firms mature and become obsolete), we are less likely to find evidence consistent with Hypotheses 2a and 2b.
+
+### 5.2.3 Control variables
+
+We control for the following firm characteristics. Firms' R & D investments are controlled for with the total R & D expenditure divided by fulltime equivalent employees. Since higher R & D intensity is likely associated with greater innovative output, i.e., more potentially patentable inventions, by holding R & D intensity constant, our independent variable logarithm of firm employees is a measure of economies of scale. We measure the importance of exports for each firm by exports per sales.
+
+Exporting firms need to be aware of national differences in IPR systems (Ginarte and Park 1997; Park 2008) in their target markets which could increase the relative importance of patents as all countries have patent systems, but a smaller share have UM systems (Kim et al. 2012). Group membership indicates whether a firm is a part of a larger group of firms. Firms that are part of a larger group, i.e., a conglomerate, may be able to leverage the knowledge and capabilities of their conglomerate to exploit IPR systems. The most robust finding in prior studies on the motives to patent is that the use differs across industries (Cohen et al. 2000; Arundel 2001; Hall et al. 2014).
+
+Furthermore, Johnson and Popp (2003) report varying grant lags across technologies.
+
+Therefore, to control for industry and technology heterogeneity, we include seven technology class dummies. We follow Eurostat's high-tech classification of
+
+---
+
+manufacturing industries (high tech, medium high tech, medium low tech, low tech) and knowledge-intensive business services (knowledge-intensive businesses, low knowledge-intensive businesses). 19 The seventh technology class consists of NACE Rev. 1.1 industries 40 (electricity, gas, steam, and hot water), 41 (supply, collection, purification, and distribution of water), and 45 (construction).
+
+### 5.3 Econometric models
+
+Our data are cross-sectional. We model the use and importance of patents and UMs as follows. First, we examine which firm characteristics are associated with the use of patents and UMs by estimating a bivariate probit model (Greene 2012, pp. 778-781). The bivariate probit model is appropriate for modeling joint determination of patent and UM use, as they are presumably interrelated.
+
+$$Use_{i}^{*, j}=\alpha^{j}+\beta_{1}^{j} \log \left( Employees _{i}\right)+\beta_{2}^{j} D\left( Short life cycle _{i}\right)+\delta^{\prime} x_{i}+\theta_{k}+\varepsilon_{i}^{j},$$
+
+$$Use_{i}^{j}=1\text{ if }Use_{i}^{*,j}>1,0\text{ otherwise}$$
+
+$$\left( \begin{array} { l } { \varepsilon _ { i } ^ {  p  } } \\ { \varepsilon _ { i } ^ {  U M  } } \end{array} \right| \log (  E m p l o y e e s  _ {  i  } ) ,  D  (  S h o r t \ l i f e \ c y c l e  _ {  i  } ) ,  x  _ {  i  } \right) \sim  N  \left[ \left( \begin{array} { l l } { 0 } \\ { 0 } \end{array} \right) , \left( \begin{array} { l l } { 1 } & {  p  } \\ {  p  } & { 1 } \end{array} \right) \right] ,$$
+
+where j ∈ { P, UM } and firm i belongs to technology class k, x is a vector of controls, ρ is the correlation between error terms, and, if ρ = 0, then the bivariate probit becomes two independent univariate probit models (Greene 2012, p. 782) . If ρ = 0, then there exists a disturbance correlation between the two equations, i.e., the unobservables affecting the concurrent choice of patent and UM use are correlated. In this case,
+
+---
+
+bivariate probit gives more efficient parameter estimates than separate probit models, which assume disturbances to be independent.
+
+Second, we study the interrelated importance of patents and UMs by estimating a bivariate ordered probit model which is a direct extension of a univariate ordered probit model (Greene and Hensher 2010) . The model is the same as the above-presented bivariate probit model except that the dependent variable is now an ordinal variable indicating the importance of patents and UMs (see section 5.2.1 ).
+
+We are not claiming any causal relationships, but we try to identify statistical associations between firm characteristics and the use and importance of IP protection methods. In particular, we are not investigating the decision to start using a specific IP protection method, but rather the prevailing status of a firm in utilizing IP protection methods (during the past three years, i.e., 2002-2004). The firms that have been innovative in the past may or may not have chosen to protect their inventions with patents or UMs or both. $^20$ Therefore, we did not want to narrow our baseline sample and estimations to only recently innovative firms (i.e., firms that introduced innovation from 2002 to 2004) or to R & D active firms (firms with R & D investments from 2002 to 2004).
+
+We acknowledge that the firms may have protected their inventions and IP with several other protection methods than UMs and patents (cf. Thomä and Bizer 2013). In order to be concise, we leave them out of the current analysis and focus on the relationship between patents and UMs.
+
+## 6 Empirical analysis
+
+### 6.1 Descriptive analysis
+
+Table 3 presents the descriptive statistics by the categories of patent and UM users. A
+
+---
+
+total of 75 % of our sample firms (2,265/3,016) used neither patents nor UMs to protect their IP. These firms are the smallest, invest the least in R & D, and are the least exportintensive, on average. A quarter of these firms report having short product and service life cycles. A minority, 4.2 % (127/3,016), of the sample firms use UMs only. These firms are smaller on average, invest less in R & D per employee, and have a lower export share of turnover than patenting firms.
+
+In line with the need for speed Hypothesis 2a, UM users (36.2 % ) frequently report short life cycles of products and services, although the differences between these users and both patent only users (32.1 % ) and patent and UM users (35.5 % ) are small. Consistent with the economies of scale and scope Hypothesis 1, the firms which use both protection methods to protect IP are large and, on average, have the most employees. Firms using patents only have, by far, the largest mean R & D expenditure.
+
+## Table. 3 Descriptive statistics
+
+Figure 7 illustrates how the use of patents and UMs is associated with firm size categories measured by the number of employees. Firms are classified into three categories based on number of employees: 1-49 employees, 50-249 employees, and more than 250 employees. The majority of sample firms do not use any protection methods in all size categories. The share of firms utilizing both protection methods increases with firm size. This observation is in line with the economies of scale and scope Hypothesis 1, although it should be noted that the industry composition is likely to differ between size categories. Similarly, the share of firms using only patents increases with firm size, whereas the proportion of firms using only UMs is the largest among medium-sized (50-250 full-time equivalent employees) enterprises (5.5 % ). The small proportions of firms using only UMs in all size categories indicates that UM
+
+---
+
+protection is a supplemental protection method for patenting firms rather than a main protection method on its own.
+
+Figure 7. The use of protection methods by firm size (employees) categories
+
+Note: Firms are classified into three size categories: 1-49 employees, 50-249 employees and more than 250 employees.
+
+Figure 8 shows that those firms that have short life cycles of products or services are more active users of patents and UMs. Among firms which have short life cycles of products and services, 16.1 % used UMs to protect intellectual capital, whereas, among firms which did not report short life cycles of products or services, 11 % used UMs. This is consistent with Hypothesis 2a.
+
+Figure 8. The use of protection methods by "need for speed"
+
+### 6.2 Econometric results
+
+### 6.2.1 The use and importance of patents and utility models
+
+First, we estimate a biprobit model to analyze the determinants of concurrent patent and UM utilization. Table 4 reports the estimated coefficients (columns 1-2) and marginal effects (columns 3-6).
+
+Table 4. The use of protection methods, bivariate probit model
+
+Notes: The dependent variables are indicator variables for the use of patents and utility models. First two columns report the estimated coefficients for bivariate probit model. Columns 3-6 present the marginal effects at means for patent and utility model
+
+---
+
+combinations. Heteroscedasticity robust standard errors in parentheses, *** p $<0.01, * *$ p $<0.05, * p<0.1$.
+
+Regarding the combined use of patents and UMs, we find support for our economies of scale and scope Hypothesis 1: firm size is positively associated with the likelihood to use both protection methods (columns 1 and 2). The observation is consistent with prior literature, which suggests larger firms to be more capable users of formal protection methods (Cohen et al. 2000; Hussinger 2006; Byma and Leiponen 2009; Hall et al. 2013; Thomä and Bizer 2013) . Firm size is also positively associated with the likelihood to use only patents (column 4) and only UMs (column 5), but the estimated marginal effect is the highest in the case of concurrent use of both protection methods. Moreover, in line with Hypothesis 2a, the estimates of Table 4 suggest that a short life cycle of products and services is associated with an increased likelihood to use only UMs or both patents and UMs. In line with our expectations, the association is more pronounced between short life cycles and the use of only UMs, an increase of 2.3 percentage points (column 5), than between short life cycles and the use of patents and UMs, an increase of 1.5 percentage points (column 3).
+
+A statistically significant positive $\rho$ , i.e., the correlation of error terms, indicates that the uses of patents and UMs are not independent of each other, conditional on the included covariates. 21 Furthermore, a positive $\rho$ is in line with the view that patents and UMs are complements rather than substitutes.
+
+The signs of control variables are in line with our expectations. We find R & D per employee to be positively associated with the likelihood to use patents and both protections methods. This is in line with the expectation that more important and costly inventions are protected with patents, whereas UMs are appropriate protection methods for incremental improvements (Johnson 2002; Beneito 2006; Kim et al. 2012) . The
+
+---
+
+analysis of marginal effects indicates that the positive association between R&D and the likelihood to use utility UMs is, in particular, driven by firms that use both patents and UMs, rather than by firms that use only UMs, since the estimated marginal effect for the former is more than six times larger than that of the latter.
+
+The observed positive association between export intensity and the likelihood to use patents or both protection methods is consistent with the fact that exporting firms need to protect their inventions in target markets which more often have regular patent systems than two-tiered patent systems (e.g., the US and the UK) (Kim et al. 2012) . The non-significant association between export intensity and likelihood to use only UMs highlights the importance of the UM system for firms mainly operating in Germany. It is likely that German firms which export within the EU protect their inventions with European patents granted by the EPO.
+
+Next, we explore the factors associated with the stated importance of patents and UMs. Table 5 shows the results of the bivariate ordered probit estimation. Generally, the estimates corroborate our findings about the use of patents and UMs presented in Table 4 . The need for speed Hypothesis 2b is supported as we find short life cycles of products and services to be associated with the importance of UMs, but not with the importance of patents. Again, the statistically significant positive $\rho$ suggests that there exists a disturbance correlation between the equations, i.e., the reported importance of patents and UMs are related. Again positive $\rho$ indicates that patents and UMs are complements rather than substitutes. The signs of the estimated coefficients of the control variables are similar between the bivariate probit model in Table 4 and the ordered probit model in Table 5 : firm size, R & D intensity, and export intensity are all positively and significantly associated with the importance of both patents and UMs. Thus, the use and importance of protection methods are closely linked, as expected.
+
+---
+
+Table 5. The stated importance of UMs and patents, bivariate ordered probit model Notes: The dependent variables are ordinal variables indicating the importance of patents and utility models as IP protection methods. 12 firms, which reported to use patents and/or utility models but did not answer the importance question, are omitted. Heteroscedasticity robust standard errors in parentheses, *** p $<$ 0.01, ** p $<$ 0.05, * p $<$ 0.1.
+
+### 6.2.2 Robustness checks22
+
+First, various prior studies which have utilized CIS data have focused on subsamples of “ recent innovators ” (e.g., Hussinger 2006; Thomä and Bizer 2013; Hall et al. 2013). Hence, we re-estimated the use of protection method models (Table 4) for the subsample of innovating firms (ones that introduced a product or process innovation during 2002-2004). The results are reported in Table A.1 in the Appendix. As in Table 4, we find firm size to be positively associated with the use of patents and UMs, and the estimated marginal effects are of similar magnitude. Also, short life cycles of products and services are found to be positively associated with the use of only UMs. The estimated marginal effects provide weak support that having a short life cycle of products and services is negatively associated with the use of patents only. In the case of the importance of patents and UMs, when we restrict the sample to innovators, we still find firm size to be positively associated with the use of patents and UMs. The positive association between UMs and short life cycles of products and services is also found, but the effect is somewhat weaker statistically.
+
+Second, since UMs are arguably somewhat cheaper to acquire than patents (Königer 2009) , we control for reported financial constraints, i.e., if the firm reported lack of financing (internal or external) to be an obstacle for innovation. Another reason to control for financial constraints is that patents may be more important than UMs in convincing investors (especially international investors who may not be equally familiar
+
+---
+
+with UM protection) and acquiring financial resources (Mann and Sager 2007; Hoenen et al. 2014; Hoenig and Henkel 2015) . We also control for the level of competition in the firm's main market. $^23$ The association between competition and the use of patents and UMs is unclear. On the one hand, stronger competition may induce firms to “escape competition” by innovating (Aghion et al. 2005) and by protecting their IP more intensively. On the other hand, more intense competition in fast-moving markets may induce firms to forego IP protection altogether and focus on lead time advantages. Nevertheless, the results are robust for these controls on financial constraint and competition; firm size remains significantly positively associated with the use of and importance of patents and UMs. Short life cycle of products and services is also found to be positively associated with the use and importance of UMs, but not with the use and importance of patents. Hypothesis 2a, concerning the positive significant association between short product life cycle and the likelihood to use (only) UMs, gets further support.
+
+Interestingly, we find a statistically significant positive association between the reported financial constraints and the use and importance of patents and no association with the financial constraints and the use and importance of UMs (Columns 1 and 2 in Table A.3). In particular, having financial constraints is positively associated with the likelihood to use only patents and not UMs (Column 4 in Table A.3). It is estimated that firms reporting financial constraints are 1.9 percentage points more likely to use only patents than firms which do not report financial constraints. This could be in line with the view that patents are more important than UMs in acquiring external financing, as firms signal the quality of their invention with patents (Mann and Sager 2007; Hoenen et al. 2014; Atal and Bar 2014; Hoenig and Henkel 2015). On the other hand, it could also indicate that patents protect more significant inventions which require additional
+
+---
+
+financing for further development and commercialization. Nevertheless, our results do not support the view that financial constraints push firms to protect their inventions with “cheap” UMs instead of “expensive” patents.
+
+Although none of the estimated coefficients for competition categories is statistically significant in Tables A.3 and A.4 , the signs suggest a pattern that having zero competitors is associated with a lower likelihood to use patents and UMs than having more than 15 competitors (the reference group in estimations), whereas having 1-15 competitors is associated with a higher likelihood to use patents and UMs. This is consistent with the “ escape competition effect ” and in line with the inverted U-shaped association between competition and innovation (Aghion et al. 2005) .
+
+### 6.2.3 Limitations
+
+An important limitation of the study is the focus on firm-level instead of invention-level utilization of patents and UMs (Hussinger 2006; Hall et al. 2013; Heger and Zaby 2013) . In reality, patent and UM protection are utilized at the invention level (i.e., project -level) and different inventions require different protection as they might be of a different inventive step and have life cycles of a different length. Thus, when a firm has several product lines and services, the stated use and importance of IP protection methods probably reflect the overall importance of separate protection methods, i.e., “aggregated preferences.” In other words, firms with multiple product lines and services may have short life cycles in their main market, but use patents and UMs to protect complementary products. Thus, although we control for R & D intensity per employee and firm size, the patterns that we observe might be driven by omitted variable bias: the number of product lines and services.
+
+One source of bias is that our measure for short life cycle of products is a coarse proxy and is subjectively assessed by survey respondents. It is likely that this leads to
+
+---
+
+attenuation bias, i.e., the estimated coefficient for short life cycle indicator is biased towards zero. Another possible source of bias is reverse causality: firms that are innovative protect their inventions with patents and UM, which gives them competitive advantage and helps them grow. Thus, the association between firm size and the use of patents and UMs is an endogenous process. Consequently, our estimates of this association are likely to be biased upwards. With the current data, we are not able to account for this endogeneity problem.
+
+Our study focuses on a period before the German Supreme Court 2006 decision to abolish the inventive step difference between UMs and patents. Therefore, the external validity of our results might be diminished since the current German system is different from the system prior to the decision. As shown in Figure 2 , the legal change did not lead to a sudden drop in UM filings, indicating that the lower inventive step requirement has not been the dominant motive to file UMs. Moreover, Figure A.1 in the Appendix shows that the registration lags of German UMs have been consistently shorter than grant lags of German patents even after the 2006 amendment. These observations are in line with Radauer et al.'s (2015) finding that the speed of protection was mentioned as the most important motive to file UMs, instead of protection for minor inventions. However, an important topic for further research is to analyze how the use of German UMs has evolved over time and how legal changes have affected filing activity.
+
+## 7 Discussion and conclusion
+
+We study the use and the relative importance of patents and UMs among German firms. The results suggest that larger firms tend to take advantage of both patent and UM systems. These observations indicate economies of scale and scope and are consistent with the view that larger firms have more resources and capabilities to exploit IPR
+
+---
+
+systems. Furthermore, in line with the publicly expressed justification of the German UM system as “ fast IP right ” , we find short life cycles of products and services to be associated with the use and stated importance of UMs.
+
+Our study is related to the ongoing process of harmonization of IPR systems within the European single market (see European Commission 2011). Obviously, differences in national IPR systems increase transaction costs. Thus, it is likely that, for many inventors and SMEs, the patent system remains too complex and expensive and, therefore, hardly accessible. In this type of environment, larger and more experienced firms have the upper hand. In addition to a deepening harmonization of IPR systems in Europe and across the world, policy could focus on increasing the awareness of UMs as alternative protection methods and concurrently avoid an IPR environment that is excessively complex for SMEs and individual inventors. Especially firms and inventors from those EU member states which do not provide UM protection might be unaware of the UM option in other EU member states. In contrast, firms and inventors from the EU member states with UM systems are likely to gain a better understanding of the (dis)advantages of utility models relative to patents as legal protection mechanism. Hence, increasing the awareness of national IPR protection methods remains of the utmost importance in leveling the playing field between firms in the European single market.
+
+Two-tiered patent systems provide a variety of interesting theoretical and empirical questions which have yet to be answered. A natural way to deepen the analysis of the relative importance of patents and UMs is to study patenting behavior with patent and UM data (e.g., PATSTAT) which are linked to data on applicant characteristics. As large firms seem to be active users of the UM system, an important topic for future research is how the UM system affects industry dynamics and
+
+---
+
+competition between large incumbents and smaller entrants. Future studies could also investigate the underlying strategic motives to use UMs and how these motives vary across countries and industries.
+
+## Acknowledgements
+
+We thank Cristiano Antonelli, Mika Kortelainen, Ari Hyytinen, Mika Maliranta, Jaana Rahko, Knut Blind, Keld Laursen, Pierre Mohnen, Pia Björkwall and two anonymous referees for valuable comments. Earlier versions of this paper have been presented at the XXXVIII Annual Meeting of the Finnish Economic Association in Pori, the Micro&IO and Econometrics Workshops of the Finnish Doctoral Programme in Economics in Helsinki, at the Jyväskylä University School of Business and Economics breakfast seminar, at the TU Berlin Innovation Research Colloquium and at the Strategic Management Society Conference in Berlin. The financial support from the Yrjõ Jahnsson Foundation and the OP-Pohjola group research foundation is gratefully acknowledged. The authors would like to thank ZEW, Mannheim for sharing the data.
+
+## Notes
+
+1. The Trade-Related Aspects of Intellectual Property Rights (TRIPS) Agreement is Annex 1C of the Marrakesh Agreement Establishing the World Trade Organization, signed in Marrakesh, Morocco on 15 April 1994. See: https://www.wto.org/english/docs_e/legal_e/27trips_01_e.htm
+
+2. UMs are often considered a low cost alternative to patents due to lower administrative and maintenance fees. However, Radauer et al. (2015) point out that the cost savings aspect of UM might be overstated because administrative costs are only a small part of total patenting costs; in practice, patent attorneys are frequently hired to draft applications, and their fees are much higher than administrative application costs. See also van Pottelsberghe and Francois (2009) and Königer (2009) .
+
+3. In Europe, the harmonization of IPR systems has been a top priority (see European Commission 2011), but national UM systems have been ranked low on the agenda since the European Union member states suspended the proposal for a “community utility model”(European Commission 1995) in March 2000 (European Commission 2002). However, after the unitary patent system is in place and working, it is likely that more attention will be directed to the harmonization of national UM systems.
+
+4. https://www.muenchen.ihk.de/de/rceht/patentrecht-designsehutz/weleher-schutz-fuer-was/patent-und-gebrauchsmusterrecht Accessed on 28 January 2016.
+
+---
+
+5. http://www.whip.de/patentanwalt/gebrauchsmuster_anmelden.html#allgemeine_informationen Accessed on 28 January 2016.
+
+6. This section is mainly based on information retrieved from the DPMA's webpage.
+
+7. http://dpma.de/english/utility_models/index.html Accessed on 25 September 2015.
+
+8. The source of data is the PATSTAT April 2016 edition. Populations consist of German priority patent and UM filings. PCT applications are excluded.
+
+9. We report the grant lags for 2000-2010 instead of 2000-2014 due to a truncation problem: the more recent a patent filing is, the more likely it is still pending and does not enter grant lag mean and median calculations. See Figure A.1 in the Appendix.
+
+10. https://register.dpma.de/DPMAregister/Uebersicht?lang-en Accessed on July, 13, 2015.
+
+11. Upon request, publication of the UM can be postponed for up to 15 months beginning from the filing date, but, of course, protection commences with publication of the UM. See §§ 8 Abs. 1 GebrMG, 49 Abs. 2 PatG. See also: http://www.whip.com/germany/german_utility_model.html Accessed on July 13, 2015. http://www.vo.eu/en/news/item/476/german_utility_model_alternative_to_national_patent Accessed on July, 13, 2015.
+
+12. In practice, the duration of most patents is also less than 20 years, since, in many cases, applicants let patents lapse by not paying the renewal fees (Guellec and van Pottelsberghe 2007; Clarhoff et al. 2015). Guellec and van Pottelsberghe (2007) report that 50 % of patents granted by the EPO lapse within first 10 years and only 8 % of EPO patents are renewed until the statutory term.
+
+13. Also referred to as "branching off" by Radauer et al. (2015).
+
+14. http://dpma.de/english/utility_models/procedure/index.html Accessed on July, 13, 2015.
+
+15. http://dpma.de/english/utility_models/procedure/index.html Accessed on September 25, 2015.
+
+16. Gebrauchsmutergesetz (GebrMG) (adapted on July 31, 2009) §19.
+
+17. Question 13.1. (Formale Maßnahmen) in the MIP 2005: “Hat Ihr Unternehmen in den Jahren 2002-2004 eine der folgenden Schutzmaßnahmen für geistiges Eigentum genutzt?”
+
+18. A four-point Likert scale (totally agree, agree, disagree, totally disagree).
+
+19. http://ec.europa.eu/eurostat/statistics-explained/index.php/Glossary:Iliphtech_classification_of_manufacturing_industries Accessed on September 16, 2015.
+
+20. It is possible that a firm had patented an invention in the 1980s and that this patent was still used to protect the firm's IP in the period from 2002-2004.
+
+21. Separate probit models would not take into account this correlation of unobserved determinants of patents and UMs and would therefore be a less efficient alternative. However, we also estimated separate probit models and obtained similar results.
+
+---
+
+22. The results of robustness checks are presented in the Appendix.
+
+23. In the MIP2005 questionnaire, the firms were asked, “How many competitors has your firm in its main market?”: 0, 1-5, 6-15, 15-. We constructed dummy variables to control for the level of competition.
+
+## References
+
+Aghion, P., Bloom, N., Blundell, R., Griffith, R. and Howitt, P. 2005 “Competition and innovation: an inverted U-relationship.” Quarterly Journal of Economics 120: 701–728.
+
+Allison, J., Lemley, M. and Walker, J. 2004. "Extreme Value or Trolls on Top? The Characteristics of the Most Litigated Patents." University of Pennsylvania Law Review 158 (1): 1-37.
+
+Anton, J. and Yao, D. 2004. "Little Patents and Big Secrets: Managing Intellectual Property." RAND Journal of Economics 35 (1): 1-22.
+
+Arundel, A. 2001. "The relative effectiveness of patents and secrecy for appropriation." Research Policy 30 (4): 611-624.
+
+Arundel, A. and Steinmueller, E. 1998. "The use of patent databases by European small and medium-sized enterprises." Technology Analysis & Strategic Management 10 (2): 157-173.
+
+Atal, V. and Bar, T. 2014. "Patent quality and a two-tiered patent system." The Journal of Industrial Economics 62(3): 503-540.
+
+Bayus, B., 1994. "Are Product Life Cycles Really Getting Shorter?" Journal of Product Innovation Management 11: 300-308.
+
+Bayus, B., 1998. "An Analysis of Product Lifetimes in a Technologically Dynamic Industry." Management Science 44: 763-775.
+
+Beneito, P. 2006. "The innovative performance of in-house and contracted R&D in terms of patents and utility models." Research Policy 35: 502-517.
+
+Björkwall, P. 2009. “Nyttighetsmodeller : Ett ändamålsenligt innovationsskydd?” Doctoral dissertation. Svenska handelshögskolan, Helsinki.
+
+Blind, K., Edler, J., Frietsch, R. and Schmoch, U. 2006. "Motives to patent: Empirical evidence from Germany." Research Policy 35: 655-672.
+
+Boztosun, N. 2010. "Exploring the utility models for fostering innovation." Journal of Intellectual Property Rights 15: 429-439.
+
+---
+
+Byma, J. and Leiponen, A. 2009. "If you cannot block, you better run: Small firms, cooperative innovation, and appropriation strategies." Research Policy 38: 14781488.
+
+Cao, S., Lei, Z. and Wright, B. 2014. "Speed vs. length of patent protection: Evidence from innovations patented in U.S. and China." Mimeo.
+
+Cohen, W., Nelson, R., and Walsh, J. 2000. “Protecting their intellectual assets: Appropriability conditions and why U.S. manufacturing firms patent (or not).” NBER Working Paper No. 7552.
+
+Cremers, K., Ernicke, M., Gaessler, F., Harhoff, D., Helmers, C., McDonagh, L., Schliessler, P. and van Zeebroeck, N. 2013. “Patent litigation in Europe.” ZEW Discussion Paper No. 13-072.
+
+Cremers, K., Gaessler, F., Harhoff, D. and Helmers, C. 2014. "Invalid but infringed? An analysis of Germany's bifurcated patent litigation system." ZEW Discussion Paper No. 14-072.
+
+Cummings, P. 2010. "From Germany to Australia - Opportunity For a Second Tier Patent System in the US." Michigan State Journal of International Law 18 (2): 297-322.
+
+Denicolò, V. and Franzoni, L. 2004. "The contract theory of patents." International Review of Law and Economics 23: 365-380.
+
+de Rassenfosse, G. 2012. “How SMEs exploit their intellectual property assets: evidence from survey data.” Small Business Economics 39: 437-452.
+
+Encaoua, D., Guellec, D. and Martinez, C., 2006. "Patent systems for encouraging innovation: Lessons from economic analysis." Research Policy 35: 1423-1440.
+
+European Commission. 1995. "The Protection of Utility Models in the Single Market." Green Paper COM(95), 370 final. Brussels, 19.07.1995.
+
+European Commission. 2002. “Summary report of replies to the questionnaire on the impact of the Community utility model with a view to updating the Green Paper on protection by the utility model in the internal market.” SEC(2001)1307.
+
+European Commission. 2011. “A Single Market for Intellectual Property Rights: Boosting creativity and innovation to provide economic growth, high quality jobs and first class products and services in Europe.” COM(2011) 287 final, Brussels, 24.5.2011.
+
+Farrell, J. and Shapiro, C. 2008. "How strong are weak patents?" American Economic Review 98 (4): 1347-1369.
+
+---
+
+Franzoni, C. and Scellato, G. 2010. "The grace period in international patent law and its effect on the timing of disclosure." Research Policy 39: 200-213.
+
+German and Patcnt and Trademark Office. 2013. “Utility Models: An information brochure on utility model protection.” http://dpma.de/docs/service/veroeffentlichungen/broschueren_en/utilitymodels_ engl.pdf (Accessed 8.9.2015)
+
+Ginarte, J. and Park, W. 1997. "Dctcrminants of patent rights: A cross-national study." Research Policy 26: 283-301.
+
+Graham, S. and Hegde, D. 2015. "Disclosing patents' secrets." Science 347 (6219): 236-237.
+
+Greene, W. 2012. Econometric Analysis. Seventh Edition. Pearson Education Limited.
+
+Greene, W. and Hensher, D. 2010. Modeling ordered choices. Cambridge University Press.
+
+Grimpe, C. and Kaiser, U. 2010. "Balancing Internal and External Knowledge Acquisition: The Gains and Pains from R&D Outsourcing." Journal of Management Studies 47 (8): 1483–1509.
+
+Grosse Ruse-Khan, H. 2013. "Utility Model Protection in Pakistan – An Option for Incentivising Incremental Innovation." A study for the WIPO.
+
+Grosse Ruse-Khan, H. 2012. “The international legal framework for the protection of utility models.” Paper prepared for the WIPO Regional Seminar on the Legislative, Economic and Policy Aspects of the Utility Model System, Kuala Lumpur (Malaysia), 3-4 September 2012.
+
+Guellec, D. and van Pottelsberghe de la Potterie, B. 2007. The economics of the European patent system. Oxford University Press.
+
+Guellec, D., Martinez, C. and Zuniga, P. 2012. “Pre-emptive patenting: securing market exclusion and freedom of operation.” Economics of Innovation and New Technology 21 (1): 1-29.
+
+Hall, B., Helmers, C., Rogers, M. and Sena, V. 2014. "The choice between formal and informal intellectual property: A review." Journal of Economic Literature 52 (2): 375-413.
+
+Hall, B., Helmers, C., Rogers, M. and Sena, V. 2013. "The importance (or not) of patents to UK firms." Oxford Economic Papers 65 (3): 603-629.
+
+Harhoff, D., Scherer, F.M. and Vopel, K. 2003. "Citations, family size, opposition and the value of patent rights." Research Policy 32 (8): 1343-1363.
+
+---
+
+Harhoff, D., Rudyk, I. and Stoll, S. 2015. "Deferred patent examination." DRUID working paper. Presented at DRUID15 Conderence in Rome 15.-17.6.2015.
+
+Heger, D. and Zaby, A. 2013. “The heterogeneous costs of disclosure and the propensity to patent.” Oxford Economic Papers 65 (3): 630-652.
+
+Hoenen, S., Kolympiris, C., Schoenmakers, W. and Kalaitzandonakes, N., 2014. “The diminishing signaling value of patents between early rounds of venture capital financing.” Research Policy 43: 956–989.
+
+Hoenig, D. and Henkel, J. 2015. "Quality signals? The role of patents, alliances, and team experience in venture capital financing." Research Policy 44: 1049-1064.
+
+Holgersson, M. 2013. “Patent management in entrepreneurial SMEs: a literature review and an empirical study of innovation appropriation, patent propensity, and motives.” R&D Management 43 (1): 21-36.
+
+Hu, A. and Jefferson, G. 2009. "A great wall of patents: What is behind China's recent patent explosion?" Journal of Development Economics 90: 57-68.
+
+Hussinger, K. 2006. "Is silence golden? Patents versus secrecy at the firm level." Economics of Innovation and New Technology 15 (8): 735-752.
+
+Janis, M. 1999. "Second tier patent protection." Harvard International Law Journal 40 (1): 151-219.
+
+Jell, F., Henkel, J. and Hoisl, K. 2013. "The relationship between patenting motives and pendency durations." Mimeo.
+
+Johnson, D. 2002. “’Learning-by-Licensing’: R&D and Technology Licensing in Brazilian Invention.” Economics of Innovation and New Technology 11 (3): 163177.
+
+Johnson, D. and Popp. D. 2003. “Forced out of the closet: The impact of the American inventors protection act on the timing of patent disclosure.” RAND Journal of Economics 34 (1): 96–112.
+
+Johnson, M., Bialowas, A., Nicholson, P., Mitra-Kahn, B., Man, B. and Bakhtari, S. 2015. “The economic impact of innovation patents.” IP Australia Economic Research Paper 05.
+
+Kim, Y., Lee, K., Park, W. and Choo, K. 2012. “Appropriate intellectual property protection and economic growth in countries at different levels of development.” Research Policy 41 (2), 358-375.
+
+Kingston, W. 2001. "Innovation needs patent reform." Research Policy 30: 403-423.
+
+---
+
+Kultti, K., Takalo, T. and Toikka, J. 2007. "Secrecy versus patenting." RAND Journal of Economics 38(1): 22-42.
+
+Königer, K. 2009. “Registration without examination: Utility model – a useful model?” In Patents and Technological Progress in a Globalized World edited by Amicorum, L. and Straus, J. MPI Studies on Intellectual Property, Competition and Tax Law Volume 6.
+
+Lemley, M., Lichtman, D. and Sampat, B. 2005. "What to do about bad patents?" Regulation 28 (4): 10-13.
+
+Mann, R. and Sager, T., 2007. "Patents, venture capital, and software start-ups." Research Policy 36: 193-208.
+
+Maskus, K. and McDaniel, C. 1999. "Impacts of the Japanese patent system on productivity growth." Japan and the World Economy 11: 557-574.
+
+Maurseth, P. Botolf and Svensson, R. 2014. "Micro evidence on international patenting." Economics of Innovation and New Technology 23 (4): 398-422.
+
+Mazzoleni, R. and Nelson, R., 1998. "The benefits and costs of strong patent protection: a contribution to the current debate." Research Policy 27: 273-284.
+
+Moser, P. 2005. "How Do Patent Laws Influence Innovation? Evidence from Nineteenth-Century World's Fairs." American Economic Review 95 (4): 12141236.
+
+Park, W., 2008. "International patent protection: 1960-2005." Research Policy 37: 761766.
+
+van Pottelsberghe de la Potterie, B. and Francois, D. 2009. "The cost factor in patent systems." Journal of Industry, Competition and Trade 9: 329-355.
+
+Prud’homme, D. 2014. “Creating a ‘model’ utility model system: A comparative analysis of the utility model systems in Europe and China.” IP Key Project Working Paper Series. Available at: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2541900
+
+Radauer, A., Rosemberg, C., Cassagncau-Francis, O., Goddar, H. and Haarmann, C. 2015. "Study on the economic impact of the utility model legislation in selected Member States: Final Report." A study tendered by the European Commission – DG Internal Market and Services in 2013, MARKT/2013/065/D2/ST/OP.
+
+Suthersanen, U. 2006. "Utility models and innovation in developing countries." The International Centre for Trade and Sustainable Development Issue Paper 13, UNCTAD.
+
+---
+
+Thomä, J. and Bizer, K. 2013. “To protect or not to protect ? Modes of appropriability in the small enterprise sector.” Research Policy 42 (1): 35-49.
+
+Wagner, S. 2006. "Make-or-buy decisions in patent related services." Munich School of Management Discussion Papers in Business Administration, No. 2006-16.
+
+Wang, H., He, J. and Mahoney, J. 2009. “Firm-specific knowledge resources and competitive advantage: the roles of economic- and relationship-based employee governance mechanisms.” Strategic Management Journal 30 (12): 1265-1285.
+
+Wernerfelt B. and Karnani, A. 1987. "Competitive strategy under uncertainty." Strategic Management Journal 8 (2): 187-194.
+
+van Zeebroeck, N. and van Pottelsberghe de la Potterie. "Filing strategies and patent value." Economics of Innovation and New Technology 20 (6): 539-561.
+
+---
+
+Figure 1
+
+![Figure](figures/figure_001.png)
+
+Figure 2
+
+![Figure](figures/figure_003.png)
+
+---
+
+Figure 3
+
+![Figure](figures/figure_001.png)
+
+Figure 4
+
+![Figure](figures/figure_003.png)
+
+---
+
+Figure 5
+
+![Figure](figures/figure_001.png)
+
+Figure 6
+
+![Figure](figures/figure_003.png)
+
+---
+
+Figure 7
+
+![Figure](figures/figure_001.png)
+
+Figure 8
+
+![Figure](figures/figure_003.png)
+
+---
+
+Table 1
+
+<table><tr><td></td><td>Patent</td><td>Utility model</td></tr><tr><td>Maximum term of protection</td><td>20 years</td><td>6-15 years</td></tr><tr><td>Inventive step requirement</td><td>Yes</td><td>Typically lower</td></tr><tr><td>Novelty requirement</td><td>Global</td><td>Global or local</td></tr><tr><td>Examination at the patent office</td><td>Yes</td><td>No*</td></tr><tr><td>Publication lag from filing</td><td>18 months</td><td>Some months</td></tr><tr><td>Grant lag</td><td>Years</td><td>Typically months</td></tr><tr><td>Process inventions</td><td>Yes</td><td>No</td></tr></table>
+
+
+Table 2
+
+<table><tr><td></td><td>Observations</td><td>Mean</td><td>St.</td><td>Median</td><td>Min</td><td>Max</td></tr><tr><td colspan="7">Grant/registration lags 2000-2010</td></tr><tr><td>Grant lag of patents</td><td>125319</td><td>1364.91</td><td>970.44</td><td>1025</td><td>204</td><td>5776</td></tr><tr><td>Registration lag of utility models</td><td>156657</td><td>185.07</td><td>215.99</td><td>107</td><td>8</td><td>3710</td></tr><tr><td colspan="7">Publication lags 2000-2014</td></tr><tr><td>Patents</td><td>578598</td><td>540.01</td><td>105.50</td><td>554</td><td>44</td><td>5769</td></tr><tr><td>Utility models</td><td>194242</td><td>176.58</td><td>203.87</td><td>101</td><td>3</td><td>3553</td></tr></table>
+
+
+---
+
+Table 3
+
+<table><tr><td></td><td colspan="3">No patents or utility models</td><td colspan="10">Theuse of patents and utility models</td></tr><tr><td></td><td colspan="2">No patents or utility models</td><td colspan="3">Only utility models</td><td colspan="3">Only patents</td><td colspan="3">Parents and utility models</td><td colspan="3">Total</td></tr><tr><td></td><td>N</td><td>Mean</td><td>s.d.</td><td>N</td><td>Mean</td><td>s.d.</td><td>N</td><td>Mean</td><td>s.d.</td><td>N</td><td>Mean</td><td>s.d.</td><td>N</td><td>Mean</td><td>s.d.</td></tr><tr><td>Employees</td><td>2265</td><td>174.92</td><td>1322.21</td><td>127</td><td>214.34</td><td>943.63</td><td>249</td><td>309.07</td><td>597.74</td><td>375</td><td>1839.66</td><td>11724.51</td><td>3016</td><td>402.10</td><td>4330.27</td></tr><tr><td>RsD-3 per employee (€)</td><td>2265</td><td>3539.75</td><td>26101.05</td><td>127</td><td>1464.54</td><td>7709.04</td><td>249</td><td>3038.13</td><td>61539.98</td><td>375</td><td>14642.45</td><td>23650.01</td><td>3016</td><td>7600.22</td><td>34691.19</td></tr><tr><td>Export share</td><td>2265</td><td>0.889</td><td>0.192</td><td>127</td><td>0.214</td><td>0.248</td><td>249</td><td>0.354</td><td>0.282</td><td>375</td><td>0.338</td><td>0.282</td><td>3016</td><td>0.147</td><td>0.240</td></tr><tr><td>Group member dummy</td><td>2265</td><td>0.344</td><td>0.498</td><td>127</td><td>0.598</td><td>0.492</td><td>249</td><td>0.711</td><td>0.454</td><td>375</td><td>0.751</td><td>0.444</td><td>3016</td><td>0.58</td><td>0.493</td></tr><tr><td>Short life cycle dummy</td><td>2265</td><td>0.250</td><td>0.433</td><td>127</td><td>0.362</td><td>0.483</td><td>249</td><td>0.321</td><td>0.468</td><td>375</td><td>0.355</td><td>0.479</td><td>3016</td><td>0.274</td><td>0.446</td></tr></table>
+
+
+---
+
+Table 4
+
+<table><tr><td rowspan="2">Model Dependent variable Estimate</td><td colspan="2">Bivariate probit</td><td rowspan="2">Patent=1, UM. M.E. (3)</td><td rowspan="2">Patent=1, UM=0. M.E. (4)</td><td rowspan="2">Patent=0, UM=1. M.E. (5)</td><td rowspan="2">Patent=0, UM=0. M.E. (6)</td></tr><tr><td>Patent Coeff. (1)</td><td>UM Coeff. (2)</td></tr><tr><td>D(Short life cycle)</td><td>0.031 (0.070)</td><td>0.186*** (0.067)</td><td>0.015* (0.008)</td><td>-0.008 (0.008)</td><td>0.023*** (0.009)</td><td>-0.030* (0.016)</td></tr><tr><td>log(Employees)</td><td>0.198*** (0.021)</td><td>0.164*** (0.020)</td><td>0.024*** (0.002)</td><td>0.016*** (0.003)</td><td>0.008*** (0.002)</td><td>-0.048*** (0.005)</td></tr><tr><td>log(R&amp;D per employee)</td><td>0.125*** (0.008)</td><td>0.078*** (0.008)</td><td>0.013*** (0.001)</td><td>0.012*** (0.001)</td><td>0.002** (0.001)</td><td>-0.027*** (0.002)</td></tr><tr><td>log(Export intensity)</td><td>1.482*** (0.172)</td><td>0.884*** (0.176)</td><td>0.155*** (0.022)</td><td>0.140*** (0.022)</td><td>0.018 (0.021)</td><td>-0.313*** (0.040)</td></tr><tr><td>D(Group member)</td><td>-0.010 (0.072)</td><td>-0.078 (0.067)</td><td>-0.006 (0.008)</td><td>0.004 (0.008)</td><td>-0.009 (0.008)</td><td>0.012 (0.016)</td></tr><tr><td>Constant</td><td>-2.199*** (0.136)</td><td>-2.120*** (0.136)</td><td></td><td></td><td></td><td></td></tr><tr><td>Technology classes (7)</td><td colspan="2">Yes</td><td></td><td></td><td></td><td></td></tr><tr><td>p</td><td colspan="2">0.767*** (0.024)</td><td></td><td></td><td></td><td></td></tr><tr><td>Log-likelihood</td><td colspan="2">-1869.07</td><td></td><td></td><td></td><td></td></tr><tr><td>Observations</td><td colspan="2">3016</td><td></td><td></td><td></td><td></td></tr></table>
+
+
+---
+
+Table 5
+
+<table><tr><td rowspan="2">Model Dependent variable Estimate</td><td colspan="2">Bivariate ordered probit</td></tr><tr><td>Patent Coeff. (1)</td><td>UM Coeff. (2)</td></tr><tr><td>D(Short life cycle)</td><td>0.009 (0.067)</td><td>0.150** (0.065)</td></tr><tr><td>log(Employees)</td><td>0.184*** (0.019)</td><td>0.147*** (0.018)</td></tr><tr><td>log(R&amp;D per employee)</td><td>0.130*** (0.008)</td><td>0.081*** (0.008)</td></tr><tr><td>log(Export intensity)</td><td>1.508*** (0.167)</td><td>0.928*** (0.172)</td></tr><tr><td>D(Group member)</td><td>0.009 (0.068)</td><td>-0.053 (0.065)</td></tr><tr><td>Technology classes (7)</td><td colspan="2">Yes</td></tr><tr><td>Constant cut1</td><td>2.228 (0.129)</td><td>2.066 (0.125)</td></tr><tr><td>Constant cut2</td><td>2.376 (0.127)</td><td>2.212 (0.124)</td></tr><tr><td>Constant cut2</td><td>2.686 (0.131)</td><td>2.599 (0.130)</td></tr><tr><td>p</td><td colspan="2">0.721*** (0.024)</td></tr><tr><td>Log pseudolikelihood</td><td colspan="2">-2818.01</td></tr><tr><td>Observations</td><td colspan="2">3004</td></tr></table>
+
+
+---
+
+## Appendix
+
+Figure A.1. Grant, registration, and publication lags
+
+![Figure](figures/figure_002.png)
+
+Note: Data source is the PATSTAT 2016 April edition. Priority patents and UMs filed at the German patent office 2000-2014 are reported. PCT applications are excluded. For UMs, the registration lag is the same as publication lag. Y-axis measures days from the filing date. Decreasing grant lag of patents is caused by truncation as pending patents are not taken into account when calculating medians and means.
+
+---
+
+Table A.1. The use of protection methods for the subsample of innovators, bivariate probit model
+
+<table><tr><td rowspan="2">Model</td><td colspan="2">Bivariate probit</td><td rowspan="2">Patent=1, UM=1</td><td rowspan="2">Patent=1, UM=0</td><td rowspan="2">Patent=0, UM=1</td><td rowspan="2">Patent=0, UM=0</td></tr><tr><td>Patent</td><td>UM</td></tr><tr><td>Estimate</td><td>Coeff.</td><td>Coeff.</td><td>M.E.</td><td>M.E.</td><td>M.E.</td><td>M.E.</td></tr><tr><td></td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td><td>(5)</td><td>(6)</td></tr><tr><td>D(Short life cycle)</td><td>0.019 (0.081)</td><td>0.186** (0.078)</td><td>0.021 (0.016)</td><td>0.027* (0.015)</td><td>0.033** (0.013)</td><td>-0.027 (0.025)</td></tr><tr><td>log(Employees)</td><td>0.223*** (0.026)</td><td>0.188*** (0.024)</td><td>0.045*** (0.005)</td><td>0.024*** (0.005)</td><td>0.008** (0.004)</td><td>-0.077*** (0.008)</td></tr><tr><td>log(R&amp;D per employee)</td><td>0.111*** (0.012)</td><td>0.053*** (0.010)</td><td>0.018*** (0.002)</td><td>0.017*** (0.002)</td><td>0.003* (0.002)</td><td>-0.032*** (0.003)</td></tr><tr><td>log(Export intensity)</td><td>1.545*** (0.207)</td><td>0.748*** (0.211)</td><td>0.246*** (0.040)</td><td>0.237*** (0.045)</td><td>-0.035 (0.033)</td><td>-0.449*** (0.066)</td></tr><tr><td>D(Group member)</td><td>-0.041 (0.088)</td><td>-0.107 (0.083)</td><td>-0.018 (0.017)</td><td>0.005 (0.017)</td><td>-0.013 (0.013)</td><td>0.026 (0.028)</td></tr><tr><td>Technology classes (7)</td><td colspan="2">Yes</td><td></td><td></td><td></td><td></td></tr><tr><td>Constant</td><td>-2.603*** (0.165)</td><td>-2.341*** (0.149)</td><td></td><td></td><td></td><td></td></tr><tr><td>p</td><td colspan="2">0.710*** (0.031)</td><td></td><td></td><td></td><td></td></tr><tr><td>Log-likelihood</td><td colspan="2">-1374.71</td><td></td><td></td><td></td><td></td></tr><tr><td>Observations</td><td colspan="2">1675</td><td></td><td></td><td></td><td></td></tr></table>
+
+
+Notes: The dependent variables are indicator variables for the use of patents and utility models. First two columns report the estimated coefficients for bivariate probit model. Columns 3-6 present the marginal effects for patent and utility model combinations. Heteroscedasticity robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.
+
+---
+
+Table A.2. The importance of UMs and patents for the subsample of innovators, bivariate ordered probit model
+
+<table><tr><td rowspan="2">Model Dependent variable Estimate</td><td colspan="2">Bivariate ordered probit</td></tr><tr><td>Importance of patents Coeff. (1)</td><td>Importance of UMs Coeff. (2)</td></tr><tr><td>D(Short life cycle)</td><td>-0.030 (0.076)</td><td>0.140* (0.075)</td></tr><tr><td>log(Employees)</td><td>0.201*** (0.023)</td><td>0.157*** (0.021)</td></tr><tr><td>log(R&amp;D per employee)</td><td>0.113*** (0.012)</td><td>0.055*** (0.010)</td></tr><tr><td>log(Export intensity)</td><td>1.547*** (0.197)</td><td>0.785*** (0.201)</td></tr><tr><td>D(Group member)</td><td>-0.022 (0.083)</td><td>-0.061 (0.079)</td></tr><tr><td>Technology classes (7)</td><td>Yes</td><td></td></tr><tr><td>Constant cut1</td><td>2.094 (0.170)</td><td>1.788 (0.156)</td></tr><tr><td>Constant cut2</td><td>2.207 (0.170)</td><td>1.894 (0.155)</td></tr><tr><td>Constant cut2</td><td>2.532 (0.174)</td><td>2.306 (0.161)</td></tr><tr><td>p</td><td colspan="2">0.681*** (0.030)</td></tr><tr><td>Log pseudolikelihood</td><td colspan="2">-2129 30</td></tr><tr><td>Observations</td><td colspan="2">1664</td></tr></table>
+
+
+Notes: The dependent variables are ordinal variables indicating the importance of patents and utility models as IP protection methods. 12 firms, which reported to use patents and/or utility models but did not answer the importance question, are omitted. Robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.
+
+---
+
+Table A.3. The use of protection methods and additional controls, bivariate probit model
+
+<table><tr><td rowspan="2">Model</td><td colspan="2">Bivariate probit</td><td colspan="4">Marginal effects after bivariate probit model</td></tr><tr><td>Patent</td><td>UM</td><td>Patent=1, UM=1</td><td>Patent=1, UM=0</td><td>Patent=0, UM=1</td><td>Patent=0, UM=0</td></tr><tr><td>Estimate</td><td>Coeff.</td><td>Coeff.</td><td>M.E.</td><td>M.E.</td><td>M.E.</td><td>M.E.</td></tr><tr><td></td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td><td>(5)</td><td>(6)</td></tr><tr><td>D(Short life cycle)</td><td>0.039</td><td>0.193**</td><td>0.017*</td><td>-0.009</td><td>0.025***</td><td>-0.034</td></tr><tr><td></td><td>(0.073)</td><td>(0.070)</td><td>(0.009)</td><td>(0.009)</td><td>(0.010)</td><td>(0.018)</td></tr><tr><td>log(Employees)</td><td>0.215***</td><td>0.177***</td><td>0.028***</td><td>0.018***</td><td>0.009***</td><td>-0.055***</td></tr><tr><td></td><td>(0.023)</td><td>(0.021)</td><td>(0.003)</td><td>(0.003)</td><td>(0.003)</td><td>(0.005)</td></tr><tr><td>log(R&amp;D per employee)</td><td>0.120***</td><td>0.072***</td><td>0.014***</td><td>0.012***</td><td>0.001</td><td>-0.027***</td></tr><tr><td></td><td>(0.009)</td><td>(0.008)</td><td>(0.001)</td><td>(0.001)</td><td>(0.001)</td><td>(0.002)</td></tr><tr><td>log(Export intensity)</td><td>1.429***</td><td>0.841***</td><td>0.161***</td><td>0.144***</td><td>0.014</td><td>-0.320***</td></tr><tr><td></td><td>(0.180)</td><td>(0.183)</td><td>(0.024)</td><td>(0.025)</td><td>(0.024)</td><td>(0.044)</td></tr><tr><td>D(Group member)</td><td>-0.010</td><td>-0.100</td><td>-0.008</td><td>0.006</td><td>-0.013</td><td>0.015</td></tr><tr><td></td><td>(0.074)</td><td>(0.070)</td><td>(0.009)</td><td>(0.009)</td><td>(0.009)</td><td>(0.017)</td></tr><tr><td>D(Financial constraint)</td><td>0.139**</td><td>0.022</td><td>0.011</td><td>0.019**</td><td>-0.007</td><td>-0.023</td></tr><tr><td></td><td>(0.067)</td><td>(0.066)</td><td>(0.008)</td><td>(0.009)</td><td>(0.008)</td><td>(0.016)</td></tr><tr><td colspan="7">Competition categories</td></tr><tr><td>0 competitors</td><td>-0.119</td><td>-0.187</td><td></td><td></td><td></td><td></td></tr><tr><td></td><td>(0.209)</td><td>(0.217)</td><td></td><td></td><td></td><td></td></tr><tr><td>1-5 competitors</td><td>0.167</td><td>0.062</td><td></td><td></td><td></td><td></td></tr><tr><td></td><td>(0.097)</td><td>(0.090)</td><td></td><td></td><td></td><td></td></tr><tr><td>6-15 competitors</td><td>0.040</td><td>0.041</td><td></td><td></td><td></td><td></td></tr><tr><td></td><td>(0.112)</td><td>(0.105)</td><td></td><td></td><td></td><td></td></tr><tr><td>15- competitors</td><td>ref.</td><td>ref.</td><td></td><td></td><td></td><td></td></tr><tr><td>Technology classes (7)</td><td colspan="2">Yes</td><td></td><td></td><td></td><td></td></tr><tr><td>Constant</td><td>-2.863***</td><td>-2.341***</td><td></td><td></td><td></td><td></td></tr><tr><td></td><td>(0.156)</td><td>(0.149)</td><td></td><td></td><td></td><td></td></tr><tr><td> $\rho$ </td><td colspan="2">0.760*** (0.025)</td><td></td><td></td><td></td><td></td></tr><tr><td>Log-likelihood</td><td colspan="2">-1741.85</td><td></td><td></td><td></td><td></td></tr><tr><td>Observations</td><td colspan="2">2704</td><td></td><td></td><td></td><td></td></tr></table>
+
+
+Notes: The dependent variables are indicator variables for the use of patents and utility models. First two columns report the estimated coefficients for bivariate probit model. Columns 3-6 present the marginal effects for patent and utility model combinations. Financial constraint is a dummy variable, 1 if firm reported a financial constraint as n obstacle for innovation in 2004 and 0 otherwise. Competition categories are dummi s for the number of competitors: 0, 1-5, 6-15, 15-. The reference group is more than 5 competitors. Heteroscedasticity robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.
+
+---
+
+Table A.4. The importance of UMs and patents and additional controls, bivariate ordered probit model
+
+<table><tr><td rowspan="2">Model</td><td colspan="2">Bivariate ordered probit</td></tr><tr><td>Importance of patents</td><td>Importance of UMs</td></tr><tr><td>Estimate</td><td>Cueff. (1)</td><td>Cueff. (2)</td></tr><tr><td>D(Short life cycle)</td><td>0.011 (0.069)</td><td>0.151** (0.067)</td></tr><tr><td>log(Employees)</td><td>0.199*** (0.020)</td><td>0.157*** (0.019)</td></tr><tr><td>log(R&amp;D per employee)</td><td>0.126*** (0.008)</td><td>0.076*** (0.008)</td></tr><tr><td>log(Export intensity)</td><td>1.472*** (0.173)</td><td>0.882*** (0.177)</td></tr><tr><td>D(Group member)</td><td>0.012 (0.071)</td><td>-0.069 (0.067)</td></tr><tr><td>D(Financial constraint)</td><td>0.125** (0.063)</td><td>0.017 (0.063)</td></tr><tr><td colspan="3">Competition categories</td></tr><tr><td>0 competitors</td><td>-0.019 (0.221)</td><td>-0.177 (0.229)</td></tr><tr><td>1-5 competitors</td><td>0.118 (0.094)</td><td>0.049 (0.087)</td></tr><tr><td>6-15 competitors</td><td>0.017 (0.109)</td><td>0.047 (0.103)</td></tr><tr><td>15- competitors</td><td>ref.</td><td>ref.</td></tr><tr><td>Technology classes (7)</td><td colspan="2">Yes</td></tr><tr><td>Constant cut1</td><td>2.439 (0.239)</td><td>2.266 (0.243)</td></tr><tr><td>Constant cut2</td><td>2.589 (0.239)</td><td>2.411 (0.244)</td></tr><tr><td>Constant cut2</td><td>2.910 (0.241)</td><td>2.813 (0.246)</td></tr><tr><td> $\rho$ </td><td colspan="2">0.714*** (0.025)</td></tr><tr><td>Log pseudolikelihood</td><td colspan="2">-2664.54</td></tr><tr><td>Observations</td><td colspan="2">2692</td></tr></table>
+
+
+Notes: The dependent variables are ordinal variables indicating the importance of patents and utility models as IP protection methods. 12 firms, which reported to use patents and/or utility models but did not answer the importance question, are omitted. Financial constraint is a dummy varibale, 1 if firm reported a financial constraint as an obstacle for innovation in 2004 and 0 otherwise. Competition catecories are dummies for the number of competitors: 0, 1-5, 6-15, 15-. Heteroscedasticity robust standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.
+

processed/00164_W2945582986/document.md

@@ -0,0 +1,6 @@
+Good Governance of Private Standardization and the Role of Tort Law - European Review of Private Law View Good Governance of Private Standardization and the Role of Tort Law by - European Review of Private Law Good Governance of Private Standardization and the Role of Tort Law 27 2
+
+This article enquires into the potential of tort law to control private standardization and foster good governance in regulatory decisionmaking. Private standardization has been characterized as a political game of winners and losers: while for some firms it brings about opportunities for product development, innovation and market access, for others it means switching costs and barriers to trade. With so much at stake, firms have strong incentives to influence standardization and ensure that it meets their narrow private interests. This dynamic puts pressure on the integrity and quality of private standardization and has led policy-makers to require standards development organizations (SDOs) to adhere to good governance principles such as stakeholder participation, transparency and the use of state-of-the-art scientific research. Drawing on case law from the United States and Europe regarding the liability for negligent standardization, the article finds that tort law currently offers limited incentives for SDOs to comply with good governance norms. The degree to which compliance with such norms can be required appears to fundamentally depend on an ex post weighing of interests under the circumstances. This balancing, the article argues, should at least involve consideration of (i) the magnitude of risk private standardization is concerned with; (ii) the existing internal rules and procedures for private standardization; (iii) the costs concerned with the (re)organization of such rules and procedures; and (iv) the character and societal benefit of private standardization.
+
+European Review of Private Law
+

processed/00164_W2945582986/metadata.json

@@ -0,0 +1,45 @@
+{
+  "source_file": "/tmp/tmppfrqyjm8/00164_W2945582986_Good_Governance_of_Private_Standardization_and_the_Role_of_Tort_Law.pdf",
+  "total_pages": 1,
+  "pages": [
+    {
+      "page_number": 1,
+      "elements": [
+        {
+          "bbox": [
+            45,
+            63,
+            546,
+            165
+          ],
+          "label": "para",
+          "reading_order": 0,
+          "text": "Good Governance of Private Standardization and the Role of Tort\nLaw - European Review of Private Law View Good Governance of\nPrivate Standardization and the Role of Tort Law by - European\nReview of Private Law Good Governance of Private Standardization\nand the Role of Tort Law 27 2"
+        },
+        {
+          "bbox": [
+            44,
+            178,
+            550,
+            635
+          ],
+          "label": "para",
+          "reading_order": 1,
+          "text": "This article enquires into the potential of tort law to control private\nstandardization and foster good governance in regulatory decision-\nmaking. Private standardization has been characterized as a political\ngame of winners and losers: while for some firms it brings about\nopportunities for product development, innovation and market access,\nfor others it means switching costs and barriers to trade. With so much\nat stake, firms have strong incentives to influence standardization and\nensure that it meets their narrow private interests. This dynamic puts\npressure on the integrity and quality of private standardization and has\nled policy-makers to require standards development organizations\n(SDOs) to adhere to good governance principles such as stakeholder\nparticipation, transparency and the use of state-of-the-art scientific\nresearch. Drawing on case law from the United States and Europe\nregarding the liability for negligent standardization, the article finds that\ntort law currently offers limited incentives for SDOs to comply with good\ngovernance norms. The degree to which compliance with such norms can\nbe required appears to fundamentally depend on an ex post weighing of\ninterests under the circumstances. This balancing, the article argues,\nshould at least involve consideration of (i) the magnitude of risk private\nstandardization is concerned with; (ii) the existing internal rules and\nprocedures for private standardization; (iii) the costs concerned with the\n(re)organization of such rules and procedures; and (iv) the character\nand societal benefit of private standardization."
+        },
+        {
+          "bbox": [
+            46,
+            649,
+            290,
+            672
+          ],
+          "label": "para",
+          "reading_order": 2,
+          "text": "European Review of Private Law"
+        }
+      ]
+    }
+  ],
+  "metadata": {}
+}

processed/00169_W2991243598/document.md

@@ -0,0 +1,1262 @@
+DOI: 10.1111/jwip.12139
+
+ORIGINAL ARTICLE
+
+![Figure](figures/figure_002.png)
+
+# Inventions and patenting in Africa: Empirical trends from 1970 to 2010
+
+Gregory D. Graff$^{1,2,3}$ - Philip G. Pardey$^{3,4}$
+
+1Department of Agricultural and Resource Economics, Colorado State University, Fort Collins, Colorado
+
+2Department of Technology Management and Economics, Chalmers University of Technology, Gothenburg, Sweden
+
+3International Science and Technology Practice and Policy (InSTePP), University of Minnesota, St. Paul, Minnesota
+
+4Department of Applied Economics, University of Minnesota, St. Paul, Minnesota
+
+Correspondence
+
+Gregory D. Graff, Department of Agricultural and Resource Economics, Colorado State University, B328 Andrew G. Clark Building, 1200 Center Avenue Mall, Fort Collins, CO 80523-1172.
+
+Email: gregory.graff@colostate.edu
+
+Funding information
+
+Bill and Melinda Gates Foundation; U.S.
+
+National Institutes of Health
+
+## Abstract
+
+Economic development is increasingly dependent upon on utilizing new knowledge to innovate and create value, even in traditional industries and in low-income countries. This analysis uses evidence on patent families to assess innovation activity throughout sub-Saharan Africa. We find patent activity in sub-Saharan Africa—both by African inventors and by foreign inventors—is comparable to similar regions around the world, when conditioned on economic size. Patent filings in Africa have grown, particularly, since the mid-1990s, but at different rates within different African jurisdictions. Types of technologies being patented in Africa have remained stable over 30 years, with most in pharmaceuticals, chemistry, biotechnology, and engineering. The majority of patent filings in Africa are from Europe, the United States, and other high income countries. Yet, in South Africa, between 15% and 20% of patent filings are by residents of South Africa, and 3% are from other developing and emerging economies. Only a small share of inventions globally are made in sub-Saharan Africa, but for those inventions that do arise in Africa, foreign filings are made widely outside of Africa.
+
+## KEYWORDS
+
+Africa, intellectual property rights, international technology trans fer, patent families, patent offices
+
+This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and
+
+reproduction in any medium, provided the original work is properly cited.
+
+© 2019 The Authors. The Journal of World Intellectual Property published by John Wiley & Sons Ltd
+
+J World Intellect Prop. 2019;1-25.
+
+wileyonlinelibrary.com/journal/jnip 1
+
+---
+
+2 Wiley-
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+## 1 | INTRODUCTION
+
+The world's first patent statute—adopted in the city-state of Venice in 1474—was an early preindustrial economic development policy. By contrast, the globalized patent system of today may seem far removed from the economic development goals of low- and middle-income countries, such as those of sub-Saharan Africa (hereafter, interchangeably as Africa). This is due, in large part, to the increasing asymmetry in technological capabilities that have grown, since the Industrial Revolution, between inventors in leading industrialized countries and technological followers in all other countries. Yet, economic development is increasingly a question of utilizing knowledge to innovate, solve problems, and create value, even in traditional industries and even in least-developed countries. This is evidenced by the contributions of new knowledge—in the form of vaccines, improved crop varieties, or mobile telephone and data services—to the livelihoods of millions. Innovation has played an important role in meeting several of the Millennium Development Goals, and the need for further innovation is integral to the framing of the Sustainable Development Goals (United Nations, 2017).
+
+Scholars and policymakers are divided on the role of patents in economic development policy (Barton et al., 2002; Maskus, 2000; Siebeck, Evenson, Lesser, & Primo Braga, 1990). The negotiation, adoption, and implementation of common minimum standards under the Trade Related Intellectual Property Rights (TRIPS) agreement as part of the World Trade Organization (WTO) Treaty provided ample context for this debate over the past three decades (Blakeney & Mengistie, 2011; Diwan & Rodrik, 1991), yet it is a debate that has run for far longer (May & Sell, 2006).
+
+Some argue that intellectual property rights (IPRs) in developing countries are counterproductive to economic
+
+development. The first and strongest argument in this vein is based on the observation that, since more applications
+
+come from inventors in high-income countries, a developing country's patent system is most likely to issue patent
+
+rights to foreigners. Therefore, a stronger patent system serves mainly to transfer wealth from domestic consumers
+
+to inventors in high-income countries (Maskus, 2000).
+
+The second argument against patents is related to the first, but focuses on domestic producers, holding that
+
+stronger patent regimes in developing countries create difficult conditions for domestic industry to compete
+
+against global technological leaders. Under a weaker patent regime, developing-country firms have greater freedom
+
+to imitate technologies invented in wealthier countries. The developing-country imitators can then move, at lower
+
+cost, through the crucial phases of catching up to the global technological frontier. Once they have caught up, it is
+
+conceded, IPRs can then be accordingly strengthened. This pattern, it is argued, has been followed repeatedly in
+
+history, first by firms in Germany and the United States catching up to those in Britain and France in 19th century,
+
+and successively by firms in Japan, Korea, and most recently in China catching up to those in the West.
+
+A third major argument against patents in developing countries has focused on humanitarian issues of access to
+
+technologies that meet fundamental human needs, such as food security and essential medicines (Gold & Lam,
+
+2003; Kapczynski, Chaifetz, Katz, & Benkler, 2005; Orsi, Camara, & Coriat, 2006). Stronger patent regimes in
+
+developing countries, it is argued, tend to increase prices of food and medicines, particularly for the poorest of
+
+consumers for whom these categories make up a large share of typical household budgets. Furthermore, agriculture
+
+and health-care represent some of the most widespread forms of economic activity in developing countries: with
+
+large segments of the population deriving their livelihoods from smallholder agriculture and with malnutrition,
+
+poverty, and infectious disease creating disproportionate public health and economic burdens. In sub-Saharan
+
+Africa, such conditions certainly are observed, and, for example, high profile objections erupted in South Africa in
+
+the late 1990s regarding patents on antiretroviral human immunodeficiency virus drugs (Fisher & Rigamonti, 2005;
+
+Ostergard, 1999).
+
+Other scholars caution, however, that developing-country policymakers should not too-readily neglect patent
+
+systems, given the increasingly important role that knowledge plays in economic growth. Foremost, it is countered,
+
+under a weak patent system domestic industry, entrepreneurs, or publicly funded researchers choose not to invest
+
+resources in innovative effort, given the lack of incentives from the lack of domestic protections
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+-Wiley_3
+
+(Chen & Puttitanun, 2005; Krattiger, Mahoney, & Nelsen, 2007). Second, with weaker IPR protections in developing countries, innovators in more developed countries, those who have greater technological capacity to innovate, see fewer incentives to invest resources and efforts in creating technologies that meet the different, and sometimes idiosyncratic, needs of producers or consumers in developing countries (Chen & Puttitanun, 2005; Diwan & Rodrik, 1991). A third argument, and one that was widely advanced in advocating for the TRIPS agreement under the WTO treaty, holds that foreign direct investments, particularly of the sort which is anticipated to result in greater technology transfer, capacity development, and economic growth in the receiving countries, are more likely to be made into those countries with patent systems strong enough to afford the foreign investor with sufficient protections.
+
+Yet, as is often the case with such policy debates, the reality of the situation is more complex. Empirical analysis by Gould and Gruben (1996), Maskus (2000), Chen and Puttitanun (2005), and others, have found that the strength of the IPR systems in developing countries—as measured by IPR indexes such as the ones developed by Rapp and Rozek (1990) or Ginarte and Park (1997)—tend to exhibit a U-shaped relationship with respect to level of economic development. In practical terms, the tendency for a weak IPR system at lower levels of development is due to low levels of institutional capacity to innovate and to domestic political support for maintaining the freedom to imitate foreign technologies, both by low-tech firms and by consumers who desire low prices. Strengthening of the IPR system, it is argued, comes about as the economy grows and as domestic political support for protection of inventions intensifies, both from leading domestic firms that begin making inventions of their own, for which they seek protection, and by domestic consumers who increasingly demand higher-quality products that embody or require higher levels of technology from abroad.
+
+Using 25 years of historical data for 64 countries, Chen and Puttitanun (2005) estimate that an upturn in the
+
+strengthening of IPRs begins around a relatively low threshold of US$854 per capita gross domestic product (GDP)
+
+(in 1995 dollars). In 1980, only South Africa and a handful of smaller countries in sub-Saharan Africa—including
+
+Namibia, Botswana, Gabon, Mauritius, and the Seychelles—met or surpassed this GDP-per-capita threshold. By
+
+2000, the number of countries in sub-Saharan Africa above that threshold had more than doubled. By 2010, the
+
+average per capita GDP across sub-Saharan Africa as a whole exceeded the threshold (US$1,222 in 2010 dollars).
+
+In 2015, the roster of countries with per capita GDP above the estimated threshold (US$1,328 in 2015 dollars)
+
+included Kenya, Cameroon, Cote d'Ivoire, Sao Tome and Principe, Ghana, Djibouti, Sudan, Zambia, Nigeria, Angola,
+
+Congo, Equatorial Guinea, Cabo Verde, and Swaziland: a set of countries that together accounted for over 82% of
+
+Africa's GDP and 46% of the population.
+
+As the single country in sub-Saharan Africa with middle-income status, South Africa's firms and scientific
+
+institutions have long had capacity to generate inventions and to protect them, both domestically and in foreign
+
+jurisdictions (Naidoo, 2010). Now, increasingly other parts of sub-Saharan Africa are reaching a turning point. As
+
+countries across the subcontinent exceed Chen and Puttitanun's (2005) estimated threshold of economic
+
+development, innovation capacity emerges, domestic invention begins to increase, and, along with that, domestic
+
+utilization of the patent system, if available, grows. Yet, more empirical analysis is needed to ground policy
+
+discussions of knowledge creation and utilization in sub-Saharan Africa.
+
+This analysis provides a set of quantitative benchmarks for how patents—and the inventions they protect—have
+
+been playing a role in sub-Saharan Africa. For the period 1980–2010, we explore overall patent filing trends
+
+utilizing summary data from the World Intellectual Property Organization (WIPO, 2016). For the same time period,
+
+we also focus on patent filing trends in biological inventions for health and agriculture (e.g., genetic resources,
+
+vaccines, microorganisms, seeds, living modified organisms, etc.) which have been particularly controversial in
+
+developing countries (Blakeney & Mengistie, 2011; Boettiger, Graff, Pardey, Van Dusen, & Wright, 2004; Castle,
+
+2009; Krattiger et al., 2007), utilizing detailed data from the International Science and Technology Practice and
+
+Policy (InSTePP) Global Genetics Patent Database (InSTePP, 2016). These patent data are analyzed to determine
+
+how extensively patents are being used to protect inventions in sub-Saharan Africa, by whom, and in which
+
+industries. We explore how invention and filing rates in sub-Saharan Africa compare with invention and filing rates
+
+---
+
+4 WILEY-THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+in other countries and regions around the world. We seek to understand how the patent system is being used to mediate the flow of technology transfers into and out of sub-Saharan Africa, and even among the countries in subSaharan Africa. We are also curious to understand what types of technologies are being patented in sub-Saharan Africa and whether the mix of technologies has changed over time. Finally, we are particularly interested in how patenting rates over biological inventions for essentials in health and agriculture compare with overall patenting rates in Africa. From these detailed interrogations of available patent data, we establish what historical practice has been in sub-Saharan Africa, and we draw a policy implications regarding the patent system and its emerging role in encouraging and enabling future knowledge-driven economic growth.
+
+## 2 | BACKGROUND
+
+### 2.1 | Inventions and how they are patented
+
+Initially, when an invention is made, information about that new technology is known only to its inventors. The inventors then—often in close consultation with their employer or legal counsel—make decisions about whether and where to file applications for patent protection. If they choose not to file, then, obviously, information on their invention will not show up in patent data; they may choose instead to keep information about their new technology secret, or they may choose to disclose information about the new technology in a scientific article or technical publication (Hall, Helmers, Rogers, & Sena, 2014). If the inventors do choose to file one or more patent applications, then a rich source of data becomes available, in the form of the primary technical information disclosed in each patent application, and in the form of secondary information that can be derived from the patterns of where and when patent applications were filed, and by whom (Griliches, 1990; Pavitt, 1985).
+
+Most patent jurisdictions are national in scope, with a national patent office that receives and reviews patent
+
+applications and issues patents, which are then enforced by the courts within the borders of that nation. In several
+
+parts of the world—such as in Europe and in Africa—groups of countries have joined together to create regional
+
+patent offices, at least for purposes of receiving and reviewing applications. Yet, even in these cases when patents
+
+are issued by a regional office, those patents are accepted within each member country as a national right. An
+
+invention is under patent protection in a given country only to the extent that a patent is currently in force in that
+
+country.
+
+### 2.2 | Patent families
+
+Given this national nature of patent rights and the international rules long-established under the Paris Convention
+
+of 1883, inventors initiate the patenting process by making an initial or priority patent application at a patent office
+
+of their choosing. Today 175 countries are members of the Paris Convention, including most African countries.
+
+Inventors in both member and nonmember countries tend to follow the typical pattern of filing an initial application
+
+at the patent office of the country in which they reside and therefore the country in which they made the invention.
+
+However, this is not required and is not always the case. If the inventors wish to seek patent protection for their
+
+invention in other markets, they can file applications at those other national or regional patent offices, either
+
+directly or via the facilitating mechanism of the Patent Cooperation Treaty administered by WIPO (see Viksnins &
+
+McCrackin, 2007). If inventors file in multiple patent offices, the follow-on (or parallel) patent applications
+
+reference back to the priority patent application that was made at the office of first filing.
+
+These documents that result from the patenting process for a given invention collectively make up what is
+
+known as a patent family for that invention: a set of one or more related patent documents at one or more patent
+
+offices that all reference back to the same priority application and therefore represent the same underlying
+
+invention (EPO, 2017; Martinez, 2011; WIPO, 2017). It should be clear that a patent family, as observed in the
+
+patent data, can continue to grow over several years as additional patent applications are filed and published and as
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+WILEY
+
+5
+
+new patents are issued for a given invention, potentially in several different patent offices. For the purpose of this
+
+analysis, we rely on the systematic identification of patent families in global patent data provided by INPADOC, a
+
+public service of the European Patent Office (EPO, 2017).
+
+All patent families can, by definition, be characterized by their office of first filing, that patent office at which the
+
+initial or priority patent application on the invention was filed. Domestic patent families result when protection is
+
+sought in just one jurisdiction and thus consist of patent documents from just a single patent office, while
+
+international patent families result when protection is sought in more than one jurisdiction and consist of patent
+
+documents from multiple patent offices. WIPO has adopted the somewhat more general term foreign-oriented
+
+patent families, which is defined as a patent family having at least one filing at an office that is different from the
+
+office of the applicant's origin, that is, the country of residence of the first-named applicant on the patent
+
+application (WIPO, 2017). Throughout the following analyses, unless otherwise indicated, we consider the patent
+
+family as the measure of an invention.
+
+### 2.3 | African patent institutions
+
+A basic understanding of the structure and geographic coverage of the main patent offices in sub-Saharan Africa is necessary to understand and interpret the data on patenting activity in Africa, as well as to appreciate the institutional responses that have emerged from the formation of African patent policies. While these institutions handle patent filings being made in sub-Saharan Africa, both by residents and by foreigners, they also influence patent filings made in foreign patent offices by residents of sub-Saharan Africa. This is due to a general tendency for inventors to make an initial patent application at their own domestic or regional patent office and then, from the basis of that priority application, as a sort of "springboard," to launch into subsequent foreign filings abroad. Thus, these African institutions dictate the architecture of the patent data and influence the rates of patent filings coming into Africa and going out of Africa.
+
+Three patent offices in sub-Saharan Africa represent most of the countries and the majority of economic activity of the subcontinent. These include the national patent office of South Africa, in Pretoria (Barratt, Snyman, & Lutchman, 2018; Naidoo, 2010; Pechacek, 2012), and two regional patent offices—the African Regional Intellectual Property Organization (ARIPO), based in Harare, Zimbabwe (Adewopo, 2002; ARIPO, 2016; Nwauche, 2003), and the Organisation Africaine de la Propriété Intellectuelle (OAPI), based in Yaounde, Cameroon (Adewopo, 2002; Botoy, 2001; Nwauche, 2003; OAPI, 2015)—for which member countries are shown in Figure 1. The vast majority of patent filings in sub-Saharan Africa occur at one of these three main offices. South Africa and the member countries of these two regional patent offices encompass 56% of the population and 60% of the economic activity as measured by GDP of sub-Saharan Africa. Countries with independent national patent offices make up another 37% of GDP of sub-Saharan Africa. Of these, Nigeria and Angola, which are both ARIPO observer states, represent perhaps the most conspicuous gaps in the patent system in sub-Saharan Africa. Together these two countries account for 20% of sub-Saharan Africa’s population and fully one-third of its GDP. Yet, the national patent offices for Nigeria and Angola report just a handful of patent filings to WIPO and thus do not appear to have effective patent systems. The few remaining countries that are entirely unaffiliated with the regional patent offices—such as the Democratic Republic of Congo, Madagascar, and South Sudan—are among the least developed countries and, likewise, account for only a handful of patent filings (For brief histories and more details on Africa’s major patent offices, see the Supporting Information Appendix, Section I.)
+
+### 2.4 | Patent data sources for sub-Saharan Africa
+
+To ascertain trends in patenting—both of inventions made by residents of sub-Saharan Africa and of inventions
+
+made elsewhere but filed in one or more of the patent jurisdictions of sub-Saharan Africa—we rely upon two
+
+---
+
+6 WILEY —— THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+![Figure](figures/figure_002.png)
+
+FIGURE 1 Africa's major patent offices and their members
+
+sources of patent data (see the Supporting Information Appendix, section 2 for greater detail on data sources and methods). WIPO’s Statistics Database provides summary statistics of annual counts of patent publications and patent families across all technologies and all patent offices (WIPO, 2016). The WIPO database contains summary data that include annual counts of patent families that originate from each country and annual counts of patent families that are filed at each reporting patent office around the world. The data also include counts of annual patent publications at each reporting office by technology type. These data provide context regarding the full scope of patenting activities in sub-Saharan Africa. The WIPO Statistics Database reports 173,079 total patent family filings in South Africa, 4,813 total filings in ARIPO, and 7,106 total filings in OAPI from 1980 to 2010.
+
+The second data source is the InSTePP Global Genetics database (InSTePP, 2016), based upon data originally
+
+compiled from Thomson Innovation (today Derwent Innovation of Clarivate Analytics), which provides detailed
+
+records of selected patent filings related to biological subject matters, including biological research tools, nucleic
+
+acids (DNAs and RNAs), proteins, and associated biological materials, such as biologics, biotherapeutics, or
+
+biopesticides, as well as genetic resources, breeding materials, and modified living organisms. This database has
+
+detailed information on individual patent publications (both applications and granted patents), organized into
+
+patent families, and can therefore connect a range of characteristics for each invention, including country of
+
+origin, the jurisdictions of all patent family filings, patent assignees, technology classifications, and industry of
+
+application, allowing for detailed analyses of invention and filing trends of biological inventions in sub-Saharan
+
+Africa. The InSTePP Global Genetics database identifies 43,696 total biological patent family filings in South
+
+Africa (with data coverage for years 1980–2005), 2,161 total biological patent family filings in ARIPO (with data
+
+coverage for years 1980–2005), and 2,251 total biological patent family filings in OAPI (with data coverage for
+
+years 1980–2002).
+
+Due to data reporting limitations as well as the long lags naturally involved in foreign-oriented patent family
+
+growth and reporting—particularly for filings into smaller patent jurisdictions—an historic window of 1980–2010
+
+was chosen that allows for maximum coverage and overlap between the WIPO and InSTePP data. A major
+
+challenge for the analysis of intellectual property in sub-Saharan Africa is incomplete data reporting
+
+(see discussion in Supporting Information Appendix, section 2.2). In addition, for some analyses below the
+
+timeframe ends even earlier—in 2003, 2005, or 2008—again due to incomplete data or truncation specific to
+
+each variable.
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+-Wiley_1_7
+
+## 3 | UNDERSTANDING AFRICA IN A GLOBAL CONTEXT
+
+How do patent activity levels of sub-Saharan Africa compare with those of other comparable countries or regions around the world? Generally, we expect larger and more developed economies both to generate more inventions internally and to attract more patent filings from abroad. To examine the African economies in a relative context, controlling for market size, Figure 2 plots the count of foreign-oriented patent families against the GDP of a country (or countries) served by a given patent office. Counting foreign-oriented patent families provides a somewhat normalized representation of inventions, since the inventions being represented must conform to the patent eligibility policies of at least two jurisdictions and has been considered valuable enough to warrant filing in at least one additional jurisdiction. Figure 2a measures foreign-oriented patent families originated by country or region, and Figure 2b measures the count of foreign-oriented patent family filings received by a national or regional patent office. In both panels, we use averages calculated over 5 years from 2004 to 2008, before the impacts of the global financial crisis of 2008–2009 were felt, to smooth out some of the volatilities and idiosyncrasies of annual data. This analysis includes independent national patent offices and regional patent offices. For those countries that are members of regional patent offices, country-level data are aggregated together and represented as a region, to enable comparison of the patenting activities of those regional offices according to the collective sizes of the markets they represent.
+
+### 3.1 | Where foreign-oriented patent families originate
+
+The numbers of foreign-oriented patent families originating from within a country or region is related to the overall size of that country's or region's economy, following a double logarithmic trend that appears roughly J-shaped (Figure 2a). Starting at the high end of the curve, with the largest economies, the United States, Europe, and Japan are also the sources of origin of the largest numbers of foreign-oriented patent families. These are then followed by a set of other large economies, including Korea, which tracks closely to its neighbor Japan, as well as Canada, Australia, and the so-called BRICS, consisting of Brazil, Russia (which is represented here as part of the Eurasian Patent Office), India, China, and South Africa. Below these economies, in terms of market size, the variation from (the J shaped) trend in the numbers of inventions originating within countries grow wider. We see a group of small but relatively wealthy countries, including Bermuda, Barbados, Bahamas, and Malta, with relatively more inventions per GDP, causing the upturn that forms the hook of the “J” in the J-shaped trend. There are certainly many smalland mid-sized economies originating few or no foreign-oriented patent families. However, above a certain size, all economies are originating foreign-oriented patent families, with a lower envelope (see dashed line in Figure 2a) defining a minimum number of inventions per GDP that appears to be strongly increasing in GDP.
+
+African countries and regional patent offices (labeled with boldface font in Figure 2a) are found in different
+
+segments of this global landscape in terms of originating foreign-oriented patent families. South Africa clearly fits
+
+its characterization as belonging among the BRICS. About the same number of foreign-oriented patent families
+
+were originated in South Africa as in Brazil, even though the GDP of South Africa is significantly smaller than the
+
+GDP of Brazil. South Africa accounts for many more inventions than Argentina or Indonesia, economies that are
+
+closer to it in size in terms of GDP. The two regional African patent offices, ARIPO and OAPI, originate fewer
+
+foreign-oriented patent families than many comparably sized economies, but they are not altogether outliers. OAPI
+
+members, for example, account for a regional economy comparable in GDP to that of Pakistan, and they originate a
+
+similar number of foreign-oriented patent families. Nigeria has a GDP comparable to the GDP of all ARIPO member
+
+states combined. Nigeria also accounts for a similar, albeit somewhat lower, number of foreign-oriented patent
+
+families as all ARIPO member states combined. It is notable, however, that most of the foreign-oriented patent
+
+families that originate in Nigeria are not, in fact, being filed in Nigeria; they are, however, being filed elsewhere in
+
+the world, particularly in the United States and the United Kingdom. At the lower extreme, there are several sub Saharan African countries with independent patent offices that originate essentially no foreign-oriented patent
+
+---
+
+8 WILEY —— THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+![Figure](figures/figure_002.png)
+
+FIGURE 2 Continued.
+
+---
+
+GRAFF AND PARDEY
+
+![Figure](figures/figure_001.png)
+
+-Wiley——
+
+families, lying along the horizontal axis of Figure 2a. These include Angola, Ethiopia, Madagascar, Cabo Verde, and Burundi. Finally, we note that Mauritius appears to be approaching the club of small island states of Bermuda, Barbados, the Bahamas, Malta, and others. In fact, Mauritius originates a greater number of foreign-oriented patent families than Nigeria, than all OAPI member states combined, or than all ARIPO member states combined.
+
+### 3.2 | Where foreign-oriented patent families are filed
+
+The numbers of foreign-oriented patent families that come to be filed in a given country or region is likewise
+
+related to the overall size of that country's or region's economy, with a few exceptions, as illustrated in Figure 2b.
+
+There are, in fact, three notable trends in the relationship between the logarithm of GDP and the logarithm of the
+
+numbers of foreign-oriented patent families received.
+
+The main trend, representing what may be described as the global "mainstream" of the patent filing system, is
+
+an almost linearly increasing relationship, involving the majority of countries, between the log of GDP and the log of
+
+foreign-oriented patent families received. In essence, the larger the economy, the more filings it receives, all else
+
+being equal. This is due both to receiving more filings on inventions from abroad, as well as having more domestic
+
+inventions, filed at home, that also end up getting filed abroad (and thus being defined as "foreign-oriented"). This
+
+trend is visible in Figure 2b, as the diagonal running from lower left to upper right. In the extreme upper right,
+
+Europe, the United States, Japan, and other large economies like Canada and Australia receive the most patent
+
+family filings globally. Many middle- and low-income countries also lie along this main trend line, with the smaller
+
+economies receiving proportionately fewer foreign-oriented patent filings.
+
+In terms of count of foreign-oriented patent family filings received, South Africa lies, in fact, just above this main
+
+trend line, aligning almost exactly with Argentina (even though, as seen in Figure 2a, South Africa originates more
+
+inventions that Argentina). Israel, Hong Kong, and New Zealand, although somewhat smaller economies than South
+
+Africa, receive about the same number of foreign-oriented patent family filings, which might be expected for
+
+countries with relatively higher levels of economic development. The other BRICS, with larger economies than
+
+South Africa, receive proportionately larger numbers of foreign-oriented patent family filings but are also within
+
+trend. Brazil, Mexico, and India lie closely together, but they are also very comparable to Australia and Canada.
+
+China receives almost as many foreign-oriented patent family filings as the United States, Europe, or Japan.
+
+The second notable trend in Figure 2b, is that all the regional patent offices lie below the primary trend line,
+
+meaning that the regional offices receive comparatively fewer foreign-oriented patent family filings relative to the
+
+aggregate GDP of the economies they serve. This is likely due to several factors. Regional offices are each, to
+
+varying degrees, complemented or augmented by filings made at the national patent offices of their member
+
+countries. Thus, the count of patent families filed at the regional office likely underrepresents total filings within the
+
+FIGURE 2 Relationships between economic size and patent family filings, 2004–2008. (a) Foreign-oriented patent families originated. (b) Foreign-oriented patent families received. Relationship between GDP and (a) the numbers of foreign-oriented patent families originated from national and regional filing offices, and (b) the number of foreign-oriented patent families received at the identified national and regional filing offices. Data represent log of average annual counts of foreign-oriented patent families originated or received, respectively, over 5 years from 2004 to 2008 (from WIPO) is plotted against log of average national GDP at market prices in billions of current U.S. dollars over the same years (from World Bank). For regional patent offices, inventions originating by country and GDP of all member states are aggregated. Regional patent offices are designated with square symbols. SubSaharan African countries with independent patent offices are designated with triangular symbols. Names of all sub-Saharan African entities in bold. (a) Members that joined the European Patent Office after 2008 are considered separately from the EPO in this analysis. Following WIPO Statistical Database definitions, the origin of a patent family is defined by the country of residence of the first listed inventor or applicant on the priority application. GDP, gross domestic product; WIPO, World Intellectual Property Organization. Source: World Bank (2016) and WIPO Statistics Database (2016)
+
+---
+
+10 WILEY-
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+region. Because of the nature of the WIPO summary data, it is not possible to determine how many of the foreignoriented patent family filings at different offices represent the same inventions. Anticipating that the resulting over count would be a more serious problem, this analysis chose the lesser of possible biases and dropped the counts of filings at the member states' national offices. Second, regional patent offices have tended to be an institutional response of countries with less-developed and therefore weaker patent systems, which relates to the third observed trend, below. The Eurasian Patent Organization (EAPO) includes the nine member states of the Commonwealth of Independent States, all of which are former Soviet Republics: Russia, Turkmenistan, Belarus, Tajikistan, Kazakhstan, Azerbaijan, Kyrgyzstan, Armenia, and Moldova. Notably, among the former republics of the Soviet Union, Ukraine, and Uzbekistan, as well as the Baltics, are not members of EAPO. The Gulf Cooperation Council Patent Office (GCCPO) is made up of six countries on the Arabian Peninsula: Saudi Arabia, Bahrain, Kuwait, Oman, Qatar, and United Arab Emirates. Among the regional offices, the two African regional offices, ARIPO and OAPI, conform to the observed relationship between size of the markets represented and the number of foreignoriented patent families filed. ARIPO is shifted downward from the primary trend line to an extent that is very much comparable to how the European Patent Office is shifted. Even OAPI does not deviate as far from the primary trend line as does the GCCPO, which appears to lie the furthest away.
+
+The third trend involves a diverse set of middle- and low-income countries situated close to the bottom axis,
+
+meaning that they receive very few foreign-originated patent applications relative to other countries with similarly
+
+sized economies. This suggests that the patent systems in these countries are weaker than the conditional norm
+
+(i.e., after controlling for economy size) and, therefore, are not considered worthwhile offices in which to file for
+
+patent protection, all else being equal. However, there may be other factors involved. Many of these are energy rich
+
+countries. Others are countries that have experienced civil conflict, or for other reasons have been less integrated
+
+with the global economy. Three sub-Saharan African countries that independently operate their own patent offices
+
+—Nigeria, Mauritius, and Burundi—all fall within this cluster. Of these, Mauritius may be considered too small to
+
+matter as a destination market, even though it proves to be a source of inventions.
+
+These summary statistics reveal that the patent activity levels of sub-Saharan African countries—both in terms
+
+of inventions they originate and the filings they receive—are comparable to similarly sized countries or regions
+
+around the world. The differences among sub-Saharan African countries are greater than the differences between
+
+them and comparable countries around the world. South Africa aligns with global patent activity trends,
+
+conditioning on the size of its economy, comparing particularly closely to it BRICS counterparts. The member
+
+countries of ARIPO and OAPI generate and receive fewer inventions than comparably sized developed countries,
+
+but still perform at a level comparable to the other regional patent offices. Only a handful of African countries with
+
+independent patent offices—the largest of which is Nigeria—are relatively uninvolved in patenting activities relative
+
+to global trends. Still, given the size of its economy, Nigeria does give rise to inventions being filed abroad. And, the
+
+patent activity profile of Mauritius is comparable to other small, higher-income island states.
+
+## 4 | PATENT FILINGS IN SUB-SAHARAN AFRICA
+
+The comparative assessment in the previous section indicates that the only jurisdictions in sub-Saharan Africa to
+
+receive appreciable annual patent filings are the three main offices of South Africa, ARIPO, and OAPI. Annual filings
+
+in all other independent national offices are so small as to barely register. We therefore focus in this section on
+
+patent family filing trends in the three main patent offices.
+
+### 4.1 | Patent family filing trends
+
+Annual counts of the initial filings for patent families in the three main offices of South Africa, ARIPO, and OAPI, are
+
+plotted in Figure 3a, 3b, and 3c, respectively. The count of total new patent families includes both domestic-only
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+-Wiley_11
+
+![Figure](figures/figure_003.png)
+
+FIGURE 3 Inventions Filed in sub-Saharan Africa, 1980–2010. (a) Total inventions filed, by office. (b) Biological inventions filed, by office. (c) Biological inventions as percent of total inventions filed, by office. Data are counts of inventions filed in sub-Saharan Africa's three main patent offices. (a) Includes total initial filings of patent families, in all technologies, according to WIPO Patent Statistics Database. (b) Includes initial filings of patent families involving biological subject matter inventions, according to the InSTePP Global Genetics Patent Database. (c) Plots the share that biological patents represent of total filings. InSTePP, International Science and Technology Practice and Policy; WIPO, World Intellectual Property Organization. Source: WIPO Statistics Database (2016) and InSTePP Global Genetics Database (2016)
+
+and foreign-oriented patent families, to represent overall filing activities in these jurisdictions (see also Supporting
+
+Information Appendix, Table A1).
+
+The first major observation is that filings in South Africa are significantly greater than those in the two regional
+
+jurisdictions. In fact, filings in South Africa are greater than in all other sub-Saharan African jurisdictions combined. In
+
+2000, South Africa's GDP was 55% larger than the combined GDPs of the ARIPO member states and 155% larger than
+
+the combined GDPs of the OAPI member states; yet, in that same year, South Africa reported 2,350% more patent family
+
+filings than ARIPO (7,206 compared with 307) and 1,790% more than OAPI (7,206 compared with 403) (Figure 3a).
+
+---
+
+12 WILEY-
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+For South Africa, the volatility in filing levels from 1983 through 1992 corresponds to the period of economic sanctions, the end of the apartheid government, and prevailing market uncertainties over the change in government. Annual filing rates began to recover after 1992, as Nelson Mandela's new government proved stable. Starting from very low levels, filing activity at the regional patent offices of ARIPO and OAPI grew, at least modestly, throughout, although OAPI experienced some stagnation during the 1980s. At both ARIPO and OAPI, filings began to increase in the late-1990s, and by the mid-2000s both were registering around 500 annually.
+
+Patent family filings for biological inventions (Figure 3b) follow similar trends, dominated by the numbers of patent families filed in South Africa and with much lower levels of filings in the other offices. It appears that innovation in the bioeconomy makes up a significant share of all innovation being pursued and applied in subSaharan Africa. Figure 3c indicates the annual share of patent families on biological subject matter, as identified in the InSTePP Global Genetics Patent Database, out of patent families overall, according to WIPO Statistical Database. According to this, biological inventions filed in South Africa make up about 30% of the total, at least since the mid-1990s. Similarly, at the OAPI and ARIPO offices the annual count of biologically oriented patent families, according to InSTePP data, make up 40–50% of the annual count of total patent families, according to WIPO data (Figure 3c). This relatively large share may reflect the heavy dependence of these economies on agriculture and natural biological resources. The large proportion of biological inventions may also help to explain the high degree of concern that African policymakers and others have expressed over policies governing the patentability of technologies involving biological subject matters (Taylor & Cayford, 2003).
+
+### 4.2 | Technology and industry trends
+
+Of the filings received by patent offices located throughout sub-Saharan Africa, the kinds of technologies and the
+
+industries they serve indicate the nature of the economic activity that is being influenced by patents. It also
+
+indicates which industry stakeholders are more likely to seek to influence patent policies.
+
+In analyzing technology and industry trends revealed in patent data, it is crucial to keep in mind that even in
+
+high-income countries, not all technologies are equally amenable to patent protection, and therefore not all
+
+industries equally utilize patent protection as a means of appropriating returns on investments in innovation.
+
+Particularly high propensities to patent have been observed in pharmaceuticals and chemicals (Levin et al., 1987;
+
+Pakes, Simpson, Judd, & Mansfield, 1989). It may be expected that such trends are even more pronounced in low income countries with less diversity of industry and generally newer patent systems. Thus, while patent data
+
+reveals only a partial picture of knowledge-driven economic activity, it is helpful in identifying technologies and
+
+industries that are actively innovating.
+
+### 4.2.1 | Overall technology categories
+
+WIPO summarizes, for each country, annual counts of patent grants by category of technology, utilizing the International Patent Class (IPC) codes. From these data, we see that in relative terms the broad categories of technologies being protected with patents at the three main patent offices in sub-Saharan Africa have remained quite stable over the past 30 years (see Supporting Information Appendix, Figure A2). In South Africa the mostprevalent technology categories have not changed since 1980, with the same ten broad categories collectively representing about 60% of total annual patent grants. At the ARIPO and OAPI offices, the ten most prevalent categories make up between 65% and 75% of annual publications.
+
+Six of the ten top technology categories are common across the three main patent offices in Africa. These six
+
+include pharmaceuticals, organic fine chemistry, basic materials chemistry, chemical engineering, biotechnology,
+
+and civil engineering (Supporting Information Appendix, Figure A2, comparing A2a, A2b, and A2c). Pharmaceuticals
+
+technology is the single largest category at all three offices. At South Africa’s patent office, pharmaceuticals make
+
+up 10–15% of annual patent publications in the latest years (Supporting Information Appendix, Figure A2-a). At the
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+-WILEY___{13}
+
+ARIPO and OAPI offices, the share of patent publications in pharmaceuticals has increased over time, coming to make up 20-25% of total patent publications in recent years (Supporting Information Appendix, Figure A2b,c). Four categories describing chemistry—organic fine chemistry, basic materials chemistry, chemical engineering, and food chemistry—make up 20-30% of publications at each of the three offices in recent years. Biotechnology is also found among the top 10 categories at all three offices, making up 4-6% of publications in the most recent years at all three offices.
+
+### 4.2.2 | Industries of application for biological inventions: The scope of Africa's bioeconomy
+
+The IPC-based technology categories reported by WIPO relate to the technical characteristics of an invention. This,
+
+however, does do not necessarily identify the industry in which that technology is being applied, despite significant
+
+efforts to create concordances mapping patent classes to industries of application (Evenson, Putnam, & Kortum,
+
+1991; Johnson, 2002; Lybbert & Zolas, 2014; Verspagen, van Moergastel, & Slabbers, 1994). The InSTePP Global
+
+Genetics Patent Database draws upon Derwent World Patent Index (DWPI) Manual Code designations that enable
+
+a categorization of patent families by industry of application. While the IPC-based WIPO category named
+
+"biotechnology" represents only about four to 6% of annual patent grants, the InSTePP database, indicates that
+
+30–40% of inventions filed in sub-Saharan Africa involve biological subject matters (Figure 3c). The difference is
+
+likely due to the much wider definition used by InSTePP of biological innovations, following more expansive
+
+definitions of the "bioeconomy" that have been espoused in policy discussions (European Commission, 2012;
+
+OECD, 2009; White House, 2012). Presumably, a large number of the patent publications classified by WIPO in
+
+pharmaceuticals, organic chemistry, food chemistry, and others are included in the InSTePP data because they are
+
+involved with or related to biology.
+
+Of the biological inventions identified in the InSTePP data, fully 80% carried a DWPI Manual Code indicating an application of that invention in the pharmaceuticals industry, while 20% carried a DWPI Manual Code indicating an application in veterinary medicine (Table 1). Most of those indicated as "Veterinary" are also identified as "Pharmaceuticals." Meanwhile, 17% of biological inventions indicated an application in "Agriculture." Other industries of application include "Energy" (6%), "Food and beverage manufacturing" (6%), "Paper and textile manufacturing" (4%), and "Environmental conservation and remediation" (4%).
+
+TABLE 1 The industry of application for inventions involving biological subject matter filed at sub-Saharan African patent offices 1970–2010, according Derwent World Patent Index Manual Code designations
+
+<table><tr><td>Industry of application</td><td>Sub-Saharan African patent publications involving biological subject matter (count)</td><td>Share  $^{a}(\%)$ </td></tr><tr><td>Pharmaceuticals</td><td>43,386</td><td>80.0</td></tr><tr><td>Industrial chemicals</td><td>16,926</td><td>31.2</td></tr><tr><td>Veterinary</td><td>11,705</td><td>21.5</td></tr><tr><td>Agriculture</td><td>9,398</td><td>17.3</td></tr><tr><td>Bioenergy</td><td>3,320</td><td>6.1</td></tr><tr><td>Food and beverage</td><td>3,188</td><td>5.8</td></tr><tr><td>Pulp, paper, and textile</td><td>2,275</td><td>4.1</td></tr><tr><td>Environment</td><td>2,165</td><td>3.9</td></tr></table>
+
+
+Source: InSTePP Global Genetics Patent Database (2016).
+
+
+3Shares out of 54,194 patent publications on inventions involving biological subject matter, filed with patent offices in sub-Saharan Africa. Shares sum to more than 100% because patent publications are counted in more than one industry of application. See the Supporting Information Appendix for list of DWPI Manual Codes used to designate each industry of application.
+
+---
+
+14 WILEY-
+
+GRAFF AND PARDEY
+
+### 4.3 | Countries of origin
+
+Where do the inventions being filed in sub-Saharan Africa originate? What share of the inventions filed in subSaharan Africa comes from domestic inventors, from inventors in other countries of Africa, and from inventors outside Africa? Who is making the most use of the patent protections provided by the patent offices of Africa? Do we see evidence of the patent system being used to mediate South-South technology transfers? To assess country of origin, WIPO's Statistical Database considers the country from the listed address of the first inventor or applicant.
+
+### 4.3.1 | Countries of origin of patent filings in South Africa
+
+At the national patent office of South Africa, according to summary statistics from WIPO, the greatest share of
+
+patents filed are part of patent families that originated in Europe. Europe accounts for about 40% of total patent
+
+family filings in South Africa over the 25 years (Figure 4a). Among these, Germany is the largest European source
+
+of inventions (12% of total inventions), followed by the United Kingdom (10%) and France (5%). The United States,
+
+at about 30% of total patent family filings in South Africa over the 25 years, accounts for fewer patent families than
+
+Europe as a whole, but for more than any single European country.
+
+Of the inventions filed in South Africa, 17% originated domestically from within South Africa. While this share may appear modest, it is comparable to the shares of domestic inventions observed in a number of high-income countries, such as New Zealand, Canada, or Sweden, which receive similarly high proportions of patent filings from abroad. It is not unusual in a smaller economy for domestic inventors to account for a minor percentage of total patent family filings received. The high shares of filings by domestic inventors observed in the larger patent offices, such as the United States, Europe, Japan, and China, are due to a combination of the large size of those economies— thus the relatively large share of world inventions accounted for by domestic inventors—and a “home market bias” in filing propensities demonstrated by inventors everywhere. To illustrate how it might be considered the norm for a national patent office to serve more foreign inventors than domestic, Baumol (1993) relates a thought experiment of an imaginary world made up of just ten countries with equally high income levels, all actively investing in R&D and all trading openly with one another: In such a world, the shares of inventions at each country's patent office would consist, on average, of 10% domestic inventions and 90% foreign inventions.
+
+The number of inventions filed in South Africa by residents of South Africa has remained remarkably stable, at
+
+around 1,000 inventions per year from 1980 to 2005 and has not fluctuated nearly as much as the number of
+
+inventions being file in South Africa from abroad, which presumably occurred in response to changes in economic
+
+and political conditions. In recent years, domestic filings have amounted to about 15% of total filings. In addition,
+
+the rate of domestic invention filings by residents of South Africa has also stagnated: the total number of inventions
+
+filed in South Africa by South African inventors in 2005 was in fact slightly lower than the number of inventions
+
+filed 20 years earlier, in 1985.
+
+A relatively small share of the inventions protected in South Africa come from other developing and emerging
+
+economies. The share has grown from about zero as recently as 1995 to about 3% by 2005 (the latest year from
+
+which WIPO summary statistics are available). This mostly consists of inventions originating in the other "BRICS"
+
+nations—the large emerging economies of Brazil, Russia, India, and China, (often considered together with South
+
+Africa) several of which are beginning to invest more heavily in public science and private R&D. The percentage of
+
+inventions from other countries of sub-Saharan Africa protected in South Africa is close to zero, with the absolute
+
+number of inventions from other African countries in the single digits.
+
+The distribution of where inventions filed in South Africa originate has remained remarkably stable over a
+
+25-year period (Figure 4a). While absolute numbers have risen and fallen, the relative shares from each of the
+
+countries or regions of origin have not changed significantly over this time. Inventors in South Africa account for
+
+about 15%. Inventors in other developing and emerging economies now account for about 3% of the patent families
+
+---
+
+GRAFF AND PARDEY
+
+15
+
+![Figure](figures/figure_002.png)
+
+FIGURE 4 Countries of origin of all patent families filed at the patent offices of South Africa, ARIPO, and OAPI. (a) South Africa, 1980–2005. (b) ARIPO, 1980–2010. (c) OAPI, 1980–2003. Other BRICs include Brazil, Russia (including Soviet Union before 1991), India, China (including Hong Kong). ARIPO, African Regional Intellectual Property Organization; OAPI, Organisation Africaine de la Propriété Intellectuelle. Data Source: WIPO Statistics Database (2016)
+
+filed in South Africa annually. Inventors in Europe and the United States account for the bulk, at about 70%.
+
+Inventors in other high-income countries account for the remaining 12% of inventions filed in South Africa.
+
+### 4.3.2 | Countries of origin of patent filings at ARIPO
+
+At ARIPO, the total number of inventions being filed is less than one-tenth of those being filed at South Africa's
+
+patent office. Yet, the pattern of countries of origin of inventions being filed displays several similarities with South
+
+Africa (Figure 4b). For one, the largest share of patent families at ARIPO, an average of 40% annually, originate
+
+from inventors in Europe. Also similarly, the largest single country of origin of patent families at ARIPO is the
+
+United States, accounting for an average of 27% of new patent families each year.
+
+---
+
+16 WILEY-
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+Among European countries, the largest source of inventions filed at the ARIPO office is the United Kingdom,
+
+accounting for an average of 15% of total inventions filed each year, followed by France (5%) and Germany (4%).
+
+This likely reflects the legacy of trade relations and business ownership ties with the United Kingdom for many of
+
+the ARIPO member countries, as well as similarities of language and legal systems that reduce the transaction costs
+
+of adapting a patent application for a British or European Patent Office filing to become a filing at ARIPO.
+
+Collectively, developing countries account for 17% on average, the same share as observed in South Africa.
+
+Almost 10% of the inventions filed at ARIPO each year come from inventors in South Africa, and another 5% of
+
+inventions filed come from one of the other BRICs (Brazil, Russia, India, or China). Perhaps the most acute difference
+
+between ARIPO and South Africa as an office, is the almost complete absence at ARIPO of "domestic" inventions
+
+originating from inventors in ARIPO member countries. Each year less than 10 applications for new inventions come
+
+from within the 19 ARIPO member countries while in South Africa a majority of the filings received by the patent
+
+office originate from inventors in South Africa (see Supporting Information Appendix, Figure A4).
+
+In summary, inventors in Europe and the United States together account for two thirds, at about 67% annually,
+
+of the total number of patent filings. And, inventors in other high income countries account for the remaining 16%.
+
+Inventors in developing and emerging economies account for an average of about 17% of the patent families filed at
+
+ARIPO annually.
+
+### 4.3.3 | Countries of origin of patent filings at OAPI
+
+At OAPI, annual filings of new patent families (or inventions) are even fewer than those at ARIPO, but there is a longer history. The countries of origin of inventions filed in OAPI (Figure 4c) reflect both OAPI's legacy of close ties with France as well as some of the larger global trends reflected in the origins of patent families filed in South Africa and ARIPO. In the 1980s, a very high share of inventions filed in OAPI—as much as 50% in some years—were by inventors in France. The French share declined by 1990, but still Europeans collectively have continued to account for an average of 47% of inventions filed at OAPI. France has continued to account for fully one-third (33%). Inventors in the United States have come to account, since 1990, for another 31% of inventions filed at OAPI.
+
+One difference between OAPI and ARIPO is the relatively higher share of resident inventions originating from
+
+inventors in OAPI member countries. Resident inventions account for an average of 5% of the inventions filed at
+
+OAPI since 1990. Conversely, however, inventions from South Africa account for far fewer, just 1.5%, and, likewise,
+
+inventions from the other BRICS (Brazil, Russia, India, China) account for just 1.6%. Inventions from developing and
+
+emerging economies overall account for just 10% of patent families filed at OAPI since 1990.
+
+### 4.4 | Patent applicants/assignees of filings made in sub-Saharan Africa
+
+Having established that the bulk of patent filings made in sub-Saharan Africa originate from countries outside of
+
+Africa, with a particularly large share coming from Europe and the United States, we now inquire, more specifically,
+
+into what sorts of organizations are utilizing the patent systems in sub-Saharan Africa. To what extent do the
+
+inventions protected in sub-Saharan Africa come from companies, from individual inventors, or from public sector
+
+and academic research institutions?
+
+Under the patent laws of most countries, the rights created by a patent are initially and fundamentally granted
+
+to the individual inventor(s) who created the invention. In the process of applying for the patent, each individual
+
+inventor then has an option to transfer or assign those rights to another legal entity, typically to the company or
+
+other organization that employed the inventor. In most situations, this transfer or assignment of patent rights is
+
+governed by terms of the contract between employer and employee, although in some countries the rights and
+
+obligations for employee's assignment of inventions to employers is governed by statute (Graff, 2007).
+
+The WIPO database, summarizes patent statistics at a national level, which, unfortunately, is not practical for
+
+summarizing the characteristics or locations of the organizations described as "applicants." The InSTePP Global
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+WILEY 17
+
+Genetics Patent Database, however, contains detailed information at the level of individual patent publication
+
+records, including the name and address of the applicant or assignee organization(s). These can then be aggregated
+
+by patent family. When applicant/assignee data are not provided for a filing in an African jurisdiction, the identity of
+
+the applicant/assignee is imputed from other publications in the same patent family, such as those filed in Europe or
+
+the United States, offices which regularly do provide applicant/assignee information.
+
+From the InSTePP Global Genetics Patent Database, we found 104,443 designations of an applicant or assignee on
+
+the 54,194 patent families filed in sub-Saharan Africa, with a mean of 1.92 applicants/assignees per patent family. Of
+
+these, the dominant share, at 88.2%, were companies. An additional 6.4% were identified to be public sector entities—
+
+including government agencies or laboratories, academic institutions, and nonprofit research foundations or hospitals.
+
+The remaining 5.4% indicated only the names of the individual inventors, meaning either that a designation of
+
+assignment to an employer organization had not yet been submitted or processed at the time of publication or that
+
+those individuals were working independently of any such employer (see Supporting Information Appendix, Figure A3).
+
+Tallying up the leading applicant/assignee organizations accounting for patenting activity in sub-Saharan Africa,
+
+out of the 104,443 designations of an applicant or assignee, the identities of the top 20 (listed in Table 2) reflect the
+
+leading countries of origin and the leading technologies and industries identified in the previous sections of our
+
+analysis. More than half of the top 20 applicant/assignees, including all of the first nine, are global pharmaceutical
+
+TABLE 2 Top 20 patent applicant/assignee organizations by count of patent publications on inventions involving biological subject matter filed at sub-Saharan African patent offices 1970-2010
+
+<table><tr><td></td><td></td><td></td><td>Sub-Saharan African patent publications on inventions involving biological subject matter (count)</td></tr><tr><td>Rank</td><td>Assignee/applicant name</td><td>Headquarters</td><td>3,995</td></tr><tr><td>1.</td><td>Pfizer</td><td>New York, NY, USA</td><td>3,239</td></tr><tr><td>2.</td><td>SanofiAventis</td><td>Paris, France</td><td>3,239</td></tr><tr><td>3.</td><td>GlaxoSmithKline</td><td>London, UK</td><td>2,434</td></tr><tr><td>4.</td><td>Merck</td><td>Newark, NJ, USA</td><td>2,203</td></tr><tr><td>5.</td><td>Roche</td><td>Basel, Switzerland</td><td>2,159</td></tr><tr><td>6.</td><td>Novartis</td><td>Basel, Switzerland</td><td>2,133</td></tr><tr><td>7.</td><td>Bayer</td><td>Cologne, Germany</td><td>2,071</td></tr><tr><td>8.</td><td>AstraZeneca</td><td>London, UK</td><td>1,385</td></tr><tr><td>9.</td><td>Johnson \&amp; Johnson</td><td>New Brunswick, NJ, USA</td><td>1,245</td></tr><tr><td>10.</td><td>DuPont</td><td>Wilmington, DE, USA</td><td>1,054</td></tr><tr><td>11.</td><td>BASF</td><td>Mannheim, Germany</td><td>992</td></tr><tr><td>12.</td><td>Unilever</td><td>London, UK</td><td>972</td></tr><tr><td>13.</td><td>Akzo Nobel</td><td>Amsterdam, The Netherlands</td><td>955</td></tr><tr><td>14.</td><td>Eli Lilly</td><td>Indianapolis, IN, USA</td><td>919</td></tr><tr><td>15.</td><td>Bristol-Myers Squibb</td><td>New York, NY, USA</td><td>914</td></tr><tr><td>16.</td><td>Dow</td><td>Midland, MI, USA</td><td>864</td></tr><tr><td>17.</td><td>Boehringer</td><td>Frankfurt, Germany</td><td>834</td></tr><tr><td>18.</td><td>Council of Scientific and Industrial Research (CSIR)</td><td>Pretoria, South Africa</td><td>667</td></tr><tr><td>19.</td><td>Procter \&amp; Gamble</td><td>Cincinnati, OH, USA</td><td>609</td></tr><tr><td>20.</td><td>Abbott Laboratories</td><td>Chicago, IL, USA</td><td>529</td></tr></table>
+
+
+Data Source: InSTePP Global Genetics Patent Database.
+
+---
+
+$^{18}$ WILEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+corporations. At least five of the top 20 are multinational chemical corporations (Akzo Nobel, Bayer, BASF, Dow, and DuPont). Two are multinational corporations focused on consumer and health-care products (Unilever and Procter & Gamble). Only one of the top 20 is an African organization (the public sector Council of Scientific and Industrial Research, of South Africa).
+
+## 5 | INVENTIONS BY SUB-SAHARAN AFRICANS
+
+Sub-Saharan African inventors have made only a small contribution to invention and patenting rates globally, both
+
+overall and specifically in biological subject matters. This is not surprising, given the levels of economic
+
+development in sub-Saharan Africa (see Figure 2a). However, to understand the nascent knowledge-creation
+
+capacities of sub-Saharan Africa and prospects for knowledge-driven economic development, it is helpful to analyze
+
+in more detail the nature and sources of those few patented inventions that have been observed.
+
+Patent families originating from the countries of sub-Saharan Africa through 2010 are shown for all
+
+technologies in Figure 5a and for just biological inventions in Figure 5b. Figure 5a shows that overall invention rates
+
+in South Africa and in the other countries of sub-Saharan Africa have not grown significantly since the 1980s. These
+
+have held to around 1,000–1,200 inventions per year by inventors in South Africa and about 100 inventions
+
+per year by inventors across the rest of sub-Saharan Africa.
+
+![Figure](figures/figure_007.png)
+
+FIGURE 5 Inventions made in Africa. (a) Total inventions made in Africa. (b) Biological inventions made in Africa. Panel a plots new patent families for which the first inventor or first applicant listed is located in subSaharan Africa. (b) Plots new patent families in biological subject matters with any inventor residing in sub-Saharan Africa.Data Sources: WIPO Statistics Database (a) and InSTePP Global Genetics Database (b)
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+WILEY 19
+
+The situation, however, is somewhat different with respect to patenting in genetics and the biosciences (Figure 5b). Biological inventions account for a small share of overall inventions made in Africa—roughly 5% in South Africa and 2% in the rest of sub-Saharan Africa (comparing levels in Figure 5b vs. Figure 5a). The share has been growing, however, beginning in the early 1990s and continuing through the end of the available data series in 2008.
+
+A cumulative tally is provided in Table 3 that counts how many inventions (patent families) have originated from
+
+the countries of sub-Saharan Africa through 2010—both overall and just in biological subject matters. The cumulative
+
+total of inventions by South African inventors over the time period of this analysis is 34,276, according to WIPO data,
+
+far greater than the number of patented inventions from the rest of Africa combined. Of that total, just 755 (or 2%)
+
+involve biological subject matters, according to InSTePP data. Of the biological inventions that originated in South
+
+Africa, 649, or 86%, were assigned to a South African entity. The remaining 14% were presumably invented by
+
+residents of South Africa employed by companies with operations in South Africa but based outside South Africa.
+
+Collectively, inventors in the largely Anglophone member countries of the ARIPO made and sought to patent just 1,601 inventions overall from 1980 to 2010, according to WIPO data. Of these inventions, 126 (about 8% of total inventions) were specifically in the fields of genetics or biology, according to InSTePP data. And, of those, 81 (64%) were assigned to domestic entities within the ARIPO countries (Table 3). Collectively, inventors residing in the largely Francophone member countries of OAPI made and sought to patent 522 inventions overall between 1980 and 2010 (although WIPO's data coverage for OAPI drops off after 2006). Just 56 inventions involved biological subject matter (equivalent to 11% of total inventions) according to InSTePP data. Thirty one (55%) of these biological inventions went to applicants or assignee organizations based in an OAPI member country (Table 3). Finally, inventors in other countries that are not members of either regional patent office made and filed for protection on 365 inventions over the three decades to 2010, according to WIPO data. We find 60 inventions from these countries in the InSTePP data involving genetics or biology (16% of total inventions) and, of these, 40 (or 67%) went to applicant or assignee organizations based in the inventors' home country (Table 3). Yet, the cumulative shares of African assignees at the bottom of Table 3 misses an important trend: the number of African assignees of biological inventions has grown from virtually none as recently as the mid-1990s, both in South Africa and in the rest of sub-Saharan Africa (Supporting Information Appendix, Figure A6).
+
+Only a very small share globally of inventions has come from in Africa. Of the global total of 1,093,038 genetic or biological inventions made between 1970 and 2010, identified in the InSTePP Global Genetics database, only 997, or about 0.1%, were made by residents of sub-Saharan African; three quarters of these were made in South Africa, and one quarter, or roughly 250 inventions originated in the rest of the countries of sub-Saharan Africa. In contrast, 228,882 or 23% of the world's patented biological inventions were made by residents of the United States. Yet, for those inventions made in sub-Saharan Africa, foreign patent filings are quite global. While the majority of South African inventions (see Supporting Information Appendix, Figure A4) are filed only in South Africa, a reasonable share are also filed abroad. In fact, there is a small share of inventions made by South Africans that are only filed abroad and never filed in South Africa.
+
+In tracking individual filings on inventions made by South Africans (Supporting Information Appendix, Figure A5-a), we see that over one-third of patents sought by South Africans are in developing and emerging economies. Before 1995, South African inventors made a small number of filings in other independent national offices in sub-Saharan Africa, but that practice appeared to end with the end of the apartheid government and/or the adoption of the TRIPS agreement in the 1990s. A small number of filings by South African inventors have also been made annually in ARIPO since the 1990s, but virtually none in OAPI. The share of South African inventions being filed in the other BRICS has grown steadily since the 1990s. In the 1980s and 1990s almost half of filings made by South African inventors were in Europe. But, the share being filed in Europe has decreased overall, shifting to focus more on the European Patent Office (EPO) and away from national offices. The share of filings in the United States, Canada, and other OECD countries have remained relatively stable over the three decades analyzed in Figure A5-a (see Supporting Information Appendix).
+
+Invention from the rest of sub-Saharan Africa (Supporting Information Appendix, Figure A5-b) saw a relative
+
+surge of activity in the 1980s, although the absolute numbers were quite small. A couple phenomena that arguably
+
+account for that trend include extensive national office filings in Zimbabwe and Zambia as well as several hundreds
+
+---
+
+20 WILEY-
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+TABLE 3 Inventions made and owned by residents of sub-Saharan Africa
+
+<table><tr><td>Country of invention</td><td>Total inventions, 1980–2010 (WIPO) (count)</td><td>Biological inventions, 1970–2010 (InSTePP) (count)</td><td>Biological inventions assigned to a domestic organization or firm, 1970–2010 (InSTePP) (count)</td><td>Share of biological inventions assigned to a domestic organization or firm (%)</td></tr><tr><td>South Africa</td><td>34,276</td><td>755</td><td>649</td><td>86</td></tr><tr><td colspan="5">ARIPO member countries</td></tr><tr><td>Botswana</td><td>15</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Gambia</td><td>1</td><td>9</td><td>3</td><td>33</td></tr><tr><td>Ghana</td><td>16</td><td>22</td><td>6</td><td>27</td></tr><tr><td>Kenya</td><td>116</td><td>46</td><td>29</td><td>63</td></tr><tr><td>Lesotho</td><td>5</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Liberia</td><td>51</td><td>2</td><td>2</td><td>100</td></tr><tr><td>Malaw</td><td>17</td><td>1</td><td>0</td><td>0</td></tr><tr><td>Mozambique</td><td>1</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Namibia</td><td>79</td><td>6</td><td>6</td><td>100</td></tr><tr><td>Rwanda</td><td>2</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Sao Tome and Principe</td><td>889</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Sierra Leone</td><td>44</td><td>10</td><td>5</td><td>50</td></tr><tr><td>Sudan</td><td>38</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Swaziland</td><td>7</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Uganda</td><td>8</td><td>6</td><td>3</td><td>50</td></tr><tr><td>Tanzania</td><td>7</td><td>7</td><td>1</td><td>14</td></tr><tr><td>Zambia</td><td>44</td><td>4</td><td>2</td><td>50</td></tr><tr><td>Zimbabwe</td><td>261</td><td>9</td><td>7</td><td>78</td></tr><tr><td>ARIPO members overall</td><td>1,601</td><td>126</td><td>81</td><td>64</td></tr><tr><td colspan="5">OAPI member countries</td></tr><tr><td>Benin</td><td>28</td><td>3</td><td>3</td><td>100</td></tr><tr><td>Burkina Faso</td><td>12</td><td>3</td><td>1</td><td>33</td></tr><tr><td>Cameroon</td><td>89</td><td>19</td><td>11</td><td>58</td></tr><tr><td>Central African Republic</td><td>5</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Chad</td><td>7</td><td>2</td><td>0</td><td>0</td></tr><tr><td>Comoros</td><td>4</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Congo</td><td>31</td><td>3</td><td>0</td><td>0</td></tr><tr><td>Côte d'Ivoire</td><td>82</td><td>6</td><td>1</td><td>17</td></tr><tr><td>Equatorial Guinea</td><td>3</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Gabon</td><td>24</td><td>4</td><td>3</td><td>75</td></tr><tr><td>Guinea</td><td>57</td><td>3</td><td>0</td><td>0</td></tr><tr><td>Mal</td><td>29</td><td>1</td><td>0</td><td>0</td></tr><tr><td>Mauritania</td><td>28</td><td>1</td><td>0</td><td>0</td></tr><tr><td>Niger</td><td>30</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Senegal</td><td>71</td><td>14</td><td>7</td><td>50</td></tr><tr><td>Togo</td><td>23</td><td>0</td><td>0</td><td>–</td></tr><tr><td>OAPI members overall</td><td>522</td><td>56</td><td>31</td><td>55</td></tr><tr><td colspan="5">Other countries</td></tr><tr><td>Angola</td><td>2</td><td>1</td><td>0</td><td>0</td></tr><tr><td>Burund</td><td>19</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Cabo Verde</td><td>16</td><td>1</td><td>1</td><td>100</td></tr><tr><td>Djibouti</td><td>0</td><td>3</td><td>1</td><td>33</td></tr><tr><td>Eritrea</td><td>5</td><td>0</td><td>0</td><td>–</td></tr><tr><td>Ethiopia</td><td>8</td><td>9</td><td>1</td><td>11</td></tr></table>
+
+
+(Continues)
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+-Wiley_21
+
+TABLE 3 (Continued)
+
+<table><tr><td>Country of invention</td><td>Total inventions, 1980–2010 (WIPO) (count)</td><td>Biological inventions, 1970–2010 (InSTePP) (count)</td><td>Biological inventions assigned to a domestic organization or firm, 1970–2010 (InSTePP) (count)</td><td>Share of biological inventions assigned to a domestic organization or firm (%)</td></tr><tr><td>Madagascar</td><td>14</td><td>11</td><td>8</td><td>73</td></tr><tr><td>Mauritius</td><td>224</td><td>7</td><td>7</td><td>100</td></tr><tr><td>Nigeria</td><td>73</td><td>30</td><td>11</td><td>37</td></tr><tr><td>Seychelles</td><td>120</td><td>6</td><td>5</td><td>83</td></tr><tr><td>Other countries overall</td><td>365</td><td>60</td><td>40</td><td>67</td></tr></table>
+
+
+Abbreviation: InSTePP, International Science and Technology Practice and Policy.
+
+Sources: WIPO Statistics Database and InSTePP Global Genetics Patent Database.
+
+of inventions that originated from Sao Tome and Principe but were filed first at the national patent office in Austria.
+
+This anomalous pattern could be attributed even to a single firm or even a single highly prolific inventor in Sao
+
+Tome and Principe with a specific connection to Austria. Since 1992, invention rates across the rest of sub-Saharan
+
+Africa have resulted in <100 filings per year, although the patent families resulting from these African inventions
+
+show they have been filed quite broadly. About half of these patent filings have been in developing or emerging
+
+countries: whether at ARIPO or OAPI, in other developing countries, or in the BRICS. The other half have been filed
+
+in high-income countries (Supporting Information Appendix, Figure A5-b).
+
+## 6 | SUMMARY AND CONCLUSIONS
+
+This investigation of the patenting landscape in sub-Saharan Africa gives some empirical grounding to discussions of the role of patents in Africa. Even as the subcontinent has experienced steady economic growth, especially since 2000, the rates of domestic inventions that utilize the patent system show evidence of having only begun to increase modestly. To the extent that patents are playing a role in the development of sub-Saharan Africa, it appears that the first order effect is largely one of facilitating transfers of technology into Africa from abroad. Yet, within this larger picture, there is significant variation in how extensively patents are being used to protect inventions in sub-Saharan Africa.
+
+In South Africa, both foreign and domestic filings are quite extensive, and very much on par with global trends,
+
+given the size of the South African economy. In the regional patent offices of ARIPO and OAPI, both domestic and
+
+foreign filings are detectible, although at much lower levels. Still, they are comparable with what is seen in other
+
+economies of comparable sizes and levels of development around the world. It is the complete absence of patent
+
+activity observed in some of the larger economies—such as Nigeria and Angola—that is more conspicuous. While
+
+we see a small but steady stream of inventions from Nigeria, they are only patented in foreign jurisdictions like the
+
+United States or Europe, not at home in Nigeria.
+
+Over the three decades covered by this analysis, filings in all three of the main offices in sub-Saharan Africa have grown, particularly since the mid-1990s. Yet, rates of patented inventions made by residents of sub-Saharan Africa have, at best, only held steady over this time period. Although data for more recent years is not as reliable, and there are hints in the data, such as the increasing numbers of sub-Saharan African assignees, that there may be an upturn occurring in African innovation in the most recent decade. About 1,200 patent families per year have been originating from inventors in South Africa, with the majority of these inventions only filed domestically in South Africa's patent office. Another 100 inventions per year originate from inventors throughout the rest of subSaharan Africa, although interestingly with patent family filings made quite widely around the world. Altogether,
+
+---
+
+22 WILEY - THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+when controlling for size of economies, both the invention and filing rates in sub-Saharan Africa are comparable to
+
+those in other national and regional patent offices around the world.
+
+The patterns of patent filings indicate that the main use of the patent system in sub-Saharan Africa thus far has
+
+been for the transfer of technologies into sub-Saharan Africa. The inventions being patented in sub-Saharan Africa
+
+primarily originate in Europe, the United States, and other OECD countries such as Japan, Canada, and Australia.
+
+The top applicants of biological inventions are almost all multinational pharmaceutical, chemical, or consumer goods corporations based in Europe and the United States. Still, public sector research organizations account for
+
+6.4% of the biological inventions filed in sub-Saharan Africa, and South Africa's public-sector Council of Scientific
+
+and Industrial Research is one of the top 20 applicants for biological inventions filed in sub-Saharan Africa. Thus,
+
+the patent system may be beginning to play a role in encouraging commercially relevant innovation from the
+
+domestic research capacity in sub-Saharan Africa which is predominantly public sector.
+
+Yet, a secondary effect of the patent system may be to facilitate the transfer of new knowledge within and
+
+among the countries in sub-Saharan Africa, and between sub-Saharan Africa and other developing countries in
+
+what may be considered “South-South” technology transfer. About 10% of the patent families filed at ARIPO each
+
+year are from South Africa, although there does not seem to be a similar linkage between South Africa and OAPI.
+
+Another 10% of the patent families filed at ARIPO each year are from the other BRICS and other developing
+
+countries. South Africa's inventors make as many new filings in other BRICS as they do at home in South Africa
+
+each year. Half of the filings by inventors in other sub-Saharan countries are in developing countries or the BRICS.
+
+The types of technologies being protected in sub-Saharan Africa are those that tend to be most amenable to
+
+patent protections, including pharmaceuticals, chemicals, biotechnology, and some engineering. These categories
+
+are consistent with that observed in higher-income countries.
+
+From these analyses, we can draw several policy relevant implications regarding the patent systems and
+
+knowledge-driven economic growth in sub-Saharan Africa:
+
+- • The balance of intellectual trade between sub-Saharan Africa and the rest of the world is clearly one of net
+
+imports of technology to Africa.
+
+• However, the patent system is providing opportunities for inventors in sub-Saharan Africa to protect their
+
+inventions in both other developing countries, in large emerging economies, and even in high-income economies.
+
+• To some extent “South-South” knowledge flows are beginning to be mediated by the patent systems of the
+
+respective developing and emerging economies involved.
+
+• The types of technologies that tend to be patented in sub-Saharan Africa are the same types that tend to be
+
+patented in high-income countries: those technologies with higher patenting propensity as a means of
+
+appropriating returns to invention. This lends us to advance a testable hypothesis for future research, that
+
+foreign filings into sub-Saharan Africa are part of families larger than average, and thus sub-Saharan African
+
+jurisdictions tend to play host to fewer yet stronger patents than their higher income peers.
+
+• The practical reality of knowledge flows into and out of sub-Saharan Africa tend to go through "knowledge
+
+gateway" countries, with both the legal and commercial capacity to facilitate economic exchanges with the global
+
+economic mainstream. The most important of these is South Africa, but other secondary gateways, such as
+
+Mauritius, are observed as well.
+
+• While invention and patent filing within Africa are nascent, both data reporting and formal institutional
+
+capacities may be lagging relative to actual levels of innovation activity within the economies of Africa.
+## ACKNOWLEDGEMENTS
+
+The authors sincerely thank Devon Phillips and Connie Chan-Kang for capable research assistance, and Mats
+
+Lundqvist for advice. This analysis arose as a convergence of work under the HarvestChoice project, supported by
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+WILEY 23
+
+the Gates Foundation, and the project "International Assessments of Patenting in Genetics," supported by Grant No.: 5R01 HG004041-03 from the U.S. National Institutes of Health.
+
+## CONFLICT OF INTERESTS
+
+The authors declare that there are no conflict of interests.
+
+## ORCID
+
+Gregory D. Graff http://orcid.org/0000-0002-9140-0901 Philip G. Pardey http://orcid.org/0000-0002-8012-1341
+
+## REFERENCES
+
+Adewopo, A. (2002). The global intellectual property system and sub-Saharan Africa: A prognostic reflection. The University of Toledo Law Review, 33(4), 749-771.
+
+ARIPO. (2016). ARIPO: Who we are & what we do. ARIPO Secretariat, Harare, Zimbabwe. Retrieved from https://www.aripo.org/
+
+Barratt, A., Snyman, P., & Lutchman, S. (2018). Researching South African law. GLOBALEX, Hauser Global Law School program, NYU School of Law. Retrieved from http://www.nyulawglobal.org/Globalex/South_Africa.htm
+
+Barton, J., Alexander, D., Correa, C., Mashelkar, R., Samuels, G., & Thomas, S. (2002). Integrating intellectual property rights and development policy. London: Commission on Intellectual Property Rights.
+
+Baumol, W. J. (1993). Entrepreneurship, management, and the structure of payoffs. Cambridge, MA: MIT Press.
+
+Blakeney, M., & Mengistie, G. (2011). Intellectual property and economic development in sub-Saharan Africa. The Journal of World Intellectual Property, 14(3-4), 238–264.
+
+Boettiger, S., Graff, G., Pardey, P., Van Dusen, E., & Wright, B. (2004). Intellectual property rights for plant biotechnology: International aspects. In Christou, P. & Klee, H. (Eds.), Handbook of Plant Biotechnology (pp. 1089-1113). Chichester: John Wiley & Sons, Ltd.
+
+Botoy, I. E. (2001). Potential and substantial benefits of the TRIPS agreement to the member countries of the African intellectual property organization in the patent field: An African’s perspective. Journal of World Intellectual Property, 4, 91–103.
+
+Castle, D. (2009). The role of intellectual property rights in biotechnology innovation. Northampton: Edward Elgar Publishing.
+
+Chen, Y., & Puttitanun, T. (2005). Intellectual property rights and innovation in developing countries. Journal of Development Economics, 78(2), 474-493.
+
+Diwan, I., & Rodrik, D. (1991). Patents, appropriate technology, and north-south trade. Journal of International Economics, 30(1-2), 27-47.
+
+EPO. (2017). Patent Families at the EPO. European Patent Office.
+
+European Commission. (2012). Innovating for sustainable growth: A bioeconomy for Europe. Luxembourg: Publications Office of the European Union.
+
+Evenson, R. E., Putnam, J., & Kortum, S. (1991). Estimating patent counts by industry using the Viksnins and McCrackin (2007)
+
+Yale-Canada concordance (Final Report to the National Science Foundation).
+
+Fisher, W. W., & Rigamonti, C. P. (2005). The South Africa AIDS controversy: A case study in patent law and policy. Cambridge, MA: Harvard Law School.
+
+Ginarte, J. C., & Park, W. G. (1997). Determinants of patent rights: A cross-national study. Research Policy, 26(3), 283-30.
+
+Gold, E. R., & Lam, D. K. (2003). Balancing trade in patents: Public noncommercial use and compulsory licensing. Journal of World Intellectual Property, 6, 5-31.
+
+Gould, D. M., & Gruben, W. C. (1996). The role of intellectual property rights in economic growth. Journal of Development Economics, 48(2), 323-350.
+
+Graff, G. (2007). Echoes of Bayh-Dole: A survey of intellectual property and technology transfer policies in emerging and developing economies. In Krattiger, K., Mahoney, R., & Nelsen, L. et al. Intellectual property management in health and agricultural innovation: A handbook of best practices (pp. 169–196). Davis, CA: Public Intellectual Property Resource for Agriculture (PIPRA). www.iphandbook.org
+
+Griliches, Z. (1990). Patent statistics as economic indicators. Journal of Economic Literature, 28(4), 1661-1707.
+
+Hall, B., Helmers, C., Rogers, M., & Sena, V. (2014). The choice between formal and informal intellectual property: A review. Journal of Economic Literature, 52(2), 375-423.
+
+---
+
+$^{24}$ WILEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+GRAFF AND PARDEY
+
+InSTePP. (2016). Global Genetics Patent Database. International Science and Technology Practice and Policy (InSTePP), University of Minnesota. Retrieved from http://www.instepp.umn.edu/international-assessments-patenting-geneticsand-genomics-health-and-agriculture
+
+Johnson, D. K. N. (2002). The OECD technology concordance (OTC). Paris: Organisation for Economic Co-operation and Development.
+
+Kapczynski, A., Chaifetz, S., Katz, Z., & Benkler, Y. (2005). Addressing global health inequities: An open licensing approach for university innovations. Berkeley Technology Law Journal, 20(2), 1031-1114.
+
+Krattiger, A., Mahoney, R., & Nelsen, L., et al. (2007). Intellectual property management in health and agricultural innovation: A handbook of best practices. Davis, CA: Public Intellectual Property Resource for Agriculture (PIPRA). Retrieved from www.iphandbook.org.
+
+Levin, R. C., Klevorick, A. K., Nelson, R. R., Winter, S. G., Gilbert, R., & Griliches, Z. (1987). Appropriating the returns from industrial research and development. Brookings Papers on Economic Activity, 3, 783-831.
+
+Lybbert, T. J., & Zolas, N. J. (2014). Getting patents and economic data to speak to each other: An "algorithmic links with probabilities' approach for joint analyses of patenting and economic activity. Research Policy, 43(3), 530-542.
+
+Martinez, C. (2011). Patent families: When do different definitions really matter? Scientometrics, 86, 39-63.
+
+Maskus, K. (2000). Intellectual property rights in the global economy. Washington, DC: Institute for International Economics.
+
+May, C., & Sell, S. K. (2006). Intellectual property rights: A critical history. Boulder: CO: Lynne Rienner Publishers.
+
+Naidoo, T. (2010). Globalization and South African patent law. Point, 2, 12-15.
+
+Nwauche, E. (2003). An evaluation of the African regional intellectual property right systems. Journal of World Intellectual Property, 6, 101-138.
+
+OAPI. (2015). Specificités du système. Organisation Africaine de la Propriété Intellectuelle (OAPI). Retrieved from http:// www.oapi.int/index.php/en/oapi/presentation/specificites-du-systeme
+
+OECD. (2009). The Bioeconomy to 2030: Designing a Policy Agenda. Paris: Organisation for Economic Co-operation and Development.
+
+Orsi, F., Camara, M., & Coriat, B. (2006). AIDS, TRIPS and TRIPS plus': The case for developing and less developed countries. intellectual property rights, innovation, governance and the institutional environment. Cheltenham: Edward Elgar Publishing.
+
+Ostergard, R. L., Jr (1999). The political economy of the South Africa-United States patent dispute. Journal of World Intellectual Property, 2, 875-888.
+
+Pakes, A., Simpson, M., Judd, K., & Mansfield, E. (1989). Patent renewal data. Brooking Papers: Microeconomics, 1989, 331-401.
+
+Pavitt, K. (1985). Patent statistics as indicators of innovative activities: Possibilities and problems. Scientometrics, 7(1-2), 77-99.
+
+Pechacek, J. (2012). The past, present, and future of South Africa's patent system. Cybaris, 3(2), 188-215.
+
+Rapp, R. T., & Rozek, R. P. (1990). Benefits and costs of intellectual property protection in developing countries. Journal of World Trade, 24(5), 75-102.
+
+Siebeck, W. E., Evenson, R. E., Lesser, W., & Primo Braga, C. A. (1990). Strengthening of intellectual property in developing countries: A survey of the literature (World Bank Discussion Paper No. 112). Washington, DC: The World Bank.
+
+Taylor, M. R., & Cayford, J. (2003). American patent policy, biotechnology, and african agriculture: The case for policy change. Harvard Journal of Law and Technology, 17, 321.
+
+United Nations. (2017). Sustainable development goals. Retrieved from https://sustainabledevelopment.un.org/
+
+Verspagen, B., van Moergastel, T., & Slabbers, M. (1994). MERIT concordance table: IPC-ISIC (rev. 2). Maastricht, The Netherlands: Maastricht Economic Research Institute on Innovation and Technology.
+
+Viksnins, A., & McCrackin, A. (2007). A Guide to International Patent Protection. In Krattiger, A., Mahoney, R., & Nelsen, L. et al. Intellectual Property Management in Health and Agricultural Innovation: A Handbook of Best Practices (pp. 927–940). Davis, CA: Public Intellectual Property Resource for Agriculture (PIPRA). Retrieved from www.iphandbook.org.
+
+White House (2012). National bioeconomy blueprint. Washington, DC: The White House. Retrieved from https://www. whitehouse.gov/
+
+WIPO (2016). Industrial property statistics. Retrieved from http://www.wipo.int/ipstats/en/statistics/glossary.html
+
+WIPO (2017). World Intellectual Property Indicators-2016. Geneva: World Intellectual Property Organization.
+
+## AUTHOR BIOGRAPHIES
+
+Gregory D. Graff is professor of innovation and entrepreneurship in the Department of Agricultural and Resource Economics at Colorado State University. He is also a fellow of the International Science and Technology Practice and Policy (InSTePP) center at the University of Minnesota. His research looks at the
+
+---
+
+GRAFF AND PARDEY
+
+THE JOURNAL OF WORLD INTELLECTUAL PROPERTY
+
+WILEY
+
+25
+
+policy and economics of scientific research, technological innovation, and entrepreneurship, including the
+
+complex roles played by intellectual property rights. Dr. Graff conducted analysis for this study while on
+
+sabbatical at Chalmers University of Technology in Gothenburg, Sweden.
+
+Philip G. Pardey is professor of science and technology policy in the Department of Applied Economics at the
+
+University of Minnesota. He is also the Director of Global Research Strategy for the College of Food
+
+Agricultural and Natural Resource Sciences and the Minnesota Agricultural Experiment Station and directs the
+
+University's International Science and Technology Practice and Policy (InSTePP) center. His research deals with
+
+productivity measurement and assessment, the finance and conduct of R&D globally, methods for assessing the
+
+economic impacts of research, and the economic and policy (especially intellectual property) aspects of genetic
+
+resources and the biosciences. He currently co-directs a Gates Foundation project, HarvestChoice (www.
+
+HarvestChoice.org), designed to inform and guide investments intended to stimulate productivity growth in
+
+African agriculture.
+
+## SUPPORTING INFORMATION
+
+Additional supporting information may be found online in the Supporting Information section.
+
+How to cite this article: Graff GD, Pardey PG. Inventions and patenting in Africa: Empirical trends from 1970 to 2010. J World Intellect Prop. 2019;1-25. https://doi.org/10.1111/jwip.12139
+

processed/00182_W4289767568/document.md

@@ -0,0 +1,4 @@
+In order to continue, we need to verify that you're not a robot. This requires JavaScript. Enable JavaScript and then reload the page.
+
+JavaScript is disabled
+

processed/00182_W4289767568/metadata.json

@@ -0,0 +1,34 @@
+{
+  "source_file": "/tmp/tmph_r8lu2p/00182_W4289767568_The_development_and_implementation_of_GRI_Standards_practice_and_policy_issues.pdf",
+  "total_pages": 1,
+  "pages": [
+    {
+      "page_number": 1,
+      "elements": [
+        {
+          "bbox": [
+            45,
+            63,
+            536,
+            108
+          ],
+          "label": "para",
+          "reading_order": 0,
+          "text": "In order to continue, we need to verify that you're not a robot. This\nrequires JavaScript. Enable JavaScript and then reload the page."
+        },
+        {
+          "bbox": [
+            45,
+            126,
+            398,
+            168
+          ],
+          "label": "para",
+          "reading_order": 1,
+          "text": "JavaScript is disabled"
+        }
+      ]
+    }
+  ],
+  "metadata": {}
+}

processed/00275_W2234817904/document.md

@@ -0,0 +1,746 @@
+4
+
+# Prospects for Improving U.S. Patent Quality via Postgrant Opposition
+
+Bronwyn H. Hall, University of California at Berkeley, NBER Stuart Graham, Georgia Institute of Technology Dietmar Harhoff, Ludwig-Maximilians-Universität München, Centre for Economic Policy Research David C. Mowery, University of California at Berkeley, NBER
+
+## Executive Summary
+
+The recent surge in U.S. patenting and the expansion of patentable subject matter has increased patent office backlogs and raised concerns that, in some cases, patents of insufficient quality or with inadequate search of prior art are being issued. At the same time, patent litigation and its costs are rising. This paper explores the potential of a postgrant review process modeled on the European opposition system to improve patent quality, reveal overlooked prior art, and reduce subsequent litigation. We argue that the welfare gains to such a system may be substantial.
+
+This is an agency in crisis, and it' s going to get worse if we don 't change our dynamic. It doesn 't do me any good to pretend there's not a problem when there is.
+
+—James E. Rogan, appointed director of the U.S. Patent and Trademark Office in December 2001, as quoted in the Los Angeles Times, February 7, 2003.
+
+## I. Introduction
+
+Beginning in the 1980s, a series of administrative, judicial, and legislative actions strengthened the economic value of U.S. patents and extended their coverage in areas such as computer software and business methods. Partly as a result of these changes, patent applications in the United States continue to grow (see figure 4.1), and the resources available to the patent office have not kept pace. The fraction of patents granted within two years of application has fallen from 85 percent in the early 1990s to about 70 percent in the late 1990s. At the April 2002 hearings before the House Subcommittee on Courts, the Internet and Intellectual Property, Commissioner Rogan stated, “Average pendency surpassed 24 months in 1999, and we expect it to average 26.5 months this year” (Rogan 2002).
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+116
+
+Hall, Graham, Harhoff, and Mowery
+
+![Figure](figures/figure_002.png)
+
+Figure 4.1
+
+USPTO utility patents 1965-2002.
+
+A second consequence of these changes has been an increase in patent litigation and a more recent increase in patent litigation rates. In a study of patent litigation between 1978 and 1995, Lanjouw and Schankerman (2002) found that the rate of litigation rose only slightly between the 1978–1984 and 1991–1995 periods, from nineteen to twenty-one suits per thousand patents, with some variation across technology areas. Somaya (2002) suggests that this rate rose again in the late 1990s. In a new and comprehensive study of patent litigation focusing on cases that terminated in 1998–2000, Allison et al. (2003) report a litigation rate of approximately thirty-two suits per thousand patents. Whether or not litigation per patent issued has increased substantially, the fact remains that the absolute amount of litigation has grown enormously, increasing both the private and public costs of the system as a whole.
+
+Although many of the administrative, legal, and judicial changes in policy affecting patent validity, examination, and value were undertaken at the behest of the U.S. business community, concerns have been raised about the potential economic burdens of low-quality patents in an environment of greater deference to the rights of the patent holder (Merges 1999, Barton 2000, Kingston 2001) . It is therefore not surprising that a number of experts have suggested that the U.S. patent examination system does not currently impose a sufficiently rigorous review
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+117
+
+of patent and nonpatent prior art, resulting in the issuing of patents of considerable breadth and insufficient quality, and that this problem has worsened in recent years. Many of these critics advocate the reform or extension of procedures that would enable interested parties other than U.S. Patent and Trademark Office (USPTO) examiners to bring relevant information to bear on this process either before or shortly after the issue of a patent.
+
+The present paper reviews the prospects for improving the operation of the patent system and lowering its cost by changing the administrative process at the USPTO. At present, the primary administrative procedure for a challenge to the validity of a U.S. patent is the reexamination proceeding, which may be initiated by any party during the life of the patent. A more elaborate and adversarial procedure for challenging the validity of patents in the immediate aftermath of their issue is the opposition proceeding used by the European Patent Office (EPO). Several scholars, both legal and economic, have advocated the introduction of such a system in the United States or a strengthening of the re-examination system to improve the quality of patents and to increase the likelihood that relevant prior art is brought to bear, especially in new subject matter areas. (See Hall 2003 for a summary of these views.) In this paper, we consider the likely effects of introducing such a postgrant review process in the U.S. patent system, focusing in particular on the ability of such a process to improve the quality of patents and reduce the length of time that the current reliance on litigation requires to ascertain the validity of the relatively valuable patents contested in court. Our assessment of the benefits and costs of a U.S. postgrant patent review system draws on our previous work (Graham et al. 2003) , as well as that of Janis (1997) on alternatives to the reexamination procedures and Levin and Levin (2002) for models of strategic litigation and opposition behavior.
+
+We begin by reviewing the current causes of concern about the operation of the U.S. patent system, focusing on the patent quality issues that such a system might be designed to improve. We then describe the institutional background and administrative structure of the U.S. and European patent systems. Readers who are already familiar with the patent system or who are not interested in the institutional detail may wish to skip this section. The final sections of the paper compare the operation of the two patent systems and then present a simple welfare analysis of the expected costs and benefits of introducing a more adversarial administrative challenge system.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+118
+
+Hall, Graham, Harhoff, and Mowery
+
+## II. The Problem
+
+### Patent Quality
+
+What do we mean by “ patent quality ” ? From the perspective of economic welfare, granting the property right described by a patent involves trading off the gain from providing an incentive for innovation against the deadweight loss implied by the possibility of a monopoly during the patent term. The statutory definition of a patentable invention is that it be novel and nonobvious, and that it have utility. $^1$ Both the economic and legal views suggest that high-quality patents describe an invention that is truly new, rather than an invention that is already in widespread use but not yet patented. $^2$ See table 4.1 for some examples of patents that appear to violate this definition, mostly because there is prior art that is not easily searchable in written form.
+
+Besides the three statutory requirements, a fourth criterion for granting a patent on an invention is that the patent application must disclose sufficient details about the invention. Patents thus serve another social purpose: these disclosures in the published patent can facilitate knowledge spillovers to others who might use or improve the invention. Another criterion for a high-quality patent therefore is that it enable those skilled in the art to comprehend the invention well enough to use the patent document for implementation of the described invention. This dimension of patent quality, however, is less likely to be affected by postgrant opposition proceedings.
+
+From a social welfare perspective, an important characteristic of a high-quality patent is that there be relatively little uncertainty over the breadth of its claims, i.e., over what specific features of a technical advance are claimed under the terms of the patent, as well as whether these claims are likely to be upheld in legal proceedings following the issue of the patent. Uncertainty about the validity of a patent has several potential costs: such uncertainty may cause the patent holder to underinvest in the technology, it could reduce investment by potential competitors in competing technical advances, and it may lead to costly litigation after both the holder and potential competitors have invested sizable amounts of money.
+
+### Consequences of Low Patent Quality
+
+Although some scholars, notably Lemley (2001), have argued that the costs of having higher quality patents may exceed the benefits, recent
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+119
+
+Table 4.1
+
+Some examples of patents whose validity has been questioned
+
+<table><tr><td>Patent number</td><td>Year granted</td><td>Assignee</td><td>Title</td><td>Description</td></tr><tr><td>5,241,671</td><td>1993</td><td>Encyclopædia Britannica (Compton)</td><td>Multimedia search system using a plurality of entry paths that indicate interrelatedness of information.</td><td>Broad claims that cover a multimen diala CD-ROM that searches multiple databases of text, images, and video.</td></tr><tr><td>5,797,127</td><td>1998</td><td>Walker Asset Management LP/Priceline</td><td>Method, apparatus, and program for pricing, selling, and exercising options to purchase airline tickets.</td><td>An algorithm for constructing airline ticket option prices based on past demand, implemented on a computer.</td></tr><tr><td>5,937,468</td><td>1999</td><td>Lucent</td><td>Size/cosine table lookup on a computer.</td><td>Use of a computer algorithm to perform the table lookup familiar from math textbooks to obtain the sine or cosine of an angle.</td></tr><tr><td>5,963,916</td><td>1999</td><td>Intouch Group, Inc.</td><td>Network apparatus and method for preview of music products and compilation of market data.</td><td>Patent on storing music on a server (with backup servers) and letting users listen to it by clicking on a list of the music available.</td></tr><tr><td>6,257,248</td><td>2001</td><td>NA</td><td>Two-hand hair-cutting method.</td><td>Cutting or styling hair using instruments such as scissors or combs in both hands.</td></tr><tr><td>6,368,227</td><td>2002</td><td>NA</td><td>Method of swinging on a swing.</td><td>Swinging a swing sideways or in a circular motion instead of back and forth by pulling on the chains (re-examined in 2003).</td></tr></table>
+
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+120
+
+Hall, Graham, Harhoff, and Mowery
+
+experience suggests that there are some unintended consequences, in the form of complicating property rights and feedback effects. In this section, we review the arguments for increasing patent quality.
+
+Low-quality patents can create considerable uncertainty among inventors or would-be commercializers of inventions and slow either the pace of innovation or investment in the commercialization of new technologies. Lerner (1995) has shown that fear of litigation may cause smaller entrant firms to avoid areas where incumbents hold large numbers of patents. Such entry avoidance may be rational and even welfare enhancing if the incumbents' patents are known for certain to be valid, but low-quality patents held by incumbents may also deter entry into a technological area if the costs of invalidating the patents are too high. In these circumstances, technological alternatives may not be commercialized, and consumer welfare suffers.
+
+The lack of relatively rapid processes for resolving patent validity and ensuring higher patent quality may also slow the pace of invention in fields characterized by “cumulative invention,” i.e., those in which one inventor’s efforts rely on previous technical advances or advances in complementary technologies. But if these previous technical advances are covered by patents of dubious validity or excessive breadth, the costs to inventors of pursuing the inventions that rely on them may be so high that they discourage such cumulative invention. On the other hand, large numbers of low-quality patents may dramatically increase the level of fragmentation of property rights covering priorgeneration or complementary technologies, thus raising the transaction costs for inventors obtaining access (e.g., through licenses) to these technologies. Finally, the issue of a large number of low-quality patents will increase uncertainty among inventors concerning the level of protection enjoyed by these related inventions, which in turn will make it more costly and difficult for inventors to build on these related inventions in their own technical advances.
+
+The issuance of low-quality patents is also likely to spur significant increases in patent applications, further straining the already overburdened examination processes of the USPTO. A kind of vicious circle may result, in which cursory examinations of patent applications result in the issue of low-quality patents, which triggers rapid growth in applications, further taxing the limited resources of the USPTO, and thus further limiting the examination of individual applications, and further degrading the quality of patents.
+
+Recent decisions by the court of appeals for the federal circuit (CAFC), the specialized appeals court for patent cases, concerning the
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+121
+
+validity of important patents (those deemed sufficiently valuable by the patent holder or a competitor to litigate and appeal) create still another reason for serious consideration of a nonjudicial process for postissue validity challenges. For example, the CAFC ruled in 2002 that the USPTO had incorrectly rejected two applications for “obviousness,” arguing that if an examiner rejects an application using “general knowledge,” that knowledge “must be articulated and placed on the record.” 3 According to deputy commissioner Esther Kepplinger, “[W]e can’t reject something just because it’s stupid.” 4 This decision could significantly weaken the level of scrutiny provided by the already costly and overcrowded patent-litigation system. A system that enables third parties (including competitors) to bring such knowledge (in the form of written prior art) to bear on the patent could help in making an obviousness determination.
+
+The U.S. patent system, no less than those of other advanced industrial economies, is very much a political creation. Its development and frequent alterations at the hands of the U.S. Congress reflect changes in the balance of political power among corporate inventors, independent inventors, the broader research community, and the general public. It is hardly an accident, after all, that the most recent series of changes in the U.S. and international patent systems were undertaken by U.S. political actors during the throes of the competitiveness crisis of the 1980s. And the increased economic stakes in intellectual property resulting from these policy changes, as well as the broader shift to a knowledge-based economy, are likely to complicate reforms of such potentially far-reaching scope as the introduction of a more elaborate system for postgrant validity challenges. Indeed, the re-examination process that we discuss below was considerably weakened during debate over its passage by the U.S. Congress. Nevertheless, consideration of any such reforms must begin with an analysis of their operation in other industrial-economy patent systems. The remainder of this paper examines the EPO opposition system and compares its operation and outcomes with those of the existing administrative process for patentvalidity challenges in the United States: the “re-examination.”
+
+## III. Institutional Background
+
+The U.S. and European patent systems have similar aims and requirements for patentability, but they differ in the allowable subject matter and in their administrative procedures. As we noted earlier, the U.S. system requires that an invention (process, machine, manufacture, or
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+122
+
+Hall, Graham, Harhoff, and Mowery
+
+composition of matter) satisfy four requirements to be patentable: adequate disclosure, novelty, usefulness, and nonobviousness. In Europe, firms and individuals have been able since 1978 to submit a single application to the European Patent Office that specifies up to twenty-four national jurisdictions in which they desire patent protection for an invention. Under the EPO regime, the patentability requirements—adequate disclosure, novelty, industrial application, and inventive step—are broadly similar but not identical to those of the United States.
+
+The median time between application and patent issue in the U.S. system during the 1980s and 1990s was eighteen months to two years (with a long list of patent applications with a much longer wait time), and in the EPO, it was slightly more than four years. As part of the patent system harmonization legislated in the American Inventors Protection Act of 1999, the United States instituted a policy of publication eighteen months after application in November 2000 for many patents with applications pending in jurisdictions outside the United States. In contrast, EPO applications have always been published with an eighteen-month lag time, regardless of whether the patents have been issued.
+
+Both systems have a postgrant procedure through which the validity of the patent can be challenged by other parties, but the two patent systems' postgrant challenge procedures differ significantly. In both systems, interested parties can also bring suit in court over infringement and validity (with some restrictions on when a suit can be filed). We discuss these administrative processes for postgrant challenges in the following sections.
+
+## USPTO Examination and Re-examination Procedures
+
+In the United States, inventors may claim a utility patent by making application to the USPTO. Before a patent is issued, the USPTO is charged with ensuring that the invention is adequately specified; covers patentable subject matter; and is useful, novel, and nonobvious. Procedurally, the application must be filed within one year of the invention's public use or publication, contain an adequate description with one or more claims, and be accompanied by the payment of a fee.
+
+The USPTO patent examiner is the arbiter of the patentability, novelty, usefulness, and nonobviousness requirements cited above, judg-
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+123
+
+ing these standards against the “prior art,” i.e., prior inventions, in the field. Prosecution of the patent has been characterized as a give-andtake affair, with negotiation and renegotiation between the patentee and the examiner that ordinarily continues for two to three years (Merges et al. 1997) . The costs of prosecuting a patent through the USPTO range from $5,000 to $100,000 (including the USPTO issue fee), depending on the nature of the technology.
+
+Re-examination, originally envisioned as an alternative to expensive and time-consuming litigation, was created by the 1980 Bayh-Dole Act. The legislative history of this act suggests that the re-examination was intended to be a mechanism that would be less expensive and less time consuming than litigation. $^5$ During the legislative process, however, the act was purged of its intended adversarial characteristics, reducing the usefulness of the procedure for opponents of a given patent.
+
+Procedurally, the re-examination proceeding permits the patent owner or any other party to notify the USPTO and request that the grounds on which the patent was originally issued be reconsidered by an examiner. Initiation of a re-examination requires that some previously undisclosed new and relevant piece of prior art be presented to the agency. Under the statute, a relevant disclosure must be printed in either a prior patent or prior publication—no other source can serve as grounds for the re-examination.
+
+After initiation by notification and the payment of a fee to the USPTO, the re-examination goes forward only if the USPTO finds a substantial new question of patentability. Such a determination was intended by law makers to prevent the reopening of issues deemed settled in the original examination (Merges 1997). The USPTO must make this determination within three months of the request and, having made the determination, must notify the patent owner.
+
+When the owner is not the re-examination proponent (about half the cases), the patentee is allowed to file a response to the newly discovered prior art within two months. If the owner chooses to respond, the requester is afforded an opportunity to reply within two months. By choosing not to respond, the owner can limit the requester's participation in the process. The re-examination is an ex parte proceeding between the patent owner and the USPTO that provides limited opportunities for third-party involvement. $^6$
+
+The party requesting a re-examination is entitled to notify the USPTO of the triggering prior art, to receive a copy of the patentee's reply to the re-examination (if any), and to file a response to that reply. The
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+124
+
+Hall, Graham, Harhoff, and Mowery
+
+owner's role in the process is much more involved: the re-examination statute contemplates a second examination, with the same type of giveand-take negotiation between owner and patent office that occurred during the initial issuance of the patent. The examiner remains the final arbiter of the process, and it is not uncommon for the original examiner to be assigned the follow-up re-examination, thus putting the question of whether prior art was overlooked in the hands of the same government official who was responsible for ensuring that no prior art was overlooked in the previous search.
+
+Once the re-examination goes forward, however, the statute requires that the commissioner make a validity determination. The original patent is afforded no statutory presumption of validity in the proceeding, although the practice of assigning re-examinations to the original examiner may produce such a presumption. The re-examination cannot be abandoned or postponed to await the result of concurrent litigation proceedings, although it may be stayed during other USPTO proceedings, including re-issue or interferences. A re-examination may result in the cancellation of all or some of the claims in a patent or the confirmation of all or some of the claims. Nothing in the re-examination procedure can expand the scope of the original patent's claims, but claims may be amended or new claims added during the renegotiation between the patent owner and the examiner.
+
+In summary, for parties seeking to invalidate an issued patent, the re-examination procedure involves considerable costs and risks. The filing fee for the re-examination is substantial, and practitioners estimate the average costs of a re-examination at $ 10,000 to $ 100,000, depending on the complexity of the matter. Although the costs of a reexamination are lower than those of litigation ( $ 1 to $ 3 million), the third-party challenger in re-examination is denied a meaningful role in the process, and the patent holder maintains communications with the examining officer, offering amendments or adding new claims during the re-examination. Re-examination may make it more difficult for challengers to prevail in patent-validity litigation because juries tend to give added weight to re-examined patents. The court of appeals for the federal circuit has indicated that claims confirmed by the reexamining officer present added barriers to a successful contest. $^7$ As a result, challengers face powerful incentives to forego re-examination in favor of litigation, a process that may well be more expensive, more time consuming, and less expert in testing postissue validity.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+125
+
+## Patent Litigation in the United States
+
+In the United States, postissue validity can also be tested in court. Procedurally, litigation differs markedly from the re-examination procedure. Unlike the re-examination procedure, litigation is an adversarial proceeding in which the litigant may elect to have the case heard by either a judge or a jury. Because patent suits generally arise from a charge from the patent owner of infringement, the patent owner exerts considerable control over the timing of enforcement and litigation in a patent dispute. $^8$
+
+Legal standards create a relatively hostile environment in the federal courts for challengers seeking to invalidate an issued patent. Under the statute, patents are “born valid,” thus enjoying a presumption of validity during the court proceedings. Furthermore, the evidentiary standard for proving that a claim is invalid is clear and convincing evidence, a standard considerably higher than the mere preponderance of proof required in the typical civil suit. Because judges and juries may have limited technical expertise, these presumptions and evidentiary barriers create high costs for challengers. The “propatent” judicial philosophy promulgated by the CAFC since its creation has compounded these barriers. According to one study, successful challenges to patent validity fell from 50 percent to 33 percent in the years after the creation of the CAFC (Lemley and Allison 1998).
+
+Direct costs in litigation are also high compared with those of reexamination. Estimates of legal costs in patent litigation run from $ 500,000 to $ 3 million per suit, depending on the amount at risk (AIPLA 2001) or to $ 500,000 per claim at issue, per side (Barton 2000). One important driver of these costs is the extensive use of pretrial discovery. The lag between filing a patent suit and reaching a resolution can also be considerable. One study estimates the average length of a district court patent suit at thirty-one months (Magrab 1993).
+
+But, in fact, very few patent suits actually go to trial, as reported in Lanjouw and Schankerman (2002) , who find that approximately 95 percent of all patent suits settle either before or during trial: 78 percent settle even before the pretrial hearing, an additional 16 percent settle before trial, and 1 percent settle during trial. The median length of time to settlement is eight months, sixteen months, and twenty-five months, respectively, implying that the average or median time to a litigation outcome is somewhat less than the thirty-one months reported by
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+126
+
+Hall, Graham, Harhoff, and Mowery
+
+Magrab. This does not mean that litigation is inexpensive. According to the surveys conducted by the AIPLA, about half of the estimated legal costs of litigation are incurred before the end of the discovery phase (AIPLA 2001). It also does not mean that the social costs of a patent are avoided because settlement before trial is likely to lead to a collusive outcome.
+
+## EPO Examination and Opposition Procedures
+
+Patent protection for European signatories to the European Patent Convention (EPC) can be obtained by filing several national applications at the respective national patent offices or by filing one EPO patent application at the European Patent Office. The EPO application designates the EPC member states for which patent protection is requested. 9 The total cost of a European patent amounts to approximately €29,800, roughly three times as much as a typical national application. 10 Thus, if patent protection is sought for more than three designated states, the application for a European patent is less expensive than independent applications in several jurisdictions. This cost advantage has made the European filing path particularly attractive for applicants selling goods and services in multiple European markets. Increases in the number of patent applications and grants have given the EPO a level of economic importance that now resembles that of the USPTO.
+
+EPO patent grants are issued for inventions that are novel, mark an inventive step, are commercially applicable, and are not excluded from patentability for other reasons. After the filing of an EPO application, The Hague EPO office produces a search report for the applicant. The search report describes the state of prior art regarded as relevant according to EPO guidelines for the patentability of the invention, i.e., it contains a list of references to prior patents and/or nonpatent sources. Unlike the procedure in the U.S. system, applicants at the EPO are not required to supply a full list of prior art (see Michel and Bettels 2001, p. 191ff). Within six months after the announcement of the publication of the search report in the EP bulletin, applicants can request the examination of their application. This request is a compulsory prerequisite for the patent grant. If examination is not requested, the patent application is deemed to be withdrawn. Eighteen months after the priority date, the patent application is published. At this point, the application is normally under examination; thus, the patent owner is generally required to reveal some information about his or her in-
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+127
+
+vention before the grant of the patent and even if no patent is ever issued.
+
+After examination (if requested) has been performed, the EPO either informs the applicant that the patent will be granted as specified in the original application or requires the applicant to agree to changes in the application that are necessary for the patent grant. In the latter case, a negotiation process similar to that in the U.S. system follows. Once the applicant and the EPO have agreed on the scope of the allowable subject matter, the patent is issued for the designated states and is translated into the relevant national languages. If the EPO declines to grant a patent, the applicant may file an appeal. Within nine months after the patent has been granted, any third party can oppose the European patent centrally at the EPO by filing an opposition against the granting decision. The outcome of the opposition procedure is binding for all designated states. If opposition is not filed within nine months after the grant, the patent's validity can be challenged only under the legal rules of the respective designated countries. The EPO opposition procedure is thus the only centralized challenge process for European patents.
+
+An EPO patent opponent must file an opposition with the EPO and present evidence that the prerequisites for patentability were not fulfilled; i.e., the opponent must show that the invention lacked novelty and/or an inventive step or that the disclosure was poor or insufficient. At the EPO, an opposition division determines the outcome. The examiner who granted the patent is a member of the three-person opposition chamber but may not be the chairperson. The opposition procedure can have one of three outcomes: the patent may be upheld without amendments, it may be amended, or it may be revoked. Data on opposition outcomes reported in Graham et al. (2003) indicate that revocation occurs in about one-third of all opposition cases.
+
+Another interesting aspect of the opposition procedure concerns the restrictions imposed by this process on the opponent's ability to settle out of court. Once an opposition is filed, the EPO can choose to pursue the case on its own, even if the opposition is withdrawn. Thus, the opponent and patent holder may not be free to settle their case outside the EPO opposition process once the opposition is filed. This provision of the opposition proceeding may discourage its use by opponents seeking to force patent holders to license their patents.
+
+Both the patent holder(s) and the opponent(s) may appeal the outcome of the opposition procedure. The appeal must be filed within two months after receipt of the decision of the opposition division, and it
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+128
+
+Hall, Graham, Harhoff, and Mowery
+
+must be substantiated within an additional two months. The board of appeal affords the final opportunity at the EPO to test the validity of the contested European patent. Both parties can bring expert witnesses to the proceedings, and various options are available for extending deadlines. Graham et al. (2003) report that the median duration of the challenge procedures (opposition and any appeal) is about three years, although there is considerable variation in the duration of individual cases (the interquartile range is also about three years).
+
+The official fee for filing an opposition is €613; for filing an appeal against the outcome of opposition, the fee is €1,022. However, the total costs to an opponent or the patent holder are much higher. Estimates by patent attorneys of the costs of an opposition range between €15,000 and €25,000 for each party. Patent attorneys we interviewed agreed that opponents have limited ability to drive up the patent holders' costs by filing an opposition (in contrast to litigation in the United States), because attorney fees are regulated in most European countries, including Germany, where many patent lawyers who have the required EPO registration reside.
+
+## Patent Litigation in Europe
+
+One desirable feature of the EPO opposition system is its centralized structure within the fragmented European legal system for patent challenges. As we noted earlier, patent litigation affecting EPO patents is conducted at the national level. The centralized nature of the EPO opposition process thus arguably is more important in this context than in the United States, where the federal courts operate as a more unified system. Nonetheless, a full evaluation of the effects of the EPO opposition system requires some consideration of the possibilities for litigation, which are not precluded by the opposition proceeding.
+
+There have been few systematic studies of patent litigation within the various European nations, and we therefore confine ourselves to a brief review of the few known facts. Outcomes in the local litigation processes involving EPO patents are restricted to the “ local ” level; for example, the patent may be invalidated in Spain, but this finding does not affect its validity in Italy. During the past decade, national patent courts have increasingly taken evidence and decisions from litigation in other European nations into account, but no systematic study has analyzed such legal spillover effects (Stauder 1996, Stauder et al. 1999) .
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+129
+
+Table 4.2
+
+Estimated patent costs in the United States and Europea
+
+<table><tr><td></td><td>United States</td><td>Europe</td></tr><tr><td>Application</td><td>34,000</td><td>22,903</td></tr><tr><td>Fees  $^{b}$ </td><td>4,000</td><td>4,624</td></tr><tr><td>Legal costs</td><td>30,000</td><td>5,914</td></tr><tr><td>Translation</td><td>NA</td><td>12,366</td></tr><tr><td>Renewal (10 years)</td><td>6,000</td><td>9,140</td></tr><tr><td colspan="3">Re-examination/opposition</td></tr><tr><td>Fees</td><td>2,520</td><td>1,075</td></tr><tr><td>Legal costs</td><td>10,000-100,000</td><td>21,505</td></tr><tr><td>Litigation</td><td>0.5M-33M</td><td>54K-540K</td></tr></table>
+
+
+a Numbers are in 2002 dollars using an exchange rate of 0.93 euros to the dollar. Figures are approximate. b These fees are for an entity that is not small.
+
+The differences among national jurisdictions within Europe are enormous, requiring substantial investments in each national suit and driving up the costs of challenging the national patents emerging from an EPO grant in several of the designated states. The costs of litigation in any national court have been estimated to be between €50,000 and €500,000, depending on the complexity of the case. This cost structure makes an attack at the European level with the opposition procedure particularly attractive for a current or potential competitor of the patent holder. The litigation rate (computed as the number of cases for which a suit is filed divided by the number of patents) in most European countries is roughly 1 percent, slightly lower than the 1.9 percent reported for the United States (Stauder 1996, 1989; Lanjouw and Schankerman 2001) . However, recent estimates by Cremers (2003) suggest that the litigation rate may be as high in Germany as in the United States, on the order of 2.1 percent. $^11$ At this juncture, all that can be said is that the quantitative evidence is too sparse to conclude from these figures that the existence of the opposition mechanism leads to a reduction in litigation or in litigation cost. Table 4.2 summarizes the costs associated with each system.
+
+## IV. Comparing the Two Systems
+
+In Graham et al. (2003) , we compared the operation of the postissue re-examination and opposition systems for challenging patent validity in the United States and drew some conclusions about the differences
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+130
+
+Hall, Graham, Harhoff, and Mowery
+
+between an ex parte system like the current U.S. system and the inter partes system used in Europe. First, the U.S. re-examination procedure differs dramatically from the EPO opposition procedure in almost all its features. The two most important are that re-examinations are much less common, with an overall average rate of 0.2 percent, in contrast to the European opposition rate of about 8 percent, and that the identity of the party requesting a re-examination in the U.S. system is the patent owner in at least 44 percent of the cases, lowering the effective rate even more. This characteristic of re-examination hardly qualifies it as the sort of adversarial procedure that EPO oppositions represent.
+
+We also found that EPO oppositions resolved validity challenges more slowly than USPTO re-examination proceedings. Indeed, opposition proceedings in some cases (and almost certainly in important, complex cases with numerous opponents, appeals, etc.) may well take as much time to be resolved as with litigation in the U.S. system. Nonetheless, the higher frequency of EPO opposition compared to U.S. reexamination or litigation is at least consistent with the hypothesis that the opposition process handles many more legal disputes over patent validity than are addressed by the U.S. re-examination process and at a lower cost than the U.S. litigation process. $^12$
+
+Our analysis of the outcomes of the re-examination and opposition system outcomes confirmed that the adversarial nature of the opposition system was more likely to lead to outcomes unfavorable to the patent holder. In table 4.3 , we show the distribution of outcomes for all re-examinations and all oppositions of patents over the past twenty years. It is clear from the table that patent revocation is much more likely when a patent is opposed in Europe (one in three is revoked) than when a patent is re-examined in the United States (only one in ten is revoked). Combined with the lower probability of re-examination, the overall probability that a patent is revoked via a postgrant administrative challenge is 3 percent in Europe and essentially zero (0.02 percent) in the United States. Conversely, re-examination is more likely than opposition to lead to amendment of the patent, whether or not the patent owner initiated the process.
+
+Our analysis also indicated that patent amendment, rather than revocation, is more likely for oppositions in relatively new fields of inventive activity, for more complex patents, or for oppositions in which numerous opponents participate. Because we lack evidence on the
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+131
+
+Table 4.3 Outcomes from oppositions (EPO) and re-examinations (USPTO)
+
+<table><tr><td rowspan="2">Outcome</td><td colspan="2">Opposition  $^{1}$ </td><td colspan="2">Re-examination  $^{1}$ </td><td colspan="2">Re-examination, excluding owner- requested re-examinations  $^{b}$ </td></tr><tr><td>Total number</td><td>Total share</td><td>Total number</td><td>Total share</td><td>Total number</td><td>Total share</td></tr><tr><td>No change to patent</td><td>5,590</td><td>22.4%</td><td>716</td><td>23.9%</td><td>476</td><td>25.9%</td></tr><tr><td>Patent amended</td><td>6,466</td><td>33.0%</td><td>1,993</td><td>66.4%</td><td>1,151</td><td>62.7%</td></tr><tr><td>Patent revoked</td><td>6,655</td><td>35.1%</td><td>291</td><td>9.7%</td><td>209</td><td>11.4%</td></tr><tr><td>Closed / no outcome</td><td>1,753</td><td>9.6%</td><td>0</td><td>0.0%</td><td>0</td><td>0.0%</td></tr><tr><td>Total with an outcome</td><td>20,464</td><td>100.0%</td><td>3,000</td><td>100.0%</td><td>1,836</td><td>100.0%</td></tr><tr><td>Pending</td><td>3,221</td><td>22.4%</td><td>902</td><td>23.1%</td><td>472</td><td>20.5%</td></tr><tr><td>Total</td><td>23,685</td><td></td><td>3,902</td><td></td><td>2,308</td><td></td></tr></table>
+
+
+*Opposition outcomes are for all patents granted by the EPO 1980–1999, oppositions filed 1980 to September 2000. *Re-examination outcomes are for all re-examinations filed between 1981 and 1998. Each re-examination appears only once. In the cases where there is more than one re-examination request, the outcomes have been combined.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+132
+
+Hall, Graham, Harhoff, and Mowery
+
+extent to which oppositions are followed by litigation in the European patent system, we were unable to determine whether the lack of any speed advantage for oppositions in resolving patent disputes quickly is offset by a reduction of litigation rates associated with oppositions. The EPO system may offer few advantages over the U.S. system for postissue patent challenges, but we cannot address this issue without analyzing litigation data for both the U.S. and European systems. Any truly comprehensive assessment of the social costs and benefits of the two challenge systems requires that we consider both patent office processes of postgrant challenge (opposition or re-examination) and legal system litigation. Nevertheless, we present a simple version of such an analysis in the next section of the paper.
+
+Although an EPO opposition must be filed within nine months of patent issue, it does not reach a conclusion more rapidly than the U.S. re-examination procedure if we estimate the total time lag as the length of time from patent application date to final resolution. The average lag time between application date and the initiation of a challenge is substantially greater within the U.S. re-examination system than in the EPO opposition system, but this difference reflects the different time limits on the initiation of such proceedings. Should we conclude from these comparative data that the longer lag times in the EPO opposition system imply a lengthier period of uncertainty and legal expense, and therefore a higher welfare burden within the innovation systems of these economies? Such a conclusion is unfounded because it relies on a characterization of the re-examination and opposition proceedings as analogous in their characteristics, rigor, and outcomes. The data presented above on the identity of the parties initiating re-examinations, as well as the abundant evidence discussed earlier of significant procedural differences between the re-examination and opposition processes, should invalidate any such analogies. Any such comparison of challenges must incorporate data on the next stages of these challenges, which in both Europe and the United States involve litigation.
+
+One of the concerns often raised about both litigation and postgrant administrative challenges is that they may be used by firms with deep pockets to harass smaller firms and independent inventors. 13 Although re-examination requests are slightly more likely if the patent is held by an independent inventor, there is little if any evidence that independent-inventor patents are significantly more prone to EPO oppositions than other patents, which means that the opposition system
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+133
+
+in Europe is not being used by large players to harass small inventors. Nevertheless, the possibility of some strategic use of the system by competing firms cannot be ruled out. For example, Harhoff and Hall (2002) find in the hair care industry that German firms systematically oppose the patents of their multinational competitors and are not opposed in turn. Whether this finding is due to greater expertise and knowledge of the prior art or to greater familiarity with the use of the opposition system as a weapon is not clear.
+
+## V. Welfare Gains from Improved Postgrant Review
+
+Would introducing an adversarial challenge system simply add another level of costly litigation to what already exists? Or does it have the potential to reduce the subsequent litigation? Answering these questions is a complex task, given the difficulty of predicting the responses of firms and individuals to a rather radical change in the administration of the patent system. Nevertheless, it is possible to perform some simple welfare computations based on plausible assumptions about the value of patent validity and the known costs of both the European and U.S. patent systems. As Levin and Levin (2002) argue in their discussion of the same issue, successful patent opposition is likely to reduce subsequent patent litigation substantially, thus increasing social welfare, but opposition itself brings forward cases that would not have gone to litigation and affects the terms on which firms are able to license the technology embedded in patents. They conclude that, on balance, substantial welfare gains are likely from the introduction of an opposition system. In this section of the paper, we present our own analysis of the consequences of this change and reach a similar conclusion.
+
+Successful opposition should reduce litigation because invalid patents could not then be litigated, although this might require a somewhat circumscribed appeals process that does not allow recourse to the courts. Unsuccessful opposition may still lead to litigation later and, unless barred by statute, successful opposition might also lead to later litigation on the part of the former patent holder. The net result of this scenario is fewer suits filed, and possibly fewer collusive settlements based on the threat of a suit, although there might be an increase in collusive settlements based on the threat of opposition.
+
+The computation shown below makes these ideas more precise. We compute the social benefits and costs of introducing an opposition
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+134
+
+Hall, Graham, Harhoff, and Mowery
+
+system under a few simple stylized assumptions. Although the precise numbers should be viewed somewhat sceptically, the order of magnitudes are such that substantial social gains may accrue from such a system. The total cost of the system per year is assumed to be given by
+
+$$C=c_{0} \phi N$$
+
+where $c_o$ is the cost of an opposition, $\phi$ is the rate of opposition, and $N$ is the total number of patents issued in a year. We assume a range of costs between $100,000 per opposition (the European estimate) and $500,000 (the cost assumed by Levin and Levin [2002] ), and a range of opposition rates between 2 percent (very conservative) and 10 percent, which is approximately the rate in Europe.
+
+Computing the benefit of the system requires some assumptions about the effects of opposition outcomes on the costs associated with avoiding litigation, collusive presuit settlements, and the exercise of monopoly power. We write the benefit as
+
+$$B=b_{o} \phi N$$
+
+where $b_o$ is the cost avoided by each opposition, $\phi$ is the rate of opposition, and $N$ is the total number of patents issued in a year. The variable $b_o$ has three components, corresponding to the three possible outcomes of the opposition: and $N$ revocation, amendment, or rejection. To estimate these components, we assume that the opposition and appeals boards make the correct decision when they revoke or amend a patent or when they reject an opposition.
+
+In the case of revocation, we can assume that the patent should never have been granted and that its existence may create excessive market power. Suppose that in litigation, the same correct outcome would be obtained; i.e., the patent right would be destroyed. As long as the respective case would have ended up in litigation, the opposition will simply pre-empt later litigation. The welfare effect is then to reduce the number of cases in litigation, at the cost of an opposition proceeding. We assign an average social value of $ 2 million to avoided litigation, based on the estimates provided by AIPLA (2001) for legal costs plus an addition for the direct costs to the firms involved and the cost of court services.
+
+The parties may have an incentive to settle; thus, not all cases will actually be litigated (Lanjouw and Schankerman 2002) . But in this case, if the correct outcome would have been a revocation, a settlement be-
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+135
+
+tween the parties will be socially inefficient because it typically maintains the patent for the proprietor and allows the other party in litigation to have an exclusive license. This situation amounts to a case in which the two parties collusively maintain patent protection, which may impose considerable welfare costs on society. 14 The welfare costs of such an agreement depend on the value of the patent right and demand conditions. As our base case, we assume that high-value patents that are attacked in the opposition procedure have an average value of $ 4 million and therefore a monopoly welfare loss of $ 2 million. 15 For example, Harhoff, Scherer, and Vopel (2003) compute an average value of about (1977) DM 400,000 for patents of German proprietors. They also estimate regressions indicating that opposed patents that survived opposition are worth about ten times this amount. We assume that patents that were opposed but did not survive opposition would generate the same level of profits as patents that withstood opposition. In 2003 terms (assuming a 4 percent growth rate), the average value of an attacked patent would then be roughly 5.76 million euros [= (1.04 $^{27}$ 400,000)/2]. A conservative estimate of the value of the attacked patent is therefore $ 4 million on average, with welfare losses in the monopoly case of $ 2 million.
+
+In the case of amendment (where claims are usually narrowed), a similar argument holds, although the avoided litigation cost is likely to be much smaller. The results from opposition rejection are more ambiguous. It may reduce uncertainty about the patent validity and therefore reduce subsequent litigation, but this is by no means certain. In our base case, we assume no effect, and we also evaluate a variant where there is increased social cost due to an increase in litigation probability when opposition is rejected.
+
+In table 4.4 , we show the cost-benefit computation under different scenarios. Three panels are shown in the table, each corresponding to a set of assumptions about outcome probabilities. The first uses the probability that a U.S. patent is found valid during litigation, as reported by Allison and Lemley (2002) . The second uses the observed opposition outcome probabilities for the EPO system, and the third uses the observed re-examination outcome probabilities of the USPTO system, both given in table 4.2 . The latter choice is very conservative, and an opposition system is unlikely to lead to patent revocation probabilities as low as 11 percent. For each of these three outcome scenarios, we report five computations, three using a (comparatively) low opposition cost and two using the higher estimate of $ 500,000 that was used
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+136
+
+Hall, Graham, Harhoff, and Mowery
+
+Table 4.4
+
+Welfare computation under different scenarios
+
+<table><tr><td rowspan="2">Scenario</td><td rowspan="2">Cost of an opposition per patent (thousands)</td><td colspan="3">Outcome probabilities</td><td colspan="3">Benefits (avoided cost)</td><td rowspan="2">Total benefit per patent (thousands)</td><td rowspan="2">Cost-benefit ratio</td></tr><tr><td>Revocation (probability)</td><td>Amendment Rejection (probability) (probability)</td><td>Revocation (probability)</td><td>Amendment Rejection (thousands)</td><td>Amendment Rejection (thousands)</td><td>Total benefit per patent (thousands)</td></tr><tr><td>Validity probability from $100$ Allison and Lemley 2002; low cost.</td><td>0.450</td><td>0.300</td><td>0.250</td><td>2,000</td><td>$0$</td><td>$0$</td><td>$0$</td><td>$900</td></tr><tr><td>Validity probability from $100$ Allison and Lemley 2002; low cost.</td><td>0.450</td><td>0.300</td><td>0.250</td><td>$2,000</td><td>$300</td><td>$0$</td><td>$990</td><td>9.9</td></tr><tr><td>Validity probability from $500$ Allison and Lemley 2002; high cost.</td><td>0.450</td><td>0.300</td><td>0.250</td><td>$2,000</td><td>$0$</td><td>$0$</td><td>$9900</td><td>1.8</td></tr><tr><td>Validity probability from $100$ Allison and Lemley 2002; low cost; rejection raises cost.</td><td>0.450</td><td>0.300</td><td>0.250</td><td>$2,000</td><td>$0$</td><td>$-$200</td><td>$850</td><td>8.5</td></tr><tr><td>Validity probability from $500$ Allison and Lemley 2002; high cost; rejection raises cost.</td><td>0.450</td><td>0.300</td><td>0.250</td><td>$2,000</td><td>$0$</td><td>$-$200</td><td>$850</td><td>1.7</td></tr><tr><td>Opposition outcome probabilities; low cost.</td><td>$100$</td><td>0.350</td><td>0.330</td><td>0.320</td><td>$2,000</td><td>$0$</td><td>$0$</td><td>$700</td><td>7.0</td></tr><tr><td>Opposition outcome probabilities; low cost.</td><td>$100$</td><td>0.350</td><td>0.330</td><td>0.320</td><td>$2,000</td><td>$300</td><td>$0$</td><td>$799</td><td>8.0</td></tr></table>
+
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+137
+
+Table 4.4
+
+(continued)
+
+<table><tr><td rowspan="2">Scenario</td><td rowspan="2">Cost of an opposition per patient (thousands)</td><td colspan="3">Outcome probabilities</td><td colspan="3">Benefits (avoided cost)</td><td rowspan="2">Total benefit (thousands)</td><td rowspan="2">Cost-benefit ratio</td></tr><tr><td>Revocation (probability)</td><td>Amendment Rejection (probability) (probability) (probability)</td><td>Revocation (probability)</td><td>Amendment Rejection (thousands)</td><td>Amendment Rejection (thousands)</td><td>Total benefit per patient (thousands)</td></tr><tr><td>Opposition outcome probabilities; high cost. Opposition outcome probabilities; low cost; rejection raises cost.</td><td>$500</td><td>0.350</td><td>0.330</td><td>0.320</td><td>$2,000</td><td>$0$</td><td>$0$</td><td>$700</td><td>1.4</td></tr><tr><td>Opposition outcome probabilities; low cost.</td><td>$100</td><td>0.350</td><td>0.330</td><td>0.320</td><td>$2,000</td><td>$0$</td><td>$-$200</td><td>$636</td><td>6.4</td></tr><tr><td>Opposition outcome probabilities; high cost; rejection raises cost.</td><td>$500</td><td>0.350</td><td>0.330</td><td>0.320</td><td>$2,000</td><td>$0$</td><td>$-$200</td><td>$636</td><td>1.3</td></tr><tr><td>Re-examination outcome probabilities; low cost.</td><td>$100</td><td>0.110</td><td>0.630</td><td>0.260</td><td>$2,000</td><td>$0$</td><td>$0$</td><td>$220</td><td>2.2</td></tr><tr><td>Re-examination outcome probabilities; low cost.</td><td>$100</td><td>0.110</td><td>0.630</td><td>0.260</td><td>$2,000</td><td>$300</td><td>$0$</td><td>$409</td><td>4.1</td></tr><tr><td>Re-examination outcome probabilities; high cost.</td><td>$500</td><td>0.110</td><td>0.630</td><td>0.260</td><td>$2,000</td><td>$0$</td><td>$0$</td><td>$5220</td><td>0.4</td></tr><tr><td>Re-examination outcome probabilities; low cost; rejection raises cost.</td><td>$100</td><td>0.110</td><td>0.630</td><td>0.260</td><td>$2,000</td><td>$0$</td><td>$-$200</td><td>$168</td><td>1.7</td></tr><tr><td>Re-examination outcome probabilities; high cost; rejection raises cost.</td><td>$100</td><td>0.630</td><td>0.260</td><td>$2,000</td><td>$0$</td><td>$0$</td><td>$200</td><td>$168</td><td>0.3</td></tr></table>
+
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+138
+
+Hall, Graham, Harhoff, and Mowery
+
+by Levin and Levin (2002). We also experiment with assuming a social cost for rejection and an avoided cost for patent amendment as well as for patent revocation.
+
+Almost all the scenarios yield cost-benefit ratios well in excess of unity, with the exception of some of those that use the re-examination outcome probabilities. The lowest ratios for each panel are for the high opposition cost cases. We conclude that unless the opposition system is very expensive to operate and yields results similar to those now obtained with the re-examination system, it would be likely to generate substantial welfare gains.
+
+## VI. Conclusions and Additional Questions
+
+The determinants and characteristics of patent challenge procedures are important issues in any assessment of intellectual property systems of the United States or other industrial economies. In a knowledgebased economy, intellectual property systems are constantly challenged by the advance of technology, a process that, among other things, creates new artifacts to which the necessarily backward-looking patent system must respond. A knowledge-based economy also is one in which the high political salience of national and global intellectual property systems means that they are the focus of political lobbying to strengthen, adapt, or weaken specific features of intellectual property regulation, administration, and law to favor particular interests. Both these forces have been at work within the U.S. intellectual property system during the past quarter-century; a period of significant strengthening of patent holder rights has triggered a debate over the appropriate level and limits of such rights. This debate has important transatlantic and global repercussions and analogues.
+
+As the knowledge-based economy has evolved in the United States and elsewhere, it has become clear that such evolution brings with it increasing attention to the ownership of knowledge in the form of intellectual property by firms and governments. Together with the increasing importance of software in all areas, which in itself is an impetus to the growth of the knowledge-based economy, we have an expansion of the subject matter base that must be considered by patent offices everywhere. Such expansion is not a new phenomenon. It has tended to happen whenever important changes take place in technological regimes, but it does lead to two kinds of adjustment problems: first, debates over the validity of the subject matter extension, such as those
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+139
+
+that have occurred between the United States and Europe over business method patents, and second, a concern that overly broad and invalid patents may issue early in the development of the technology because of the lack of prior art in the relevant patent office databases, even though such prior art may exist among those who practice the technology. The second problem, which may be defined broadly as the difficulty of determining the validity of a patent application in new and less well understood technologies, is the one that an inter partes postgrant review or re-examination system might be designed to address.
+
+To understand how such a system might work, the analysis in this paper highlights several interesting features of the patent challenge systems of the U.S. and EPO systems. First, the current U.S. re-examination procedure differs dramatically from the EPO opposition procedure in almost all its features, of which the most significant are the identity of the party requesting a re-examination (the patent owner in more than 40 percent of the cases) and the outcomes (which rarely include revocation of the patent). These characteristics of re-examination hardly qualify it as the sort of adversarial procedure that EPO oppositions represent, but they are not surprising when we consider the differences in structure between the two procedures: (1) a three-examiner panel including the original examiner, but not as chair, in the EPO versus a single examiner, often the same as the original examiner, in the USPTO; (2) the prohibition in later litigation of questions that could have been raised in a re-examination in the United States; and (3) the inter partes nature of the proceeding at the EPO versus the ex parte nature of the proceeding at the USPTO.
+
+Keeping in mind the significant differences between the re-examination and opposition processes, our comparative analysis suggests that EPO oppositions are not significantly swifter in resolving challenges than the USPTO re-examination proceedings, as might be expected given their more adversarial nature. Indeed, opposition proceedings in some cases (and almost certainly in important, complex cases with numerous opponents, appeals, etc.) may well take as much time to be resolved as does litigation in the U.S. system. Nonetheless, the higher frequency of opposition (which is presumably due to the lower cost associated with opposition compared to the cost of litigation in the United States) within the EPO system suggests that this process handles many more legal disputes over patent validity than are addressed by the U.S. re-examination process.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+140
+
+Hall, Graham, Harhoff, and Mowery
+
+Our analysis also indicates that patent amendment, rather than revocation, is more likely for oppositions in relatively new fields of inventive activity, for more complex patents, or for oppositions in which numerous opponents participate. We lack evidence on the extent to which oppositions are followed by litigation in the European patent system, so we cannot determine whether the lack of any advantage for oppositions in resolving patent disputes quickly is offset by a reduction of litigation rates associated with oppositions. The EPO system may offer few advantages over the U.S. system for postissue patent challenges, but we cannot address this issue without analyzing litigation data for both the U.S. and European systems. Any comprehensive assessment of the social costs and benefits of the two challenge systems requires a consideration of both the patent office processes of postgrant challenge (opposition or re-examination) and legal system litigation. Nevertheless, the preliminary computations presented here suggest that the social gains from such a system might be substantial.
+
+## Notes
+
+We appreciate the extremely helpful comments on an earlier draft by the editors of this journal, Adam Jaffe and Josh Lerner.
+
+- 1. See Lunney (2001) for an argument that the nonobviousness test has been weakened
+since the creation of the Federal Circuit Court of Appeals in 1982.
+2. Presumably, if the invention has already been reduced to practice by others, the poten-
+tial gain from incenting an inventor is zero, so we are left only with the deadweight loss
+from monopoly.
+3. This decision presumably made it more difficult to reject patents such as U.S. 6368227,
+the patent on a swinging method that uses a technique known by children for decades
+but not placed on the record. Note that this particular patent has been subject to a re-
+examination request by the U.S. patent commissioner because of the publicity it received.
+The problem with patents like this one is not necessarily that they are enforceable in the
+courts but that they clog the system and raise its total cost.
+4. As quoted in the Los Angeles Times, February 7, 2003.
+5. Our evidence (Graham et al. 2003) suggests that the average reexamination takes less
+than two years, slightly shorter than the average duration of a patent lawsuit (thirty-
+one months), but this difference is not large (especially in view of the high variance
+of the “average duration” estimate for a trial). Some observers have criticized the re-
+examination system for not providing a fast and cheap alternative to trial.
+6. An alternative re-examination procedure, the inter partes re-examination, was enacted
+by the U.S. Congress in 1999 (see the American Inventors Protection Act, codified
+in 35 USC 311-318. Several commentators have questioned the efficacy of the inter partes
+re-examination on the grounds that it allows the third-party requestor limited opportuni-
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+141
+
+ties of involvement; prevents any adverse findings of the USPTO from being appealed to the courts; and also precludes the raising of any questions of validity on grounds that were, or may have been, raised during the inter partes re-examination from being litigated in the courts (Neifeld 2000). The USPTO reports no inter partes re-examination requests in 2000 and one in 2001, suggesting that the procedure has been little used.
+
+7. Kaufman Company v. Lantech, Inc., 807 F.2d 970 (CAFC 1986), suggesting that evidentiary burdens are likely higher for challengers after re-examination.
+
+8. This owner initiation occurs in many cases in which declaratory validity determinations are being sought by a challenger third party. These suits, which make the patentee the defendant, are often initiated only after a demand by the patent holder that the challenger stop infringing on the patent, thus putting the initial move in the hands of the patent holder.
+
+9. The Convention on the Grant of European Patents, also referred to here as the European Patent Convention (EPC) was enacted in October 1973. It is the legal foundation for the establishment of the EPO. The full text of the convention is available at http:// www3.european-patent-office.org/dwld/epc/epc_2000.pdf.
+
+10. At the time of writing, the exchange rate was approximately 0.93 euros per dollar. As in other patent systems, the official patent office fees are a relatively small part of the costs (in this case, €4,300. Professional representation before the EPO amounts to €5,500 on average, whereas translation into the languages of eight contracting states requires €11,500. Renewal fees for a patent maintained for ten years amount to roughly €8,500. See “Cost of an average European patent as of 1.7.99,” http://www.europeanpatent-office.org/epo/new/kosten_e.pdf (accessed January 14, 2002).
+
+11. Using data from court filings in Mannheim and Duesseldorf, Cremers (2003) identifies 715 patent infringement cases involving 905 patents during the 1993–1995 period. These courts account for about 55 to 60 percent of all cases filed in Germany. The threeyear average of EPO patents issued since 1980, which included Germany as one of the covered regions, is about 71,000. This figure is for a litigation probability of about 2.1 percent (assuming that the process is stationary).
+
+12. The latter statement is premised on the argument that more oppositions than suits are filed partly because they are lower cost.
+
+13. In fact, according to data reported by Lanjouw and Schankerman (2002), it is the other way around in the United States: small firms and independent inventors are far more likely to file infringement suits than are larger firms.
+
+14. This is the reason why the European Patent Office is entitled to pursue opposition cases, even if the two parties wish to settle the case. Thus, this feature of an opposition system can be expected to increase social welfare.
+
+15. With linear demand, the welfare loss from monopoly will be one-half the monopoly rents.
+
+## References
+
+AIPLA (American Intellectual Property Law Association) 2001. "Report of Economic Survey," Washington, DC: American Intellectual Property Law Association.
+
+Allison, J. R., and M. Lemley. 2002. "The Growing Complexity of the United States Patent System." Boston University Law Review 82(1): 77-144.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+142
+
+Hall, Graham, Harhoff, and Mowery
+
+Allison, J. R., M. A. Lemley, K. A. Moore, and R. D. Trunkey. 2003. "Valuable Patents." Draft Manuscript, University of Texas at Austin, University of California at Berkeley, and George Mason University.
+
+Barton, J. H. 2000. "Reforming the Patent System." Science 287: 1933-34.
+
+Cremers, K. 2003. "Determinants of Patent Litigation in Germany." Paper presented at the ZEW Workshop on the Empirical Economics of Innovation and Patenting, Mannheim, Germany, March 14-15, 2003.
+
+Graham, J. H. S., B. H. Hall, D. Harhoff, and D. C. Mowery. 2003. "Patent Quality Control: A Comparison of U.S. Re-examination and European Patent Oppositions." In W. M. Cohen and S. A. Merrill, eds. Patents in the Knowledge-Based Economy. Washington, DC: National Academy Press: pp. 74-119.
+
+Hall, B. H. 2003. "Business Method Patents and Innovation." Paper prepared for the Atlanta Federal Reserve Bank Conference on Business Method Patents, Sea Island, Georgia. April 2003.
+
+Harhoff, D., and B. H. Hall. 2002. "Intellectual Property Strategy in the Global Cosmetics Industry." Unpublished Paper. Ludwig-Maximilians Universitatet and University of California at Berkeley.
+
+Harhoff, D., and M. Reitzig. 2001. "Determinants of Opposition Against EPO Patent Grants—The Case of Biotechnology and Pharmaceuticals." London: CEPR Discussion Paper No. 3645.
+
+Harhoff, D., F. M. Scherer, and K. Vopel. 2003. "Citations, Family Size, Opposition and the Value of Patent Rights-Evidence from Germany." Research Policy 32(8):1343-63.
+
+Janis, M. D. 1997. "Rethinking Reexamination: Toward a Viable Administrative Revocation System for U.S. Patent Law." Harvard Journal of Law and Technology 11(1): 1-122.
+
+Kaufman Company v. Lantech, Inc., 807 F.2d 970 (CAFC 1986).
+
+Kingston, W. 2001. "Innovation Needs Patents Reform." Research Policy 30: 403-23.
+
+Lanjouw, J. O., and M. Schankerman. 2001, December. "Enforcing Intellectual Property Rights." Working paper no. 8656. Cambridge, MA: National Bureau of Economic Research.
+
+Lanjouw, J. O., and M. Schankerman. 2002, August. "Enforcing Patent Rights: An Empirical Study." Washington, DC: Brookings Institution and LSE.
+
+Lemley, M. A. 2001. "Rational Ignorance at the Patent Office." Northwestern University Law Review 95(4): 1495-1532.
+
+Lemley, M. A., and J. R. Allison. 1998. "Empirical Evidence on the Validity of Litigated Patents." American Intellectual Property Law Association Quarterly Journal 185: 26.
+
+Lerner, J. 1995. "Patenting in the Shadow of Competitors." Journal of Law and Economics 38(2): 463-96.
+
+Levin, R. C., and J. Levin. 2002. "Patent Oppositions." New Haven, CT, and Palo Alto, CA: Yale University and Stanford University.
+
+Lunney, G. S., Jr. 2001. "e-Obviousness." Michigan Telecommunications Technology Law Review 7: 363-422.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+
+---
+
+Improving U.S. Patent Quality
+
+143
+
+Magrab, E. B. 1993. "Patent Validity Determinations of the ITC: Should U.S. District Grant Them Preclusive Effect?" Journal of the Patent & Trademark Office Society 75(125): 127-35.
+
+Merges, R. P. 1999. "As Many as Six Impossible Patents Before Breakfast: Property Rights for Business Concepts and Patent System Reform." Berkeley High Technology Law Journal 14: 577-615.
+
+Michel, J., and B. Bettels. 2001. "Patent Citation Analysis—A Closer Look at the Basic Input Data from Patent Research Reports." Scientometrics 51: 181-201.
+
+Neifeld, R. 2000. "Analysis of the New Patent Laws Enacted November 29, 1999." Journal of the Patent and Trademark Office Society 82: 181.
+
+Rogan, James E. 2002. Statement before the Subcommittee on Commerce, Justice, State and the Judiciary Committee on Appropriations, U.S. House of Representatives, Ashburn, VA April 23, 2002. Available at <http://www.uspto.gov/web/offices/com/ speeches/houseapprop.htm>
+
+Stauder, D. 1989. "Patent- und Gebrauchsmusterverletzungsklagen in der Bundesrepublik Deutschland, Großritannien, Frankreich und Italien" (Patent and Utility-model Infringement Lawsuits in the Federal Republic of Germany, Great Britain, France and Italy). Schriftenreihe zum Gewerblichen Rechtsschutz (Series on Intellectual Property Law) 89, Max-Planck-Institute for the Study of International Patents, Copyrights, and Anti-trust Law. Cologne.
+
+Stauder, D. 1996. Aspekte der Durchsetzung gewerblicher Schutzrechte: Fachkundiger Richter, schnelles Verfahren und europaweites Verletzungsverbot (Aspects of the Enforcement of Commercial Patent Rights: Expert Judges, Expedited Procedure and European-wide Infringement Standards). In J. Straus, ed., Aktuelle Herausforderungen des geistigen Eigentums (Current Challenges of Intellectual Property). Cologne.
+
+Stauder, D., P. von Rospatt, and M. von Rospatt. 1999. "Protection transfrontaliere des brevets européens (Trans-border Protection of European Patent Rights)." Revue Internationale de Droit Economique (International Review of Law and Economics) 1: 119-33.
+
+Straus, J. 1996. "Die Aufrechterhaltung eines europäischen Patents in Geändertem Umfang im Einspruchsverfahren und ihre Folgen (The Legal Status of a European Patent After Amendment in the Opposition Process and its Consequences)." In J. Straus (ed.), Aktuelle Herausforderungen des geistigen Eigentums (Current Challenges of Intellectual Property). Cologne, 171-84.
+
+This content downloaded from 012.045.190.190 on July 25, 2016 23:37:14 PM
+
+
+All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c)
+
+
+1
+

processed/00278_W2200657106/document.md

@@ -0,0 +1,311 @@
+# Measuring patent similarity by comparing inventions functional trees
+
+Gaetano Cascini$^{1}$ and Manuel Zini$^{2}$
+
+$^{1}$ University of Florence, Italy, gaetano.cascini@unifi.it
+
+2 drWolf srl, Italy, mlzini@drwolf.it
+
+Abstract: The estimation of the conceptual distance between patents is a critical issue for Computer-Aided patent portfolio analysis systems, an emerging class of computer tools for supporting R & D analyses and decisions, patent infringement risk evaluation, technology forecasting. The aim of the present work is the introduction of an original algorithm for patent comparison: since typical text analyses are biased by the writer's style, the inventions similarity is here estimated by comparing the components and their hierarchical and functional interactions automatically extracted by means of a custom software tool. The whole procedure is clarified with an exemplary application in the field of electric current circuit breakers.
+
+Keywords: Patent mining, document similarity, plagiarism
+
+## 1. Introduction
+
+Computer-aided patent portfolio analysis is an emerging topic in the scientific community and attracts interests from several disciplines, since it deals with economical, technical, management, life science issues [1-4] .
+
+Indeed, computers have been used for patent searches and analyses since the '90s, but most of the applications were limited to statistical computations by means of bibliometric methods. Indeed, these techniques are still adopted as a relevant source of information [5] . This is mainly due to a heritage of traditional practices when statistical techniques were adopted to examine the effect of technology development in economic, national and international contexts or to plan a corporate technology activity at a corporate level [2] .
+
+The introduction of text-mining algorithms has created new opportunities for identifying complex relationship among patent documents. Besides, up to now, the researchers in this field have dedicated major attention to Information Extraction purposes in order to capture relevant information from patents, while still limited
+
+Please use the following format when citing this chapter:
+
+Cascini, G. and Zini, M., 2008, in IFIP International Federation for Information Processing, Volume 277; ComputerAided Innovation (CAI); Gaetano Cascini; (Boston: Springer), pp. 31-42.
+
+---
+
+32
+
+G. Cascini and M. Zini
+
+studies exist about patent comparison and trend extraction, except applications of general purpose clustering algorithms [6].
+
+Nowadays, computer-based systems for patent analysis are assuming more and more specialist roles and will cover soon a wider range of application areas like:
+
+- - generation of new research directions for biomedical studies [1] ;
+- comparison of the morphology portfolio of different technologies [2] ;
+- evaluation of the R & D landscape and business opportunities [3] ;
+- evaluation of the risk of patent infringement [4] .
+All the above mentioned activities require the estimation of the conceptual distance between patents, but all the approaches proposed so far are based on keywords comparisons (e.g. co-occurrences of terms and/or multi-words), while the nature of patent contents is poorly taken into account. An even more critical issue is the dependence of these techniques to the language style of the writer; as a result, very often it happens that patents of the same inventor or company are clustered together despite their different contents, while conceptually close inventions are considered distant from each other just because they adopt a different terminology.
+
+In the present paper the authors, also thanks to previous experiences in the field of plagiarism detection, propose a novel technique for assessing patent similarity as a means for avoiding the impact of the language style on patent comparison.
+
+In the next chapter we report a brief survey of plagiarism detection techniques; then the third section describes the proposed algorithm also resuming some previous works relevant for the present application. Then an exemplary application of the proposed similarity metric is shown, by comparing the results of the automatic analyses performed by means of a prototype software to the results obtained by humans in the field of electrical circuit interruption devices. Finally, in chapter five, the discussion is focused on the capabilities and the sore points of the proposed technique.
+
+## 2. State of the art techniques for plagiarism detection
+
+Plagiarism is a growing problem and has recently received a lot of attention. The increase in availability of material in digital form has made plagiarism much easier.
+
+However, the use of digital media also means that there are greater opportunities to trace plagiarism by means of dedicated software tools. Automated plagiarism detection as a subject has not yet achieved the same degree of scientific maturity as other subjects in the Text-Mining field, but a growing number of publications [7] , websites and recently available products on this matter [8, 9] indicates that both the scientific and the industrial communities have started to recognize and acknowledge the existence of a recent problem which is yet awaiting its systematic solution [10] .
+
+---
+
+Measuring patent similarity by comparing inventions functional trees
+
+33
+
+There are several approaches to automatically identify plagiarism in different types of documents. The SCAM tool developed by Shivakumar [11] is based on building unions of word sets and counting domain-specific keywords in them. Plagiarism is then revealed via unexpected or otherwise suspicious occurrences of such keywords.
+
+In some works, plagiarism detection has been regarded as a special case of duplicate document detection, which is both a necessary and difficult task in the management of large scale and very large scale databases (possibly multi-media databases). A variety of data mining methods and text-based techniques for such purposes have been proposed and investigated [12] .
+
+Comparing whole document checksums is simple and suffices for reliably detecting exact copies; however, detecting partial copies is subtler; in some works, for example in [13] , an approach based on multiple fingerprints evaluation is used to detect partial copies. These techniques mostly rely on the use of k-grams, i.e. contiguous sub-strings of characters with length k. The process requires to divide a document into k-grams, and extract a hash value from each one [14, 15] . The result of this process is a fingerprint that represents the document in each of its sub-parts of length k, further exploited for comparison. Such a procedure, however, does not take into account the behavioral pattern of the plagiarist. In [16] , the edit distance is introduced as a similarity metric between chunks of text.
+
+In [17] an hypothetical behavioral pattern of the plagiarist is taken into account. The authors hypothesize that the behavior of the plagiarist consists in the repetition of three prototypical actions: insertion, deletion and substitution. This actions can be performed at any level of the document structure, phrase, paragraph or chapter. Distance between documents is then evaluated recursively exploiting the Levensthein edit distance [18] .
+
+All this approaches take into account plagiarism as an operation on text to be considered a mere sequence of characters, with no attempt to capture the likely semantic nature of plagiarism.
+
+The main limit of plagiarism detection algorithms, as a means for identifying similar inventions and patent infringements, is their focus on the language of the description instead of the structure of the invention. Still some lessons learned can be readapted to the specific situation.
+
+In facts, an acknowledged measure of similarity is expressed in the form
+
+$$SIM_{ij}=\frac{keywords_{ij}+keywords_{ji}}{keywords_{i}+keywords_{j}}\quad (1)$$
+
+where $keywords_{ij}$ is the number of occurrences of keywords of the document $i$ found in the document $j$ and $keywords_i$ is the overall number of keywords extracted from the document $i$ .
+
+An exemplary attempt to reuse in a novel form such a typical plagiarism assessment metric is proposed in [4] , where the authors measure patent similarity
+
+---
+
+34
+
+G. Cascini and M. Zini
+
+by comparing the number of shared SAO triples (Subject Action Object), instead of the keywords alone. The main advantage of the SAO-based approach is that patents are compared in terms of functions delivered by the elements of the invention and general terms are filtered out.
+
+Nevertheless, we observe that while taking into account syntactical information this comparison is still too dependent on the mere text and, as such, it depends more on the writer's style than on the actual `semantics' of the described invention.
+
+In this paper we propose an alternative approach which is not based on text comparison but on the comparison of the structural and functional architecture of the invention disclosed in a patent.
+
+## 3. A new approach to measure patent similarity
+
+As discussed above, the main limit of the traditional techniques for estimating the conceptual distance between two patents is the dependence on the language style of the inventor.
+
+In order to clarify this concept let's consider the following excerpts:
+
+- - US4,713,635: “For example, the barrier portion or insert 107 includes a rib
+or tongue 109 that is aligned with rib or tongue portions 111, 113, and
+115.”
+- US4,056,798: “One end of the cradle 48 forms a tongue member 50 which
+is releasably secured within an apertured latch 52 of the trip mechanism
+42. […] . . ] This deflection causes the bimetal element to engage a hook-
+shaped projection 66 of the latch 52, pulling the latch 52 to the right and
+causing the tongue 50 of the cradle 48 to be disengaged from the latch 52.”
+In both patents, a tongue member is a feature of the disclosed invention and can be considered as a subsystem of a further element of the invention (the barrier portion 107 and the cradle 48 respectively). The property of being a subsystem is expressed by means of totally different locations: <component i> “includes” <component j> and <component i> “of the” <component j>. It is worth to notice that the adoption of a SAO-based comparison criterion does not allow to identify this kind of similarity, whatever is the richness and quality of its synonyms list. Similar remarks can be applied also to functional and positional interactions.
+
+In the present paper the authors suggest to evaluate the similarity between two patents by comparing their functional tree [19] , i.e. the hierarchical architecture of the invention's components and their functional interactions. In facts, working with the functional tree allows to identify conceptual similarities like the example presented above and to limit the influence of the language style. Moreover, the algorithm described hereafter allows to focus the comparison on a subset of components and/or interactions according to the peculiarity score proposed in [20] .
+
+---
+
+Measuring patent similarity by comparing inventions functional trees
+
+35
+
+### 3.1 Previous works: automatic functional analysis of patents and extraction of invention peculiarities
+
+The authors are working on the development of new techniques and algorithms for patent analysis and comparison [19-23] . As a result of these previous experiences a prototype software system (named PatAnalyzer) has been developed with the following functionalities:
+
+- - identify the components of the invention;
+- classify the identified components in terms of detail/abstraction level and their
+compositional relationships in terms of supersystem/subsystem links;
+- identify positional and functional interactions between the components both
+internal and external to the system;
+- build a thesaurus of “alternative denominations” of the functional elements
+identified in a given set of patents (hereafter called project );
+- identify the most relevant components of each patent for a given project
+according to a ranking criterion which combines the detail level of the
+description with the Inverse Document Frequency , i.e. the “rarity” of each
+synset of the Thesaurus.
+In Figure 1 the exemplary results related to the patent US6,064,024 are shown: the conceptual map visualizes the components of the inventions, their hierarchical and functional interactions, as well as their relevance score by means of a color code.
+
+It is worth to notice that the score assigned to the components of each invention allows to select a subset of sentences from the description and the claims of the patent where the top-ranked components are mentioned. In [20] it was demonstrated that such a subset of sentences is sufficient for a “person skilled in the art” for understanding what the core of the patent is about.
+
+In this paper, the top-ranked components and their hierarchical and functional relationships are adopted as a means to compare the inventions of a given project in order to estimate their similarity, as described in the following paragraph.
+
+### 3.2 Comparing the functional tree of two inventions
+
+In this work it is assumed that two technical systems belonging to the same field of application, sharing the same components, structured with the same architecture and characterized by the same functional interactions are conceptually identical. As a consequence, the similarity between two patents is estimated by comparing their components, hierarchical relationships and functions.
+
+---
+
+36
+
+G. Cascini and M. Zini
+
+![Figure](figures/figure_002.png)
+
+Figure 1. Portion of the conceptual map of the patent US6,064,024: the arrows with no label represent hierarchical relationships (the component at the tip of the arrow is a subsystem of the components at the tail of the arrow); labeled arrows represent functional and positional interactions; the colors highlight the relevance score of each component.
+
+---
+
+Measuring patent similarity by comparing inventions functional trees
+
+37
+
+Such a comparison is made also taking into account the alternative denominations of each component, by means of the Thesaurus built according to the rules defined in [20] . More precisely, while comparing the functional trees of two inventions, two nodes are considered equivalent if they belong to the same synset in the Thesaurus of the project .
+
+Then the following formula is applied:
+
+$$SIM_{ij}=\alpha\frac{|\Gamma(i)\subset\Gamma(j)|+|\Gamma(j)\subset\Gamma(i)|}{|\Gamma(i)|+|\Gamma(j)|}+\beta\frac{|C(i)\subset C(j)|+|C(j)\subset C(i)|}{|C(i)|+|C(j)|}\quad (2)$$
+
+where $\Gamma(i)$ is the set of hierarchical and functional interactions belonging to the $i$ th patent; $\Gamma(i) \subset \Gamma(j)$ stands for the hierarchical and functional interactions of the $i$ -th patent appearing also in the functional tree of the $j$ -th patent; $C(i)$ is the list of components belonging to the $i$ -th patent; $\alpha$ and $\beta$ are coefficients to weight the mutual relevance of interactions and components.
+
+It is worth to note that the formula (2) can be applied to the whole set of components and interactions extracted from each patent or to a subset of topscored components and their interactions. Thus, three parameters must be arbitrarily set to evaluate the similarity between two patents: $\alpha$ , $\beta$ and $\gamma$ , where the latter represents the threshold score for components selection ( $y=0$ means that the whole hierarchical/functional tree is considered to estimate the patent similarity, while $\gamma=1$ means that only the component with the highest score and its interactions are taken into account).
+
+Whatever is the value assumed by $\alpha$ , $\beta$ and $\gamma$ , the similarity matrix of a given project , built in analogy of the incidence matrix proposed in [24] , is a symmetric square matrix $N \times N$ ( $N$ being the number of documents analyzed in the project ) constituted by the similarities of each patents' pair. In other words each cell $(i,j)$ contains the estimated similarity among the $i$ -th and the $j$ -th patent.
+
+The rules to define the most suitable values for $\alpha$ , $\beta$ and $\gamma$ are still under validation; nevertheless some general directions have already been developed and are briefly discussed in section 5 .
+
+## 4. Exemplary application: electrical circuit interruption devices
+
+In order to clarify the procedure described in section 3.2 and to demonstrate its capabilities this chapter reports an exemplary application in the field of electrical circuit interruption devices.
+
+On the base of a previous experience with ABB SACE (www.abb.com), an evolutionary analysis of electrical circuit breakers has been made at the MTI Lab of the University of Florence. A set of 85 patents (ABB project ) was selected as a combination of two citation trees, i.e. the patents cited from US6,064,024 and US6,373,016 following backward citations up to three levels from the source
+
+---
+
+38
+
+G. Cascini and M. Zini
+
+patent. Figure 2 shows a portion of the citation tree related to the former patent: each arrow represents a backward citation.
+
+In order to perform such an evolutionary analysis each patent has been analyzed by a technician and a summary has been extracted in the form ProblemSolution (figure 2 , close up). In other words, the citation trees have been manually translated into a Problem-Solution tree which can be used with similar purposes of the OTSM-TRIZ Problem Flow Network proposed in [25] .
+
+![Figure](figures/figure_004.png)
+
+Figure 2. Excerpt from the Patent Citation tree and close up on the Problem-Solution notations associated to a pair of citing-cited patents.
+
+The details of this evolutionary analysis are outside the scopes of the present paper and will be presented in a next publication. Besides, such an extensive analysis carefully operated by humans constitutes a valuable resource to validate the proposed similarity score.
+
+According to the above mentioned procedure, the 85 patents have been processed by PatAnalyzer, thus producing a corresponding number of conceptual maps (as the functional tree shown in figure 1 ) and related relevance scores. Moreover the analysis has generated a thesaurus specifically dedicated to the ABB project (figure 3 ). The similarity matrix (figure 4 ) has been calculated assigning the same weight to components and interactions (i.e. $\alpha = 1$ , $\beta = 1$ ) and discarding those components with a score lower than one fifth of the maximum score of the entire project (i.e. $\gamma = 0.2$ ). The whole computation (functional tree identification, thesaurus generation, definition of the similarity matrix) has required less than 60 minutes on a standard laptop.
+
+---
+
+Measuring patent similarity by comparing inventions functional trees
+
+39
+
+![Figure](figures/figure_002.png)
+
+Figure 3. Portion of the Thesaurus graph of the ABB project: the arrows point from hypernyms to hyponyms, (red) dashed lines represent alternative denominations. The box in the right-low corner lists the patents containing the selected word/multi-word.
+
+![Figure](figures/figure_004.png)
+
+Figure 4. Similarity matrix for the 85 patents of the ABB project: entire matrix (above) and close up on a reduced number of patents (below). The cells of the matrix can be colored according to one or more thresholds defined by the user (here 1 % and 2 % respectively) in order to highlight similar patents and clusters.
+
+---
+
+40
+
+G. Cascini and M. Zini
+
+The resulting similarities have been compared with the outputs of the analysis manually performed by two operators as described above. Such a comparison has revealed a consistent coherence among the two set of results.
+
+As an exemplary demonstration the following patents have revealed relevant matched features: US6,064,024, US5,763,847, US5,130,504, US4,424,428. Indeed all those inventions are characterized by the adoption of a permanent magnet aimed at the deviation and elongation of the electric arc (see also figure 2 ):
+
+- - US6,064,024: “ […] . . . ] Thus the permanent magnet's strong field will always be oriented to enhance
+the potentially weak self magnetic field as described with respect to the embodiment in FIG. 1 .
+Therefore the resultant Lorentz force acting on the arc will always be strong enough to drive the arc
+off the contact pads 30 and 32 and along stationary contact 17 even when the self magnetic field is
+weak (low current) […] . . ] ” .
+- US5,763,847: “ […] . . . ] As the arc travels into the arc extinguishing chamber 34, it also interacts with
+the individual magnetic fields produced by permanent magnet 54 in each of the first type of splitter
+plates 40. […] . . . ]. The interaction of the arc current with this magnetic field around each plate causes
+the arc 77 to move in circles on the surface of the splitter plate casing 44. Thus the arc energy is not
+constricted to one spot on the casing surface as occurred in previous arc chambers, thus erosive
+effects of arcs impinging the splitter plates are reduced in the present design. […] . . ] ” .
+- US5,130,504: “ […] . . . ] The permanent magnets 80-88 are polarized across the width thereof to
+establish a magnetic field B (FIGS. 10 and 11) directed front-to-rear through the respective arc
+chambers, the plates 54 and 90 forming a magnetic path around the outside of the switching
+apparatus and an air gap across the respective arc extinguishing chambers. […] . . ] ” .
+- US4,424,428: “ […] . . . ] The magnetic field of magnet 38, which is present when the arc appears, leads
+to rotation of the arc along the annular tracks formed by contact surfaces 34, 36 and rapid
+extinction of the arc in a well known manner […] . . ] ” .
+A selection of the paragraphs containing the components and interactions contributing to the similarity score (in this case “ permanent magnet ” and related denominations like “ interior magnet ” , figure 3 ) has been judged sufficient for a person skilled in the art to understand the role of the component and to assess the originality of the solution.
+
+## 5. Discussion and conclusions
+
+The technique proposed in this paper defines the similarity of two patents as the match in terms of functional structure of the inventions, instead of the traditional frequency of keywords co-occurrences. By doing so, patents are grouped into more appropriate conceptual classes and more intrinsically homogeneous clusters can be produced. As explained before, keywords co-occurrence analysis deals with the patents as a whole and considers only the frequency of co-citations. Thus, the result of grouping may be superficial or even spurious since those statistics would not reveal the internal structural relationships between patents.
+
+Besides the proposed algorithm allows to find analogies between inventions described with totally different locations, while general poorly informative elements (according to a ranking which depends on the specific project and not to a general terminology classification) are neglected.
+
+---
+
+Measuring patent similarity by comparing inventions functional trees
+
+41
+
+An open issue is the definition of the rules to assign a proper value to the weights $\alpha$ (interactions), $\beta$ (components) and the threshold $\gamma$ . According to the analyses performed so far, the components part of the formula (2) can lead to wrong estimations of the similarity when dealing with a patent having a reduced number of components: in these cases the similarity score is zero when the relevance score of the components is low, since there are no opportunities for matching other patents. Vice versa, if the relevance score of the components of an invention characterized by a reduced number of elements is high, the patent will result highly similar with many patents of the project. Besides, the similarity between patents with a reduced number of components is more suitably assessed by the interactions part of the formula (2). Inversely, in case of inventions with a high number of components described in the patent, also the components part of (2) significantly contributes to the similarity assessment.
+
+A further emerging note is that the contribution of the interactions to the overall similarity score inversely depends on the value assigned to $\gamma$ , i.e. the relevance/peculiarity threshold defining the number of components to be considered from each patent in order to perform the comparison. In other words, hierarchical and functional interactions between components provide relevant contributions for similarity assessment if a wider portion of the functional tree is considered for each patent under evaluation, while in a selection limited to the topscore elements from each patent the similarity is mostly evaluated in terms of components.
+
+The authors are involved in a more extensive validation of the proposed algorithm with the aim of providing more detailed guidelines for the definition of the most suitable parameters $\alpha$ , $\beta$ and $\gamma$ for a given set of patents.
+
+## References
+
+1. Fluck J., Zimmermann M., Kurapkat G., Hofmann M.: Information extraction technologies for the life science industry. Drug Discovery Today: Technologies, vol. 2, Issue 3, pp. 217224 (2005).
+
+2. Yoon B., Park Y.: A systematic approach for identifying technology opportunities: Keywordbased morphology analysis. Technological Forecasting and Social Change, vol. 72, Issue 2, pp. 145-160 (2005).
+
+3. Fabry B., Ernst H, Langholz J., Köster M.: Patent portfolio analysis as a useful tool for identifying R&D and business opportunities—an empirical application in the nutrition and health industry. World Patent Information, vol. 28, Issue 3, pp. 215-225 (2006).
+
+4. Bergmann I., Butzke D., Walter L., Fuerste J. P., Moherle M. G., Erdmann V. A.: Evaluating The Risk of Patent Infringement By Means of Semantic Patent Analysis: The Case of DNAChips, Proceedings of the R&D Management Conference, 4-6 July 2007, Bremen, Germany, ISBN: 0-9549916-9-9 (2007).
+
+5. Daim T. U., Rueda G., Martin H., Gerdstri P.: Forecasting emerging technologies: Use of bibliometrics and patent analysis. Technological Forecasting and Social Change, vol. 73, Issue 8, pp. 981-1012 (2006).
+
+6. Trappey A. J. C., Hsua F. C., Trappey C. V., Lin C.: Development of a patent document classification and search platform using a back-propagation network. Expert Systems with Applications, vol. 31, Issue 4, pp. 755-765 (2006).
+
+---
+
+42
+
+G. Cascini and M. Zini
+
+7. Clough P.: Plagiarism in natural and programming languages: An overview of current tools and technologies. Internal Report CS-00-05, University of Sheffield (2000). Available at http://ir.shef.ac.uk/cloughie/papers/plagiarism2000.pdf. Last access 27 Apr 2008.
+
+8. EVE Plagiarism Detection System. http://www.canexus.com. Last access 27 Apr 2008.
+
+9. Turnitin. http://www.turnitin.com/static/home.html. Last access 27 Apr 2008.
+
+10. Barrett R., Malcolm J., Lyon C.: Are we ready for large scale use of plagiarism detection tools? Proceedings of the 4th Annual LTSN-ICS Conference, NUI Galway, pp. 79-84 (2003).
+
+11. Shivakumar N.: Detecting digital copyright violations on the internet. PhD thesis Stanford University (1999). Available at http://infolab.stanford.edu/~shiva/thesis.html, Last access 27 Apr 2008.
+
+12. Lopresti D.: A comparison of text-based methods for detecting duplication in document image databases. Proceedings of Document Recognition and Retrieval VII (IS and T SPIE Electronic Imaging) San Jose (USA), pp. 210–221, January (2000).
+
+13. Schleimer A. A. S., Wilkerson D. S., Aiken A.: Winnowing: local algorithms for document fingerprinting. Proceedings of the 2003 ACM SIGMOD International Conference on Management of data. ACM 1-58113-634-X/03/06. (2003). Available at http://theory.stanford.edu/~aiken/publications/papers/sigmod03.pdf. Last access 27 Apr 2008.
+
+14.Heintze N.: Scalable document fingerprinting. In Proceedings of the 1996 USENIX Workshop on Electronic Commerce, pp. 191-200 (1996). Available at http://citeseer.ist.psu.edu/348631.html. Last access 27 Apr 2008.
+
+15. Koala Document Fingerprinting (KDF). http://www-2.cs.cmu.edu/afs/cs/user/nch/www/koalainfo.html. Last access 27 Apr 2008.
+
+16. Mandreoli P. F., Martoglia R.: Un metodo per il riconoscimento di duplicati in collezioni di documenti. Proceedings of the Eleventh Italian Symposium on Advanced Database Systems, SEBD. (2003).
+
+17. Zini M., Fabbri M., Moneglia M., Panunzi A.: Plagiarism Detection Through Multilevel Text Comparison. In Proceedings of the Second International Conference on Automated Production of Cross Media Content for Multi-Channel Distribution table of contents, pp. 181-185, ISBN: 0-7695-2625-X (2006).
+
+18. Levenshtein V.: Binary codes capable of correcting deletions, insertions and reversals. Soviet Physics-Report, SEBD(10):707–710 (1966).
+
+19. Cascini G.: System and Method for performing functional analyses making use of a plurality of inputs. Patent Application 02425149.8, European Patent Office, 14.3.2002, International Publication Number WO 03/077154 A2 (18 September 2003).
+
+20. Cascini G., Russo D., Zini M.: Computer-Aided Patent Analysis: finding invention peculiarities. Proceedings of the 2nd IFIP Working Conference on Computer Aided Innovation, Brighton (MI), USA, 8-9 October, 2007, in Trends in Computer-Aided Innovation, Springer, pp. 167-178, ISBN 978-0-387-75455-0 (2007).
+
+21. Cascini G., Neri F.: Natural Language Processing for patents analysis and classification. Proceedings of the TRIZ Future 4th World Conference, 3-5 November 2004, Florence, Firenze University Press, ISBN 88-8453-221-3 (2004).
+
+22. Cascini G., Agili A., Zini M.: Building a patents small-world network as a tool for Computer-Aided Innovation. Proceedings of the 1st IFIP Working Conference on Computer Aided Innovation, Ulm, Germany, November 14-15 (2005).
+
+23. Cascini G., Russo D.: Computer-Aided analysis of patents and search for TRIZ contradictions. International Journal of Product Development, Special Issue: Creativity and Innovation Employing TRIZ, vol. 4(1-2) (2007).
+
+24. Yoon B., Park Y.: A text-mining-based patent network: Analytical tool for high-technology trend. The Journal of High Technology Management Research, vol. 15, Issue 1, pp. 37-50 (2004).
+
+25. Khomenko N., De Guio R., Lelait L., Kaikov I.: A Framework for OTSM-TRIZ Based Computer Support to be used in Complex Problem Management. International Journal of Computer Application in Technology (IJCAT), vol. 30 Issue 1-2 (2007).
+