Add files using upload-large-folder tool

processed/00006_W2138453742/document.md

@@ -0,0 +1,424 @@
+Journal of Economic Perspectives-Volume 4, Number 3-Summer 1990-Pages 75-96
+
+# Innovation and Cooperation: Implications for Competition and Antitrust
+
+Thomas M. Jorde and David J. Teece
+
+N obel Laureate Robert Solow and his colleagues on MIT's Industrial Productivity Commission recently noted (Dertouzos, Lester, and Solow, 1989, p. 7): "Undeveloped cooperative relationships between individuals and between organizations stand out in our industry studies as obstacles to technological innovation and the improvement of industrial performance” and later (p. 105) that “interfirm cooperation in the U.S. has often, though not always, been inhibited by government antitrust regulation.” These striking conclusions warrant further exploration.
+
+Unfortunately, industrial organization textbooks still discuss horizontal cooperation and competition almost exclusively in terms of standard cartel theory. (On the other hand, vertical cooperation/contracting is viewed differently, and some textbooks provide treatments of supplier-buyer relationships in which cooperation is viewed as enhancing efficiency.) Both in the textbooks and in policy discussion among economists, cooperation among competitors is highly suspect, being perhaps the last bastion of what was once referred to as the “inhospitality tradition” in antitrust. As a result, very little literature addresses how cooperation among competitors can promote competition, notwithstanding that cooperation among competitors may sometimes be essential if innovating firms are to compete in today's increasingly global markets (Imai and Baba, 1989) . Such cooperation is already important in Japan and in Europe. $^1$
+
+1For instance, cooperative R&D and related activities have been important to the success of the Western German machine tool industry. The industry formed a strong association that has a research and teaching institute at Aachen. The West German industry has been described as
+
+■ Thomas M. Jorde is Professor of Law, University of California, Berkeley, California. David J. Teece is Mitsubishi Bank Professor, Walter A. Haas School of Business, University of California, Berkeley, California.
+
+---
+
+76 Journal of Economic Perspectives
+
+This paper begins by describing the nature of the innovation process. We then explore socially beneficial forms of cooperation that can assist the development and commercialization of new technology, and suggest modifications to current U.S. antitrust law that would remove unnecessary impediments to organizational arrangements that support innovation and stimulate competition in the United States. The modifications we propose would create “safe harbors” for various forms of cooperative activities among competitors in unconcentrated markets, and they would permit cooperation in concentrated markets if commercialization and appropriability were thereby facilitated. These modifications would bring U.S. antitrust laws closer to what is already in place in Europe and Japan and would promote competition more assuredly than would existing law. $^2$
+
+We have no illusion that our proposed changes, standing alone, would dramatically improve the performance of U.S. industry, though specific industries might be transformed. However, the changes we propose in antitrust have the attraction that they do not require the expenditure of public funds. In short, we see existing law as a self-imposed impediment to U.S. economic performance. $^3$
+
+## The Nature of Innovation
+
+Innovation is the search for, and the discovery, development, improvement, adoption and commercialization of new processes, new products, and new organizational structures and procedures. 4 It involves uncertainty, risk taking, probing and reprobing, experimenting, and testing. It is an activity in which “dry holes” and “blind alleys” are the rule, not the exception. Many of these aspects are well-known and have been frequently analyzed in the economics literature.
+
+"groups of clubs" (Collis, 1988, p. 95) because of the nature of the cooperation displayed. The Italian machine tool industry around Modena is similarly organized, as is the Italian textile industry and the Danish furniture industry. A review of examples of cooperative activity abroad is part of the authors' ongoing research.
+
+2 There is no necessary conflict between promoting cooperation and competition, if the cooperation improves efficiency or advances innovation. As Schumpeter (1942, p. 85) pointed out, when compared to competition among firms with similar products and technologies, the competition that counts “comes from the new commodity, the new technology, the new source of supply.... This kind of competition is as much more effective than the other as bombardment is in comparison with forcing a door, and so much more important that it becomes a matter of comparative indifference whether competition in the ordinary sense functions more or less promptly.”
+
+3 As this Journal of Economic Perspectives issue goes to press, the House Judiciary Committee approved the “National Cooperative Production Amendments of 1990” (H.R. 4611), a bill that incorporates many of the changes we suggest in this article and which we have been advocating since 1988. We discuss the provisions of H.R. 4611 and additional antitrust changes that we believe would advance innovation and U.S. competitiveness later in this article.
+
+$^{4}$ Dosi (1988) provides an excellent review of the innovation literature.
+
+---
+
+Thomas M. Jorde and David J. Teece 77
+
+However, other aspects of innovation, particularly its organizational requirements, have not been sufficiently explored. The traditional serial model that has served as the basis for current antitrust policy is described below. Its inadequacies are then addressed in light of the “ simultaneous ” nature of the process, which is particularly relevant in certain industries, like microelectronics, experiencing high rates of technological change. 5
+
+## The Traditional Serial Model
+
+Traditional descriptions of the innovation process commonly break it down into a number of stages which proceed sequentially and theoretical treatments of R & D in industrial organization reflect this model. According to this view, the innovation process proceeds in a linear and predictable fashion from research to development, design, production, and then finally to marketing, sales, and service (Grossman and Shapiro, 1986, p. 319; Tirole, 1988, p. 389). In simple models, there is not even any feedback or overlap between and among stages.
+
+If the serial model adequately characterizes innovation today, then it is mainly the innovation which occurs in some scale-intensive industries. The initial development of nylon at Dupont perhaps fits this model. The Manhattan Project during World War II is also illustrative. The serial model does not address the many small but cumulatively important incremental innovations that are at the heart of technological change in many industries, especially well-established industries like semiconductors, computers, and automobiles. The serial model of innovation is an analytic convenience which no longer adequately characterizes the innovation process, except in special circumstances.
+
+The serial model has enabled economists to model innovation as a vertical process. Inasmuch as antitrust policy toward vertical restraints is very permissive, many economists and legal scholars do not understand how U.S. antitrust laws could stand in the way of the various kinds of standard and non-standard contracting often needed to support the commercialization of innovation. But as we shall see, matters are not so simple.
+
+## The Simultaneous Model
+
+The simultaneous model of innovation recognizes the existence of tight linkages and feedback mechanisms which must operate quickly and efficiently, including links between firms, within firms, and sometimes between firms and other organizations like universities. From this perspective, innovation does not necessarily begin with research; nor is the process serial. But it does require rapid feedback, mid-course corrections to designs, and redesign. $^6$ This concep-
+
+$^{5}$ This argument is presented at greater length in D. Teece $(1989 a)$.
+
+6This process has also been termed "cyclic" (Gomory, 1987, p. 72). The popular press has even begun to recognize and discuss the simultaneous nature of innovation and effective commercialization. See "A Smarter Way to Manufacture," Business Week, April 30, 1990, 110-117 (discussing "concurrent engineering").
+
+---
+
+78 Journal of Economic Perspectives
+
+tualization recognizes aspects of the serial model — such as the flow of activity, in certain cases through design to development, production and marketing — but also recognizes the constant feedback between and among activities, and the involvement of a wide variety of economic actors and organizations that need not have a simple upstream-downstream relationship to each other. 7 It suggests that R & D personnel must be closely connected to the manufacturing and to marketing personnel and to external sources of supply of new components and complementary technologies, so that supplier, manufacturer and customer reactions can be fed back into the design process rapidly. In this way new technology, whether internal or external, becomes embedded into designs which meet customer needs quickly and efficiently.
+
+The simultaneous model visualizes innovation as an incremental and cumulative activity that involves building on what went before, whether it is inside the organization or outside the organization, and whether the knowledge is proprietary or in the public domain. The simultaneous model also stresses the importance of the speed of the design cycle, and flexibility. IBM followed this model in developing its first PC, employing alliances with Microsoft and others to launch a successful personal computer system. Sun Microsystems and NeXT Computer launched themselves in this way and have remained in this mode for subsequent new product development. Microprocessor development at Intel often follows this logic too.
+
+When innovation has this character, the company which is quickest in product design and development will appear to be the pioneer, even if its own contribution to science and technology is minimal, because it can be first to “design in” science and technology already in the public domain. Both small and large organizations operate by this model, reaching out upstream and downstream, horizontally and laterally to develop and assemble leading edge systems.
+
+In short, much innovation today is likely to require lateral and horizontal linkages as well as vertical ones. As we discuss below, and particularly for small firms, innovation may require accessing complementary assets which lie outside the organization. If innovating firms do not have the necessary capabilities in-house, they may need to engage in various forms of restrictive contracts with providers of inputs and complementary assets. The possibility that antitrust laws could be invoked, particularly by excluded competitors, thus arises. Lying in the weeds to create mischief for unsuspecting firms engaged in socially desirable but poorly understood business practices are plaintiffs' attorneys and their expert economists entreating the courts to view reality through the lens of monopoly theory and modern variants such as raising rivals. These theories
+
+7Moreover, the linkage from science to innovation is not solely or even preponderantly at the beginning of typical innovations, but rather extends all through the process. “Science can be visualized as lying alongside development processes, to be used when needed” (Kline and Rosenberg, 1986). Design is often at the center of the innovation process. Research is often spawned by the problems associated with trying to get the design right. Indeed, important technological breakthroughs can often proceed even when the underlying science is not understood.
+
+---
+
+Innovation and Cooperation: Implications for Competition and Antitrust 79
+
+have been honed in the context of a hypothetical world of unchanging technology. If new technology does arrive it often falls like manna from heaven; behavior which is anticompetitive in the static context may be procompetitive in a dynamic one. Because the study of innovation is largely outside the mainstream of economic research and antitrust jurisprudence, the possibility of expensive and distracting litigation followed by judicial error is significant.
+
+Paradoxically, the giant integrated enterprises are not most heavily at risk. Instead, most at risk are mid-sized enterprises that have developed and commercialized important innovations, because such firms are likely to have some market power (under orthodox definitions) and have the need to engage in complex forms of interfirm cooperation. Because of these risks, managers may choose to forego socially desirable arrangements and investments, and innovation and the competition it engenders will be attenuated.
+
+## Organizational Requirements of Innovation
+
+Whether innovation is serial or simultaneous, it requires the coordination of various activities. The serial model suggests a rather simple organizational problem; the simultaneous model a more complex one, often employing various forms of non-standard contracting. To the extent that economists employ just the serial model, they greatly oversimplify the organizational challenges which innovation provides and underestimate potential antitrust problems. Also, they probably exaggerate the importance of research and downplay the importance of other factors. As discussed below, except in special cases, a firm's R & D capability is for naught if it cannot organize the rest of the innovation process efficiently and effectively, particularly if that innovation is taking place in an already-established industry.
+
+For innovations to be commercialized, the economic system must somehow assemble all the relevant complementary assets and create an interactive and dynamically efficient system of learning and information exchange. The necessary complementary assets can conceivably be assembled by administrative processes, or by market processes, as when the innovator simply licenses the technology to firms that already own the relevant assets, or are willing to create them. These organizational choices have received scant attention in the context of innovation. Indeed, the serial model relies on an implicit belief that armslength contracts between unaffiliated firms in the vertical chain from research to customer will suffice to commercialize technology. In particular, there has been little consideration of how complex contractual arrangements among firms can assist commercialization — that is, translating R & D capability into profitable new products and processes. The one partial exception is a tiny literature on joint R & D activity (Grossman and Shapiro, 1986; Ordover and Willig, 1985); but this literature addresses the organization of R & D and not the organization of innovation. 8
+
+$^{8}$ For a more complete statement of our own views on this, see Teece $(1977,1989 b)$.
+
+---
+
+80 Journal of Economic Perspectives
+
+If innovation takes place in a regime of tight appropriability—that is, if the technological leader can secure legal protection, perhaps by obtaining an ironclad patent (Teece, 1986)—and if technology can be transferred at zero cost as is commonly assumed in theoretical models, the organizational challenge that is created by innovation is relatively simple. In these instances, the market for intellectual property is likely to support transactions enabling the developer of the technology to simply sell its intellectual property for cash, or at least license it to downstream firms who can then engage in whatever valueadded activities are necessary to extract value from the technology. With a well-functioning market for know-how, markets can provide the structure for the requisite organization to be accomplished.
+
+But in reality, the market for know-how is riddled with imperfections (Arrow, 1962) . Simple unilateral contracts, where technology is sold for cash, are unlikely to be efficient (Teece, 1980, 1982) . Complex bilateral and multilateral contracts, internal organization, or various hybrid structures are often required to shore up obvious market failures (Williamson, 1985; Teece, 1986) . This section will examine various market failures and the institutional arrangements which can ameliorate them.
+
+## Technology Transfer Efficiency
+
+The transfer of technology among the various activities that constitute innovation is not costless. This is especially true if the know-how to be transferred cannot be easily bundled and shipped out in one lot — which is clearly the case when the development activity must proceed simultaneously and when the knowledge has a high tacit component. 9 In these instances, the required transfer of technology cannot be separated from the transfer of personnel, which is typically difficult if the contractual relationship is arms-length and non-exclusive.
+
+Besides the problems of getting technology-driven concepts to market, there is the converse problem of getting user-driven innovations to developers. In some industries, users other than the manufacturers conceive of and design innovative prototypes. The manufacturers' role in the innovation process is somehow to become aware of the user innovation and its value, and then to manufacture a commercial version of the device for sale to other users. User-dominated innovation accounts for more than two-thirds of first-to-market innovations in scientific instruments and in process machinery used in semiconductor and electronic subassembly manufacture (von Hippel, 1988). Clearly, user innovation requires two kinds of technology transfer: first from user to manufacturer, and then from the manufacturer to the developer-user and other users.
+
+Mirroring the role that users play in stimulating innovation upstream is the role that suppliers play in stimulating downstream innovation. For example, a good deal of the innovation in the automobile industry, including fuel injection,
+
+$^{9}$ For a review of the characteristics of know-how, see Winter (1987) and Teece $(1989 ~b)$.
+
+---
+
+Thomas M. Jorde and David J. Teece 81
+
+alternators and power steering, has its origins in upstream component suppliers. Bendix and Bosch developed fuel injection and Motorola the alternator. The challenge to the manufacturer then becomes how to “design in” the new components and how to avoid sole source dependency. As discussed below, deep and enduring relationships need to be established between component developer-manufacturers and suppliers to ensure adoption and diffusion of the technology. 10 These relationships, while functionally vertical, could well turn out to be viewed as horizontal by a court. Unless the courts have an adequate model of innovation and competition presented to them, beneficial contractual arrangements with attendant restraints could well be viewed negatively.
+
+## Scale, Scope, and Duplication Issues
+
+Successful new product and process development innovation often requires horizontal and lateral as well as vertical cooperation. It is well understood that horizontal linkages can help overcome scale barriers in research; they can also assist in defining technical standards. But it is common to assert that if firms need to engage in joint research to achieve these economies, the maintenance of competition requires that firms participating in joint research work go their own way with respect to related activities such as manufacturing. However, a requirement that firms participating in a joint research arrangement commercialize the technology independently can impose an unnecessary technology transfer burden. As discussed above, the imposition of a market interface between “research” and “commercialization” activities will most assuredly create a technology transfer challenge, a loss of effectiveness and timeliness, and higher costs.
+
+Collaborative research also reduces what William Norris, CEO of Control Data Corporation, refers to as “ shameful and needless duplication of effort ” (David, 1985) . Independent research activities often proceed down identical or near-identical technological paths. This is sometimes wasteful and can be minimized if research plans are coordinated. The danger of horizontal cooperation, on the other hand, is that it may reduce diversity. This concern is legitimate and is commonly stressed by economists. $^11$ Unquestionably, a system
+
+10 A related set of vertical relationships involving innovation has been remarked upon by Rosenberg (1972, pp. 98–102) in his treatise on technology and American economic growth. The machine tool industry in the 19th century played a unique role both in the initial solution of technical problems in user industries, such as textiles, and as the disseminator of these techniques to other industries, such as railroad locomotive manufacture. Rosenberg's description suggests that the users played a role in the development of new equipment. He notes that before 1820 in the United States, one could not identify a distinct set of firms that were specialists in the design and manufacture of machinery. Machines were either produced by users or by firms engaged in the production of metal or wooden products. Machinery-producing firms were thus first observed as adjuncts to textile factories. However, once established, these firms played an important role as the transmission center in the diffusion of new technology.
+
+11 Nalebuff and Stiglitz (1983) argue that the gains from competition may more than offset the losses from duplication. Also, Sah and Stiglitz (1989) show that in a model with ex post Bertrand competition where there is knowledge of which research projects others are undertaking, the number and range of research projects undertaken will be a constrained Pareto optimum.
+
+---
+
+82 Journal of Economic Perspectives
+
+of innovation that converges on just one view of the technological possibilities is likely to close off productive avenues of inquiry.
+
+However, a private enterprise economy without horizontal coordination and communication offers no guarantee that the desired level of diversity is achieved at the lowest cost. In addition, cooperation need not be the enemy of diversity. If firms can coordinate their research programs to some degree, duplication can be minimized without the industry converging on a single technological approach. Indeed, Bell Labs has been noted for the very considerable internal diversity it has been able to achieve, at least in the pre-divestiture period.
+
+## Rent Dissipation Issues
+
+Innovation has well-known free rider and public good characteristics. Know-how leakage and other spillovers impair incentives to innovate by redistributing benefits to others, particularly competitors and users. To maintain adequate incentives to invest in innovative activity, without providing government subsidies, free riding must be curtailed. This is how economists justify patents, copyrights, trade secrets, and other aspects of intellectual property law.
+
+The organizational form in which innovation takes place, interacting with the protection provided by intellectual property law (Teece, 1986), will affect the degree of rent dissipation which the innovator experiences. If the innovation has value and intellectual property protection is effective, an innovator specializing just in early stage activity is in a good position to capture a portion of the returns from innovation.
+
+But surveys show that intellectual property law has a limited ability to provide protection from imitation, 12 even though there have been recent efforts by the courts to tighten enforcement. For a sample of 48 patented product innovations in the chemical, drug, electronics and machinery industry, one group of researchers found that within four years of their introduction, 60 percent of the patented successful innovations in the sample were imitated (Mansfield et al, 1982) . Not surprisingly, the social returns to innovation are greater than the private returns. Underinvestment in innovative activities is to be expected.
+
+A “research joint venture” may not do enough to overcome appropriability problems, unless many potential competitors are in the joint venture. Thus, a
+
+12See Levin, Klevorick, Nelson, and Winter (1987). These researchers surveyed R&D managers in various industries. The survey shows that, on a seven-point scale (1 = not at all effective, 7 = very effective) for 18 industry categories with 10 or more respondents, managers in only chemicals (specifically drugs, plastic materials, inorganic chemicals, and organic chemicals) and petroleum refining rated process patents effectiveness higher than 4 on the scale, and only these same chemical industries and steel mills rated product patents higher than 5. These findings make very clear that managers have little confidence that patents suffice as mechanisms to protect intellectual property from free riders. The results also show that other methods of appropriation such as first mover advantages (lead time and learning curve advantages), secrecy, and investment in sales or service support were more effective.
+
+---
+
+Innovation and Cooperation: Implications for Competition and Antitrust 83
+
+single firm or even a consortium with good intellectual property protection will often need to bolster its market position and its stream of rents by other strategies and mechanisms. These mechanisms include building, acquiring, or renting (on an exclusive basis) complementary assets and exploiting first-mover advantages. We use the term complementary assets to refer to those assets and capabilities that need to be employed to package new technology so that it is valuable to the end user. 13 Broad categories of complementary assets include complementary technologies, manufacturing, marketing, distribution, sales, and service.
+
+It is essential to distinguish further between generic and specific complementary assets. Generic assets include general purpose facilities and equipment and nonspecific skills; they tend to be disembodied and codified and hence easy to transfer. Specific assets, on the other hand, include highly differentiated system and firm-specific assets and skills. Specific assets and capabilities are typically embedded in the organization; or even if not embedded in the organization (like a specialized machine) are of reduced value in a different organizational context. In a sense, specific assets represent the firm's particular assemblage of physical assets and prior learning. Accordingly, they are difficult for competitors to replicate.
+
+Thus, when imitation of aspects of a firm's technology is easy, it is essential for firms to be world-class—or to be linked to partners who are world-class—in the less imitatable complementary activities. Accordingly, the best defense against product imitators may well be the development of a less easily imitatable superior manufacturing process to make the product, or it may be the firm's superior service capability. In short, because a firm's comparative advantage in research does not necessarily coincide with an advantage in the relevant complementary assets, the expert performance of the innovator's contractual partners in certain key activities complementary to the easily imitatable activities is often essential if the innovator is to capture a portion of the profits that the innovation generates. The antitrust laws must be shaped so that they do not impair such beneficial linkages.
+
+In this regard, many British and American firms responsible for important product innovations have captured very little value from innovations for which they have been responsible because of their weaknesses in manufacturing. Often competitors can quickly reverse engineer new products. Once the new product design is apparent to competitors, success in the marketplace is determined by manufacturing costs and quality. In these circumstances, firms that are excellent at manufacturing—and this excellence is often harder to replicate than a new product is to reverse engineer—can garner practically all of the profits associated with the new product designs. Hence it is critical that
+
+13 There has been almost no treatment in the economic literature of the concept of complementary assets. It does not map easily into the familiar concept of indivisibilities, which is perhaps the closest analogue. For a more complete treatment, see Teece (1986).
+
+---
+
+84 Journal of Economic Perspectives
+
+innovating firms protect themselves from such outcomes by developing or somehow uniquely accessing the requisite complementary assets. The next section explains why cooperation may be necessary for firms to perform this function.
+
+## Governance Alternatives
+
+The previous section has argued that innovation often requires firms to enter complex contracts and relationships with other firms to bring technology to the market, and to hold imitators at bay. This section considers in more detail the range of organizational alternatives available to the innovator to generate, coordinate and control such complementary assets.
+
+Consider first the price mechanism. Theoretical treatments generally assume that the requisite coordination and control can be achieved by the invisible hand. Efficient levels of investment in complementary assets are brought forward at the right time and place by price signals. Entrepreneurship is automatic and costless. This is the view implicit in textbook presentations; in turn, the textbook view seems implicit in U.S. antitrust law.
+
+However, many economists seem to have what Tjalling Koopmans calls an “overextended belief” regarding the efficiency of competitive markets as a means of allocating resources in a world characterized by ubiquitous uncertainty. Market failures are likely to arise because of the ignorance which firms have with respect to their competitors' future actions, preferences, and states of technological information (Koopmans, 1957, part II). In reality, nothing guarantees that investment programs are made known to all concerned at the time of their inception. This uncertainty is especially high for the development and commercialization of new technology. Accordingly, innovating firms need to achieve greater coordination than the price system alone appears to be able to bring about.
+
+A second mechanism for effectuating coordination is the administrative processes within the firm. A company's internal organization can serve to shore up some market imperfections and provide some of the necessary coordination. As Alfred Chandler (1977) has explained, the modern multidivisional business enterprise “took over from the market the coordination and integration of the flow of goods and services from the production of raw materials through the several processes of production to the sale to the ultimate consumer ... administrative coordination replaced market coordination in an increasingly large portion of the economy.” Oliver Williamson (1985) has developed an elegant and powerful framework to explain the relative efficiencies of markets and administrative processes. However, one property of large integrated structures is that they have the potential to become excessively hierarchical and less responsive to market needs (Teece, 1989c). Accordingly, at least for some aspects of innovative activity, smaller organizations are often superior.
+
+---
+
+Thomas M. Jorde and David J. Teece 85
+
+In between pure market and full administrative solutions are many intermediate and hybrid possibilities, including interfirm agreements. Interfirm agreements can be classified as unilateral (where $A$ sells $X$ to $B$ ) or bilateral (whereby $A$ agrees to buy $Y$ from $B$ as a condition for making the sale of $X$ , and both parties understand that the transaction will be continued only if reciprocity is observed). Such arrangements can also be multilateral.
+
+An especially interesting interfirm agreement is the strategic alliance, which can be defined as a bilateral or multilateral relationship characterized by the commitment of two or more partner firms to a common goal. A strategic alliance might include (1) technology swaps, (2) joint R & D or co-development, and/or (3) the sharing of complementary assets, such as where one party does manufacturing and the other distribution for a co-developed product. If the common goal was simply price-fixing or market-sharing, such an agreement might constitute a cartel, especially if the agreement included substantially all members of an industry.
+
+By definition, a strategic alliance can never have one side receiving cash alone; it is not a unilateral exchange transaction. Nor do strategic alliances include mergers, because alliances by definition cannot involve acquisition of another firm's assets or controlling interest in another firm's stock. Alliances need not involve equity swaps or equity investments, though they often do. Strategic alliances without equity typically consist of contracts between or among partner firms that are nonaffiliated. Equity alliances can take many forms, including minority equity holdings, consortia, and joint ventures. Such interfirm agreements are usually temporary, and are assembled and disassembled as circumstances warrant. Typically, only a limited range of the firm's activities are enveloped in such agreements, and many competitors are excluded.
+
+Strategic alliances, including consortia and joint ventures, are often an effective and efficient way to organize for innovation, particularly when an industry is fragmented. Interfirm cooperation preserves market selection and responsiveness; in a sense, it is the pure private enterprise solution. The case for planning and industrial policy recedes if a degree of operational and strategic coordination can be attained through private agreements. The benefits associated with less hierarchical structures can be obtained without incurring the disadvantages of insufficient scale and scope.
+
+## Antitrust Treatment of Interfirm Agreements
+
+Current U.S. antitrust law needlessly inhibits interfirm agreements designed to develop and commercialize new technology. The problem is that the legal standards for interfirm agreements are ambiguous. While “rule of reason” analysis will generally be applied to contractual arrangements designed to advance innovation, the elements of rule of reason analysis are quite muddled.
+
+---
+
+86 Journal of Economic Perspectives
+
+In addition, although current law seems to recognize a “safe harbor” for mergers and acquisitions between firms that will have less than 20 percent market share, it does not recognize a similar safe harbor for horizontal contractual arrangements among firms.
+
+The Clayton Act also permits private parties to sue for treble damages for alleged antitrust injuries, and allows state attorney generals to recover treble damages on behalf of persons residing in the state. Successful plaintiffs can also recover attorneys' fees. These remedies are available only in the United States. They provide a powerful incentive for plaintiffs to litigate, and given the current state of the law, a powerful disincentive for businesses to form cooperative innovation arrangements and strategic alliances. While measuring the missed opportunities for cooperative innovation caused by the threat of treble damage litigation is difficult, we believe the loss is substantial. Moreover, these disincentives work to the particular detriment of small and medium-sized innovative firms in industries where the innovative process is simultaneous.
+
+Congress has recognized that these provisions may inhibit technological innovation, and the National Cooperative Research Act (NCRA) of 1984 took two significant steps to remove legal disincentives to cooperative research. First, the NCRA provides that “joint research and development ventures” must not be held illegal per se, and that such ventures instead should be “judged on the basis of [their] reasonableness, taking into account all relevant factors affecting competition, including, but not limited to, effects on competition in properly defined, relevant research and development markets.” Second, the NCRA establishes a registration procedure for joint research and development ventures, limiting antitrust recoveries against registered ventures to single damages, interest, and costs, including attorney's fees. Thus, Congress eliminated the threat of treble damages for litigation challenging cooperative R&D arrangements, provided that the parties to the arrangement first register their venture. But R&D is only a small piece of the innovation puzzle.
+
+In our view, the NCRA is not sufficiently permissive. The substantive protections provided by the NCRA—guaranteed rule of reason treatment and reduction of damages—extend only to research, and downstream commercial activity “reasonably required” for research and narrowly confined to marketing intellectual property developed through a joint R & D program. Treatment of other agreements designed to facilitate innovation is thus left uncertain, to be determined only by interpretation of the “reasonably required” standard. The NCRA unwisely precludes joint manufacturing and production of innovative products and processes, which is often necessary to provide the cooperating ventures with significant feedback information to aid in further innovation and product development, and to make the joint activity profitable. The NCRA implicitly accepts the serial and not the simultaneous model of innovation.
+
+In addition, the NCRA gives little guidance concerning the substantive content of its rule of reason approach. While the Act did require that markets be defined in the context of research and not the products that might result from it, the NCRA fails to specify factors to be considered within rule of reason
+
+---
+
+Innovation and Cooperation: Implications for Competition and Antitrust 87
+
+analysis. It simply requires consideration of “ all relevant factors affecting competition, ” paying no special attention to the special characteristics of the innovation process in a quickly changing industry.
+
+Finally, while the NCRA's elimination of treble damages for registered ventures is an important step forward, cooperating firms are still not protected from antitrust litigation. Even after the NCRA, antitrust law still permits private plaintiffs to engage in treble damage litigation against cooperative arrangements facilitating commercialization. Moreover, single damages are still available even against those registered under NCRA. The cost of defending antitrust suits is not materially reduced by the exceedingly narrow circumstances in which the Act permits an award of attorneys' fees to prevailing defendants. Moreover, the threat of litigation, with attendant managerial distraction, can be extremely damaging to the competitive performance of a fast-paced industry.
+
+Businesses seem to have recognized the limited nature of the steps taken by the NCRA. Not surprisingly, only 111 separate cooperative ventures registered under the NCRA between 1984 and June 1988. Our review of these filings indicates that they are very modest endeavors that are aimed at solving industry problems and are not of great competitive moment. We believe that if an approval procedure existed under which procompetitive arrangements could obtain exemptions from further antitrust exposure to private damage actions, then many more competitively beneficial ventures would utilize the NCRA.
+
+In contrast to this picture of U.S. antitrust law, the antitrust and business environment in Japan and Europe is more hospitable to strategic alliances and cooperative arrangements for innovation. The basic Japanese attitude is that joint R & D activities are procompetitive and thus should not be touched by the Antimonopoly Act. Significantly, the literal Japanese translation of “R & D” — kenkyu kaihatsu — implicitly includes commercialization; there is no semantic distinction between the concepts of R & D and commercialization.
+
+In Japan, the Fair Trade Commission is responsible for executing and enforcing the Antimonopoly Act of 1947, which (like the Sherman Act) broadly prohibits unreasonable restraints of trade. While the Act provides no specific legislative exemption for joint innovation arrangements, Japan's FTC has been able to exempt cooperative innovation efforts from the scope of the law by virtue of its power as the primary enforcer of the Act. $^14$ FTC policy also states
+
+14 The basic administrative policy outlining the standards by which such joint innovation efforts are to be scrutinized is contained in a report of Japan's Fair Trade Commission (1984, 37-39). The report states that the evaluation of the anticompetitive effect of joint R&D at the product market stage will depend significantly “on the competition and market shares among the participants and the market structure of the industry to which the participants belong…. . . In cases where the market shares of the participants are small … the effects will be small.” Although “small” is not defined in the report, Japan's Merger Guidelines state that the FTC is not likely to closely examine cases in which the combined market share of the merging parties is less than 25 percent. See H. Iyori and A. Yesugi (1983, pp. 86–88). Our discussions with MITI and FTC officials confirm that the horizontal merger safe harbors would be equally applicable to cooperative contractual arrangements.
+
+---
+
+88 Journal of Economic Perspectives
+
+that if anticompetitive effects are alleged, the procompetitiveness benefits of innovation must be balanced, too. Balancing will take place not only within a particular market but also across markets (FTC, 1984), because “there is a possibility of the emergence of competition at the intersection of industrial sectors as a result of joint R&D between firms in different sectors.”
+
+In considering anticompetitive effects of cooperative innovation arrangements, Japan's FTC analyzes market shares and market structure. The FTC specifically recognizes the needs of innovators and articulates procompetitive justifications that include: (1) the difficulty of single-firm innovation; (2) the faster innovation created by cooperation and specialization between joint participants; (3) the pursuit of innovation in new fields by utilizing shared technology and know-how; and (4) enhancement of the technological level of each participant through the interchange of technology.
+
+When MITI seeks to promote cooperative R&D activities (for example, as authorized by the Act for Facilitation of Research in Key Technology, or the Research Association for Mining and Manufacturing Technology Act), the FTC is consulted in advance. Once the FTC clears an activity, it is extraordinarily unlikely to pursue antitrust remedies at a future time. Significantly, treble damages are not available to private parties seeking to enforce Japanese antitrust laws, and private suits for single damages are very rare and usually unsuccessful. Thus, Japanese firms cooperating on innovation and commercialization of innovation have little to fear from Japanese antitrust laws.
+
+Under this type of antitrust environment, it is not surprising that collaboration for innovation is frequent. Although regular statistics are not kept in Japan, because there is no reporting requirement for collaborative research and commercialization activities, a Fair Trade Commission report issued in 1984 contains statistics suggestive of the quantity and variety of joint innovation activities in Japan. The survey results indicate that joint R & D projects among corporations in the same industrial sector, which might be classified as horizontal collaboration, represent 19.1 percent of total projects. 15
+
+The antitrust environment shaping cooperation in the European Community is also markedly different from the United States. In 1968, the European Commission issued a “ Notice of Cooperation between Enterprises ” which indicates that horizontal collaboration for purposes of R & D is normally outside the scope of antitrust concerns as defined in Articles 85 and 86 of the EEC Treaty. The Commission has consistently taken a favorable position on R & D
+
+15Questionnaires were sent to 484 manufacturing corporations in the fields of electronics, telecommunications, automobiles, chemicals, ceramics, steel and nonferrous metals, whose stocks were listed in Tokyo and Osaka Stock Exchanges. Data was provided by 242 corporations, representing 1.9 percent of the total manufacturing industry that engage in R&D activities in terms of the number of corporations and 16.7 percent in terms of sales. As to the nature of the joint R&D projects, 54.3 percent of the total cases were developmental research. Basic and applied research were 13.6 and 32.1 percent respectively. In the case of large corporations with capital of more than 10 billion yen, the total basic and application research amounted to 52.1 percent.
+
+---
+
+Thomas M. Jorde and David J. Teece 89
+
+agreements unless the large entities involved imply serious anticompetitive consequences.
+
+In 1984, the European Commission adopted Regulation No. 418/85 (hereafter Reg. 418) expanding the favorable antitrust treatment of R & D. For firms whose total market share does not exceed 20 percent, it provides blanket exceptions for horizontal R & D arrangements, including commercialization — which the Commission views as “ the natural consequence of joint R & D ”— up to the point of distribution and sales. 16 In addition, under Article 85(3), the Commission is authorized to grant exemptions for cooperative efforts that do not fall within the automatic safe harbor. Such exemptions may be granted when a horizontal agreement contributes to economic or technological progress in the research, production, or distribution of goods, and when procompetitive features outweigh anticompetitive aspects.
+
+## Proposed Modifications to U.S. Antitrust Law
+
+To insure that antitrust law is responsive to the needs of innovating firms and does not inhibit U.S. firms from competing effectively in global markets experiencing rapid technological change, we believe the following changes are in order:
+
+First, the rule of reason should be clarified to take specific account of the appropriability regime, the pace of technological change, the diversity of sources of new technology, the need to access complementary assets and technologies, and the need to have cheek-by-jowl cooperation to manage the innovation process simultaneously rather than serially.
+
+Second, a safe harbor defined according to market power should be expressly adopted that would shield from antitrust liability interfirm agreements that involve less than 20 to 25 percent of the relevant market.
+
+Third, market definition should be tailored to the context of innovation and should focus primarily on the market for know-how; specific product markets become relevant only when commercialization is included within the scope of the cooperative agreement. Even then, the extent of appropriability should be factored in when analyzing product market issues. The geographic market should be presumed to be worldwide, with the burden upon the challenger to demonstrate otherwise.
+
+16 Regulation No. 418/85 of 19 December 1984 on the application of Art. 85(3) of the Treaty to categories of research and development agreements, O. J. Eur. Comm. (No. L 53) 5 (1985), entered into force March 1, 1985, and applicable until December 31, 1997. The statutory framework of Reg. 418 is complex. It applies to three categories of agreements involving R&D joint research and development of products or processes and joint exploitation of the results of the R&D (2) joint exploitation of the results of R&D product or processes pursuant to a prior agreements between the same parties; and (3) joint research and development of products without joint exploitation should the agreement fall within the purview of Art. 85(1). Under Reg. 418, joint exploitation is interpreted to mean joint manufacturing and licensing to third parties. Joint distribution and sales, however, are not covered and required individual exemptions pursuant to Art. 85(3).
+
+---
+
+90 Journal of Economic Perspectives
+
+Fourth, antitrust law should not bias the selection of interfirm organizational forms; at a minimum, integration by contract or alliance should be treated no less favorably than full mergers.
+
+Fifth, the NCRA should be amended to include joint commercialization efforts to exploit innovation.
+
+Sixth, an administrative procedure should be created, involving both the Justice and Commerce Departments, to allow evaluation and possible certification of cooperative arrangements among firms with higher market shares, when dynamic efficiency gains are likely and rivalry robust. We favor providing the opportunity for firms to either simply register and receive relief from treble damages as with the NCRA, or to apply for a certificate of exemption from the Justice and Commerce Departments that would provide even more protection. However, the quid pro quo would be greater disclosure and scrutiny of business plans. The firms themselves would choose which path to take.
+
+Seventh, private antitrust suits challenging cooperative innovation arrange- ments should be limited to equitable relief, and attorneys' fees should be awarded to the prevailing party.
+
+The first four of these proposals could be accomplished by courts interpreting the rule of reason and the National Cooperative Research Act. We hope courts will not hesitate to employ the tools of evolutionary, common law interpretation and development to achieve these changes. However, to achieve the complete package of substantive and procedural changes most quickly, and thus assure certainty and predictability, legislation is the best overall solution. At a U.C. Berkeley Conference on “ Antitrust, Innovation and Competitiveness ” in October 1988, we distributed a draft of legislation that combined a “ registration ” and “ certification ” approach for cooperative commercialization ventures. Shortly thereafter, Congressmen Edwards (H.R. 1025) and Congressman Fish (H.R. 2264) advanced a “ registration ” approach to cooperative commercialization efforts and Congressmen Boucher and Campbell (H.R. 1024) proposed a “ certification ” approach. After three hearings on these bills, Chairman Jack Brooks of the House Judiciary Committee introduced and the Judiciary Committee passed the National Cooperative Production Amendments of 1990 (H.R. 4611). H.R. 4611 would amend the National Cooperative Research Act to extend its registration approach to joint production ventures. $^17$ At the same time, Attorney General Richard Thornburgh and Commerce Secretary Robert
+
+17Professor Jorde testified on July 26, 1989, in favor of both a registration and certification approach. See “Legislative Proposals to Modify the U.S. Antitrust Laws to Facilitate Cooperative Arrangements to Commercialize Innovation” (with David Teece), in Hearings Before the Subcommittee on Economics and Commercial Law, Committee on the Judiciary, U.S. House of Representatives (July 26, 1989). Legislation advancing a registration approach for production joint ventures has also been introduced in the Senate by Senators Patrick Leahy (D-VT) and Strom Thurmond (R-SC) (S.1006). Three aspects of H.R. 4611 bear noting. First, relevant market definition under rule of reason analysis would specifically consider the worldwide capacity of suppliers. Second, foreign participation in a production joint venture would be limited to 30 percent of the voting securities or equity interests, and all production facilities would have to be located in the United States or its
+
+---
+
+Innovation and Cooperation: Implications for Competition and Antitrust 91
+
+Mosbacher announced the Bush Administration's support of a registration approach for production joint ventures.$^18$
+
+As mentioned above, we support both a registration and certification approach. We do not see them as alternatives. Rather, we believe they should be combined into a single, two-track approach. Firms could choose the level and then form of protection most appropriate for their joint activity. Greater disclosure could buy greater protection.
+
+The case for these changes rests on three fundamental pillars. The first is that the innovation process is terribly important to economic growth and development, because it yields social returns in excess of private returns, and because innovation is a powerful spur to competition. Hence, if antitrust policy is going to err, it ought to do so by facilitating innovation, rather than inhibiting it. This principle is well-understood in Europe and Japan.
+
+Second, economic theory tells us that if certain organizational arrangements are exposed to governmentally-imposed costs while others are not, firms will substitute away from the burdened forms (in this context, interfirm agreements) and in favor of the unburdened forms (in this context, hierarchy), even when the former are potentially economically superior. According to Aoki (1989), the slowdown in total factor productivity in the United States can be attributed in large part to a mismatch between organizational form and the requirements of new technology; in particular, he is concerned that hierarchical solutions are overused, at least in the United States. As we have explained at some length above, we are concerned that present laws do not give full recognition to the interorganizational requirements of the innovation process; failure to do so is damaging when innovation must proceed according to the simultaneous model.
+
+Third, cartelization of industries experiencing rapid technological change, and which are open to international trade and investment, is very difficult. So long as these industries remain open and innovative, antitrust policy should err on the side of permitting rather than restricting interfirm contracts.
+
+Beneficial cooperation will eventually expand if antitrust laws are revised along the lines we propose. The response may not be immediate, particularly with respect to consortia, because the experience base in U.S. industry in this area is thin, because of our antitrust history, and because U.S. firms, at least in the postwar period, have been large relative to their foreign competitors. Accordingly, the need to cooperate has not been as powerful in the past as it is now. However, once organizational learning accumulates, we expect consortia to begin to flourish even in the absence of government funding. We also expect
+
+territories. Third, apparently production joint ventures would not be limited to efforts designed to commercialize joint R&D nor need they be related to innovation.
+
+18See Department of Justice release, "Thornburgh Mosbacher Send Revision Legislation to Congress" (May 7, 1990) (supporting and detailing "legislation designed to facilitate joint production ventures"), reported at Antitrust and Trade Regulation Report, p. 701 (Vol. 58, No. 1465) (May 10, 1990).
+
+---
+
+92 Journal of Economic Perspectives
+
+the reinforcement of bilateral alliances already common in U.S. industry. We briefly discuss the kinds of activities that might take place.
+
+## Cooperative Manufacturing and Commercialization
+
+In a number of circumstances, cooperative activity beyond early stages will benefit innovating firms. As discussed, sometimes this is true because of scale, risk, and appropriability considerations. Sometimes it is true because prohibition of cooperative commercialization imposes a significant technology transfer problem, for instance from the research joint venture (if there is one) back to the funding companies. In most cases, firms will not wish to cooperate all the way from research through to commercialization. But in some instances they will, or they will wish to cooperate simply on a downstream production venture. When cartelization of the industry is not a threat, we see no reason for antitrust restraints.
+
+The now defunct U.S. Memories, Inc. consortium wanted to invest $500 million to $1 billion to develop and manufacture for its members and for the market advanced dynamic random access memories (DRAMs). With fabrication facilities costing hundreds of millions, acting alone is beyond the financial resources of many companies in this industry who might otherwise wish to have some control over their DRAM supply. This proposed consortium had to contend with a number of difficulties, including threats of third party litigation (Jorde and Teece, 1989b) . While antitrust was not the main reason for the failure of this enterprise, the antitrust environment did nothing to help it succeed. A certification procedure would have provided important certainty to this venture, and others like it. A registration procedure would provide less certainty, but still would be a significant advance over current antitrust law.
+
+Similarly, in the area of superconductors, it is likely that the real challenges will come not in developing superconductors, but in their commercialization. Applying superconductors in systems like railroads, computers, and electricity distribution will require great amounts of time, resources, and capital — probably greater than any single business can muster internally. Accordingly, a public policy stance that treats only early stage activity as potentially requiring cooperation is misguided and will thwart both early and later stage activities. Most firms will not have much incentive to engage in early stage, joint development if later stage, stand-alone commercialization appears too expensive to accomplish profitably.
+
+## Cooperative Innovation Designed to Achieve Catch-Up
+
+Cooperative activities in Japan and Europe have frequently been motivated by a desire to catch up with the world's technological frontier, which in the postwar years was usually the technology of U.S. firms. However, U.S. firms are increasingly slipping behind the frontier. For instance, U.S. firms are now behind in areas like ceramics and robotics, and in products like VCRs, facsimiles, and HDTV. Just as foreign firms have found cooperative ventures useful
+
+---
+
+Thomas M. Jorde and David J. Teece 93
+
+for catch-up in the past, U.S. firms could utilize cooperation for this purpose. For example, U.S.-based firms, acting together and with foreign firms, may still have a slender chance of competing in the market for high definition televisions (HDTV) and related products expected to evolve in the 1990s. In the absence of cooperative interfirm agreements, we doubt that development of HDTV systems is possible in the United States. If America's potential “reentrants” to the consumer electronics business combine to attempt reentry, they cannot be sure of avoiding serious antitrust problems involving treble damages, particularly if they are successful.
+
+At minimum, the legislative changes proposed would facilitate unfettered information exchange and strategic coordination with respect to reentry strategies. If such efforts facilitated profitable reentry into high technology businesses when reentering would otherwise not occur, or would occur in a more limited and unprofitable way, we do not see why antitrust concerns ought to interfere.
+
+### Cooperation in Response to Foreign Industrial and Technology Policy
+
+In high technology industries, both European and East Asian nations have active industrial and technology policies that significantly impact market outcomes, both in their own countries and abroad. Airbus is a case in point. The dominant U.S. attitude is one of laissez-faire, and many economists are of the view that the United States should send a letter of thanks to foreign governments who subsidize exports to the United States. Such a view is insensitive to the dynamics of technological change, to the importance of cumulative learning, and to reentry costs.
+
+Some U.S. policy makers, however, favor retaliation against foreign countries which have active industrial policies. We support a modification of U.S. antitrust laws which in some circumstances would permit a competitive response by U.S. industry acting collectively. The proposals we advance to encourage greater cooperation among U.S. firms do not require government expenditures nor do they involve the government “picking winners.” But they would soften the tensions emerging in the United States between technology, antitrust and trade policies.
+
+## Conclusion
+
+The past two decades have wrought significant changes in the business environment. Markets have become globalized, sources of new technology are increasingly pluralistic, and “ simultaneous ” systems of innovation have substituted for linear, hierarchical ones. Moreover, the ability of foreign firms to utilize technology developed in the United States has increased markedly. Imitation is easier, not harder, in spite of recent court decisions which have strengthened patents.
+
+---
+
+94 Journal of Economic Perspectives
+
+Accordingly, innovative firms confront significant challenges in capturing value from new technology. Success in research and development does not automatically translate into a financial success, even if the technology developed meets a significant market need. To succeed financially, innovative firms must quickly position themselves advantageously in the appropriate complementary assets and technologies. If they are not already integrated, the best solution often involves bilateral and multilateral cooperative agreements. $^19$
+
+U.S. antitrust policy, like so much of our economic policy, has been preoccupied with static rather than intertemporal concerns. Despite important recent developments, it is informed by naive theories of the innovation process, and in particular is insensitive to the organizational needs of innovation. U.S. antitrust scholars still harbor suspicion of cooperative agreements among competitors, and do not appreciate the benefits. This suspicion fuels uncertainty about how the courts would view interfirm arrangements to promote technological progress and competition.
+
+The policy changes we advance are certainly no panacea for the severe problems U.S. high technology industry is currently experiencing. But in bringing American policy closer to Europe and Japan, we will at least purge dogma that no longer deserves a place in U.S. industrial policy. In time, reduced antitrust exposure will help clear the way for beneficial cooperation, thereby reducing incentives for mergers and acquisitions.
+
+The 1990 centennial of the Sherman Act would be a good occasion to set things right. The economics profession, which in the past has had a significant impact on the law of vertical restraints, can provide the intellectual leadership necessary to propel adjustments in the horizontal area, thereby helping to align U.S. policies with the technological and competitive realities of today's global economy.
+
+■ This paper is based in part on Jorde and Teece (1989a) and Teece (1986). We are extremely grateful for financial support from the Alfred P. Sloan Foundation, the Smith-Richardson Foundation, The Pew Foundation, and the Sasakawa Peace Foundation. We wish to thank Joseph Stiglitz, Carl Shapiro, and Timothy Taylor for valuable substantive and editorial comments. Bill Baxter, Oliver Williamson, and Dick Nelson made helpful comments on earlier drafts and oral presentations. We implicate none of the above in our conclusions.
+
+19 As Richard Nelson (1990) notes, a wide variety of new kinds of organizational arrangements is emerging to support innovation. He predicts, and we concur, that some will succeed, and some will not. Our concern is that because the requirements of innovation are not well understood in mainstream economics and in contemporary antitrust analysis, there is significant danger that the performance of U.S. firms will be impaired by outdated antitrust law.
+
+---
+
+Innovation and Cooperation: Implications for Competition and Antitrust 95
+
+## References
+
+Aoki, M., "Global Competition, Firm Organization, and Total Factor Productivity: A Comparative Micro Perspective." Paper presented at the International Seminar on Science, Technology, and Economic Growth, OECD, Paris, June 1989.
+
+Arrow, Kenneth J., "Economic Welfare and the Allocation of Resources for Invention." In National Bureau of Economic Research, ed., The Rate and Direction of Inventive Activity. Princeton: Princeton University Press, 1962, pp. 609-625.
+
+Chandler, Alfred D. Jr., The Visible Hand: The Managerial Revolution in American Business. Cambridge: Harvard University Press, 1977.
+
+Collis, David, "The Machine Tool Industry and Industrial Policy, 1955-1982." In Spence, A. Michael, and Heather A. Hazard, eds., International Competitiveness. Cambridge: Ballinger, 1988, pp.
+
+David, D., "R&D Consortia," High Technology, October 1985, p. 42.
+
+Dertouzos, Michael L., Richard K. Lester, and Robert M. Solow, Made in America: Regaining the Productive Edge. Cambridge: MIT Press, 1989.
+
+Dosi, Giovanni, "Sources, Procedures, and Microeconomic Effects of Innovation," Journal of Economic Literature, September 1988, 26, 1120-1171.
+
+Fair Trade Commission (Japan), Research and Development Activities in Private Enterprises and Problems They Pose in the Competition Policy (Minkan kigyo ni okeru kenkyu kaihatsu katsudo no jitai to kyoso seidaku jo no kaidai), 1984.
+
+Gomory, R., "Dominant Science Does Not Mean Dominant Product," Research and Development, November 1987, p. 72.
+
+Grossman, Gene M., and Carl Shapiro, "Research Joint Ventures: An Antitrust Analysis," Journal of Law and Economics, Fall 1986, 2, pp. 315-337.
+
+von Hippel, Eric, The Sources of Innovation. Cambridge: MIT Press, 1988.
+
+Imai, Ken-ichi, and Yasunori Baba, “Systemic Innovation and Cross Border Networks.” Paper presented at Seminar on the Contributions of Science and Technology to Economic Growth, OECD, Paris, June 1989.
+
+Iyori, H., and A. Yesugi, The Antimonopoly Laws of Japan, New York: Federal Legal Publications, 2nd edition, 1983.
+
+Jorde, Thomas M., and David J. Teece, "Innovation, Cooperation, and Antitrust: Bal-
+
+ancing Competition and Cooperation," High Technology Law Journal, Spring 1989a, 4, pp. 1-113.
+
+Jorde, Thomas M., and David J. Teece, "To Keep U.S. in the Chips, Modify the Antitrust Laws," Los Angeles Times, July 24, 1989b, Part II, p. 5.
+
+Kline, S. J., and Nathan Rosenberg, "An Overview of Innovation." In Rosenberg, Nathan, and R. Landau, eds., The Positive Sum Strategy. Washington, DC: National Academy Press, 1986, pp. 275-305.
+
+Koopmans, Tjalling, Three Essays in the State of Economic Science. New York: McGraw Hill, 1957.
+
+Levin, Richard, A. Klevorick, R. Nelson, and S. Winter, "Appropriating the Returns from Industrial Research and Development," Brookings Papers on Economic Activity, Winter 1987, 3, 783-820.
+
+Mansfield, E., A. Romeo, M. Schwartz, D. Teece, S. Wagner and P. Brach, Technology Transfer, Productivity, and Economic Policy. New York: W. W. Norton, 1982.
+
+Nalebuff, Barry, and Joseph Stiglitz, "Information, Competition and Markets," American Economic Review, May 1983, 72, 278-284.
+
+Nelson, Richard, "Capitalism as an Enigma of Progress," Research Policy, 1990, 19, 193-214.
+
+Ordover, Janusz, and Robert Willig, "Antitrust for High Technology Industries: Assessing Research Joint Ventures and Mergers," Journal of Law and Economics, May 1985, 28, 311-33.
+
+Rosenberg, Nathan, Technology and American Economic Growth. Armonk: M. E. Sharpe, 1972.
+
+Sah, Raaj, and Joseph Stiglitz, “Technological Learning, Social Learning and Technological Change.” In Chakravarty, S., ed., The Balance between Industry and Agriculture in Economic Development. New York: St. Martin's/International Economic Association, 1989, pp. 285–298.
+
+Schumpeter, J. A., Capitalism, Socialism and Democracy. New York: Harper Brothers, 1942.
+
+Teece, David J., "Technology Transfer by Multinational Firms: The Resource Costs of Transferring Technological Know-how," The Economic Journal, June 1977, 87, 242-261.
+
+Teece, David J., "Economies of Scope and the Scope of the Enterprise," Journal of Economic Behavior and Organization, 1980, 1, 223-247.
+
+---
+
+96 Journal of Economic Perspectives
+
+Teece, David J., "Towards an Economic Theory of the Multiproduct Firm," Journal of Economic Behavior and Organization, 1982, 3, 39-63.
+
+Teece, David J., "Profiting from Technological Innovation," Research Policy, December 1986, 285-305.
+
+Teece, David J., "Inter-organizational Requirements of the Innovation Process," Managerial and Decision Economics, 1989a, 10, 35-42.
+
+Teece, David J., "Innovation and the Organization of Industry." Unpublished working
+
+paper, Center for Research in Management, University of California at Berkeley, 1989b.
+
+Teece, David J., "Market Entry Strategies for Innovators: Avoiding Pyrrhic Victories," Strategic Management Journal, 1991.
+
+Tirole, Jean, The Theory of Industrial Organization. Cambridge: MIT Press, 1988.
+
+Williamson, Oliver E., The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting. New York: Free Press, 1985.
+
+Winter, Sidney J., "Knowledge and Competence as Strategic Assets." In Teece, David J., ed., The Competitive Challenge. Cambridge: Ballinger, 1987.
+

processed/00009_W2109614009/document.md

@@ -0,0 +1,602 @@
+RAND Journal of Economics Vol. 39, No. 3, Autumn 2008 pp. 617-635
+
+# Academic freedom, private­sector focus, and the process of innovation
+
+Philippe Aghion* Mathias Dewatripont** and Jeremy C. Stein***
+
+We develop a model that clarifies the respective advantages and disadvantages of academic and private-sector research. Rather than relying on lack of appropriability or spillovers to generate a rationale for academic research, we emphasize control-rights considerations, and argue that the fundamental tradeoff between academia and the private sector is one of creative control versus focus. By serving as a precommitment mechanism that allows scientists to freely pursue their own interests, academia can be indispensable for early-stage research. At the same time, the private sector's ability to direct scientists toward higher-payoff activities makes it more attractive for later-stage research.
+
+## 1. Introduction
+
+■ Many important innovations, in industries ranging from pharmaceuticals to computer technology, have their origins in publicly funded research conducted at universities, foundations, and other nonprofit institutions. The traditional case for government funding of such academic research, as articulated by Nelson (1959) and Arrow (1962), is a familiar one: because of knowledge spillovers and imperfect intellectual-property-rights (IPR) protection, the economic value associated with certain kinds of ideas cannot be fully appropriated by the developers of these ideas, leading to private-sector underinvestment in “ basic ” research.
+
+In recent years, there has been a substantial expansion of formal IPR protection for earlystage research. This trend is in part a result of the Bayh-Dole Act of 1980, which gives academic institutions the right to patent and commercialize discoveries made with government-sponsored research support, and which has helped to launch a boom in the creation of university technology
+
+"Harvard University, NBER, CEPR, and CIAR; paghion@harvard.edu.
+
+**ECARES, Universite Libre de Bruxelles, and CEPR; mdewat@ulb.ac.be.
+
+***Harvard University and NBER; jeremy_stein@harvard.edu.
+
+This work is supported by the National Science Foundation. We are grateful to seminar participants at numerous institutions, as well as to Estelle Cantillon, Micael Castanheira, Georg Kirchsteiger, Nicola Lacetera, Josh Lerner, Andreu Mas-Colell, Fiona Murray, and Nicolas Sahuguet for their comments. Thanks especially to Bob Gibbons and Rebecca Henderson for extremely helpful conversations during the early stages of this research.
+
+Copyright c 2008, RAND.
+
+617
+
+---
+
+618 / THE RAND JOURNAL OF ECONOMICS
+
+transfer offices. Lach and Schankerman (2004) report that the number of patents granted to university scientists increased from 500 in 1982 to more than 3100 in 1998.1
+
+According to the traditional view, one might expect this trend toward increased IPR protection to be a largely beneficial one. After all, if academia is thought of as a second-best solution to the underinvestment problem caused by insufficient appropriability, then increased appropriability and a shift of research activity to the private sector should be efficiency enhancing. However, the trend has been controversial, particularly in fields such as biotechnology and pharmaceuticals. Many authors have expressed the concern that innovation in these fields is ultimately held back, rather than encouraged, when IPR protection is granted to the sorts of ideas that have traditionally been left in the public domain. Heller and Eisenberg (1998) talk of an “anti-commons” effect associated with early IPR protection, arguing that “a proliferation of intellectual property rights upstream may be stifling life-saving innovations further downstream in the course of research and product development.” 2
+
+In this article, we develop a model that (i) clarifies the respective advantages and disadvantages of academic and private-sector research and (ii) allows one to say when — in the process of developing an idea from its very earliest stages to a finished commercial product — it is normatively optimal to make the transition from academia to the private sector. Unlike the traditional literature, our model does not rely on imperfect appropriability to generate a role for academia. This is not to say that appropriability considerations are not important in some cases; certainly, there are many sorts of early-stage ideas where IPR protection is simply not feasible. Nevertheless, in fields like biotechnology, it is becoming increasingly evident that a lack of available IPR protection is not the only relevant factor in thinking about the merits of academia versus the private sector. Moreover, although arguments based on imperfect appropriability make clear the need for basic research to be subsidized, they are less clear on why this subsidy needs to happen in a different organizational form — that is, in a university, as opposed to in a private corporation.
+
+Our model is based on authority and control-rights considerations, and emphasizes what we believe to be a fundamental tradeoff between academia and the private sector, namely the tradeoff of creative control versus focus. We take the defining characteristic of academic research to be that scientists retain the decision rights over what specific projects to take on, and what methods to use in tackling these projects. Indeed, in our model, academia effectively boils down to a commitment mechanism that ensures scientists that these decision rights will not be abrogated. In contrast, the defining characteristic of private-sector research is that decision rights inevitably reside with the owner/manager of the firm, who can (and will) largely dictate project choice and methods to the individual scientists who work for the firm.
+
+More specifically, following Aghion and Tirole (1997) , and in the general spirit of the property-rights literature (Grossman and Hart, 1986; Hart and Moore, 1990; Hart, 1995) , we argue that scientists value creative control, and will have to be paid a wage premium in order to give it up. This assumption receives striking support in recent empirical work by Stern (2004) , who studies the job market for recent PhDs in biology. By using multiple job offers, Stern is able to control for differences in ability across job candidates. After doing so, he finds that wages are substantially lower in jobs that promise scientists either some freedom to pursue their own individual research agendas, or that encourage the publication of this work.
+
+Thus, one advantage of academia is that scientists can be hired more cheaply than in the private sector. The disadvantage of academia, however, is that they may end up working on projects that they find interesting, or prestige enhancing, but that have little immediate economic value. In contrast, firms can, by virtue of their control rights, direct scientists to work on those projects that have the highest economic payoffs.
+
+It turns out that the resolution of this tradeoff depends crucially on how far from commercialization a particular line of research is. To be concrete, imagine a line of biotech
+
+1 See also Henderson, Jaffe, and Trajtenberg (1998).
+
+$^{2}$ See Murray and Stern (2004) for an empirical analysis of the anti-commons argument.
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 619
+
+research which consists of ten distinct stages, and which will yield a drug worth $10 billion if and only if all ten stages are successfully completed. At the final stage, so close to an enormous potential payoff, the wages of individual scientists are relatively insignificant, and the most important consideration is simply ensuring that every available scientist is working on the task at hand, as opposed to on some other pet project. Thus, the directedness advantage of the private sector looms large, and it is optimal to have the project be privately owned at the last stage.
+
+Now consider things from the perspective of the very first stage of the research line. It may be that even if this first stage is successful, there is only a one-in-a-thousand chance that all nine of the subsequent stages will be also. So loosely speaking, the value of succeeding in the first stage is only on the order of $ 10 million. In this case, it becomes much more important to cede creative control, so as to economize on scientists' wages: if private-sector scientists cost $ 200,000 each per year, and academic scientists cost only $ 100,000 each per year, it may well be better to locate the project in academia, even if this entails some probability of the scientists wandering off in other directions.
+
+Thus, our primary contribution is to provide a simple account of why it can be socially optimal to have earlier-stage, more “ basic ” research take place in academia, without relying on spillovers, differences in IPR protection, or any of the other frictions that are usually invoked to rationalize a role for public funding of research. In so doing, we are able to offer a precise description of the potential costs associated with any anti-commons effect. If, for whatever reason, a research line transitions to the private sector sooner than is socially optimal, the inefficiency is manifested in the fact that — because of the higher wages — a private firm may employ too few scientists to work on the line, relative to what would happen in academia.
+
+We then go on to explore several extensions of the basic model. The first of these considers the possibility of research lines “ branching out ”— that is, giving rise to multiple potential directions for further work, some of which are nearer to commercial payoff than others. To be specific, suppose that there are two potentially legitimate research projects that make use of a firm's patented idea. One is an “ applied ” project that is only two stages away from a commercial payoff, while the other is a more “ basic ” project that is five stages away from any payoff. It is possible that the ultimate payoff on the latter, more basic project is sufficiently high that, evaluated at academic-sector wages , it is not only positive net present value (NPV), but of greater NPV than the applied project. At the same time, it is also possible that, evaluated at private-sector wages , the basic project is negative NPV, for the reasons described above. If this is the case, then when a private-sector firm has the decision rights, it will allocate all of its scientists to the applied project, and completely ignore the basic project. $^3$
+
+By contrast, if the private firm had never acquired the patent, and the ideas were left freely available to academic scientists, there would naturally tend to be some progress on both projects, as individual scientists followed their own interests. Moreover, the resources invested in each of the two projects would be positive NPV, because they would now each be evaluated at academic-sector wages. There would still be some deviation from the first best inherent in this outcome—without the ability to direct scientists, academia can never ensure the optimal allocation of scientists across the two projects—but this might be better than the private solution, which simply shuts down the basic project.
+
+Another set of extensions looks at hybrid governance structures that lie in between the extremes of a totally authoritarian private sector and a completely laissez-faire version of academia. We analyze in turn (i) the case where some real authority over the choice of research strategy may be left to scientists in a private firm; and (ii) the possibility that incentive schemes are used to focus the interests of academic scientists. These variations add some interesting nuances to our story, but do not alter the central message.
+
+3 Note that this conclusion does not rest on any exogenously imposed constraints regarding the scope of private firms: it is not that a firm cannot manage multiple projects at once, it simply finds it uneconomic to do so in this example, given its wage structure.
+
+© RAND 2008.
+
+---
+
+620 / THE RAND JOURNAL OF ECONOMICS
+
+The remainder of the article is organized as follows. In Sections 2 and 3, we introduce and then solve the most basic version of the model, in which at any given stage there is only one productive way to work with an existing idea carried forward from the previous stage. In this version, the only downside to the private sector is its more expensive wage structure. As noted above, this implies that the primary welfare cost of going private too early is that too few scientists may end up working on the idea.
+
+In Section 4 , we introduce the possibility that, in addition to pushing forward an idea along an existing chain, a scientist may instead prefer to branch off and work on something that is more basic in nature, but still economically promising. Here, there is an additional inefficiency associated with going private too early, namely an excessive aversion on the part of the private sector to stepping back in this fashion. In Section 5 , we examine hybrid governance structures. Section 6 discusses the connection to related literature, and Section 7 concludes.
+
+## 2. Basic framework
+
+■ Technology. The development of an economically valuable product (e.g., a new drug) starts with an initial idea $I_0$ . This idea can be built on by subsequent scientists, in stages. If stage 1 is successful, there is a refined idea $I_1$ ; this refined idea can be further worked on to potentially generate an even-more-refined idea $I_2$ , and so forth. There are a total of $k$ stages after the initial idea. If and only if all $k$ stages are successful, there is a final idea $I_k$ which generates a marketable product with value $V$ .
+
+The probability of success at any given stage depends on (i) the number of scientists who are active at that stage and (ii) the research strategies that they pursue. When a scientist is first exposed to an idea that has been brought forward from the previous stage, he must decide what strategy he wants to adopt in working with it. In the simplest version of the model, there are two options. First, the scientist can follow a “practical” strategy, which maximizes the probability that the current idea will be refined, and hence move on to the next stage. In particular, if there are $n$ scientists at stage $j$ who begin with the idea $I_{j-1}$ and who all follow a practical strategy, there is a probability $\phi(n)$ that the idea will be refined and yield the new idea $I_j$ .
+
+$n \geq 1$ We consider two different specifications of the function $\phi(n)$ : (i) $\phi$ ( $n$ ) $= p$ for all , and $\phi(0) = 0$ ; and (ii) $\phi(n) = (1 - (1 - p)^n)$ . The first specification corresponds to the assumption that all scientists working on the practical strategy have a perfectly correlated draw from the same success/failure distribution. This makes things especially simple — because it implies that in equilibrium there will always be exactly one scientist active at each research stage — and hence provides a useful way to illustrate the intuition for some of our results. At the same time, it can be too simple for some purposes, not allowing us to see the effects that arise when $n$ is meaningfully endogenous. Hence the second specification, which corresponds to the assumption that scientists working on the practical strategy have independent draws from the same success/failure distribution, with each individual having a success probability of $p$ , so that the probability of at least one success among a group of $n$ is given by $(1 - (1 - p)^n)$ .
+
+Instead of the practical strategy, any given scientist may choose to follow the “alternative” strategy in working with an existing idea. In this case, the scientist has a zero individual probability of success, and hence contributes nothing to a group’s chances of a breakthrough. The simplest interpretation is that the alternative strategy is fundamentally worthless: that is, it amounts to the scientist spending his time on puzzle-solving activities with no hope of an economic payoff. However, another possibility—which we explore below—is that even if the alternative strategy does not advance the current line of research, it may spawn an entirely new line of inquiry instead.
+
+□ Scientists’ preferences. There is an infinite pool of potential scientists. These scientists can only pursue research activities if they are hired either by an academic institution or a private-sector firm—they cannot be self-employed. Each scientist also has an outside option R that he can obtain
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 621
+
+by working in another profession, for example, as a taxi driver. This outside option sets a floor on the wages that scientists must earn.
+
+Our key assumption is that scientists value creative independence — that is, they value the right to choose how to pursue a particular idea. Specifically, after being exposed to idea $I_{j-1}$ , each scientist at stage $j$ decides whether he would better enjoy following the practical strategy or the alternative strategy. If he is able to undertake his favored strategy, he suffers no disutility from working. In other words, if a scientist could be promised ex ante that he would always be able to follow his favored strategy, he would be willing to work for a wage of exactly $R$ . However, if the scientist has to undertake the strategy that he likes less, he suffers disutility of $z$ . So if the scientist is certain that he is going to be forced to follow the less attractive strategy, he will set a reservation wage of $R+z$ . In between these two extremes, scientists behave in a risk-neutral fashion, and require a wage premium that is proportional to the probability that they will have to undertake the less desirable strategy.
+
+Importantly, neither the scientists themselves, nor their potential employers, know the scientists' preferences over the two strategies ex ante — that is, before the scientists have had a chance to look at the previous-stage idea and think about it. That is, scientists' preferences for the practical versus alternative strategies depend on the specifics of what kind of work these strategies will entail, and these specifics in turn depend on the nature of the previous-stage idea. To take a concrete example: a particular scientist may like to do a certain very specific kind of experimental work. Ex ante , it is unclear to him how this kind of experimental work will tie in with the project at hand. But after he has digested the previous-stage idea, it will become apparent whether his preferred experimental techniques are actually useful for pushing the idea to the next stage (in which case it will turn out that he is a practical type) or not (in which case it will turn out that he is an alternative type). $^4$
+
+We assume that the ex ante probability that a scientist prefers to follow the practical strategy is given by $\alpha$ . In addition, we assume perfect correlation across all scientists at a given stage in terms of their preferences over the two strategies. In other words, either all scientists at a given stage prefer the practical strategy, or all prefer the alternative strategy. This strong perfectcorrelation assumption is not necessary for our results — any nonzero positive correlation will do — but it greatly simplifies the exposition. $^5$ Moreover, positive correlation in preferences can be thought of as reflecting the natural idea that some types of research are simply more fun for most scientists than others.
+
+□ Academia. As noted in the Introduction, we present an extremely rudimentary and strippeddown rendition of academia. We take the defining characteristic of this organizational form to be that it represents a precommitment to leave control over the choice of research strategy in the hands of individual scientists. Although this assumption would appear to be empirically well motivated, a natural question is why academia is uniquely able to make this commitment. We suspect that the nonprofit nature of academia plays a central role in this regard, a point that we develop more formally in Section 5 below. In particular, if one thinks of supervisory effort (the resources devoted to monitoring and directing scientists) as endogenous, it is plausible that academic administrators have much lower incentives to exert such effort than, for example, a corporate CEO, whose compensation can be linked to the share price. $^6$
+
+In the baseline version of the model, we set aside the possibility of incentive schemes (either implicit or explicit) in academia. In Section 5 , we effectively endogenize this assumption. We
+
+4 One can imagine other reasons why a scientist would prefer to work on the alternative project, in spite of its lower economic payoff. It may be more intellectually challenging, or it may hold the promise for greater professional prestige.
+
+5 More precisely, we need to avoid the limiting case where there is effectively a continuum of scientists with independent preferences. In this case, hiring $n$ scientists in academia is functionally equivalent to hiring $\alpha n$ scientists in the private sector — with probability one, both yield the same amount of research effort devoted to the practical strategy — and the solution at all stages of the research line will be to go with whichever option involves lower total wages.
+
+6 See Hart, Shleifer, and Vishny (1997) and Acemoglu, Kremer, and Mian (2003) for related discussions about the role of nonprofits.
+
+© RAND 2008.
+
+---
+
+622 / THE RAND JOURNAL OF ECONOMICS
+
+show that if $z$ is nonstochastic, it is generally optimal not to use incentives in academia. Intuitively, if it is desirable to induce all scientists — irrespective of their preferences — to follow the practical strategy, this is more cheaply done in the private sector, where they can simply be directed to do so, as opposed to in academia, where they have to be promised unconditional bonuses for voluntarily choosing this option.
+
+Finally, we assume that if the results from academic research are not sold to the private sector, these results are published and freely disseminated to other academic scientists. These assumptions are the only features that distinguish academia from the private sector in our model. In particular, we abstract from the question of what the exact mission of universities is or should be, or whether universities should be private or public.
+
+The outcome of any stage $j$ that takes place in academia is easy to describe. Suppose there are $n$ scientists active at this stage. Each scientist is paid a wage $w_a = R$ , and always works on his preferred strategy. This implies that with probability $\alpha$ , all $n$ scientists work on the practical strategy, and with probability $(1-\alpha)$ , all $n$ work on the alternative strategy. Thus, the ex ante probability of advancing to the next stage is given by $\alpha \phi(n)$ .
+
+□ The private sector: firms and property rights. At some stage, a firm may acquire exclusive rights to an idea. Thus the following transaction is contractually feasible: an academic scientist (or the institution for which he works) may sell his idea to an entrepreneur, and promise not to publish the idea or in any other way share the idea with anybody else. The model therefore incorporates the potential for a strong form of IPR protection at all stages of the innovation process.
+
+An important assumption is that only an entrepreneur has the funds to pay for an idea, as scientists have no funds of their own. This implies that if an entrepreneur is to take the idea forward, she will have to hire scientists as her agents at each subsequent stage of the development process. The entrepreneur can only derive utility from monetary sources, so she will only pay for an idea if she can earn a monetary return from it. Unlike the scientists, the entrepreneur has no innate preferences over research strategies. Nor does she get any utility from an idea being widely disseminated.
+
+At the time the entrepreneur hires a team of scientists to work on a given stage, she cannot know the scientists' preferences over the practical versus alternative strategies — these preferences only become evident once the scientists are inside the firm and have been given access to the idea by the entrepreneur. Yet ex post , the entrepreneur has the authority to force the scientists to work on whichever strategy she deems to be most profitable. Indeed, it is impossible for the entrepreneur to precommit to doing otherwise — this is the defining characteristic of private-sector research. For example, once it becomes clear that the practical strategy requires a specific type of experimental work (call it type A), whereas the alternative strategy involves a different kind of experimental work (call it type B), the entrepreneur will force the scientists' hands by buying laboratory equipment that is only compatible with type A work.
+
+It follows that scientists will demand a wage of $w_p = R + (1 - \alpha)z$ in order to work in the private sector. The $(1 - \alpha)z$ markup over the academic wage represents compensation for loss of creative freedom — the fact that scientists now always have to adopt the practical strategy, whether this turns out to coincide with their preferences or not.
+
+## 3. The case of a single research line
+
+■ The starting point for our analysis is the case of a single research line. We begin with the perfectly correlated draws assumption, which ensures that $n=1$ at all stages. Next, we consider the independent-draws alternative, in which $n$ is endogenous, and in which $\phi(n) = (1-(1-p)^n)$ .
+
+```bash
+□ Perfectly correlated draws: n = 1
+```
+
+The basic tradeoff between academia and private research. Consider a project which involves $k$ stages, and imagine that the first $(k-1)$ stages have been successful, so that we are now at stage
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 623
+
+$k$, with only one more success required to generate a payoff of $V$. If the last round of research is conducted in the private sector, and one scientist is hired, the expected payoff is
+
+$$E\left(\pi_{k}^{p}\right)=p V-w_{p} .\quad (1)$$
+
+If instead the last round of research is conducted in academia, and one scientist is hired, the expected payoff is
+
+$$E\left(\pi_{\varepsilon}^{a}\right)=\alpha p V-w_{a}.\quad (2)$$
+
+Thus there is a simple tradeoff: on the one hand, wages are lower in academia. On the other hand, the inability to direct scientists in academia means that the probability of success is lower than in the private sector. Comparing the two equations, it is easy to see that the private sector will yield a higher payoff than academia if and only if
+
+$$(1-\alpha)pV>(w_{p}-w_{\alpha})\quad (3)$$
+
+or
+
+$$p V>z . \quad (4)$$
+
+An important first piece of intuition is that the private sector looks relatively more attractive when $p$ and $V$ are high, that is, when the expected payoff to research is greater.
+
+Next, denote the maximum of $E(\pi_k^p)$ and $E(\pi_k^v)$ by $\Pi_k$ . Folding back to stage $(k-1)$ , we can now compare
+
+$$E\left(\pi_{k-1}^{p}\right)=p\Pi_{k}-w_{p}\quad (5)$$
+
+and
+
+$$E\left(\pi_{k-1}^{\alpha}\right)=\alpha p \Pi_{k}-w_{\alpha}.\quad (6)$$
+
+This implies that the private sector will yield a higher payoff than academia at stage (k − 1) if and only if
+
+$$p\Pi_{k}>z.\quad (7)$$
+
+Because $\Pi_k < V$ , it follows that if the private sector is value maximizing at stage $(k-1)$ , it is also value maximizing at stage $k$ . This recursive logic can be extended backward, so that at any earlier stage $i$ , we have
+
+$$E\left(\pi_{i}^{p}\right)=p\Pi_{i+1}-w_{p}\quad (8)$$
+
+and
+
+$$E\left(\pi_{i}^{\alpha}\right)=\alpha p \Pi_{i+1}-w_{\alpha}.\quad (9)$$
+
+Moreover, the private sector will generate a higher payoff than academia at stage i and all future stages if and only if
+
+$$p\Pi_{i+1}>z.\quad (10)$$
+
+Observe that as $i$ declines — that is, as we move backward to earlier and earlier stages — it becomes progressively harder for the private sector to outperform academia, because $\Pi_{i+1}$ falls. This immediately implies:
+
+Proposition 1. It cannot be value maximizing to have academia operate at later stages than the private sector.
+
+Next, we can show that academia may become indispensable at the earlier stages of a line if the total length $k$ of the line is sufficiently large. To see this, note that if the entire line is located
+
+(C) RAND 2008.
+
+---
+
+624 / THE RAND JOURNAL OF ECONOMICS
+
+in the private sector, its ex ante value, $\Omega$ (allprivate), is given by
+
+$$\Omega( allprivate )=p^{k} V-w_{p} \sum_{j=1}^{k} p^{j-1} .\quad (11)$$
+
+But this expression clearly becomes negative for $k$ sufficiently large because $p^k V$ converges to zero, whereas the expected wage bill $w_p \sum_{j=1}^k p^{j-1}$ remains bounded away from zero and increasing in $k$ . We thus have:
+
+Proposition 2. A research program with a sufficiently large number of stages $k$ will not be viable if located exclusively in the private sector.
+
+The proposition by itself does not fully establish the necessity of academia — we still need to show that for a nonempty set of parameter values, a research line that is not viable if located exclusively in the private sector can be viable if started in academia. But this latter point is easy to demonstrate. For example, suppose that $w_a=R=0$ , and consider the ex ante value $\Omega$ ( allacademic ) of a line that is located exclusively in academia:
+
+$$\Omega(allacademic)=(\alpha p)^{k}V,\quad (12)$$
+
+which is obviously positive for all $k$.
+
+More generally, it is easy to show that a necessary condition for academia to be viable one stage earlier than the private sector is that $\alpha z > R$ . Intuitively, when $\alpha z$ is large relative to $R$ , this tends to make academia relatively attractive because (i) academic scientists choose the practical strategy fairly often even without being directed to do so and (ii) there is a proportionally large wage premium in the private sector.
+
+For expositional simplicity, we have been assuming that the probability of success $p$ associated with the practical strategy is a fixed constant, and does not vary with the research stage. This is obviously restrictive. However, the fundamental result in Proposition 1— namely, that it is always optimal for academia to precede the private sector — is much more general, and holds even if the success probabilities vary by stage in an arbitrary way. This can be proven by induction. Denote the probability of success at stage $i$ by $p_i$ . According to equation (10) , if academia is preferred at stage $i$ , it must be that $p_i \Pi_{i+1} < z$ . But then the private sector cannot be preferred at stage $(i-1)$ , because this would require $p_{i-1} \Pi_i > z$ , which is impossible because $\Pi_i < p_i \Pi_{i+1} < z$ .
+
+The intuition for this result is that, as research progresses from one stage to the next, the value function is always rising fast enough to increase the relative appeal of the private sector, even if the success probabilities are falling. One might at first think that if $p_i$ is much lower than $p_{i-1}$ , it might be optimal to be in the private sector at stage ( $i-1$ ), and then revert back to academia at stage $i$ . However, the flaw in this reasoning is that if $p_i$ is low enough to make academia attractive at stage $i$ , then the value $\Pi_i$ of completing stage ( $i-1$ ) — and thereby gaining entry to stage $i$ — must be low as well. Indeed, $\Pi_i$ must be so low in this case that the private sector cannot be optimal at stage ( $i-1$ ).
+
+The socially optimal transition point. Using our recursive approach, it is straightforward to calculate the point at which it is socially optimal for a research line to make the transition from academia to the private sector. The following lemma is an immediate consequence of (10) :
+
+Lemma 1. From the perspective of a social planner, there is a unique transition point $i^*$ , such that it is optimal for stage $i^*$ to be the first stage conducted in the private sector. This transition point $i^*$ is the smallest value of $i$ that satisfies
+
+$$p\Pi_{i+1}=p^{k-i+1}V-w_{p}\sum_{j=1}^{k-i+1}p^{j-1}>z.\quad (13)$$
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 625
+
+The comparative statics properties of the optimal transition point follow from this lemma. They are intuitive, and can be summarized as follows:
+
+Proposition 3. Holding fixed the number of stages $k$ in a research line, it is optimal to have the transition to the private sector occur earlier if (i) $V$ is greater or (ii) $z$ is smaller.
+
+Given an optimal transition policy, we still need to check that the research project is ex ante positive NPV—namely that it is socially worthwhile to fund the stages prior to i ∗ in academia. If the line is managed optimally throughout, its ex ante value, which we denote by $\Omega$ ( i ∗ ), is given by
+
+$$\Omega(i^{*})=\alpha^{(^{*}-1}p^{k}V-w_{a}\sum_{j=1}^{i^{*}-1}(\alpha p)^{j-1}-w_{p}(\alpha p)^{(^{*}-1}\sum_{j=1}^{k-i^{*}+1}p^{j-1},\quad (14)$$
+
+where $i^{*}$ is the optimal transition point determined in the previous lemma.
+
+The ex ante feasibility (henceforth, EAF) constraint for the research line is then simply the condition that $\Omega\left(i^*\right)>0$ . This condition is always satisfied if $w_a=0$ , so that academic research is costless, and in much of what follows we use this assumption to keep things simple. $^7$ However, we will also briefly consider what happens when $w_a>0$ .
+
+Comparison with early privatization. As noted in the Introduction, a number of authors have expressed the concern that, in a world with full IPR protection where ideas can be sold to the private sector at any point in their development, privatization of a research line may occur sooner than is socially optimal. To provide a concrete way to think about this issue, imagine that the decision of whether to sell an academic idea to a private-sector firm rests in the hands of a university technology transfer office, so that the transition to the private sector occurs as soon as the value of the line under private management exceeds the reservation value $T$ of the technology transfer office. If we denote by $i(T)$ the first stage conducted in the private sector under this scenario, we have that $i(T)$ is the smallest value of $i$ that satisfies
+
+$$p \Pi_{i+1}=p^{k-i+1} V-w_{p} \sum_{j=1}^{k-i+1} p^{j-1}>T .\quad (15)$$
+
+Under this transition policy, the EAF constraint is modified, with $i^*$ being replaced everywhere by $i(T)$ , so that the constraint becomes
+
+$$\alpha^{i(T)-1} p^{k} V-w_{a} \sum_{j=1}^{i(T)-1}(\alpha p)^{j-1}-w_{p}(\alpha p)^{i(T)-1} \sum_{j=1}^{k-i(T)+1} p^{j-1}>0 .\quad (16)$$
+
+This condition is obviously more restrictive than the EAF condition corresponding to socially optimal transition. That is, it is harder for the research program to be ex ante positive NPV if it is managed suboptimally than if it is managed optimally.
+
+In what follows, we consider the limiting case where the reservation price $T$ is set at an arbitrarily low positive value. This implies that, absent any countervailing government policy, an idea transitions from academia to the private sector as soon as any private-sector firm finds it economic to make a nonzero bid for it. We refer to this outcome as “ early privatization, ” and denote the associated transition point $i(0)$ by $i^e$ . Clearly this is an extreme case, and we do not mean to suggest that it is the most realistic one. However, it serves as a convenient benchmark that allows one to clearly see the potential benefits of academia for early-stage research.
+
+What are the welfare costs associated with early privatization? In the current version of the model, there are two possible effects. First, supposing that the EAF constraint is always satisfied,
+
+7 A less literal way of thinking about the assumption that $w_a = 0$ is that, for whatever reason, certain research lines are always able to get funding in academia, irrespective of NPV considerations. This could be because the government agency responsible for funding these lines has other objectives besides value maximization.
+
+© RAND 2008.
+
+---
+
+626 / THE RAND JOURNAL OF ECONOMICS
+
+regardless of the timing of the transition to the private sector (this will be the case if $w_a=0$ ), then early privatization does not prevent a research program from getting started in the first place. Consequently, its only downside is that it leads to inefficiently high labor costs. Indeed, in this case, early privatization necessarily raises the ex ante odds that the research program will ultimately bear fruit, but it does so at a labor cost that is too high relative to the benefit.
+
+However, if we do not take for granted that the EAF constraint is always satisfied (because $w_a > 0$ ), then there can be a second cost of early privatization. In particular, a project that would initially get funded in academia under the socially optimal transition policy may no longer be worth funding if it is anticipated that the transition will happen too soon. In this case, the expectation of early privatization has a more drastic effect, because it completely kills off an otherwise positive-NPV research line. $^8$
+
+Are wage differentials quantitatively important? All of the above results hinge on there being a wage differential between academia and the private sector. In particular, academic scientists must be willing to work for lower wages than their private-sector counterparts, because they value creative freedom. Although this assumption fits qualitatively with both casual observation as well as with the evidence in Stern (2004) mentioned in the Introduction, it can reasonably be asked whether real-world wage differentials are quantitatively large enough to justify making them the centerpiece of our theory.
+
+Stern's (2004) estimates — which, again, are based on multiple job offers to entry-level PhD scientists — suggest differentials on the order of roughly 20 – 30 % of salary. These are certainly economically significant differences, although perhaps not enormous ones. However, for several reasons, we believe that a superficial glance at these sorts of numbers may lead one to underestimate the actual importance of academic versus private-sector cost differentials.
+
+First, consider Stern's multiple-job-offer methodology. This approach is attractive, in that it allows one to control for differences in aptitude across job candidates. But by its nature, it only reveals the wage differential for those types who are “on the cusp,” in the sense of being willing to entertain both academic and private-sector jobs. It seems likely that there are more extreme types for whom the required wage premium to go to the private sector would be much higher, but who are never observed receiving private-sector offers.
+
+In the terminology of our model, this amounts to saying that there is heterogeneity across scientists in the disutility parameter $z$ . In the presence of such heterogeneity, the benefits associated with academia are not adequately summarized by the observed wage differential. To take an extreme example, suppose that there are two types of scientists: some who are willing to go to the private sector at a 25 % wage premium, and others who would not be willing to go to the private sector at any wage — that is, who have an infinite value of $z$ . The observed wage differential will be 25 % , but this does not fully capture the benefits of academia, because in addition to allowing for the hiring of the first type of scientists at a discount, it also represents the only way to ever hire the second type.
+
+A further observation is that, as we have cast it, the basic version of the model understates the total wage bill associated with the private-sector form, because it assumes away any wages paid to private-sector research supervisors. Because it is impossible to have the benefits of focus in the private sector without such supervision, this is a significant omission if one wants to begin taking the magnitudes in the model seriously. We model the costs of supervision more explicitly in Section 5 below.
+
+Finally, it should be noted that the “branching” version of the model in Section 4 has the potential to greatly lever up the effects associated with even relatively modest wage differentials.
+
+8 We should be clear about the nature of the thought experiment we have in mind when we say that early privatization may lead to a violation of the EAF constraint and hence deter the initiation of a given research line. In this case, we are implicitly assuming that a social planner makes value-maximizing funding decisions in academia, but takes as given the inefficiency associated with early privatization. That is, the social planner can be thought of as a government agency that funds academic research, but that has nothing to say about the timing of the transition to the private sector.
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 627
+
+In this setting, the costs of higher wages in the private sector show up not only directly but also indirectly, in the form of all the foregone basic research opportunities that might have been undertaken in a lower-cost academic environment.
+
+□ Independent draws: $\phi(n) = (1 - (1 - p)^n)$ . There is one effect which is conspicuously absent from the correlated-draws version of the model. Conditional on the EAF constraint being satisfied, early privatization can never reduce the ex ante probability of success. That is, conditional on the project getting started in the first place, early privatization is necessarily a force in favor of innovation, with the only downside being that this comes at an inefficiently high labor cost. It turns out that this particular feature is an artifact of our simplifying assumption that the number of researchers at each stage is always equal to one. As we now show, when $n$ is made endogenous in a more reasonable way, early privatization can stymie innovation even conditional on the project getting off the ground. This is because the higher labor costs associated with early privatization can now lead to a reduction in the number of scientists employed at a given stage in the private sector.
+
+Analysis. As before, to solve the social planner's problem, we work backward from stage $k$ . If the last round of research is conducted in the private sector, and $n$ scientists are hired, leading to a success probability of $\phi(n) = (1-(1-p)^n)$ , the expected payoff is
+
+$$E\left(\pi_{k}^{p}\right)=\left(1-(1-p)^{n}\right) V-n w_{p} .\quad (17)$$
+
+Ignoring integer problems, the firm's first order condition implies that the optimal number of scientists, $n_k^p$ , is given by
+
+$$n_{k}^{p}=(\log(\beta V/w_{p}))/\beta,\quad (18)$$
+
+where we have defined $\beta = -\log{(1-p)} > 0$ . When (18) is negative, that is, when $\beta V/w_p < 1$ , we are at a corner, with private-sector employment and profits both equal to zero. When (18) is positive, expected stage- $k$ private-sector profit is given by
+
+$$E\left(\pi_{k}^{p*}\right)=V-\left(w_{p}/\beta\right)\left(1+\log\left(\beta V/w_{p}\right)\right).\quad (19)$$
+
+If instead the last round of research is conducted in academia, and n scientists are hired, the expected payoff is
+
+$$E\left(\pi_{k}^{a}\right)=\left(1-(1-p)^{n}\right) \alpha V-n w_{\alpha} .\quad (20)$$
+
+If we imagine that the number of academic scientists is also set at an optimal level—that is, a well-intentioned government agency chooses the aggregate level of research funding across all universities—then we have
+
+$$n_{k}^{a}=\left(\log \left(\alpha \beta V / w_{a}\right)\right) / \beta . \quad (21)$$
+
+Note that the optimal number of academic scientists $n_k^a$ can be either greater than or less than the optimal number of private-sector scientists, $n_k^p$ . This is because academic scientists are simultaneously cheaper but less productive. If (21) is positive, expected stage- $k$ profit in academia is
+
+$$E\left(\pi_{k}^{\alpha *}\right)=\alpha V-\left(w_{\alpha} / \beta\right)\left(1+\log \left(\alpha \beta V / w_{\alpha}\right)\right) .\quad (22)$$
+
+Denote the maximum of $E(\pi_k^{p*})$ and $E(\pi_k^{a*})$ by $\Pi_k$ . Proceeding recursively, it follows that at any earlier stage $i$ , so long as $\beta \Pi_{i+1}/w_p > 1$ , and $\alpha \beta \Pi_{i+1}/w_a > 1$ , respectively, private-sector and academic-sector expected profits are given by
+
+$$E\left(\pi_{i}^{p*}\right)=\Pi_{i+1}-\left(w_{p}/\beta\right)\left(1+\log\left(\beta\Pi_{i+1}/w_{p}\right)\right)\quad (23)$$
+
+and
+
+$$E\left(\pi_{i}^{a *}\right)=\alpha \Pi_{i+1}-\left(w_{\alpha} / \beta\right)\left(1+\log \left(\alpha \beta \Pi_{i+1} / w_{a}\right)\right) .\quad (24)$$
+
+(C) RAND 2008.
+
+---
+
+628 / THE RAND JOURNAL OF ECONOMICS
+
+Note that for academia to be viable at an earlier stage than the private sector, that is, for $E(\pi_i^{\rho^*})$ to be zero and $E(\pi_j^{\rho^*})$ to be positive for some $i$ , we require that $\alpha/w_a > 1/w_p$ , which is equivalent to $\alpha z > R$ . This is the same condition that we stated above for the version of the model with $n=1$ .
+
+In the Appendix, we prove the following analog to Lemma 1:
+
+Lemma Suppose that $E(\pi_k^{p*}) > E(\pi_k^{a*})$ , so that it is optimal to locate the last stage (i.e., stage $k$ ) in the private sector. Suppose also that $\alpha z > R$ , so that academia is viable at an earlier stage than the private sector. From the perspective of a social planner, there is a unique transition point $i^*$ , such that it is optimal for stage $i^*$ to be the first stage conducted in the private sector. This transition point $i^*$ is the smallest value of $i$ that satisfies $E(\pi_i^{p*}) > E(\pi_i^{a*})$ , where these two quantities are defined by the recursive equations (23) and (24) .
+
+By contrast to the social optimum, to solve for the transition point under early privatization, we simply keep folding backward to earlier stages, always staying in the private sector. If $k$ is large enough, we will eventually hit a stage $i^e$ such that $E(\pi_{i^e}^{p^s}) > 0$ , but $E(\pi_{i^e-1}^{p^s}) = 0$ . The latter condition obtains when $(\beta \Pi_{i^e}/w_p) < 1$ , so that at stage $(i^e-1)$ , a private-sector firm is at a corner solution, with $n_{i^e-1}^p = 0$ . It then follows that $i^e$ is the earliest stage at which an idea is viable in the private sector — if an idea were to wind up in the private sector earlier, no firm would ever invest positive resources in it.
+
+An additional benefit of academia. The following example illustrates the additional positive scale effect associated with academic research that arises when $n$ is endogenous. The example also shows how this scale effect alters the welfare comparison between the socially optimal transition policy and early privatization.
+
+Example Set $R=w_a=0$ , $z=1$ , $\alpha=0.5$ (implying that $w_p=0.5$ ), $p=0.10$ , $V=100$ , and $k$ $=8$ . Under the optimal transition policy, the first three stages are in academia, and the last five are in the private sector. This optimal policy yields an ex ante expected payoff of 2.08, and an ex ante probability of success of 0.078. Under early privatization, the first stage is in academia, and the last seven are in the private sector. Early privatization yields an ex ante expected payoff of 0.06, and an ex ante probability of success of 0.044.
+
+The key feature of the example is that early privatization now not only lowers the ex ante expected payoff (by definition), it also lowers the ex ante probability that, conditional on the research line getting started, it will ultimately bear fruit. This is because now, with variable n , when the idea is privatized early, relatively few scientists are hired to work on it in the initial private-sector stages, as compared to the number that would be hired in academia. Consequently, the success probabilities for the initial private-sector stages are relatively low. In the context of the example, if the idea moves to the private sector early, at $i=2$ , only $n=2.1$ scientists are hired, yielding a probability of success at this stage of 0.20. By contrast, if the idea stays in academia for the second stage, an infinite number of researchers are hired (because $w_u=0$ ), yielding a probability of success at this stage of 0.50.
+
+## 4. Branching out: the potential for new lines
+
+■ Thus far, we have assumed that at each stage there is only one economically legitimate research strategy — namely the practical strategy, which has the potential to advance the project to the next stage along the chain. In contrast, the alternative strategy has been taken to be nothing more than worthless puzzle solving. Now we modify this assumption. Although we keep the restriction that only the practical strategy helps to advance the current line of research, we allow the alternative strategy to yield new insights which may spawn wholly different lines of research. The interpretation is that when scientists turn away from the applied task of pushing the current line forward, they may not be shirking per se, but rather taking a useful step back that may ultimately generate fundamental breakthroughs.
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 629
+
+□ Adding offspring lines to the model. To embed this notion into our model, we proceed as follows. We keep all the same assumptions as before, with one modification. Now, if at any stage of the original research line, a scientist works on the alternative strategy, there is a probability $p_r$ of a revolutionary new idea which will form the basis for $\gamma$ entirely new “ offspring ” research lines, with $\gamma \geq 1$ . Each of these offspring lines has the same properties as the single lines analyzed above, although we allow for the possibility that the offspring have a greater number of stages than the original line, that is, that $k_o \geq k$ . Moreover, for computational simplicity but without any major loss of insight, we assume that the offspring lines are themselves sterile, and cannot give rise to further generations of revolutionary ideas. That is, revolutionary ideas that yield offspring can only come from the alternative strategy applied at some stage of the original parent line.
+
+In order to make things interesting, we assume that an offspring line has sufficiently many stages $k_o$ that it is not viable if it is born into the private sector. This just means that
+
+$$p^{k_{o}} V-w_{p} \sum_{j=1}^{k_{o}} p^{j-1}<0 .\quad (25)$$
+
+This assumption ensures that private-sector entrepreneurs will continue to direct scientists to stay away from the alternative strategy, and to focus all their efforts on the practical strategy, no matter how large $\gamma$ is. Simply put, the assumption implies that the private sector never has any use for the offspring lines generated by the alternative strategy, because these lines are so early stage that they are negative NPV when evaluated at private-sector wages . $^9$
+
+In contrast, we assume that an offspring line is viable if it is born in academia, and managed optimally from that point on. That is, denoting the ex ante value of an offspring line under optimal management by $\Omega_o^*$ , we assume that $\Omega_o^* > 0$ .
+
+Using a logic similar to that above, we can derive the socially optimal transition point for the parent line, $i^*_p$ . To keep things simple, we focus on the correlated-draws case where $n=1$ . The independent-draws case where $\phi(n)=(1-(1-p)^*)$ does not add any further insight in this setting, so we omit it for the sake of brevity.
+
+Lemma 3: From the perspective of a social planner, there is a unique transition point $i_p^*$ , such that it is optimal for stage $i_p^*$ to be the first stage of the parent line conducted in the private sector. This transition point $i_p^*$ is the smallest value of $i$ that satisfies
+
+$$p\Pi_{i+1}>z+p,\gamma\Omega_{\omega}^{*}.\quad (26)$$
+
+The logic is identical to that of Lemma 1 in the basic model, and the expression for the optimal transition point is the same, except that a $p,\gamma\Omega_o^*$ term has been added to the right-hand side of the inequality. The intuition is straightforward. In the basic model, it is optimal to make the transition to the private sector once the increase in value that comes from a higher probability of moving to the next stage is sufficient to outweigh the private-sector wage premium. Now, in addition to this wage premium, there is a second cost of moving to the private sector — the fact that offspring lines are never developed. Or said differently, academia now has the added benefit of letting many more flowers bloom, which makes it optimal to wait longer before moving to the private sector, all else equal.
+
+The early-privatization transition point, $i_p^e$ , is identical to that in the basic model. This is because once privatized, the remaining payoffs from the parent line are unchanged from before, as scientists are still always assigned to the practical strategy. Comparing the socially optimal transition point and the early-privatization transition point, we have:
+
+9 For the purposes of this section, we are implicitly assuming that the private sector cannot sell the rights to offspring lines back to academia, at least not at a price approaching their full value in an academic setting. This assumption seems to be empirically realistic, and can be defended based on (i) information asymmetries; (ii) free-riding problems among universities; or (iii) various bureaucratic constraints.
+
+© RAND 2008.
+
+---
+
+630 / THE RAND JOURNAL OF ECONOMICS
+
+Proposition 4. The gap between the socially optimal transition point for the parent line and the early-privatization transition point, given by ( $i^*_p - i^c_p$ ), is greater than in the basic model, and is increasing in the productivity $p_r \gamma$ of the alternative strategy.
+
+Again, the noteworthy feature of the branching model is that it is now possible to have a substantially lower rate of innovation — defined in terms of the ex ante expected number of research lines that reach successful completion — under early privatization than under the optimal policy, even when we restrict ourselves to the case where $n=1$ .
+
+□ Empirical implications of the branching model. From an empirical perspective, what is perhaps most interesting about the current version of the model is that it implies that once an idea becomes the property of a private firm, it will be developed along relatively narrow lines. That is, the private sector's ownership of a given idea will not yield as diverse an array of useful next-generation ideas as would be generated in academia .
+
+This implication seems to fit with the broad spirit of recent empirical work by Kaplan, Sensoy, and Stromberg (2006). They study the life-cycle evolution of 49 venture-capital-backed firms, beginning with their first business plans, and continuing until three years after these firms have gone public. The firms in their sample are largely in high-technology industries, with the vast majority coming from either the biotechnology or software/information-technology sectors. Perhaps the most striking of Kaplan et al.'s findings is the extent to which these innovative firms stick to their original business plans, and do not branch out into other lines of business. As they put it:
+
+While the companies grow dramatically, their business models or core businesses are remarkably stable. Only one firm changes its core line of business over the sample period.
+
+Although the results of Kaplan, Sensoy, and Stromberg (2006) are suggestive, they hardly represent a sharp test of our theory. There would clearly seem to be room for further work in this vein.
+
+## 5. Hybrid organizational forms
+
+■ Thus far, our renditions of academia and the private sector have been extreme caricatures. At one end of the spectrum, we have cast the private sector as an organizational form in which scientists have no freedom of choice whatsoever, and are always forced to follow a research strategy dictated by an entrepreneur/supervisor, whether they like this strategy or not. At the other end of the spectrum, we have cast academia as a setting in which researchers not only have absolute creative control but also face no incentives—either explicit or implicit—that might encourage them to pursue a relatively more practical strategy.
+
+We now discuss how each of these extreme assumptions might be relaxed. Doing so leads to a more nuanced and realistic view of both organizational forms, although it does not alter the main message of our model. In what follows, we focus on the basic version of the model from Section 3 , and on the expositionally simpler case where $n=1$ at all stages.
+
+Less authoritarian private-sector firms. Many private-sector firms are known for taking explicit measures to give some of their research-oriented employees a degree of creative independence. For example, both 3M and Google apparently allow certain employees one day a week to pursue their own research interests. $^10$ In an effort to model this behavior, we follow Aghion and Tirole (1997) and draw a distinction between formal and real authority in privatesector firms. The idea is that although the entrepreneur in a private firm always retains the formal right to direct her employees — by, for example, choosing the type of lab equipment they work
+
+$^{10}$ We thank Rebecca Henderson for suggesting these examples.
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 631
+
+with—she may in fact choose not to exercise this right if she is not sufficiently informed to know which is the better strategy.
+
+Suppose that a project is located in the private sector at stage $i$ . The timing of events is now as follows. First, the entrepreneur hires a scientist, and agrees to pay him a wage of $w_{pl}$ . Next, the entrepreneur invests effort in trying to become informed about the project. For an effort cost of $\theta \lambda^2/2$ , the entrepreneur has a probability $\lambda$ of becoming informed. If she is informed, she is then able to force the scientist to follow the practical strategy, as we have been assuming above. However, if she is uninformed, the entrepreneur is unable to direct the scientist, and the scientist is thus free to do what he wants, just as in academia.
+
+Observe that the basic version of the model in Section 3 is just a special case of this one, in which $\theta=0$ , so that the entrepreneur always chooses to become informed with probability $\lambda=1$ . It is also worth noting that this modelling framework could be used to explicitly link the precommitment function of academia to its nonprofit nature: if academic administrators (e.g., deans, or a university president) do not get profit-linked compensation, they will be unwilling to expend any effort on becoming informed, leading to a situation in which $\lambda=0$ — that is, in which real authority always rests with individual scientists.
+
+Given our assumptions, the payoff to the entrepreneur if she is informed at stage $i$ is
+
+$$E\left(\pi_{i}^{p}\mid\text{informed}\right)=p\Pi_{i+1}-w_{p i}.\quad (27)$$
+
+The payoff to the entrepreneur if she is uninformed at stage $i$ is
+
+$$E\left(\pi_{i}^{p}\mid\text{uninformed}\right)=\alpha p\Pi_{i+1}-w_{p i}.\quad (28)$$
+
+Therefore, the marginal value of being informed at stage $i$ is $(1-\alpha)p\Pi_{i+1}$ , and the entrepreneur's equilibrium probability of becoming informed at this stage is
+
+$$\lambda_{i}=(1-\alpha)p\Pi_{i+1}/\theta.\quad (29)$$
+
+It follows that the unconditional expected payoff at stage $i$ in the private sector is given by
+
+$$E\left(\pi_{i}^{p}\right)=\left(\lambda_{i}+\alpha\left(1-\lambda_{i}\right)\right) p \Pi_{i+1}-w_{p i}-\theta \lambda_{i}^{2} / 2,\quad (30)$$
+
+where the wage $w_{p l}$ is determined as
+
+$$w_{p i}=R+\lambda_{i}(1-\alpha) z .\quad (31)$$
+
+With equation (30) taking the place of equation (8) , the rest of the analysis from Section 3 continues to apply as stated. In particular, both the socially optimal transition point and the early-privatization point are determined using the same approach as before. A couple of new empirical implications are worth noting, however. First, from (29) , as the project moves closer to completion, the likelihood that the entrepreneur becomes informed and imposes her will on the scientist increases, because $\Pi_{t+1}$ goes up. It then follows from (31) that the scientist's wage also increases, to compensate for the fact that he has less de facto creative control. In other words, private-sector firms endogenously become more authoritarian — and less like academia — as research projects move into their later stages.
+
+Moreover, the model makes it clear that, even though we may observe some private-sector firms behaving in a less-than-fully authoritarian manner (e.g., the 3Ms and Googles of the world), it does not follow that there is no role for academia. Because the private-sector wage is set before the entrepreneur exerts effort to become informed, there is still a commitment problem in the private sector: although the probability of authority being exercised may be less than one, it can still be inefficiently high in the early stages of a research program. Thus, the precommitment associated with academia remains valuable.
+
+□ Low-powered incentives in academia. In modelling academia, we have assumed that scientists are completely free to follow their preferences, and face no incentives—either explicit
+
+(C) RAND 2008.
+
+---
+
+632 / THE RAND JOURNAL OF ECONOMICS
+
+or implicit — that might push them in the direction of the practical strategy. Although the tenure system can certainly be thought of as blunting the implicit incentives associated with career concerns (Holmstrom, 1999) , it is nevertheless hard to argue that tenured academics face no incentives whatsoever. For example, scientists can earn both professional prestige and monetary prizes if their work is highly cited, and citations in turn are likely to have some relationship (albeit a noisy one) to the underlying usefulness of the research.
+
+To introduce a meaningful role for incentives in academia, we generalize the model slightly, so that the disutility that a scientist experiences from following his less-favored strategy is now a random variable that can take on one of two values: $z_L$ with probability $\omega$ ; and $z_H > z_L$ with probability $(1-\omega)$ . The outcome of this random variable is independent of everything else in the model, and it has a mean of $\omega z_L + (1-\omega) z_H = z$ .
+
+Now suppose we want to design an incentive scheme that induces a scientist to follow the practical strategy when his disutility from doing so is $z_L$ , but not when it is $z_H$ . This scheme will have to have two properties. First, incentive compatibility requires that the scientist receive an expected bonus equal to $z_L$ (say in the form of expected prize winnings) whenever he follows the practical strategy, which happens with probability ( $\alpha + (1 - \alpha) \omega$ ). Second, the scientist's participation constraint requires that the ex ante expected wage be equal to at least $R + (1 - \alpha) \omega z_L$ , to compensate him for the $(1 - \alpha) \omega$ probability that he winds up following the practical strategy in a state of the world when it is not his favorite. $^11$
+
+Putting it all together, the ex ante expected wage bill for an academic scientist is now given by
+
+$$w_{a}=\max\{R+(1-\alpha)\omega z_{L},(\alpha+(1-\alpha)\omega)z_{L}\}.\quad (32)$$
+
+And as noted, with this incentive scheme the probability that an academic scientist pursues the practical strategy is now increased to $(\alpha + (1 - \alpha) \omega)$ . Everything in the private sector remains exactly as before: the wage $w_p = R + (1 - \alpha)z$ , and the scientist always follows the practical strategy.
+
+It is clear that if $z_L$ is close to zero, academia with this particular incentive scheme can be preferred to academia without incentives: the added wage cost is minimal, but there can be a meaningful increase (by an amount $(1-\alpha)\omega$ ) in the probability that the practical strategy is undertaken. At the same time, it can never make sense to try to use more powerful incentives to induce academic scientists to always follow the practical strategy, even when their aversion to it is strong (i.e., given by $z_H$ instead of $z_L$ ). This is because the ex ante cost of such higher-powered incentives would be
+
+$$w_{\alpha}=\max\{R+(1-\alpha)z,z_{H}\}\geq w_{p},\quad (33)$$
+
+with the inequality being strict if $z_H > R + (1-\alpha)z$ , which is the case so long as $R$ is not too large.
+
+In other words, if we are in the late stages of a research program, and it is important to always have scientists working on the practical strategy, this is more efficiently accomplished in the private sector, where they can simply be compelled to do so, rather than in academia, where this behavior has to be elicited by a system of high-powered incentives. $^12$ At the same time, the combination of an academic environment with some relatively low-powered incentives can be more efficient than the private sector in the early stages of a research program. The appeal of such
+
+11 It should be emphasized that we are effectively making the best possible case for incentives in academia, by allowing bonuses to be tied directly to strategy choice. In reality, academic incentives are likely to be more general in nature, and much less directly linked to the ultimate commercial value of a research project. In this sense, our basic model, which omits academic incentives entirely, may actually be closer to capturing the truth.
+
+12 Note that this same logic also implies that if $z$ is nonstochastic, there is no role for incentives in academia. Thus, we have effectively endogenized the no-incentives assumption that we made about academia in the course of developing the basic model.
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 633
+
+a combination is that it gently nudges those with only a mild aversion to the practical strategy in the right direction, while leaving creative independence to those who value it the most.
+
+The bottom line from this exercise is that our basic conclusions are robust to the introduction of some form of incentives in academia. At the same time, the model also suggests that to the extent that such incentives exist, it probably makes sense for them to be relatively low powered.
+
+## 6. Connection to the literature
+
+■ What is the role of academia in the innovation process? One answer is that because of knowledge spillovers and imperfect IPR protection, the value associated with certain kinds of ideas cannot be fully appropriated by the developers of these ideas, leading to private-sector underinvestment—hence the need for public funding of such “basic” research (Nelson, 1959; Arrow, 1962) . Although this story certainly has merit, it is becoming increasingly difficult to draw an unambiguous connection between the “basicness” of a line of research and the degree of appropriability of the resulting output. For example, Howitt (2000) mentions a National Science Foundation survey which finds that more than 22 % of all basic research in the United States during the period 1993–1997 was, according to the NSF's definition, performed by private enterprises. In our model, the relevant notion of basicness has nothing to do with appropriability, but rather corresponds to the number of stages remaining until a commercial payoff can be realized.
+
+Going beyond traditional appropriability arguments, Dasgupta and David (1994) take a broader and more institutional view of the role of academia. They emphasize that, as compared to the private sector, academia has a variety of distinctive rules, norms, and incentives that reward the production and rapid diffusion of knowledge. These include peer review, priority rules, and rewards based on the impact of publications (as measured, e.g., by citations).
+
+Our work can be thought of as fitting into the sort of institutional framework advocated by Dasgupta and David (1994) . However, we focus almost exclusively on a single institutional attribute of academia, namely the commitment that it embodies to allowing individual scientists to pursue their own preferred research strategies. In so doing, we largely set aside many of the other features highlighted by Dasgupta and David, including incentive schemes. $^13$
+
+By emphasizing the commitment role of academia, our theory implicitly offers a rationale for the tenure system, which has been prevalent for more than half a century, especially in researchoriented universities (McPherson and Schapiro, 1999) . Moreover, this rationale for tenure differs from that in Carmichael's (1988) well-known contribution: in Carmichael's model, only incumbent scientists are informed enough to evaluate potential new hires, and tenure ensures the incumbents that the new hires will not displace them.
+
+The focus on academia as a commitment device can also be found in the recent work of Lacetera (2005), developed contemporaneously with this article. Like we do, Lacetera adopts a control-rights perspective. However, unlike us, Lacetera does not model research as a multistage process. For our purposes, this multistage feature of the model is crucial, because it allows us to show that academia is most useful in the early stages of a research program, whereas the private sector tends to do better in the later stages. It also allows us to compare the socially optimal transition policy to early privatization, and to draw out the associated welfare implications.
+
+Another recent article which does model research as a multistage process is Hellmann and Perotti (2004) . They contrast the free flow of ideas in academia with the more controlled informational exchange that occurs in private firms. In particular, they model a commercially attractive new research program as consisting of two stages. The open exchange of ideas in academia maximizes the probability of completing the second stage, and therefore of innovating once the first stage has been successful. But this open structure also raises the risk of the first-stage idea being stolen. Hellmann and Perotti view the private firm as an institution that guards against
+
+15 In the spirit of Hart and Holmstrom (1987) , one might ask the following: if all that distinguishes academia from private firms are their respective incentive systems, why do we need two separate institutional entities to solve the underlying contracting problem, when a more sophisticated incentive scheme would presumably also do the job?
+
+© RAND 2008.
+
+---
+
+634 / THE RAND JOURNAL OF ECONOMICS
+
+such stealing, by carefully recording the property rights attached to first-stage ideas. Their article shares with ours the goal of endogenizing the choice of academic-versus private-sector research. However, instead of focusing as we do on control allocation, it emphasizes incentives to share information. And, in contrast to our analysis, it stresses the commitment powers of the private sector (in terms of its ability to restrict information flows) rather than those of academia (in terms of academic freedom).
+
+Finally, seeing organizations as differing in terms of the allocation of authority is of course not new when talking about private firms. A key element here is that individuals value creative control and are therefore ready to work at lower wages in return for more authority. This latter element is not entirely new either: Hart and Holmstrom (2002) stress it when comparing focused firms with conglomerates, arguing that the former are able to pay lower wages because of their greater commitment to pursuing the goals of their employees.
+
+## 7. Concluding remarks
+
+■ This article has provided a framework for evaluating the pros and cons of academic-as opposed to private-sector research. We have argued that even in a world where ideas can be sold to the private sector at all stages of the research process, academia — by virtue of its commitment to leaving creative control in the hands of scientists — can play a valuable role in fostering research projects that would not be viable entirely in the private sector. Moreover, we have shown that it is possible for ideas to be privatized sooner than is socially optimal, with negative consequences for the overall rate of innovation. This latter point echoes and clarifies some of the concerns raised in the policy literature about the potential for an “ anti-commons ” effect due to too-early privatization.
+
+In terms of directions for follow-up work, it might be interesting to study incentives more carefully in our framework. To take just one example, what are the pros and cons of citation-based rewards and promotions in a world where citations are only a noisy indicator of the value of a research contribution? Do such citation-based incentives help to focus scientists on the right kinds of projects, or do they simply tend to introduce inefficient fads or bandwagons into the research process?
+
+The model also has several empirical implications that might be worth investigating. Perhaps the most noteworthy comes from the branching version of the model, which predicts that a given idea will yield a more diverse array of offspring if it is located in academia rather than the private sector. We hope to see this hypothesis tested in future work.
+
+### Appendix
+
+■ Proof of Lemma 2. Let
+
+$$\begin{aligned}
+f(x) & =x-\left(w_{p} / \beta\right)\left(1+\log \left(\beta x / w_{p}\right)\right) \quad  if  \beta x / w_{p}>1 \\
+& =0 \quad  otherwise. 
+\end{aligned}\quad (A1)$$
+
+Similarly, let
+
+$$\begin{aligned}
+g(x) & =\alpha x-\left(w_{e} / \beta\right)\left(1+\log \left(\alpha \beta x / w_{a}\right)\right) \quad  if  \alpha \beta x / w_{e}>1 \\
+& =0 \quad  otherwise. 
+\end{aligned}\quad (A2)$$
+
+Because the value function $\Pi_i$ is increasing in $i$ , the proof of the lemma boils down to establishing that as $x$ increases, there is a unique point where the $f(x)$ and $g(x)$ functions cross (other than where they both take on the value zero). Under the assumption in the lemma that $\alpha z > R$ (or alternatively, that $\alpha/w_s > 1/w_p$ ), it can be seen that $g(x)$ first becomes positive at a lower value of $x$ than does $f(x)$ . So if we define the differential $\psi(x) = f(x) - g(x)$ , then for low values of $x$ , we have that $\psi(x) < 0$ . It is also clear that for sufficiently large values of $x$ , $\psi(x) > 0$ .
+
+To see that there is a single crossing point, that is, a single point where $\psi(x)=0$, note that
+
+$$\psi^{\nu}(x)=(w_{p}-w_{\alpha})/\beta x^{2}>0.\quad (A3)$$
+
+❾ RAND 2008.
+
+---
+
+AGHION, DEWATRIPONT, AND STEIN / 635
+
+In other words, the function $\psi(x)$ is convex. This implies that if it takes on negative values for small $x$ , and positive values for large $x$ , it has exactly one zero. (It must be increasing when it first crosses zero, and from this point on, convexity implies that it keeps increasing, and therefore never crosses zero again.) This establishes the lemma.
+
+## References
+
+Acemoglu, D., Kremer, M., and Mian, A. "Incentives in Markets, Firms and Governments." Working Paper no. 9802, NBER, 2003.
+
+AGHON, P., AND TIROLE, J. "Formal and Real Authority in Organizations." Journal of Political Economy, Vol. 105 (1997), pp. 1-29.
+
+ARROW, K. "Economics of Welfare and the Allocation of Resources for Invention." In R. Nelson, ed., The Rate and Direction of Inventive Activity. Princeton, NJ: Princeton University Press, 1962.
+
+CARMICHAEL, H.L. "Incentives in Academics: Why Is There Tenure?" Journal of Political Economy, Vol. 96 (1988), pp. 453-472.
+
+Dasgupta, P. and David, P. "Towards a New Economics of Science." Research Policy, Vol. 23 (1994), pp. 487521.
+
+GROSSMAN, S. AND HART, O. "The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration." Journal of Political Economy, Vol. 94 (1986), pp. 691-719.
+
+HART, O., ed. Firms, Contracts and Financial Structure. Oxford: Oxford University Press, 1995.
+
+——— and Holmstrom, B. "The Theory of Contracts." In T. Bewley, ed., Advances in Economic Theory, Cambridge, UK: Cambridge University Press, 1987.
+
+--AND --"A Theory of Firm Scope." Working Paper, Department of Economics, Harvard University, 2002.
+
+——— and Moore, J. "Property Rights and the Nature of the Firm." Journal of Political Economy, Vol. 98 (1990), pp. 1119-1158.
+
+——, Shleifer, A., and Vishny, R.W. "The Proper Scope of Government: Theory and an Application to Prisons." Quarterly Journal of Economics, Vol. 112 (1997), pp. 1126-1161.
+
+HELLER, M. AND EISENBERG, R. "Can Patents Deter Innovation? The Anticommons in Biomedical Research." Science, Vol. 280 (1998), pp. 698-701.
+
+HELLMANN, T. AND PEROTTI, E. "Circulation of Ideas: Firms versus Markets." Working Paper, Stanford University, 2004.
+
+HENDERSON, R., JAFFE, A., AND TRAJTENBERG, M. "Universities as a Source of Commercial Technology: A Detailed Analysis of University Patenting, 1965-1988." Review of Economics and Statistics, Vol. 80 (1998), pp. 119-127.
+
+HOLMSTROM, B. "Managerial Incentive Problems: A Dynamic Perspective." Review of Economic Studies, Vol. 66 (1999), pp. 169-182.
+
+HOWITT, P. "The Economics of Science and the Future of Universities." Paper presented at The 16th Timlin Lecture 2000, February 16, 2000, University of Saskatchewan, Saskatoon.
+
+KAPLAN, S., SENSOY, B., AND STROMBERG, P. "What Are Firms? Evolution from Birth to Public Companies." Working Paper, University of Chicago, 2006.
+
+LACETERA, N. "Different Missions and Commitment Power: An Institutional View of Industry-University Relations." Working Paper, MIT, 2005.
+
+LACH, S. AND SCHANKERMAN, M. "Incentives and Invention in Universities." Working Paper, London School of Economics, 2004.
+
+McPHERSON, M. AND SCHAPIRO, M. "Tenure Issues in Higher Education." Journal of Economic Perspectives, Vol. 13 (1999), pp. 85-98.
+
+MURRAY, F. AND STERN, S. "Do Formal Intellectual Property Rights Hinder the Free Flow of Scientific Knowledge? An Empirical Test of the Anti-Commons Hypothesis." Working Paper, Northwestern University, 2004.
+
+NELSON, R. "The Simple Economics of Basic Scientific Research." Journal of Political Economy, Vol. 67 (1959), pp. 297-306.
+
+STERN, S. "Do Scientists Pay to Be Scientists?" Management Science, Vol. 50 (2004), pp. 835-853.
+
+© RAND 2008.
+

processed/00012_W1999853213/document.md

@@ -0,0 +1,504 @@
+Journal of Economic Perspectives-Volume 5, Number 1-Winter 1991-Pages 3-27
+
+# An Introduction to the Law and Economics of Intellectual Property
+
+Stanley M. Besen and Leo J. Raskind
+
+Article I, Section 8, of the U.S. Constitution grants to the Congress the power: “To promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writing and discoveries.” Under this general grant, the Congress has enacted a number of statutes, including the Copyright Act [17 U.S.C.A. Sec. 101–810], the Patent Act [35 U.S.C.A. Sec. 1–376], and the Semiconductor Chip Protection Act of 1984 [17 U.S.C.A. Sec. 901–914]. In addition, the federal government has enacted the Trademark Act of 1946 (“Lanham Act”) as amended [15 U.S.C.A. Sec. 1051–1127] and there is state law regulation of trade secrets and of misappropriation of other information. These six legal regimes constitute U.S. intellectual property law. In addition, the United States has sought protection for the works of its authors and inventors in other countries by joining a number of international intellectual property conventions, including the Berne Convention for the Protection of Literary and Artistic Works, the Universal Copyright Convention, and the Paris Convention for the Protection of Industrial Property. Intellectual property issues have also begun to occupy a prominent place in discussions of the General Agreement on Tariffs and Trade.
+
+For as long as laws have aimed at protecting intellectual property, disputes have raged over which works to protect, for how long, and to what extent. The mainstream of the economics profession has generally argued that economic
+
+■ Stanley M. Besen is Senior Economist, The RAND Corporation, Washington, D.C. and Visiting Professor of Law and Economics, Georgetown University Law Center, Washington, D.C. Leo J. Raskind is Visiting Professor of Law, Brooklyn Law School, Brooklyn, New York, and Gray, Plant, Mooty, Mooty and Bennett Professor at University of Minnesota Law School, Minneapolis, Minnesota.
+
+---
+
+4 Journal of Economic Perspectives
+
+efficiency requires government support for innovative and creative activity (Arrow, 1962), but a dissenting tradition has argued that government action of any kind, including the awarding of copyrights and patents, is unnecessary to stimulate such activity (Plant, 1934a, 1934b; Breyer, 1970; Frase, 1966; Hughes, 1988).
+
+Although economists have written on topics of intellectual property for a long time, the impact of economics on public policy in this area has been slight, especially as compared to the influence of professional writings in areas such as antitrust and taxation. We believe that too few of the profession's resources have been devoted to these issues and that, of those resources that have been employed, too few have been devoted to empirical analyses. We hope that this introductory essay and the three papers that follow will stimulate interest in this subject. This introductory essay first describes some of the basic economic tradeoffs involved in intellectual property law, and then describes the framework of the law in the six areas described above: patent, copyright, semiconductor protection, trademark, trade secret, and misappropriation. It is intended both to provide thumbnail descriptions of the various intellectual property regimes to economists working in this area and to indicate where additional economic research might be useful.
+
+The other papers in this symposium provide important examples of ongoing research on the economics of intellectual property. Suzanne Scotchmer analyzes the complex effects of patent protection when innovation is cumulative. Rather than analyzing situations in which several firms vie to develop the same innovation—the approach of the “patent race” literature—her analysis examines circumstances in which only one firm can develop an initial innovation but others can also build upon it. She focuses on how the incentive to develop both the initial and subsequent inventions may be affected by the scope of patent protection.
+
+Janusz Ordover considers ways of adjusting the patent system that may help to both provide returns to the inventor, and encourage the diffusion of the innovation in the economy. His paper is part of a line of work that explores the place of the intellectual property system among the large number of institutions that affect the amount and nature of research and development that takes place.
+
+In the final paper, David Friedman, William Landes, and Richard Posner examine the law of trade secrets using what might be called the “Chicago school” approach to intellectual property protection. Taking the approach of previous work by Landes and Posner (1987, 1989) on trademark and copyright law, they argue that the availability of trade secret protection may efficiently fill some of the interstices left by the patent law and that the fact that trade secrets are protected against some types of discovery but not others may also be efficient. $^1$
+
+1 We should note here that, while we find many insights of the Friedman-Landes-Posner approach extremely useful, we are less sanguine than they about the efficiency properties of intellectual property law. Additional empirical analysis is needed to establish the validity of either view.
+
+---
+
+Stanley M. Besen and Leo J. Raskind 5
+
+## The Basic Economics of Intellectual Property
+
+The objective of intellectual property protection is to create incentives that maximize the difference between the value of the intellectual property that is created and used and the social cost of its creation, including the cost of administering the system. Several specific issues are subsumed under this general formulation.
+
+First, private producers have an incentive to invest in innovation only if they receive an appropriate return. Whether producers will have the correct incentives depends on their ability to appropriate at least some of the value that users place on those works. If potential innovators are limited in their ability to capture this value, they may not have enough incentive to invest a socially optimal amount in innovative activity.
+
+For example, producers may be unable to capture a sufficient portion of the value of their innovations if other producers can easily emulate or “ clone ” a new product (Arrow, 1962) . Similarly, if private copying by individual users is widespread, the revenues of creators of journals, computer software, and audio and videocassettes may be inadequate to support a socially optimal amount of creative activity (Besen and Kirby, 1989a) . The legal treatment of related innovations will also influence the ability of producers to capture sufficient revenues to justify undertaking research and development. If the law permits others easily to “ innovate around ” an innovation, or to produce complements to it, the incentives to create the innovation may be reduced significantly. $^2$
+
+Price discrimination allows producers to appropriate a larger share of the social benefits of their innovations and, thus, may permit some innovations that would otherwise not occur. Examples of discriminatory pricing include: charging different individual and library subscription rates for professional journals; establishing license fees for computer software that vary with the number of users; and setting prices for hardback and paperback versions of a book that do not reflect differences in their cost of production. It should be noted, however, that the “first sale” and “exhaustion” doctrines prevent copyright and patent owners from imposing resale conditions and thereby limit their ability to practice price discrimination.
+
+Second, there is the question of whether innovative activity takes place at minimum cost. The cost of creating new ideas will often depend on the extent to which innovators may borrow from, or build upon, earlier works. Copyright law, for example, limits borrowing by giving a creator the right not only to his or her own creation but also to “derivative works,” so that the costs to subsequent innovators may be increased. In a winner-take-all system like that
+
+2Spence (1976) and Dixit and Stiglitz (1977) pointed out the possibility that an innovation that is only slightly more attractive to users may “ cannibalize ” the market for an earlier product. In such cases, it can be profitable for a firm to incur the costs of developing the new product and bringing it to market, although total industry profits decline by more than consumer welfare is increased. As a result, total welfare, but not the welfare of consumers, would be increased by making it more difficult to produce close substitutes for existing products.
+
+---
+
+6 Journal of Economic Perspectives
+
+governing patents, competition to get the patent (and thus control over future innovations based on that patent) may result in an excessive amount of resources being devoted to obtaining the prize. In fact, the combined expenditures of two firms seeking the same patentable invention in a patent race may not only be larger than that of a single firm, but their combined expenditures may be greater than is socially optimal (Loury, 1979; Dasgupta and Stiglitz, 1980).
+
+Other provisions of intellectual property law can lower the costs of subsequent innovations. Copyright registration and patent issuance require authors and innovators to disclose the details of their innovations, which provides information that may help later authors and innovators to reduce their own costs. In addition, the copyright law denies protection to “any idea, procedure, process, system, method of operation, concept, principle, or discovery” and patent protection cannot be obtained for “laws of nature, natural phenomena, and abstract ideas,” no matter how great the expense necessary to bring forth these creations. If the market for licensing innovations were frictionless, there would be no need to impose these restrictions since later innovators could efficiently obtain licenses to use the works of early ones. However, the lack of complete information on which to base these transactions and the market power that might accrue to early inventors may explain the restrictions that have been imposed on protectible subject matter. These types of imperfections are addressed by Scotchmer in this symposium.
+
+A third issue, somewhat related to the second, is whether the intellectual property system strikes an appropriate balance between creating and disseminating intellectual property. Providing incentives for the creation of many new works may encourage resources to be devoted to innovative activity. However, if the new innovations are not widely used, the system may be less beneficial than one with less creativity, but where the materials created are more broadly disseminated. This issue focuses on the appropriate scope of protection. It arises, for example, in determining the optimal duration of patents (Nordhaus, 1969) and the optimal tradeoff between duration and breadth (Gilbert and Shapiro, 1990; Klemperer, 1990) . Another critical element in deciding how to strike the balance between encouraging creativity and dissemination is the extent to which creative activity responds to economic rewards. The less that innovation depends on the resources invested and the potential economic rewards, the more limited is the case for granting substantial rights to creators.
+
+## Patents
+
+A patent may be granted on any new and useful process, machine, manufacture, composition of matter, improvement and plant as well as to new, original and ornamental design for an article of manufacture (Chisum, 1989). The patent right is the most powerful in the intellectual property system,
+
+---
+
+Introduction to the Law and Economics of Intellectual Property 7
+
+enabling the patent holder (patentee) to exclude all others from making, selling, or using the subject matter of a valid patent for a term of 17 years (in the case of a design patent, 14 years). The 17-year term of pharmaceutical and medical device patents may be extended for as long as five additional years. During the term of the patent, any use of the patented subject matter requires permission of the patentee, customarily in return for the payment of a royalty. The patentee can even prevent an independent subsequent discoverer of the same subject matter from making, using or selling it. At the end of the term of protection, the subject matter enters the public domain.
+
+The scope of protection offered by a patent is determined by its claims, which are technical descriptions of the process, machine, method, or matter contained in the original patent application. U.S. patent law follows the principle that the first to invent has prior claim to the invention. In all other countries, the “first to file” rule determines the patentee.
+
+The patent statute requires that the claimed invention must be new, useful, and nonobvious to a person of ordinary skill in the art to which the invention pertains. The practical effects of the novelty and nonobviousness requirements are that the inventor must convince the examiners in the Patent and Trademark Office that the claims in the application make a new contribution to knowledge and are more than a mere variation of something already known or foreseeable as an extension of existing knowledge. Examiners have the power to determine that a claim is too broad, and to grant a patent only on a narrower claim, or to reject a claim entirely.
+
+## When Patents Are Disputed in Court
+
+A patent holder who believes that a patent has been infringed may bring suit against those who make, use, or sell the offending product. Courts may either choose to interpret the claims of the patent literally, or the “doctrine of equivalents” may be invoked and infringement will be found if there is a substantial, functional identity between the patent claims and the contested item. In fact, there has been a debate as to whether a patent effectively covers more than the literal subject matter of the claim by including the prospective technology inherent in the earlier patent. Kitch (1977) argues that the “prospect” theory of patent interpretation is predominant; Beck (1983) contends that the evidence does not support that conclusion.
+
+The supposed patent infringer may defend against the charge in four principal ways. First, the validity of the patent may be challenged as having been improvidently issued; that is, lacking the requisite requirements of novelty and nonobviousness. Second, a patent may be invalidated for fraudulent conduct; for example, by misrepresenting the prior art in the patent application. Third, the patent is invalid if the invention was patented or described in a printed publication here or abroad, or was in public use or on sale in the United States more than one year prior to the date of the patent application. Finally, there is the defense of patent misuse, a doctrine limiting the use of the
+
+---
+
+8 Journal of Economic Perspectives
+
+patent beyond its statutory scope (Kaplow, 1984). An allegation of infringement may be defeated under this doctrine if the license to use contains a condition that the licensee must purchase another product from the patent holder, that is, if a tying arrangement is created. However, the doctrine cannot be invoked unless the patent holder has market power. $^3$
+
+There are three basic remedies available to a patentee in an infringement suit. First, the court may issue an injunction barring further use of the infringing device, composition, method, process, or improvement. Second, the court may award damages equal to a reasonable royalty or lost profits. If the court deems a reasonably royalty inadequate in a particular case, it may treble the proven damages. Third, the court may award costs and attorney's fees.
+
+Patent litigation is distinctive in the federal court system in that all patent appeals from U.S. District Courts and from the Patent and Trademark Office are heard only in the Court of Appeals for the Federal Circuit. The Federal Circuit was organized in 1982 to provide greater expertise and uniformity in the interpretation of patent law and to achieve greater efficiency in case management. An interim review of the performance of this specialized court notes that the interpretation of patent law is becoming more precise and coherent, patentees are successful more frequently against challenges of validity, and the expansion of injunctive relief and modified methods of damage computation have made infringement more costly (Dreyfuss, 1989) .
+
+## Research and Policy Issues About Patents
+
+The patent system was perceived by the framers of the Constitution as a system of incentives and rewards. The patent offers the incentive of the statutory right to exclude as a means of inducing creative activity. The right, in turn, permits the holder to obtain the reward by having the legal power either to sell the right for the payment of a royalty or to retain the exclusive exploitation rights (Katz and Shapiro, 1985b). The right of the patent owner is conditioned on the disclosure of the subject matter to the public when the patent is issued.
+
+Although economic literature about the patent system is substantial, many questions are still heavily disputed. For example, there is no consensus as to the impact of patent protection on the growth of technology (Kitch, 1986); or on the optimal duration of the patent right (McFetridge and Rafiquzzaman, 1986);
+
+3 Patent and Trademark Authorization Act, approved November 19, 1988. Pending legislation, H.R. 469, 101st Cong., 2d Sess. (1990), would restrict the doctrine of patent misuse further by overruling Supreme Court antitrust cases that presume market power from the existence of a patent or copyright. In hearings on February 8, 1990, this legislation was supported by representatives of the computer industry who stated that the presumption impeded owners of intellectual property rights from appropriating the return on their creative efforts. Opponents of the legislation stated that the existing caselaw was needed to bar illegal tying arrangements in the computer and broadcasting industries. BNA, Patent, Trademark & Copyright Journal, Volume 39, No. 968, p. 289, February 15, 1990.
+
+---
+
+Stanley M. Besen and Leo J. Raskind 9
+
+and the data on whether patents have been used to facilitate cartel behavior is inconclusive (Hall, 1986).
+
+The current controversy over of the role of patent protection in the computer software industry suggests the practical importance of patent protection on a developing industry. Most of the early computer programs were marketed without patent protection. When Visicalc was developed in 1979, the Patent Office, relying upon Supreme Court case law, took the position that the mathematical algorithms in computer programs were not protectible subject matter. Subsequently, the Supreme Court in the Diehr case did find patentable subject matter in a process utilizing a computer algorithm. 4 Since that case, the Patent Office has begun to grant patents on computer programs.
+
+The uncertainty as to the scope of patent protection of computer software may limit the attraction of the patent regime for it. Recently, Refac Technology, a firm producing no software, acquired a 5 percent ownership interest in many software patents relating to spreadsheet programs in exchange for its promise to bring infringement actions. Refac was to receive a portion of all resulting royalties. It then filed some 2000 patent infringements suits against large and small software companies for current and retroactive royalties. Recently, in the first of these suits, Refac was dismissed on the legal technicality that Refac's slight ownership interest in the patent was void for having been acquired solely for the purpose of bringing abusive litigation. The court noted, however, that the actual patent owner could bring the suit (131 Federal Rules Decisions 56, May 29, 1990). This case has had the unsettling effect of causing software producers to undertake costly and time-consuming patent searches before marketing new products, an effort made difficult by the inadequacy of the present system of software classification by the Patent and Trademark Office. The suitability of patent protection of software in comparison to copyright protection, remains an unsettled question (Mennell, 1989; Sumner and Lundberg, 1989).
+
+Since computer programs are functional statements constructed to solve problems, they pose the issue of whether algorithms are patentable. Reports in 1988 and 1989 that the Patent and Trademark Office had adopted a more liberal approach to software patent applications, which might result in granting patent protection to fundamental building blocks of research, led the Office to publish a statement of guidelines to clarify its position (Official Gazette, September 5, 1989). The statement concluded that mathematical algorithms
+
+4The original line of Supreme Court precedents included Gottschalk v. Benson, 409 U.S. 63 (1972); Parker v. Flock, 437 U.S. 584 (1978). The revised decision was Diamond v. Diehr, 450 U.S. 175, 185-187 (1981). The Court held: "Excluded from ... patent protection are laws of nature, natural phenomena, and abstract ideas ... [A]n algorithm ... is like a law of nature, which cannot be the subject of a patent ... [Here] ... patent protection ... [is sought] for a process of curing synthetic rubber. Their process admittedly employs a well-known mathematical equation, but they do not seek to preempt the use of that equation. Rather, they seek only to foreclose from others the use of that equation in conjunction with all of the other steps in their claimed process."
+
+---
+
+10 Journal of Economic Perspectives
+
+are, as such, unpatentable, but that applications of such algorithms may be protectible as new processes.
+
+Two recent decisions by the Court of Appeals, Federal Circuit, have further unsettled this issue. The court recently affirmed a Patent and Trademark Office rejection of a patent application for a computer program that diagnosed the condition of a patient by aggregating a plurality of clinical laboratory tests ( In re Grams , 12 USPQ 2d 1824 [1989] ). However, even as the court held that the claim was no more than the application of a mathematical algorithm to data, it also noted that the presence of a mathematical formula as a step in a process involving mathematical steps could permit patent protection. In a later decision, In re Iwahashi (12 USPQ 2d 1908 [1989] ), the court reversed the Patent and Trademark Office for refusing to allow a patent on a program used to calculate the correlation of signals for recognition of voice patterns. Where the Office had considered the program no more than a more efficient means of calculation of correlations and, thus, unpatentable as an algorithm, the court interpreted the claim to involve the description of an apparatus in the form of a series of interrelated means. It held that such an apparatus may qualify for patent protection.
+
+These apparently contradictory decisions create considerable uncertainty and have raised fears among industry participants that overprotection of algorithms may stifle innovation by raising the costs of subsequent innovation, an example of the issue analyzed by Scotchmer in this symposium. (See 39 Pat. Trademk. & Copyr. J. 369 [1990].) To the extent that patent protection is utilized, perhaps increased federal expenditures are needed to improve the classification system and accelerate the examination process. Another alternative is to permit a patent application to be made public after, say, one year, even if the patent examination process is not completed, as is done in Britain and some other countries. This step would at least help to avoid unintentional infringements of patents that are still working their way through the examination process, although it would not deal with the “overprotection” issue.
+
+Another public policy issue that invites economic analysis is the recent extension of the term of patent protection for pharmaceuticals. The brand-name producers sought the extension because the delay in marketing patented products caused by the required review and approval of the Food and Drug Administration (FDA) encroached on the effective term of patent protection, making it more difficult to recoup their R & D investment. The makers of generic drugs, on the other hand, objected to the long FDA review and approval period required for modifications of compounds already approved under brand names, arguing that this duplicative review process inhibited their ability to compete with brand-name drugs. Economic analysis of this industry would illuminate the welfare effects of the resulting legislative compromise, which extended the duration of the patent monopoly while facilitating the subsequent entry of generic drugs (Wheaton, 1986) .
+
+---
+
+Introduction to the Law and Economics of Intellectual Property 11
+
+Patent protection is also a current policy concern as an aspect of international trade (Mossinghoff, 1984) . The current GATT negotiating round is seeking agreement on patent norms and standards (Braga, 1989) . As technological innovation is linked to concerns about the balance of payments and competitiveness, economic analysis should be extended to the international as well as the domestic aspects of patent protection.
+
+## Copyright
+
+Although works can be registered and deposited at the Copyright Office, the rights under the copyright statute exist independently of any such registration, prior examination, or other formalities (Nimmer and Nimmer, 1989; Goldstein, 1989) . Instead, copyright protection attaches to the stated subject matter when an “original work of authorship” is “fixed in any tangible medium of expression.” For example, protection attaches to the pages of a novel when the word processor causes the word order to be printed on paper, meeting the “fixation” requirement. The words on the cathode ray tube are considered an unprotectible evanescent image.
+
+Copyright law gives protection in most circumstances for the life of the creator plus 50 years. One exception is that a work prepared and published by the U.S. government, like the annual Economic Report of the President , cannot receive copyright protection. Presumably, the basis of this limitation is to encourage the dissemination of government materials to the electorate. Case law has extended this principle to preclude copyright in state laws, municipal codes, legislative hearings, and judicial opinions.
+
+Because copyright law has no parallel to the patent system's process of application and definition of claims, the scope of copyright protection is ultimately defined by litigation. Litigation involving the validity of a patent involves a review of the file of prior art and competing claims prepared by the Patent and Trademark Office. By contrast, the Copyright Office makes no independent review of the article or the circumstances of its creation, there is no official file, and the copyright owner must produce the evidence to support the validity of the copyright.
+
+For copyright, it is only necessary that the work originated with the claiming author; therefore, unlike patents, more than one valid copyright can be held on identical works. The traditional hypothetical example is that there could be multiple valid copyrights to identical versions of Keats's “Ode on a Grecian Urn,” as long as the each author produced the word order of each poem independently.
+
+Technically, a work is eligible for copyright protection if it is within one of five statutory categories, is original in the sense that it was neither already in
+
+---
+
+12 Journal of Economic Perspectives
+
+the public domain nor copied from another, and represents a modicum of intellectual activity. As already described, patents attach to the application of an idea in the form of a machine, method, or matter. In contrast, copyright is said to attach only to the expression, as distinguished from the idea.
+
+The statute also requires “authorship.” Thus, if a chimpanzee were to manipulate the keyboard of a personal computer with a graphics program and produce an attractive design, protection would not exist because the modicum of intellectual activity to support copyright is deemed lacking. As Congress has extended the subject matter of copyright from books and maps to photographs, sound recordings, motion pictures, computer programs, and so on, courts have had to determine the relevant “authorship” for each new subject (Raskind, 1990).
+
+Copyright law defines its subject matter more specifically than does the patent statute. Copyright protection is extended to literary works, musical works (including accompanying words), dramatic works (including accompanying music), pantomimes and choreographic works, pictorial, graphic and sculptural works, motion pictures and other audiovisual works, and sound recordings. The rights accorded to a copyright owner are also tightly defined. Copyright protection grants five basic rights: 1) the right to reproduce the protected work; 2) the right to prepare derivative works from the protected work; 3) the right to distribute copies; 4) the right to perform literary, musical, dramatic, choreographic works publicly, as well as pantomimes, motion pictures, and other audiovisual works; and 5) the right publicly to display literary, musical, dramatic, choreographic, pantomimes, and pictorial, graphic, and sculptural works, including the individual frame of a motion picture and other audiovisual works. Notice that sound recordings are omitted from the listed properties having the performance right, although the composer and publisher of the recorded work do have the right.
+
+In 1980, Congress expressly extended copyright protection to computer programs as literary works, defining a program as “a set of statements or instructions used directly or indirectly in a computer in order to bring about a certain result.” At the same time it provided that the owner of a copy of a copyrighted computer program could lawfully make or authorize the making of another copy or adaptation of that program, provided that either: 1) the new copy is created only as an essential step in utilization of that program; or 2) the new copy is for archival purposes only. The maker of the archival copy is required to destroy it if the maker's legal right to possession of the program ends. On any transfer or exchange of the underlying copyright program, the archival copy may be transferred only with the permission of the copyright owner of the underlying program.
+
+Extending protection to computer programs requires modifying the interpretation of the basic copyright requirement of fixation. Accordingly, cases hold that although the image generated by the program on the cathode ray screen is evanescent, the fixation requirement for a computer program is met when the
+
+---
+
+Stanley M. Besen and Leo J. Raskind 13
+
+source code is written on paper or when the object code or microcode is fixed in a computer chip.
+
+Remedies granted the copyright owner are more powerful than those provided under any other regime of intellectual property protection. Injunctions are liberally granted in copyright cases. Monetary damages measured by lost profits or the profits of the infringer may be awarded. If willful infringement is shown, the court may award statutory damages up to $50,000. In addition, infringing material along with the equipment used to produce the infringing copies may be impounded. As part of final judgment, a court may order sale or other disposition of impounded material or may order its destruction. Copyright infringement also is a misdemeanor under the federal criminal statute which carries a sanction of a fine of up to $10,000 or imprisonment for up to one year.
+
+Because copyright has its origins in the protection of books and other printed expression, it is sometimes mistakenly assumed that patent protection exclusively covers industrial products, while copyright is limited to products with a high content of creativity that address the intellect and the aesthetic sense. However, case law has extended copyright protection to a large group of commonplace functional articles, from lamp bases to ash trays and can openers, items that have nothing in common with the fine arts.
+
+For some products there is a choice between patent and copyright protection. A computer program producer may seek either copyright or patent protection at the outset. Exercising one alternative may restrict the other. If copyright is selected first and the program is registered and sold in the United States, access to foreign patent protection is lost and unless a U.S. patent application is filed within twelve months of the initial sale, patent protection here is also barred. If patent protection is chosen first, a copyright notice may also be attached. Similarly, protection of some commonplace functional articles may be sought either under the design patent regime or under copyright, but the Copyright Office takes the position that material protected by a design patent is ineligible for copyright protection.
+
+## The Limits of Copyright Protection
+
+Copyright also expressly qualifies and limits the granted rights in three general ways. First, the more modest copyright threshold of “originality” means only that the work has not been copied. Thus, independent creation of an identical work is a defense against a claim of copyright infringement.
+
+Second, the scope of protection varies with the creative content of the material. For example, copyright in a fact work (like a news story or a database) does not extend to the names, places, and events, but only to the expression in the writer's interpretation. However, all of the expression in a poem or a novel is protected by copyright.
+
+Third, reproduction of a copyrighted work is not infringing if the copying can be shown to be for a “fair use,” such as literary criticism, making a parody,
+
+---
+
+14 Journal of Economic Perspectives
+
+or classroom teaching. Moreover, the reproduction must not adversely affect the present or potential economic interest of the owner of the protected work. An example of a situation in which there is no adverse effect would be the reprinting an out-of-print work (Gordon, 1982; Raskind, 1984) .
+
+In the recent revision of the Copyright Act, an impasse developed between the higher education establishment and representatives of the publishers over a statutory provision permitting photocopying by classroom teachers. The matter was resolved by providing guidelines that undertake to define the circumstances where multiple copying by classroom teachers does not constitute infringement. The Guidelines were intended to describe a clearly permissible “safe harbor” amount of multiple copying for classroom use for which permission is not required. They were not intended to state a maximum. For copying beyond the Guidelines, courts are to decide fair use in the classroom setting on a case by case basis. There is pending litigation involving a group of major publishers and a prominent commercial copying firm which may provide the first interpretation of the scope of the Guidelines.
+
+It is generally agreed that a scholar may make one copy of protected material for study. Under a separate statutory provision, a library may make that single copy on behalf of a patron as well as for preservation, archival, or interlibrary loan purposes. This provision restricts the library to “isolated” and “unrelated” reproduction of the single copy. If the library persons become “aware or have substantial reason to believe” that they are participating in “concerted reproduction or distribution” of multiple copies, they become liable for infringement, unless the fair use provision applies. A library is further restricted from making a copy (except for archival or preservation purposes) of a musical work, a pictorial, graphic, or sculptural work, a motion picture or other audiovisual work. However, a library may make a copy of an audiovisual work dealing with news.
+
+## Transfer of Copyrights
+
+Copyright provides a legal framework governing market relations. An owner of a copyright is empowered to grant a license to reproduce, distribute, perform, or display the copyrighted work and to obtain a royalty for granting the right. In most circumstances, the outcomes are uncomplicated. A textbook author obtains the copyright in the manuscript by completing it, and the author can then transfer the copyright to the publisher in exchange for royalties. In this way, the copyright statute provides the framework for private contractual bargaining and transfer of economic interests.
+
+Although the copyright law is generally silent as to the terms on which private parties may strike bargains, there are some exceptions. For example, the Copyright Act provides for compulsory licenses with statutorily determined license fees for some uses, like the retransmission of broadcast signals by cable television systems (Besen, Manning, and Mitchell, 1978) . Moreover, a number of European countries have enacted compulsory licenses for off-air taping,
+
+---
+
+Introduction to the Law and Economics of Intellectual Property 15
+
+other forms of home recording, and photocopying, where the license fees are collected through levies on blank recording machinery or media. $^5$ Copyright owners have unsuccessfully attempted to have similar levies imposed in the United States to compensate for losses from home taping.
+
+In cases where the value of a given piece of copyrighted material is small relative to the transaction costs of licensing, copyright owners may employ collective administration of the licensing process (Besen and Kirby, 1989b) . For example, the American Society of Composers, Authors and Publishers (ASCAP) and Broadcast Music Incorporated (BMI) issue blanket licenses to radio stations to play recorded music and to live entertainers to perform individual songs in public, with the proceeds of the license fees distributed among their members. Since collective administration creates the potential for anticompetitive behavior, ASCAP and BMI have both been the subjects of government and private antitrust suits. Both now operate under court-administered consent decrees in which the government suits were settled in return for changes in licensing procedures.
+
+Similarly, the Copyright Clearance Center is an organization that centralizes photocopying authorization and royalty collection for literary material. Founded in 1977 as a non-profit entity, the Center obtains permission from copyright owners to grant licenses permitting users to photocopy registered titles at fees set by the owners and indicated by a notation printed on the materials; for example, 50 cents per page. The Center may issue a license on a transaction basis and require detailed records of copying or it may issue a blanket license based on a sample audit period. The royalties paid by users, mainly libraries and corporations, are distributed to the copyright owners. Several lawsuits by publishers against photocopiers have been settled following the agreement of the user to register with the Center.
+
+A work made “ for hire ” is another important instance in which the transfer of copyright is governed by different rules. If a work is determined to be one made “ for hire, ” the employer takes the copyright in the material prepared by the employee. Then the duration of protection of a work becomes the lesser of 75 years after publication or 100 years from the date of creation (Hardy, 1988). $^6$ Industry practices vary with regard to the uses of the “ work for hire ” provisions, yet there have been no economic studies of these practices. However, it appears that in cases when a commissioned work is to be treated as a work for hire, courts sometimes make assumptions about relative bargaining power and prevailing rates and fees that are unsupported by any evidence or analysis.
+
+5Besen and Kirby (1989a), Johnson (1985), Liebowitz (1985), and Novos and Waldman (1984) all contain analyses of the effects of private copying. For a model of library borrowing that deals with many of the same issues, see Ordover and Willig (1978).
+
+b A recent Supreme Court opinion held that in the case of a commissioned work, the author presumptively is entitled to the right, absent an express written designation of the commissioning person as the owner of the copyright. Community for Creative Non-Violence v. Reid, 109 S. Ct. 2166 (1989).
+
+---
+
+16 Journal of Economic Perspectives
+
+## Copyright and Incentives for Innovation
+
+By expanding and restricting the protection given to copyright, courts alter the incentives accorded to a potential innovator. The development of new products may be impeded by a legal rule that overprotects existing works.
+
+Among the basic rights granted the copyright owner is the right to prepare a derivative work, defined as a work based on one or more preexisting works. Others must obtain permission from the owner to prepare a derivative work from copyrighted material. The statute gives as examples of derivative works a translation, musical arrangement, and motion picture version. There may be more than one tier of derivative works as, for example, when a three-dimensional toy or a T-shirt picture is made of a character in a motion picture version of a novel. The independent preparer of a derivative work obtains no rights in the preexisting material.
+
+A substantial legal literature has probed where the line of protection should be drawn for derivative works. One view is that creators of initial or underlying works should have broad protection to have the maximum incentive to produce. On this view, infringement occurs if there is substantial similarity between the underlying work and a recast, transformed version. In addition, Landes and Posner (1989) have argued that granting control over derivative works to creators encourages early release of the underlying work since otherwise creators might delay release while they prepare derivatives to gain a marketing advantage over other producers. The other view would give narrower protection to the original work and accord the producer of the derivative work a zone of freedom from infringement as an incentive to produce new variations. Here, the theory is that the societal interest is better served by providing the incentive for others to create derivative works.
+
+Economic analysis would illuminate the social choices. In one recent case, a manufacturer of framed pictures purchased copyrighted volumes of an art history book, clipped the photographed illustrations, and mounted them for sale as framed pictures. In Mirage Editions, Inc., v. Albuquerque A.R.T. Co., (856 F.2d 1341 [9th Cir. 1988]), the court protected the copyright owner of the book, by holding that the framed pictures infringed the copyright. The owner of the book apparently never saw the market opportunity for framed pictures during the half dozen years the book was being sold. The person who saw that opportunity and who paid the asking price for the protected book was effectively denied any economic reward for entrepreneurship.
+
+A similar example with more serious technological implications involves the copyright provision (Sec. 117 ) that restricts the copying of computer software to archival uses. $^7$ Restricting duplication solely to archival uses makes it an act of copyright infringement for a programmer, whether independent
+
+7In the case of Vault Corp. v. Quaid Software, Ltd., 847 F.2d 255 (5th Cir., 1988), the court gave a broader interpretation, permitting destruction of the manufacturer's anticopying code to make a copy for archival purposes.
+
+---
+
+Stanley M. Besen and Leo J. Raskind 17
+
+contractor or employee, to copy a protected program for the purpose of modifying it to the particular needs of a user. Some consider this an undue restriction on the role of programmers as well as users (Stern, 1985) . Making it illegal to copy for innovative uses, such as analysis and study for the purpose of improving routines and enhancing functions, serves to inhibit and restrict an important activity associated with the advance of this new technology (Braunstein, Fischer, Ordover, and Baumol, 1977) , and thus raises the cost of subsequent innovation.
+
+The legal literature has tended to focus on restricting the protection of the underlying work. Economic analysis could contribute here by analyzing the impact of changing the relative costs and returns of initial creators and subsequent derivative work producers.
+
+## An Illustration: Computer Operating Systems and Interface Standards
+
+The issue of how to protect computer operating systems and screen displays provides an example of extending traditional copyright principles to new technology. These computer interfaces are of increasing importance to the industry, because of their potential for standardizing menu displays, keystroke usage, and operating systems. The Copyright Office has announced as its position that all the copyrightable expression embodied in a computer program, including screen displays, is to be considered as a single work (53 Federal Register, 21817-20, June 10, 1988). By reversing its earlier position of accepting screen displays for separate registration and deposit, the Office has placed itself at odds with the courts $^8$ and with some commentators (Menell, 1989).
+
+Computer users may benefit from the existence of standardized interfaces in a number of ways (Farrell and Saloner, 1985; Katz and Shapiro, 1985a) . The greater the degree of standardization, the larger is the array of complementary inputs (software, repair services, and the like) available to users, and the easier it is to switch from one system to another. These forces also create a tendency for only a small number of standardized features of interfaces to exist at any one time, and make the introduction of new interfaces more costly and difficult (Farrell, 1989) .
+
+Much litigation has occurred and is pending over the copyright protection of both computer operating systems and computer screen displays. However, this litigation does not seem likely to refine the analysis of visual displays. In Apple's suits against Microsoft and Hewlett-Packard, for example, the court found that the alleged infringing use of Apple's visual displays was covered by Apple's software license to the defendant (Apple Computer, Inc. v. Microsoft Corp., 717 F. Supp. 1428 [N.D. Cal. 1989]). Xerox's suit against Apple is
+
+8For example, see Manufacturers Technologies, Inc. v. Cams, Inc., 704 F. Supp. 984 (D. Conn. 1989); Digital Communications Associates v. Softklone Distributing Corp., 659 F. Supp. 449 (N.D. Ga. 1987).
+
+---
+
+18 Journal of Economic Perspectives
+
+pending a further hearing, but it is unlikely to require analysis of the scope of protection since Xerox does not allege infringement, but rather seeks to have Apple's copyright declared invalid, a novel copyright remedy (Xerox Corp. v. Apple Computer, Inc., D.C. N.Cal. CCH Copy. Rep. Par. 26, 556). A recent decision extending copyright protection to the menu command structure of a popular spreadsheet program has rekindled the controversy over the protection of screen displays (Lotus Development Corp. v. Paperback Software Intn'l., 740 F. Supp. 37 [D. Mass. 1990]). Most troublesome is the protection of the input command words (such as Move, Copy, Print) and the court's rejection of the defendant's argument that these familiar components of a screen display should be considered an unprotectible, de facto industry standard.
+
+Proponents of independent protection of screen displays have argued that substantial expenses are involved in developing interfaces and that without legal protection, too few resources will be devoted to this activity (Clapes, Lynch, and Steinberg, 1987) . In response, it has been argued that many standardized interfaces result from arbitrary choices among a number of equally good and widely-known alternatives, or conversely, that there may be only a single way to accomplish a given objective. In either case, providing intellectual property protection would grant considerable market power to the owner of the right to control the standardized interface. For this reason, Menell (1989) has argued that courts should give only “ thin ” copyright protection to screen displays by finding infringement only on a showing of actual copying.
+
+An alternative approach is to require the developer of the interface to seek protection under the patent system. Since copyright protection does not require novelty or non-obviousness, using copyright to protect computer interfaces might cause insignificant advances to receive a high degree of protection, with risks that widespread use of innovations in this industry will be delayed.
+
+Strong intellectual property protection might also create the possibility for patent or copyright races, as discussed earlier. Because of the potentially large returns to winning in the race to develop a broadly used interface standard, an excessive amount of resources may be devoted to this activity. $^9$ If protection were more limited, this effect would be attenuated. Alternatively, the degree of protection could be limited by requiring that compulsory licenses be made available on reasonable terms. Indeed, some standards organizations have been able to impose such licensing requirements as a condition for selecting a particular technology as a standard, and user groups have succeeded in lowering royalty fees by threatening not to support a standard. IBM and Unisys were recently forced to reduce their proposed fees for licensing patents needed for the manufacture of modems that conformed to a new standard (Lefton, 1990).
+
+9 There is probably less need to be concerned about resources spent in efforts to develop trivial variation on an existing standard. Even if those variations were to receive protection under the copyright law, trivial changes would be less likely to dominate the marketplace.
+
+---
+
+Introduction to the Law and Economics of Intellectual Property 19
+
+## Semiconductor Chip Protection
+
+The semiconductor chip has become the principal building block of contemporary electronics technology. By the mid-1970s, U.S. and European chips were being copied and distributed by foreign competitors (Stern, 1986). In 1978, the industry appealed to Congress for protection and hearings were begun.
+
+The resulting Semiconductor Chip Act of 1984 is the newest addition to the intellectual property system. The Chip Act is sometimes erroneously described as a piece of special or sui generis legislation, implying that the special nature of the technology made it ineligible for copyright protection. It is more accurately characterized as legislation within the copyright regime, a copyright-like statute that is fine-tuned to reflect the particular needs of this industry. 10 Thus it provides a 10-year term, shorter than either patent or copyright, codifies the industry practice of reverse engineering, and provides for more limited remedies.
+
+The two subjects eligible for protection under the Chip Act are the semiconductor chip products and the mask work. The first are integrated circuits containing transistors, resistors, capacitors and their interconnections, fabricated into a tiny, single piece of semiconductor material. A mask work is defined as a set of images, however fixed or encoded at a later stage of manufacturing, that produces the circuitry of the final chip product; essentially, these are the blueprints of the chip. For domestic producers, protection attaches on fixation and commercial exploitation. Protection for foreign products can be granted by the president upon a finding that a foreign nation extends to U.S. nationals the same protection as the United States accords to the foreign nationals. The obvious aim of this provision is to induce other countries to enact chip protection measures.
+
+As in traditional copyright, the Chip Act imposes a condition of “originality” for protection to attach, but the Chip Act standard is expressly set somewhat higher. The Chip Act states that a mask work design is not original if it is staple, commonplace, or familiar in the semiconductor industry. Although a mask work is eligible for protection upon fixation, the right may be forfeited by failure to register within two years after the first commercial exploitation. Clearly, this is unlike traditional copyright law.
+
+Two basic rights are given the owner of a protected mask work: the right to bar reproduction of the mask work by optical, electronic, or any other means; and the right to import or distribute a semiconductor chip product in which the mask work is embodied. Accordingly, importing a pirated chip is an infringement of the owner's importation right.
+
+10Statement, Leo J. Raskind, testimony before the House Subcommittee on Courts, Intellectual Property, and the Administration of Justice, Oversight Hearings on Computers and Intellectual Property, p. 3, November 8, 1989.
+
+---
+
+20 Journal of Economic Perspectives
+
+Reverse engineering, like fair use in copyright law, is a defense to a claim of infringement. Reverse engineering differs from the fair use defense in two material respects. First, the reverse engineering provision was expressly drafted to provide exemption from infringement liability despite proof of unauthorized copying and striking similarity, so long as the resulting chip product was the result of study and analysis and contained technological improvement, like decreased chip size, lower thermal output, or enhanced speed.
+
+The second difference is one of scope. Under fair use, the extent of economic injury is the primary concern. But injury is irrelevant under reverse engineering; what matters is that more than a trivial improvement was made. In the course of these hearings, legislators pressed various witnesses for guidance in distinguishing piracy from reverse engineering. The uniform industry response was that in a case of piracy, the copier would not be able to show a “ paper trail ” of initial flow charts, time sheets, computer simulations, or an improved chip (Sen. Rept. No. 425). By comparison, a firm that had done reverse engineering would have copious records of analysis and experimentation as well as a new chip with improved performance capabilities. As courts grapple with determining the scope of protection in computer programs and screen displays, the approach of the Chip Act may serve as a model for courts to find infringement only in cases of piracy (Raskind, 1985).
+
+Once infringement is established, the mask work owner has the traditional battery of remedies associated with copyright protection, with two exceptions. First, criminal penalties are not available. Second, the amount of permissible statutory damages for mask work infringement is five times greater ($250,000) than the maximum amount ($50,000) of statutory damages for copyright infringement.
+
+As the newest addition to the intellectual property regime, the Chip Act is a good candidate for further study by economists. To date, only one case has been decided under this statute. And in the case of Brooktree Corp. v. Advanced Micro Devices, Inc. (705 F. Supp. 491 [S.D. Cal. 1988]), the plaintiff was denied a preliminary injunction. A study of research and development in this industry would contribute to resolving the present debate over whether the Chip Act was a necessary or worthy form of intellectual property protection (Risberg, 1990).
+
+## Trademarks
+
+Trademark protection differs from the other regimes of intellectual property both in its legal basis and its economic function. Unlike patents and copyrights, there is no constitutional foundation for trademark protection. In fact, there are no federally created rights in trademarks at all; the federal framework provides a system of registration and of enforcement of marks whose genesis is in state law (McCarthy, 1984; Landes and Posner, 1987).
+
+---
+
+Stanley M. Besen and Leo J. Raskind 21
+
+Although trademark protection did not originate as an incentive for innovation or creativity, it now provides an economic incentive. The primary function of trademarks is to provide rules of orderly marketing by identifying products and their sources.
+
+State law attaches protection to trademarks (and other devices to identify producers) when they are adopted and used. Protection is given by the branch of tort law governing unfair competition. By registering a trademark with the Patent and Trademark Office, the registrant gains access to the federal courts and can stake out a national claim to the registered mark, a result that cannot be achieved by state law. Beginning in 1989, a trademark owner who has not yet used a mark in commerce may apply for registration of intended use and support that claim by actually using the mark within six months and filing a statement at that time.
+
+The origin of trademark protection is usually associated with the medieval guild practice of affixing an identifying mark to a goblet or like product. Accordingly, the initial purpose of trademark protection was to make it illegal to pass off the goods of another artisan as those of a guild member. As modern manufacturing and marketing practices made the consumer remote from the source of the product, however, trademarks took on the function of product identification and differentiation, as well as source identification. Protection was extended from the trademark, an arbitrary and distinctive designation, to include a tradename, a surname or descriptive word, a service mark, a collective mark (like “ union made ” ), as well as packaging and product features. Over time, the legal theory of protection was adapted to prevent a second entrant from unfairly appropriating the value of a successful trademark, service mark, or trade dress. Thus, the protection of trademarks has evolved as a form of indirect protection of the consumer by insuring that purchasing decisions are based on marks that properly identify the product and its source.
+
+The value of the trademark is enhanced both by the quality of the product and the public familiarity with it. A trademark may impose costs of advertising and establishing a reputation upon a prospective entrant to a market. Moreover, trademark protection may dampen competition by limiting the ability of competitors to copy a successful mark or packaging design, even though these features are not protected by copyright or patent.
+
+Patents require novelty; copyright requires originality; the counterpart of these terms for trademark is distinctiveness. The Lanham Act provides that no mark by which the goods of the applicant “may be distinguished from the goods of others” shall be refused registration. In this context, “distinctiveness” means that the claimed work must not be a generic description. Thus, “orange” is not eligible for protection as applied to citrus fruit or its extract. Also ineligible are geographic names to designate location, or even an actual surname if the attachment to a product would be likely to confuse or deceive consumers. Similarly, merely descriptive terms are also ineligible, like “pasteurized” as applied to milk.
+
+---
+
+22 Journal of Economic Perspectives
+
+A further condition for the registration of trademarks is that the proffered trademark use terms that are either arbitrary or fanciful as applied to a good or service. “ Kodak, ” as applied to photographic film, is the classic example of an arbitrary term being associated with a product. A term or phrase that is not distinctive or otherwise eligible for registration — like “ Big Red ” as applied to apples — may qualify for registration as applied to a hand tool, if it is found to have acquired “ secondary meaning. ” Secondary meaning is a doctrine that allows protection to attach to a word or phrase that is geographically or otherwise descriptive, requiring the claimant to prove that those words identify that product to consumers. Secondary meaning may be proven by survey evidence, for example by showing that some statistically significant segment of the consumers in a given geographic market will state that the phrase — “ Big Red ”— means to them a screwdriver manufactured in New Haven, Connecticut (Stern and Hoffman, 1962).
+
+Trademark law recognizes that a term that is not distinctive at the outset may become so through usage. If an arbitrary mark becomes successful enough to become part of the language, that mark may lose its protection because generic terms are not protectible. “Aspirin” and “Thermos” are examples of trademarks that have lost protection by becoming generic (Folsom and Teply, 1980). The board game “Monopoly” also lost protection for a time, based on survey evidence that 65 percent of respondents wanted the game without reference to who made it. Congress reversed this decision in 1984 by modifying the grounds for cancellation of a generic mark to require that the test be, in the words of 15 U.S.C. § 1064(3) [1984], “the primary significance of the registered mark to the relevant public rather than purchaser motivation.” The new statute effectively withdraws the significance of secondary meaning as a test of becoming generic. The concept will remain only as a basis of granting protection. But how the courts will determine “primary significance” and secondary meaning remains to be seen. They seem to be groping for some analytical statement in terms of the function of these protected symbols in the competitive process (Stern and Hoffman, 1962; Dreyfuss, 1990). Law in this unsettled state invites economic analysis of how trademarks affect product differentiation and consumer protection.
+
+To prove trademark infringement, it is not necessary to prove actual confusion of specific customers. Proof of the likelihood of confusion in the market circumstances satisfies the requirement, so that similarity between two marks in physical design, in sound or in commercial connotation can make the case for infringement. Strictly speaking, the Lanham Act does not grant legal rights in trademarks beyond registration. However, Section 43 provides a federal law regulating unfair trade practices involving trademarks and product designation. Section 43 broadens the tort rule against “passing off,” by making actionable any false statement of fact or any statement likely to deceive, when made in a competitive context.
+
+---
+
+Introduction to the Law and Economics of Intellectual Property 23
+
+Like the other branches of intellectual property law, the traditional remedies of injunction, damages, and fees and costs are available to the successful litigant. A successful plaintiff in a trademark infringement action is entitled to injunctive relief, to recover the defendant's profits and to receive damages for lost sales. These penalties can be further increased if there is a deliberate use of a counterfeit mark.
+
+## Trade Secrets
+
+Trade secret law covers specific business information transmitted by persons, firms, and markets (Grossman and Stiglitz, 1976; Kitch, 1980) . A trade secret has alternatively been described (Milgrim, 1989) as “any formula, pattern, device or compilation of information which is used in one's business, and which gives him an opportunity to obtain an advantage over competitors who do not know or use it.” Or in the words of the Uniform Trade Secrets Act, §1(4) (1979) , as “information including a formula, pattern, compilation, program, device, method technique, or process, that: (i) derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable by proper means by, other persons who can obtain value from its disclosure or use, and (ii) is the subject of efforts that are reasonable under the circumstances to maintain its secrecy.”
+
+Trade secret law differs from patent law in three respects. First, it is grounded in state law, so that aside from the uniformity prevailing in the 16 states that have enacted the Uniform Trade Secrets Act, the scope of protection varies among the states. A few states rest trade secret protection on a property theory, while the majority of states invoke doctrines of tort, contract, constructive trust, or unjust enrichment as a basis of trade secret protection. In addition, 25 states have made it a crime to steal a trade secret (Kitch, 1980).
+
+Second, trade secret law is also outside the inducement/disclosure framework of patent and copyright law. Protection is granted to a patent and copyright owner in return for the disclosure of the subject matter to the public. Trade secrets are, by definition, not disclosed. They are protected against discovery by “improper means,” but not against discovery of the trade secret by independent means or by reverse engineering. The incentive to “create” trade secrets, such as customer lists or chemical formulae, and to incur costs of protecting them, is derived from their value. Of course, trade secret protection may serve as an ancillary incentive for the firm to perfect a process that might be eligible for patent protection. But given the more limited prospects for protection of trade secrets, the patent affords a far more powerful incentive.
+
+A third difference between trade secret law and patent law is the subject matter and duration of protection. While protection is granted only for subject matter that represents some creative efforts under the copyright and patent
+
+---
+
+24 Journal of Economic Perspectives
+
+regimes, trade secret protection rests solely on the commercial value of the matter to the claimant (Kitch, 1980) . Thus, trade secret law includes a wider subject mater. The duration of trade secret protection is limited only by the happenstance of independent discovery or by improper disclosure. This potential for perpetual protection serves as an incentive to avoid the disclosure requirements of the patent regime. There is also an abusive potential in trade secret protection if the subject matter of a valuable but little-known expired patent were to be given further protection as a trade secret.
+
+. When a competitor has used “ improper means ” of discovery, or an employee has breached the confidential condition under which a trade secret was disclosed, the available remedies are injunctive relief and damages. The Uniform Trade Secrets Act makes illegal both the use of “ improper means ” to discover the trade secret and the “ misappropriation ” of improperly discovered material. Under the UTSA Sec. 1(1-2) (1979), improper means include “ theft, bribery, misrepresentation, breach or inducement of a breach of duty to maintain secrecy, or espionage through electronic or other means ” while misappropriation is defined in part as “ acquisition of a trade secret of another by a person who knows or has reason to know that the trade secret was acquired by improper means. ” Thus, the focus of trade secret law is “ fair dealing ” between competitors and between employer and employee with regard to the uses of specific business information.
+
+Since trade secret law is enmeshed both with the competitive process and the internal decisions of the firm, economic analysis of the properties and consequences of trade secret protection would illuminate important public policy interests. One possibility might be to examine the workings of the commercial espionage industry. A systematic investigation of the litigated trade secret cases would provide useful data (Miller, 1989) . In this issue, Friedman, Landes, and Posner provide a starting point by arguing that existing trade secret law appears to be consistent with considerations of economic efficiency.
+
+Economic analysis might also illuminate the judicial task of applying trade secret protection. In our view, judges are presently too concerned with the “ dirty trick ” aspects of competition. We believe that optimal administration of the trade secret laws requires more emphasis on the private and social costs and benefits of trade secret protection, and on economic efficiency, and correspondingly less concern with norms of fair commercial conduct.
+
+## Misappropriation
+
+Misappropriation is a legal doctrine that functions outside of trade secret law as a general common law property right against some takings of information of commercial value. The doctrine is derived from a Supreme Court
+
+---
+
+Stanley M. Besen and Leo J. Raskind 25
+
+opinion, International News Service v. Associated Press (248 U.S. 215 [1918]), barring the use of uncopyrighted news reports. During World War I, the Hearst news gathering agency copied Associated Press dispatches on the Eastern seaboard and transmitted them by wire to their midwest and west coast papers ahead of their receipt by Associated Press subscribers there. Although the relief granted by the court was limited to protection during the period of initial dissemination, the reasoning of the opinion launched a broad principle of unfair competition law by stating (at p. 240), “ . . . defendant . . . admits that it is taking material that has been acquired by complainant as the result of organization and the expenditure of labor, skill, and money, . . . and that . . . in appropriating it and selling it as its own is endeavoring to reap where it has not sown. ” This doctrine has been adopted by state and federal courts as a general rule of unfair competition, sometimes granting protection to material otherwise appropriately outside the reach of intellectual property laws (Baird, 1983). However, this doctrine has been repeatedly criticized for its lack of analytical content (Raskind, 1991).
+
+## Concluding Thoughts
+
+Increased interest in the economics of intellectual property is timely. The pace of technological change during the last few decades has forced intellectual property law into unknown areas and hard cases, straining the capabilities of courts and legislatures. How should innovations related to semiconductor chip or computer software be protected? Should genetically engineered life forms be patented? Under what conditions should videotaping a television show for home use infringe the rights of program producers?
+
+These and other issues have forced a searching re-examination of many of the premises of the intellectual property system during the last decade. Some authors have called into question some of the basic underpinnings of the law in this area, while others have concluded that the system is fundamentally sound. The papers in this symposium provide examples of both of these views.
+
+Research along these lines could play a key role in affecting America's future standard of living and economic competitiveness. After all, a nation's regime of intellectual property law sets the stage and establishes the incentives for innovation and technological change.
+
+■ This paper was prepared for the Symposium on Intellectual Property Law, funded by the RAND Corporation and The John P. Olin Foundation. The Symposium took place October 24, 1989.
+
+---
+
+26 Journal of Economic Perspectives
+
+## References
+
+Arrow, K. J., "Economic Welfare and the Allocation of Resources for Invention." In National Bureau of Economic Research, The Rate and Direction of Inventive Activity: Economic and Social Factors. Princeton: Princeton University Press, 1962.
+
+Baird, D. G., "Common Law Intellectual Property and the Legacy of International News Service v. Associated Press," University of Chicago Law Review, 1983, 50, 411-429.
+
+Beck, R. L., "The Prospect Theory of the Patent System and Unproductive Competition," Research in Law & Economics, 1983, 5, 193-201.
+
+Besen, S. M., and S. N. Kirby, "Private Copying, Appropriability, and Optimal Copying Royalties," Journal of Law & Economics, 1989a, 32, 255-280.
+
+Besen, S. M., and S. N. Kirby, "Compensating Creators of Intellectual Property: Collectives That Collect," RAND Corporation, R3751-MF, November 1989b.
+
+Besen, S. M., W. G. Manning, and B. M. Mitchell, "Copyright Liability for Cable Television: Compulsory Licensing and the Coase Theorem," Journal of Law and Economics, 1978, 21, 67-95.
+
+Braga, C. A., "The Economics of Intellectual Property Rights and the GATT," Vanderbilt Journal of Transnational Law, 1989, 22, 243-264.
+
+Braunstein, Y. M., D. M. Fischer, J. A. Ordover, and W. J. Baumol, “Economics of Property Rights as Applied to Computer Software and Data Bases,” United States Department of Commerce: PB-268 787, 1977.
+
+Breyer, S., "The Uneasy Case for Copyright: A Study of Copyright in Books, Photocopies, and Computer Programs," Harvard Law Review, 1970, 84, 281-351.
+
+Chisum, D., Patents. New York: Matthew Bender, 1989.
+
+Clapes, A. L., P. Lynch, and M. R. Steinberg, "Silicon Epics and Binary Bards: Determining the Proper Scope of Copyright Protection For Computer Programs," UCLA Law Review, 1987, 34, 1493-1594.
+
+Dasgupta, P., and J. E. Stiglitz, "Industrial Structure and the Nature of Innovative Activity," Economic Journal, 1980, 90, 266-293.
+
+Dixit, A. K. and J. E. Stiglitz, "Monopolistic Competition and Optimum Product Diversity," American Economic Review, 1977, 67, 297-308.
+
+Dreyfuss, R., "The Federal Circuit: A Case
+
+Study in Specialized Courts," N.Y.U. Law Review, 1989, 64, 1-77.
+
+Dreyfuss, R., "Expressive Genericity: Trademarks as Language in the Pepsi Generation," Notre Dame Law Review, 1990, 65, 397-424.
+
+Farrell, J., and G. Saloner, "Standardization, Compatibility, and Innovation," RAND Journal of Economics, 1985, 16, 70-83.
+
+Farrell, J., "Standardization and Intellectual Property," Jurimetrics, 1989, 30, 35-50.
+
+Folsom, R. H., and R. H. Teply, "Trademarked Generic Words," Yale Law Journal, 1980, 89, 1323-1359.
+
+Frase, R., "Comments on Hurt and Schuchman, The Economic Rationale of Copyright," American Economic Review, Papers and Proceedings, 1966, 56, 435-439.
+
+Gilbert, R., and C. Shapiro, "Optimal Patent Length and Breadth," RAND Journal of Economics, 1990, 21, 106-112.
+
+Goldstein, P., Copyright. Boston: Little, Brown, 1989.
+
+Gordon, W. J., "Fair Use As Market Failure: A Structural and Economic Analysis of the Betamax Case and Its Predecessors," Columbia Law Review, 1982, 82, 1600-1659.
+
+Grossman, S. J., and J. E. Stiglitz, "Information and Competitive Price Systems," American Economic Review, Papers & Proceedings, May 1976, 66, 246-253.
+
+Hall, C. D., "Patents, Licensing, and Antitrust," Research In Law & Economics, 1986, 8, 59-86.
+
+Hardy, I. T., "An Economic Understanding of Copyright Law's Work-Made-For-Hire Doctrine," Columbia-VLA Journal of Law and the Arts, 1988, 12, 181-227.
+
+Hughes, J., "The Philosophy of Intellectual Property," Georgetown Law Journal, 1988, 77, 287-366.
+
+Johnson, W. R., "The Economics of Copying," Journal of Political Economy, 1985, 93, 158-174.
+
+Kaplow, L., "The Patent-Antitrust Intersection: A Reappraisal," Harvard Law Review, 1984, 97, 1813-1892.
+
+Katz, M. L., and C. Shapiro, "Network Externalities, Competition, and Compatibility," American Economic Review, 1985a, 75, 424-440.
+
+Katz, M. L., and C. Shapiro, "On the Licensing of Innovations," Rand Journal of Economics, 1985b, 16, 504-520.
+
+Kitch, E. W., "The Nature And Function Of The Patent System," Journal of Law & Eco-
+
+---
+
+Introduction to the Law and Economics of Intellectual Property 27
+
+nomics, 1977, 20, 265-290.
+
+Kitch, E. W., "The Law and Economics of Rights in Valuable Information," Journal of Legal Studies, 1980, 9, 683-724.
+
+Kitch, E. W., "Patents: Monopolies Or Property Rights?" Research In Law & Economics, 1986, 8, 31-49.
+
+Klemperer, P., "How Broad Should the Scope of Patent Protection Be?" RAND Journal of Economics, 1990, 21, 113-130.
+
+Landes, W. M., and R. A. Posner, "Trademark Law: An Economic Perspective," Journal of Law & Economics, 1987, 30, 265-309.
+
+Landes, W. M., and R. A. Posner, "An Economic Analysis of Copyright Law," Journal of Legal Studies, 1989, 18, 325-366.
+
+Lefton, T., "IBM, Unisys Reduce Fees for Modem Compression," Electronic News,' January 1990, p. 1.
+
+Liebowitz, S. J., "Copying and Indirect Appropriability: Photocopying of Journals," Journal of Political Economy, 1985, 93, 945-957.
+
+Loury, G. C., "Market Structure and Innovation, Quarterly Journal of Economics, 1979, 93, 395-410.
+
+McCarthy, J. T., Trademarks and Unfair Competition, second edition. San Francisco: Lawyers' Co-Operative Publishing Co., 1984.
+
+McFetridge, D. G., and M. Rafiquzzaman, "The Scope And Duration Of The Patent Right And The Nature Of Research Rivalry," Research In Law & Economics, 1986, 8, 91-120.
+
+Menell, P., "An Analysis Of The Scope Of Copyright Protection For Application Programs," Stanford Law Review, 1989, 41, 1045-1104.
+
+Milgrim, R. M., On Trade Secrets. New York: Matthew Bender, 1989.
+
+Miller, E., "Antitrust Restrictions On Trade Secret Licensing: A Legal Review and Economic Analysis," Law & Contemporary Problems, 1989, 52, 183-209.
+
+Mossinghoff, G. J., "The Importance of Intellectual Property Protection in International Trade," Boston College International and Comparative Law Review, 1984, 7, 235-249.
+
+Nimmer, M., and D. Nimmer, On Copyright. New York: Matthew Bender, 1989.
+
+Nordhaus, W. D., Invention, Growth, and Welfare, A Theoretical Treatment of Technological Change. Cambridge: MIT Press, 1969.
+
+Novos, I. E., and M. Waldman, "The Effects of Increased Copyright Protection: An Analytical Approach," Journal of Political Economy, 1984, 92, 236-246.
+
+Ordover, J. A., and R. D. Willig, "On the Optimal Provisions of Journals qua Sometimes Shared Goods," American Economic Review, 1978, 68, 324-339.
+
+Plant, A., "The Economic Aspects of Copyright In Books," Economica, May 1934a, 1, 167-195.
+
+Plant, A., "The Economic Theory Concerning Patents for Inventions," Economica, February 1934b, 1, 67-95.
+
+Raskind, L. J., "A Functional Interpretation Of Fair Use," Journal Of The Copyright Society, 1984, 31, 601-639.
+
+Raskind, L. J., "Reverse Engineering, Unfair Competition, and Fair Use," University of Minnesota Law Review, 1985, 70, 385-415.
+
+Raskind, L. J., "The Continuing Process of Refining and Adapting Copyright Principles," Columbia-VLA Journal of Law and the Arts, 1990.
+
+Raskind, L. J., "The Misappropriation Doctrine As A Competitive Norm of Intellectual Property Law," University of Minnesota Law Review, 1991, 75.
+
+Risberg, R., "Five Years Without Infringement Litigation Under the Semiconductor Chip Act: Unmasking The Spectre of Chip Piracy In An Era of Diverse and Incompatible Process Technologies," Wisconsin Law Review, 1990, 241-277.
+
+Senate Report No. 425, 98th Cong., 2d Sess., 1984, p. 21.
+
+Spence, M., "Product Selection, Fixed Costs, and Monopolistic Competition," Review of Economic Studies, 1976, 43, 217-235.
+
+Stern, R. H., and J. Hoffman, "Public Injury and the Public Interest: Secondary Meaning in the Law of Unfair Competition," University of Pennsylvania Law Review, 1962, 110, 935-971.
+
+Stern, R. H., "Section 117 of the Copyright Act: Charter of Software User's Rights or an Illusory Promise?" Western New England Law Review, 1985, 7, 459-485.
+
+Stern, R. H., Semiconductor Chip Protection. New York: Harcourt Brace, 1986.
+
+Sumner, J. P., and S. W. Lundberg, "Patentable Computer Program Features As Uncopyrightable Subject Matter," American Intellectual Property Law Association Law Quarterly, 1989, 17, 238-255.
+
+Wheaton, J. J., "Generic Competition and Pharmaceutical Innovation: The Drug Price Competition and Patent Term Restoration Act of 1984," Catholic University Law Review, 1986, 35, 433-487.
+
+---
+
+![Figure](figures/figure_000.png)
+

processed/00013_W3124020986/document.md

@@ -0,0 +1,348 @@
+Journal of Economic Perspectives-Volume 27, Number 1-Winter 2013-Pages 3-22
+
+# The Case Against Patents
+
+Michele Boldrin and David K. Levine
+
+The case against patents can be summarized briefly: there is no empirical evidence that they serve to increase innovation and productivity, unless productivity is identified with the number of patents awarded—which, as evidence shows, has no correlation with measured productivity. This disconnect is at the root of what is called the “patent puzzle”: in spite of the enormous increase in the number of patents and in the strength of their legal protection, the US economy has seen neither a dramatic acceleration in the rate of technological progress nor a major increase in the levels of research and development expenditure.
+
+Both theory and evidence suggest that while patents can have a partial equilibrium effect of improving incentives to invent, the general equilibrium effect on innovation can be negative. The historical and international evidence suggests that while weak patent systems may mildly increase innovation with limited side effects, strong patent systems retard innovation with many negative side effects. More generally, the initial eruption of innovations leading to the creation of a new industry—from chemicals to cars, from radio and television to personal computers and investment banking—is seldom, if ever, born out of patent protection and is instead the fruit of a competitive environment. It is only after the initial stage of rampant growth ends that mature industries turn toward the legal protection of patents, usually because their internal growth potential diminishes and they become more concentrated. These observations, supported by a steadily increasing body of evidence, are consistent with
+
+■ Michele Boldrin is Joseph Gibson Hoyt Distinguished University Professor of Economics and David K. Levine is John H. Biggs Distinguished University Professor of Economics, both at Washington University in St. Louis, Missouri. They are also both Research Fellows with the Federal Reserve Bank of St. Louis. Their email addresses are mboldrin@artsci.wustl.edu and david@dklevine.com.
+
+http://dx.doi.org/10.1257/jep.27.1.3.
+
+doi=10.1257/jep.27.1.3
+
+---
+
+4 Journal of Economic Perspectives
+
+theories of innovation emphasizing competition and first-mover advantage as the main drivers of innovation, and they directly contradict “Schumpeterian” theories postulating that government-granted monopolies are crucial to provide incentives for innovation. A properly designed patent system might serve to increase innovation at a certain time and place—and some patent systems, such as the late-nineteenth century German system allowing only process but not final product patents, have been associated with rapid innovation. Unfortunately, the political economy of government-operated patent systems indicates that such systems are susceptible to pressures that cause the ill effects of patents to grow over time. The political economy pressures tend to benefit those who own patents and are in a good position to lobby for stronger patent protection, but disadvantage current and future innovators as well as ultimate consumers. This explains why the political demand for stronger patent protection comes from old and stagnant industries and firms, not from new and innovative ones. Our preferred policy solution is to abolish patents entirely and to find other legislative instruments, less open to lobbying and rent seeking, to foster innovation when there is clear evidence that laissez-faire undersupplies it. However, if that policy change seems too large to swallow, we discuss in the conclusion a set of partial reforms that could be implemented as part of an incremental strategy of reducing the harm done by the patent system.
+
+## Do Patents Encourage Productivity Growth?
+
+If there is to be any rationale for patent systems, with all their ancillary costs, it must be that they increase innovation and productivity. What is the evidence?
+
+Simply eyeballing the big trends shows that patenting has exploded over the last decades. In 1983 in the United States, 59,715 patents were issued; by 2003, 189,597 patents were issued; and in 2010, 244,341 new patents were approved. In less than 30 years, the flow of patents more than quadrupled. By contrast, neither innovation nor research and development expenditure nor factor productivity have exhibited any particular upward trend. According to the Bureau of Labor Statistics, annual growth in total factor productivity in the decade 1970 – 1979 was about 1.2 percent, while in the decades 1990 – 1999 and 2000 – 2009 it has been a bit below 1 percent. Meanwhile, US research and development expenditure has been oscillating for more than three decades in a narrow band around 2.5 percent of GDP. The recent explosion of patents, in other words, has not brought about any additional surge in useful innovations and aggregate productivity. In new industries such as biotechnology and software — where innovation was already thriving in their absence — patents have been introduced without any positive impact on the rate of innovation. The software industry is an important case in point. In a dramatic example of judge-made law, software patents became possible for the first time in the early 1990s. Bessen and Meurer, in a large body of empirical work culminating in Patent Failure (2008), have studied the consequences of this experiment and have concluded that it damaged social welfare.
+
+---
+
+Michele Boldrin and David K. Levine 5
+
+Academic studies have also typically failed to find much of a connection between patents and innovation. In Boldrin and Levine (2008b) , we conducted a metastudy gathering the 24 studies (including three surveys of earlier empirical work) we could find in 2006 that examined whether introducing or strengthening patent protection leads to greater innovation. The executive summary states: “ [T]hese studies find weak or no evidence that strengthening patent regimes increases innovation; they find evidence that strengthening the patent regime increases patenting! They also find evidence that, in countries with initially weak IP [intellectual property] regimes, strengthening IP increases the flow of foreign investment in sectors where patents are frequently used. ” Actually, the issue of promoting foreign direct investment, while a well-established empirical consequence of strengthening patent regimes, is entirely beside the point of this essay. There are a number of ways to strengthen a country's institutions and infrastructure in a way that would encourage foreign direct investment — and, in any case, foreign direct investment is not equivalent to innovation.
+
+Our conclusion was in keeping with other studies that have addressed this question. Some studies have failed to find any connection even between changes in the strength of patent law and the amount of patenting, while others fail to find a connection between patents and some measure of innovation or productivity. For example, after failing to find a single study claiming that innovation increased as a consequence of the strengthening of US patent protection in the 1980s, Gallini (2002, p. 139) wrote in this journal: “Although it seems plausible that the strengthening of US patents may have contributed to the rise in patenting over the past decade and a half, the connection has proven difficult to verify.” Similarly, Jaffe (2000) also examines many studies and concludes: “[D]espite the significance of the policy changes and the wide availability of detailed data relating to patenting, robust conclusions regarding the empirical consequences for technological innovations of changes in patent policy are few. There is widespread unease that the costs of stronger patent protection may exceed the benefits. Both theoretical and, to a lesser extent, empirical research suggest this possibility.” $^1$
+
+1 The study by Kanwar and Evanson (2001) illustrates some of the issues that arise in these kinds of studies. They have two five-year averages on 31 countries for the period 1981 – 1990. They find support for the idea that higher patent protection leads to higher research and development spending as a fraction of GDP. However, a different story seems equally plausible. Countries with a larger market can more easily pay the fixed costs of innovation. Indeed, one perspective is that their data essentially compares countries with relatively small economies, little intellectual property protection, and low R & D spending with countries with relatively larger economies, greater intellectual property protection, and higher R & D spending. For example, R & D spending as a fraction of GDP in their data ranges from a ten-year average of 0.2 percent in Jordan to 2.8 percent in Sweden. If we combine their data with GDP data from The 1990 CIA World Fact Book to take account of the size of the economy, increasing the strength of intellectual property protection from 0 to 1 to 2 on their five-point scale does increase R & D expenditure. But as intellectual property protection is increased further, the gains to R & D expenditure levels then falls. Even at the lower levels, we are probably observing primarily the effect of foreign direct investment: that is, among poor countries with near-zero intellectual property protection, increases bring in more foreign investment and in doing so directly raise R & D spending. In higher-income countries with larger economies, foreign investment is not an issue, and increases in intellectual property have little or no effect on innovation.
+
+---
+
+6 Journal of Economic Perspectives
+
+The Lerner (2002) study is especially notable because he examined all significant changes in patent law in all countries over the last 150 years. His conclusion: “Consider, for instance, policy changes that strengthen patent protection. Once overall trends in patenting are adjusted for, the changes in patents by residents of the country undertaking the policy change are negative, both in Great Britain and in the country itself. Subject to the caveats noted in the conclusion this evidence suggests that these policy changes did not spur innovation.” This, in summary, is what is currently known as the “patent puzzle”—although as we will explain, it is substantially coherent with a theory of innovation that emphasizes the gains from competition and first-mover incentives, rather than benefits from the monopoly power of patents.
+
+Evidence at the sectoral level of the US economy shows the same disconnect between patenting and productivity. In Boldrin, Correa, Levine, and Ornaghi (2011) , we carried out a sequence of statistical tests and econometric estimations on two datasets: an original microeconomic dataset obtained by combining firm-level information obtained through Compustat, the National Bureau of Economic Research, and the Bureau of Labor Statistics and an enriched version of the dataset used by Aghion, Bloom, Blundell, Griffith, and Howitt (2005) in their study of industry-level mark-ups. Conclusions must of course be drawn with care from this kind of data because, across industries, the strength of competition, patenting, and productivity are simultaneously determined and intertwined with technological change. With that reservation appropriately noted, at the industry level there is, in general, no statistically significant correlation between measures of productivity (whether measured by labor or total factor productivity) and of patenting activity (whether measured by number of patents or citations of patents).
+
+We then investigated the relationships between patents, competition, and productivity further. When we regressed measures of patents (or patent citations) on a measure of competition (as measured by the inverse of profitability) used by Aghion, Bloom, Blundell, Griffith, and Howitt (2005) , we found a positive relationship that is remarkably robust to changes in industry classification, time period, and set of sampled industries. That is, patents were more common in competitive industries. We also studied the correlation between the same measure of competitive pressure and objective measures of labor productivity growth. In our preferred specification, we found that average annual growth of productivity in the sectors with the highest level of competition is up to 2 percent bigger than in the sectors with the lowest level of competition. These are strikingly large differences when cumulated over various decades, as it is the case in our dataset. This finding of a positive correlation between competition and productivity at the sectorial level replicated a pioneering, and unfortunately forgotten, pattern reported in Stigler (1956) .
+
+The accumulated findings of no positive relationship between patenting and productivity are not conclusive, and arguments have raged over the specific data used, whether to look for a structural break in the data, how the researcher seeks
+
+---
+
+The Case Against Patents 7
+
+to correct for endogeneity, and so on. $^2$ However, it is fair to say that the sector-level, national, and cross-national evidence fail to provide any clear empirical link from patents to innovation or to productivity. This lack of connection is consistent with the view that the use of patents either as a defensive or as a rent-seeking tool is more widespread than one might have predicted. In addition, the empirical evidence is consistent with the proposition that greater competition, not patents, is the main factor leading to innovation and greater productivity.
+
+## Theory and Practice of Patents and Innovation
+
+There is little doubt that providing a monopoly as a reward for innovation increases the incentive to innovate. There is equally little doubt that granting a monopoly for any reason has the many ill consequences we associate with monopoly power—the most important and overlooked of which is the strong incentive of a government-granted monopolist to engage in further political rent seeking to preserve and expand its monopoly or, for those who do not yet have a monopoly, to try to obtain one. These effects are at least to some extent offsetting: while the positive impact of patents is the straightforward partial equilibrium effect of increasing the profits of the successful innovator to the monopolistic level, the negative one is the subtler general equilibrium effect of reducing everybody else's ability to compete while increasing for everyone the incentive to engage in socially wasteful lobbying efforts.
+
+### Downstream Innovation, Defensive Patenting, and Patent Trolls
+
+In the long run, even the positive partial equilibrium effect of patents in providing an incentive for innovation may be more apparent than real: the existence of a large number of monopolies created by past patent grants reduces the incentives for current innovation because current innovators are subject to constant legal action and licensing demands from earlier patent holders. The downstream blocking effect of existing monopoly grants on incentives for future innovation
+
+2 For a sense of these controversies, Aghion, Bloom, Blundell, Griffith, and Howitt (2005) find an “inverted-U” relationship between the extent of competition, as measured by the inverse of mark-ups, and a measure of patenting activity, based on a dataset of US patents of UK firms. In other words, they find that the maximum innovative effort (as measured by patents) occurs at some “intermediate” position between a high and low level of competition. However, Hashmi (2011) reexamines the inverted-U relationship using data from publicly traded US manufacturing firms and finds a robust positive relationship between the inverse of markups and citation-weighted patents. Correa (2012) reexamines the same dataset of UK firms and shows that the prediction of an inverted-U is overturned when allowing for the possibility that innovations follow a “memory process,” where the current probability of introducing a new innovation increases when a firm successfully innovated in the previous period. He also finds a structural break in the data in 1981, when the Court of Appeals for the Federal Circuit was established to hear appeals of patent cases. Overall, Correa finds a positive innovation–competition relationship for the memory industries before the 1982 reform, but no relationship between innovation and competition for those industries that he classifies as memory-less.
+
+---
+
+8 Journal of Economic Perspectives
+
+has greatly increased in recent decades because modern products are made up of so many different components. The recent—and largely successful—efforts of Microsoft to impose a licensing fee on the large and expanding Android phone market is but one case in point. With the exception of Motorola Mobility, all the handset manufacturers have agreed to the fee, and Motorola lost its first battle against the fee in spring 2012—fought not in court but in the more receptive domain of the US International Trade Commission (Investigation Number 337-TA744, May 18, 2012). Microsoft is attempting to charge a licensing fee solely over a patent involving the scheduling of meetings—a rarely used feature of modern smartphones. The meeting-schedule feature is but one of many thousands of patented “ideas” used in a modern smartphone, and each owner of each patent potentially can charge a licensing fee. Hence, the main dynamic general equilibrium effect of a patent system is to subject future inventions to a gigantic hold-up problem: with many licenses to be purchased and uncertainty about the ultimate value of the new innovation, each patent holder, in raising the price of his “component,” imposes an externality on other patent holders and so charges a higher than efficient licensing fee. In Boldrin and Levine (2005) and Llanes and Trento (2009), we and others have explored the theory; and many case studies involving patents (and other fractionated ownership problems) can be found in Heller (2008).
+
+To understand more about the actual effect of patents in the real world, consider the recent purchase by Google of Motorola Mobility, primarily for its patent portfolio—not for the ideas and innovations in that portfolio. Few if any changes or improvements to Google's Android operating system will result from the ownership or study of these software patents. Google's purpose in obtaining this patent portfolio is purely defensive: it can be used to countersue Apple and Microsoft and blunt their legal attack on Google. These remarks apply to the vast bulk of patents: they do not represent useful innovation at all and are just weapons in an arms race. This is not news: the same message emerged decades ago from the Levin, Klevorick, Nelson, and Winter (1987) and Cohen, Nelson, and Walsh (2000) surveys of research and development managers.
+
+One could argue that the costs of building up a patent portfolio to engage in this sort of defensive patenting are not too large: after all, it can cost as little as $15,000 to file a successful patent application, and filing applications on a larger scale might be cheaper. However, the acquisition of large patent portfolios by incumbents creates huge barriers to entry. In the smartphone market, for example, Apple is the market leader and Microsoft is unable to produce a product that appeals to consumers. Each are incumbent firms with a large patent portfolio. In this market, Google is the new entrant and innovator and, while wealthy, Google found itself lacking a large defensive patent portfolio. Hence we see both Apple and Microsoft attacking Google with patent litigations, generating hundreds of millions in wasteful legal costs and no social benefit whatsoever.
+
+Despite the fact that patents are mostly used for arms races and that these, in turn, are driven by patent trolls, there does not yet exist convincing formal models of the ways in which this interaction can inhibit innovation. In a pure arms
+
+---
+
+Michele Boldrin and David K. Levine 9
+
+race theory, if all firms get counterbalancing patent portfolios and all innovate, then they would all have innovated in the absence of patents—hence, patents do not encourage innovation. This follows because with counterbalancing patent portfolios, no firm can sue any other firm—exactly as would be the case in the absence of patents. Hence in this setting patents simply add a cost to innovation: if you wish to innovate, you must acquire an expensive patent portfolio to avoid trolls. On the other hand if a patentholder does not produce a marketable product and hence cannot be countersued—like Microsoft in the phone market or other patent trolls in other markets—then patents become a mechanism for sharing the profits without doing the work. In this scenario, not only do patents discourage innovation, but they are also a pure waste from a social standpoint.
+
+## Patents and Information Disclosure
+
+Another widely cited benefit of patent systems — although not so much in the economics literature — is the notion that patents are a substitute for socially costly trade secrecy and improve communication about ideas. From a theoretical point of view, the notion that patents are a substitute for trade secrecy fails in the simplest model. If a secret can be kept for $N$ years and a patent lasts $M$ years, then an innovator will patent when $N < M$ . In other words, ideas will be patented when it seems likely that the secret would have emerged before the patent expired and not patented if the secret can be kept. In practice, it is uncertain when the secret will leak out, but it can be shown that the basic intuition remains intact in the face of uncertainty (Boldrin and Levine 2004; Ponce 2007) . $^3$
+
+It is also the case that the extent of practical “ disclosure ” in modern patents is as negligible as the skills of patent attorneys can make it. It is usually impossible to build a functioning device or software program from a modern patent application; this is made especially clear by the fact that some patented ideas do not and cannot work. For example, US Patent 6,025,810 was granted for moving information through the fifth dimension. While detailed studies of the usefulness of disclosure in patent applications are not available, companies typically instruct their engineers developing products to avoid studying existing patents so as to be spared subsequent claims of willful infringement, which raises the possibility of having to pay triple damages. According to sworn testimony by Google's chief of Android development during the legal battles between Oracle and Google (for example, Niccolai 2012), the engineers that developed Android were unaware of Apple (or other) patents, and so were unlikely to have been helped by them. The opinion of Brec (2008), a Microsoft developer, reflects that of many practitioners:
+
+[Microsoft policy is for developers to] never search, view, or speculate about patents. I was confused by this guidance till I wrote and reviewed one of my
+
+3 A more subtle point is that secrecy may bias the type of inventive activity away from innovations that are not easily kept secret to those that can be. In this symposium, Moser offers some of the historical evidence on this point.
+
+---
+
+10 Journal of Economic Perspectives
+
+own patents. The legal claims section—the only section that counts—was indecipherable by anyone but a patent attorney. Ignorance is bliss and strongly recommended when it comes to patents.
+
+The related idea that patents somehow improve communication about ideas, thereby creating some positive externality—a notion key to the “public–private” partnership between governments and private research organizations in which the government funds the research and then gives the private organization a monopoly over what is developed in the course of research—is backed by neither theory nor evidence. It is impossible to study the history of innovation without recognizing that inventors and innovators exchange ideas as a matter of course and that secrecy occurs, when it occurs, typically in the final stages of an innovation process when some ambitious inventors hope to corner the market for a functioning device by patenting it. A good case in point is that of the Wright brothers, who made a modest improvement in existing flight technology that they kept secret until they could lock it down on patents, then used their patents both to monopolize the US market and to prevent further innovation for nearly 20 years (Shulman, 2003). The role that Marconi and his patent played in the development of the radio is altogether similar (Hong 2001), as are innumerable other stories. At the opposite extreme we have, again among many, the example of the Cornish steam engine discussed in Nuvolari (2004, 2006). Here engineers exchanged nonpatented ideas for decades in a collaborative effort to improve efficiency. The contemporary FLOSS (Free/Libre and Open Source Software) community is another successful example of how collaboration and exchange of ideas can thrive without the monopoly power granted by patents.
+
+## First-Mover Advantages and Incentives for Innovation
+
+In most industries, the first-mover advantage and the competitive rents it induces are substantial without patents. The smartphone industry—laden as it is with patent litigation—is a case in point. Apple derived enormous profits in this market before it faced any substantial competition. The first iPhone was released on June 29, 2007. The first serious competitor, the HTC Dream (using the Android operating system) was released on October 22, 2008. By that time, over 5 million iPhones had been sold, and sales soared to over 25 million units during the subsequent year, while total sales of all Android-based phones were less than 7 million. In the tablet market, the iPad has no serious competitor as of late 2012 despite having been introduced on April 10, 2010. While it is hard to prove this delayed imitation also would have occurred in the complete absence of patents, intuition suggests— and our formal model in Boldrin and Levine (2004) predicts—that there is little reason to assert patent rights while the first-mover advantage is still active. Apple did not initially try to use patents to prevent the Android phones from coming into its market and the subsequent “patents’ fight” has been taking place largely after 2010; these facts are consistent with a substantial first-mover advantage. How valuable for Apple was the delay in the Android phones entry? Largely because Apple kept its
+
+---
+
+The Case Against Patents 11
+
+first-mover advantage in spite of a large imitative entry in this market, the value of Apple stock—during a severe market downturn—rose by a factor of approximately five. While there may have been some delay in entry from the competition due to Apple’s threat—since executed—of patent litigation, the fact is that similar but less-successful devices had been available for a number of years before Apple finally cracked the market.
+
+Less anecdotal than the story of the iPhone is the survey of research and development managers in Cohen, Nelson, and Walsh (2000). Here, over 50 percent of managers indicate lead time (first-mover advantage) is important to earning a return on innovation; outside the pharmaceutical and medical instruments industry, less than 35 percent of managers indicate that patents are important.
+
+To understand patents in practice, it is necessary to examine the lifecycle of industries (for example, Jovanovich and MacDonald 1994; Scherer 1990). Typically a new, hence innovative, industry begins with a competitive burst of entries through which very many innovators try hard to get their products to market. In these early stages, many firms bring different versions of the new product to the market (think of the American auto industry in the early twentieth century or the software industry in the 1980s and 1990s) while demand for the new product grows rapidly and the quality of products is rapidly improved. At this stage of the industry lifecycle, the price elasticity of demand is typically high; what is important is not to dominate the market, but rather to get your own products quickly to market and to reduce costs. From the perspective of competing firms, your cost-reducing innovation is good for me in the same way that my cost-reducing innovation is good for you—hence, let us all imitate each other and compete in the market.
+
+As the industry matures, demand stabilizes and becomes much less price elastic; the scope for cost-reducing innovations decreases; the benefits of monopoly power grow; and the potential for additional product innovation shrinks. Typically there is a shakeout in which many firms either leave the industry or are bought out. The automobile industry is a classical historical example, but many readers will have a more vivid memory of the bursting of the dot-com bubble, which makes this point even more forcefully. At this stage of the industry lifecycle, rent seeking becomes important and patents are widely used to inhibit innovation, prevent entry, and encourage exit. If we look at patent litigation in practice—and as predicted by theories of first-mover competition (Boldrin and Levine 2004, among others)—it takes place when innovation is low. When an industry matures, innovation is no longer encouraged; instead, it is blocked by the ever-increasing appeal to patent protection on part of the insiders.
+
+While patent litigation has increased, few patents are actively used. Patent litigation often involves dying firms that have accumulated huge stockpile of patents but are no longer able to produce marketable products and that are now suing new and innovative firms. For example, Texas Instruments was one of the first producers of microchips, and many in our generation remember the capabilities of their first TI calculator. But Texas Instruments was unable to make the transition to the personal computer revolution and became, for a while, the symbol of a dying company
+
+---
+
+12 Journal of Economic Perspectives
+
+trying to stay alive by suing the newcomers. $^4$ In more recent times, Microsoft—once the giant bestriding the software industry—has been unable to make the leap to portable devices such as telephones and tablet personal computers. Thus, Microsoft now uses patent litigation to try to claim a share of the profits Google generates in this market. Back in 1991, Bill Gates said: “If people had understood how patents would be granted when most of today’s ideas were invented and had taken out patents, the industry would be at a complete standstill today . . . A future start-up with no patents of its own will be forced to pay whatever price the giants choose to impose.” Today, Microsoft lobbies across Europe and Asia for the introduction of software patents, a prize it has already obtained in its home country.
+
+The cost of litigating patents is not insubstantial either. Bessen and Meurer (2008) used stock market event studies to estimate the cost of patent litigation: they estimate that during the 1990s such costs rose substantially until, at the end of the period, they constituted nearly 14 percent of total research and development costs. A related but more difficult-to-quantify phenomenon is the rise of uncertainty caused by the legal system. A case in point is the NTP Inc. patents that were used to threaten the Blackberry network with a shutdown. In 2006, Research in Motion (RIM), the producer of Blackberry, agreed to pay $ 612.5 million to license the patent in question from NTP (Svensson 2006). The patent was later invalidated by the court — but RIM did not get its money back (Salmon 2012). Here, the behavior of a single judge cost RIM more than half a billion dollars. In this setting, it is no surprise that patent trolls hope to get rich quickly.
+
+It is easier to list the main social welfare implications of the tradeoff between costs of legal monopoly and incentives to patent holders than it is to calculate their magnitudes. Still, the provisional evidence we have suggests that the net welfare effects of the current patent system could easily be negative. It is somewhat conventional to think of welfare losses from distortions as small, with the idea that welfare triangles due to monopoly power are small being the paradigmatic case in point. Unfortunately, monopolies have no incentive to avoid large social losses even when the private gains are small. Witness, for example, the fact that patented pharmaceutical products often sell for hundreds of times the marginal cost of production, as some astonishing pricing differences between the US and the European markets show. Most revealing is the empirical study of the Quinolones family of drugs (Chaudhuri, Goldberg, and Gia 2006) . It measures the economic consequences of the introduction of pharmaceutical patents for this family of drugs and concludes that the consequence of patent protection to India will be nearly $ 300 million in welfare losses — while the gain to the pharmaceutical companies will be less than $ 20 million. $^5$
+
+4 Texas Instruments is such an important source of litigation that empirical work on patent litigation usually uses a dummy variable for TI. Empirical studies of the importance of firms no longer doing business in an industry to litigation can be found in Bessen and Meurer (2005) and Hall and Ziedonis (2007).
+
+5 Although the focus of this paper is on patents rather than copyright, it is worth noting that most of the copyright wars revolve around measures to prevent piracy, empirically a relatively minor factor as far as profits of media corporations are concerned (see for example Sinha, Machado, and Sellman 2010; Danaher, Dhanasobhon, Smith, and Telang 2010; Sanchez 2012).
+
+---
+
+Michele Boldrin and David K. Levine 13
+
+## Pharmaceuticals
+
+This brings us to the controversial issue of drug patents. The standard argument says: No patents, no drugs. The total cost of developing a new drug, including failures, is quickly approaching the $ 1 billion mark (DiMasi, Hansen, and Grabowski 2003) . So how can anyone, faced with such a gigantic fixed cost and a microscopic marginal cost of reproduction, innovate without the protection of patents? But consider the following facts: Under current law, the chemical formula and the efficacy of the cure as established by clinical trials are made available to competitors essentially for free. About 80 percent of the initial fixed cost of drug development comes from Stage III clinical trials, a public good that legislation requires be privately produced. The downstream social cost of monopoly pricing of pharmaceutical products is highest for life-saving drugs, and the cost of monopoly pricing of other pharmaceutical products is also quite high. Given all this, various economists, such as Kremer and Williams (2009) , have argued that if government intervention is indeed needed in this market, a system of prizes might be superior to the existing system of monopolies.
+
+There are four things that should be born in mind in thinking about the role of patents in the pharmaceutical industry. First, patents are just one piece of a set of complicated regulations that include requirements for clinical testing and disclosure, along with grants of market exclusivity that function alongside patents. Second, it is widely believed that in the absence of legal protections, generics would hit the market side by side with the originals. This assumption is presumably based on the observation that when patents expire, generics enter immediately. However, this overlooks the fact that the generic manufacturers have had more than a decade to reverse-engineer the product, study the market, and set up production lines. Lanjouw's (1998) study of India prior to the recent introduction of pharmaceutical patents there indicates that it takes closer to four years to bring a product to market after the original is introduced — in other words, the first-mover advantage in pharmaceuticals is larger than is ordinarily imagined. Third, much development of pharmaceutical products is done outside the private sector; in Boldrin and Levine (2008b) , we provide some details. Finally, the current system is not working well: as Grootendorst, Hollis, Levine, Pogge, and Edwards (2011) point out, the most notable current feature of pharmaceutical innovation is the huge “ drought ” in the development of new products.
+
+With these four factors in mind, it is possible to make proposals for reforming the pharmaceutical industry along with the patent system. For example, we could either treat Stage II and III clinical trials as public goods (where the task would be financed by National Institutes of Health, who would accept bids from firms to carry out this work) or by allowing the commercialization of new drugs—at regulated prices equal to the economic costs of drugs—if they satisfy the Food and Drug Administration requirements for safety even if they do not yet satisfy the current (overly demanding) requisites for proving efficacy. In other words, pharmaceutical companies would be requested to sell new drugs at “economic cost” until efficacy is proved, but they could start selling at market prices after that. (It is ensuring
+
+---
+
+14 Journal of Economic Perspectives
+
+the efficacy—not the safety—of drugs that is most expensive, time-consuming, and difficult.) In this way, companies would face strong incentives to conduct or fund appropriate efficacy studies where they deem the potential market for such drugs to be large enough to bear the additional costs. The new policy could begin with drugs aimed at rare diseases, which, because of their small potential market, are not currently worth the costs of efficacy testing; without the new policy, they might never make it to market at all. If this new progressive approval approach works for rare diseases, it could be adopted across the board. Our broader point is that, rather than just ratcheting up patent protection, there are a number of moves we could make to reduce the risks and cost of developing new drugs.
+
+## The Political Economy of Patents
+
+We do believe, along with many of our colleagues, that a patent system designed by impartial and disinterested economists and administered by wise and incorruptible civil servants could serve to encourage innovation. In such a system, very few patents would ever be awarded: only those for which convincing evidence existed that the fixed costs of innovation were truly very high, the costs of imitation were truly very low, and demand for the product was really highly inelastic. (The curious reader may check Boldrin and Levine, 2008a, for a more detailed explanation as to why these three conditions need to be satisfied to make a patent socially valuable). There is little dispute, among these same colleagues, that the patent system as it exists is very far from satisfying such requirements and it is, in fact, broken. To quote a proponent of patents, Shapiro (2007): “A growing chorus of scholars and practitioners are expressing concerns about the operation of the US patent system. While there is no doubt that the US economy remains highly innovative, and there is no doubt that the patent system taken as a whole plays an important role in spurring innovation, the general consensus is that the US patent system is out of balance and can be substantially improved.” Actually, we believe the evidence is clear that the patent system taken as a whole does not play an important role in spurring innovation. But if a well-designed and welladministered patent system could serve the intended purpose, why not reform it instead of abolishing it?
+
+To answer the question we need to investigate the political economy of patents: why has the political system resulted in the patent system we have? Our argument is that it cannot be otherwise: the “optimal” patent system that a benevolent economist–dictator would design and implement is not of this world. It is of course fine to recommend patent reform. But if political economy pressures make it impossible to accomplish that reform, or if they make it inevitable that the patent system will fail to meet its goals, then abolition—preferably by constitutional means as was the case in Switzerland and the Netherlands prior to the late nineteenth century—is the proper solution. This political economy logic brings us to advocate dismantlement of the patent system.
+
+---
+
+The Case Against Patents 15
+
+The political economy of patent protection is shaped by many players, but “consumers” are not prominent among them. On one side, the side of the potential patentees, there are individual inventors, corporate inventors, and patent trolls. Other players include the patent office, the patent lawyers who file and litigate patents, and the courts where the litigation takes place. The rules of the game are established by some combination of legislation, judicial action, and custom. But because patenting is a technical subject about which few voters know anything with clarity, interests of voters are not well represented. In many spheres of government regulation, this lack of representation for voters has often led to “regulatory capture”—as Stigler (1971) and other public choice theorists have argued—where regulators act in the interests of the regulated, not the broader public. Nowadays, if there is one “regulator” who is captured, it is the one in charge of regulating patents. To understand why, we need to understand the motivation and incentives of the relevant players.
+
+Let us start with the US Patent Office and the infamous “one-click” patent #5960411 issued to Amazon in September 1999. According to 35 U.S.C. 103, the statute under which the Patent Office operates, to obtain a patent “the differences between the subject matter sought to be patented and the prior art are such that the subject matter as a whole would have been not obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains . . .” Now consider the patent in question, which claims, among other things, a monopoly over:
+
+11. A method for ordering an item using a client system, the method comprising: displaying information identifying the item and displaying an indication of a single action that is to be performed to order the identified item; and in response to only the indicated single action being performed, sending to a server system a request to order the identified item whereby the item is ordered independently of a shopping cart model and the order is fulfilled to complete a purchase of the item.
+
+The idea of taking a single action to accomplish a goal is hardly innovative, and applying the idea of taking a single action to making a purchase is obvious to anybody who has ever used a soft drink machine. Purchases were already being made over the Internet in 1999. It was thus clear that orders would be made by a credit card, and either the credit card information would be provided at the time of the transaction, or stored in advance by the retailer. Either way, the user must identify itself when the purchase is made. Those obvious steps are exactly what Amazon describes in its patent, albeit with a few flow charts thrown into the eleven-page patent application. But through the fog of those flow charts, it is relatively easy to see that the verbal description of the single-click procedure applies equally well to what happens on the Amazon site and to what happens in front of millions of vending machines every day. The Amazon patent was reexamined by the US Patent Office starting in May 2006. After a preliminary finding that, indeed, “obvious” means “obvious” even
+
+---
+
+16 Journal of Economic Perspectives
+
+at the Patent Office, the office then reversed itself and in October 2007, reaffirmed the Amazon patent, albeit limiting its scope slightly. So we cannot dismiss such an absurd patent as an aberration.
+
+What lead the US Patent Office to interpret, essentially, the words “not obvious” as meaning “obvious”? The Patent Office is constantly under pressure from applicants and their lawyers to be more generous in issuing patents—that is, to adopt lower standards of obviousness and steeper standards for what is considered “prior art.” The following statement by David Kappos (2010), director of the US Patent Office concerning the allowance rate—what fraction of patents are accepted—is revealing: “Overall in FY 2010, the allowance rate increased to 45.6%, compared to an allowance rate of 41.3% in FY 2009 . . . So, while we still have a lot of work to do, I think we are on the right path.” Apparently, accepting a higher fraction of patents applications is defined as “the right path.” Talk about “regulatory capture”!
+
+Patent lawyers play a large role in the political economy of patents. According to Quinn (2011), who is a patent attorney, legal fees for filing a patent run upwards of $7,000 and roughly half are rejected. In 2010, according to the US Patent Office, 244,341 patents were issued, which would imply roughly $3 billion in legal fees per year. Obviously, patent attorneys as a group have a tremendous incentive to see that more patents are issued. This insight helps us understand better the role of the courts and their relatively recent reform. In 1982—lobbied by patent lawyers— Congress passed the Federal Courts Improvement Act, which moved federal patent appeals out of the regular court system to a special court system for dealing with patents. Naturally, many of the judges for this new court were chosen from the ranks of patent attorneys. For example, when a court voted, in a 1994 decision, to expand the scope of patents to software (In re Kuriappan P. Alappat, Edward E. Averill and James G. Larsen 33 F.3d 1526 [July 29, 1994]), of the six judges who voted in favor, half had previously been patent attorneys, while of the two that voted against, neither had been. The referee of the patent game is biased both materially and ideologically. As Landes and Posner (2004, p. 26) write in their discussion of the political economy of patents: “That has been the experience with the Federal Circuit; it has defined its mission as promoting technological progress by enlarging patent rights.”
+
+Notice, too, that many patent lawsuits have a public goods aspect. Consider a case in which the plaintiff is asserting that its patent has been infringed. If the plaintiff wins the lawsuit, by confirming its monopoly position it appropriates all the benefits of winning the lawsuit. A victory by the defendant, by contrast, benefits partly itself, but also other firms that might be sued by the plaintiff for patent infringement as well as consumers who would have a more competitive market. Thus, the defendant receives only a slice of the overall benefits from winning the lawsuit, and will be willing to spend less on such lawsuits than it would if it were to receive all the benefits. This dynamic is nothing but the patent court version of the (already noted) fundamental asymmetry in the distribution of economic incentives that defines the foundations of the political economy of patent law.
+
+Finally, political economy can be influenced by how standard terminology frames a problem. Landes and Posner (2004) point out that there is an “ideological”
+
+---
+
+Michele Boldrin and David K. Levine 17
+
+argument in support of stronger patent rights: supporters of free markets tend to favor institutions of private property, and patents and copyright are intellectual “property.” Hence, strengthening them is ideologically and politically consistent with the general principle that “private property is good for growth.” But as we (Boldrin and Levine 2008b) and many others elsewhere have argued, patents are just a monopoly, not property.
+
+Given this set of players and their incentives, the patent game moves naturally towards its equilibrium, as we have observed over time. Two centuries or so ago, patents were restricted in their areas of applicability and limited in both depth and duration over time; they were somewhat “reasonable,” to the extent social gains and costs seemed balanced. But we have witnessed a steady process of enlargement and strengthening of patent laws. At each stage, the main driving force was the rent-seeking efforts of large, cash-rich companies unable to keep up with new and creative competitors. Patent lawyers, patent officials, and wannabe patent trolls usually acted as foot soldiers. While this political economy process is pretty straightforward in broad terms, we are still missing an empirical, quantitative analysis of the stakes involved and of the gains and losses accruing to both the active players and to the rest of society, from the general public to the innovators that never emerged due to preexisting patent barriers.
+
+Perhaps surprisingly, despite the key importance of political economy in understanding why we have the patent system we have, economists have had relatively little to say on the subject. The few prominent papers that we know of on this subject typically build from analyses very similar to what we have presented here—but then shy away from drawing the logical conclusions.
+
+For example, Landes and Posner (2004) recognize that patent laws are mostly designed by interest groups keen to increase their monopoly rents, not aggregate welfare, and that this drove the enormous growth in patent legislation and judiciary activity during the last 30 years. The more elaborate writing by Scherer (2009) on “The Political Economy of Patent Policy Reform in the United States” follows a similar approach. It focuses on the fact that “government emphasis on patent systems increased” while academic research was starting to become more and more aware that patents are playing a minor positive role, if any at all, in creating incentives for high R&D and in fostering productivity growth. After providing a concise and very well-informed historical survey of all major changes in US patent policies over the last century or so, Scherer (p. 195) wonders why the political system would increase patent protection so much in light of the fact “that the record of debates on the enabling bill contains no solid evidence that the change would in fact stimulate R&D, and that there is no evidence of an acceleration in company-financed R&D between the 27 years before the bill was enacted and the 18 years thereafter.” He then extends the same argument to the international arena, paying particular attention to the case of pharmaceutical patents. While Scherer’s language and arguments are strongly critical of current trends in patents, he does not seek to explain why an institution, such as the patent system, that was supposed to be theoretically sound would degenerate into something so socially damaging over same 30-year
+
+---
+
+18 Journal of Economic Perspectives
+
+period that academic researchers were realizing the institution's limitations and potential dangerousness.
+
+In our view, even insightful writers such as Landes and Posner (2004) and Scherer (2009) seem unable to shake themselves free of the belief that patents are essential in fostering innovation and that any problems can be fixed with some tweaks to the patent system; they fail to seriously consider the possibility of intrinsic problems with the design of the institution itself. This belief in patents flies in the face of the structural realities: Marginal extensions of patents result in substantially higher per capita rents for the few holders of the right while marginally reducing the individual welfare of the much larger number of nonpatent holders. The rent of the monopolist is a lot higher than an individual consumer's deadweight loss, so the monopolist has an incentive to perpetuate the system while the individual consumer has no incentive to fight it. Those who possess a patent do not hold a “property right” in the conventional sense of that term, but they do hold a socially granted “monopoly” right, and will tend to leverage whatever initial rents their monopoly provides in order to increase their monopoly power until all potential rents are extracted (and, in all likelihood, also largely dissipated by the associated lobbying and transaction costs). This scenario helps explain how patents interact with the industry lifecycle—why patents are either ignored or scarcely used in new and competitive industries, while being highly valued and overused in mature and highly concentrated ones.
+
+## Conclusion
+
+In 1958, the distinguished economist Fritz Machlup in testimony before Congress famously said: “If we did not have a patent system, it would be irresponsible, on the basis of our present knowledge of its economic consequences, to recommend instituting one. But since we have had a patent system for a long time, it would be irresponsible, on the basis of our present knowledge, to recommend abolishing it.” A proposal to abolish patents may seem “pie in the sky.” Certainly, many interim measures could be taken to mitigate the damage caused by the current system: for example, properly enforcing the standard that patents should only be granted for nonobvious insights; requiring genuine disclosure of working methods in patents (the opposite of certain recent “protectionist” proposals to institute secret patents); and allowing an “independent invention” defense against claims of patent infringement. But why use band-aids to staunch a major wound? Economists fought for decades—ultimately with considerable success—to reduce restrictions on international trade. A similar approach, albeit less slow, should be adopted to phase out patents. Because policy proposals are often better digested and metabolized in small bites, here is our list of small reforms that could be easily implemented.
+
+- 1) Patents are time limited, which makes it relatively easy to phase them out by
+phasing in ever shorter patent durations. This conservative approach also
+---
+
+The Case Against Patents 19
+
+has the advantage that if reducing patent terms indeed has a measurable effect on innovation, the process can be reversed.
+
+2) Stop the rising tide that, since the early 1980s, has extended the set of what can be patented and has shifted the legal and judicial balance substantially in favor of patent holders.
+
+3) Because competition fosters productivity growth, antitrust and competition policies should seek to limit patents when they are hindering innovation. This policy may be of particular relevance for high-tech sectors, from software to bioengineering, to medical products and pharmaceuticals.
+
+4) Current international trade negotiations that affect patents often occur as part of either the Agreement on Trade-Related Aspects of Intellectual Property Rights ( TRIPS ), which was signed in 1995 as part of the World Trade Organization negotiations, or as part of the World Intellectual Property Organization, an agency of the United Nations. The nature of these agreements and organizations is well indicated by the use of the propaganda term “intellectual property” in their titles. In both cases, these talks are often focused on how to prevent ideas from high-income countries from being used in low-income countries—what we would characterize as essentially a neomercantilist approach toward free trade in goods and ideas. We should be highly cautious about this agenda. Within a couple of decades, the “balance of trade in ideas” between the US and European economies and emerging economies in Asia might easily equalize or reverse. Engaging in “mercantilism of ideas” may seem favorable to certain large US firms now, but such rules may become costly to the US economy if they are applied to protect patents held in the future by producers in the now-developing Asian economies.
+
+5) If the US economy is to have patents, we may want to start tailoring their length and breadth to different sectoral needs. Substantial empirical work needs to be done to implement this properly, although a vast legal literature is already pointing in this direction.
+
+6) Patents should not be granted based only on technological insights, but should also take economic evidence into account. For example, if an invention is easy to copy or has a high fixed cost, then patent protection to provide an incentive for the inventor may be more suitable. Ultimately, patents should be awarded only when strictly needed on economic grounds, as spelled out earlier.
+
+7) We advocate returning to the rule prior to the Bayh – Dole Act of 1980 according to which the results of federally subsidized research cannot lead to patents, but should be available to all market participants. This reform would be particularly useful for encouraging the dissemination of innovation and heightening competition in the pharmaceutical industry.
+
+8) In several industries, notably pharmaceuticals, it would be useful to rethink all of the government policies that bear on incentives for invention. The broad point is that there are a number of ways to reduce the risks and cost of developing new drugs, rather than just trying to ratchet up patent protection.
+
+---
+
+20 Journal of Economic Perspectives
+
+In general, public policy should aim to decrease patent monopolies gradually but surely, and the ultimate goal should be the abolition of patents. After six decades of further study since Machlup's testimony in 1958 has failed to find evidence that patents promote the common good, it is surely time to reassess his conclusion that it would be irresponsible to abolish the patent system. The patent system arose as a way to limit the power of royalty to award monopolies to favored individuals; but now its primary effect is to encourage large but stagnant incumbent firms to block innovation and inhibit competition.
+
+- ■ We are grateful to the editors, the referees, and to Richard Stallman for a careful reading
+and comments.
+## References
+
+Aghion, Phillipe, Nick Bloom, Richard Blundell, Rachel Griffith, and Peter Howitt. 2005. “Competition and Innovation: An Inverted-U Relationship.” Quarterly Journal of Economics 120(2): 701–728.
+
+Boldrin, Michelle, Juan Correa Allamand, David K. Levine, and Carmine Ornaghi. 2011, “Competition and Innovation.” In Cato Papers on Public Policy , Vol. 1, edited by J. A. Miron, 109–172. Cato Institute.
+
+Boldrin, Michele, and David K. Levine. 2004. “Rent Seeking and Innovation.” Journal of Monetary Economics 51(1): 127–60.
+
+Boldrin, Michele, and David K. Levine. 2005. “The Economics of Ideas and Intellectual Property.” Proceedings of the National Academy of Sciences 102(4): 1252–56.
+
+Boldrin, Michele, and David K. Levine. 2008a. “Perfectly Competitive Innovation.” Journal of Monetary Economics 55(3): 435–53.
+
+Boldrin, Michele, and David K. Levine. 2008b. Against Intellectual Monopoly. Cambridge University Press.
+
+Bessen, James, and Michael J. Meurer. 2005. “The Patent Litigation Explosion.” BU School of Law Working Paper 05–18, Boston University.
+
+Bessen, James, and Michael J. Meurer. 2008. Patent Failure: How Judges, Bureaucrats, and Lawyers Put Innovators at Risk. Princeton University Press.
+
+Brec, Eric. 2008. “NIHilism and Other Innovation Poison.” MSDN Blogs, November 1. http:// blogs.msdn.com/b/eric_brechner/archive /2008/11/01/nihilism-and-other-innovation -poison.aspx.
+
+Chaudhuri, Shubham, Pinelopi K. Goldberg, and Panie Gia. 2006. "Estimating the Effects of Global Patent Protection in Pharmaceuticals: A Case Study of Quinolones in India." American Economic Review 96(5): 1477–1514.
+
+Cohen, Wesely M., Richard R. Nelson, and John P. Walsh. 2000. "Protecting Their Intellectual Assets: Appropriability Conditions and Why U.S. Manufacturing Firms Patent (or Not)." NBER Working Paper 7552.
+
+Correa, Juan A. 2012. “Innovation and Competition: An Unstable Relationship.” Journal of Applied Econometrics 27(1): 160–66.
+
+Danaher, Brett, Samita Dhanasobhon, Michael D. Smith, and Rahul Telang. 2010. “Converting Pirates without Cannibalizing Purchasers: The Impact of Digital Distribution on Physical Sales
+
+---
+
+Michele Boldrin and David K. Levine 21
+
+and Internet Piracy." March 3. Available at SSRN: http://ssrn.com/abstract=1381827 or http://dx .doi.org/10.2139/ssrn.1381827.
+
+DiMasi, Joseph A., Ronald W. Hansen, and Henry G. Grabowski. 2003. "The Price of Innovation: New Estimates of Drug Development Costs." Journal of Health Economics 22(2): 151–85.
+
+Gallini, Nancy T. 2002. "The Economics of Patents: Lessons from Recent U.S. Patent Reform." Journal of Economic Perspectives 16(2): 131-154.
+
+Gates, Bill. 1991. "Challenges and Strategy." Memo, Microsoft Corporation, May 16. http:// www.std.com/obi/Bill.Gates/Challenges.and .Strategy.
+
+Grootendorst, Paul, Aidan Hollis, David K. Levine, Thomas Pogge, and Aled M. Edwards. 2011. “New Approaches to Rewarding Pharmaceutical Innovation.” CMAJ: Canadian Medical Association Journal 183(6): 681–85.
+
+Hall, Bronwyn H., and Rosemarie Ham Ziedonis. 2007. "An Empirical Analysis of Patent Litigation in the Semiconductor Industry." http://citeseerx.ist .psu.edu/viewdoc/summary?doi=10.1.1.69.5271.
+
+Hashmi, Aamir Rafique. 2011. “Competition and Innovation: The Inverted-U Relationship Revisited.” Available at SSRN: http://ssrn.com /abstract=1762388 or http://dx.doi.org/10.2139 /ssrn.1762388.
+
+Heller, Michael. 2008. The Gridlock Economy: How Too Much Ownership Wrecks Markets, Stops Innovation, and Costs Lives. Basic Books.
+
+Hong, Sungook. 2001. Wireless: from Marconi's Black-Box to the Audion. MIT University Press.
+
+Jaffe, Adam B. 2000. "The U.S. Patent System in Transition: Policy Innovation and the Innovation Process." Research Policy 29(4-5): 531-57.
+
+Jovanovic, Boyan, and Glenn M. MacDonald. 1994. "The Life Cycle of a Competitive Industry." Journal of Political Economy 102(2): 322-47.
+
+Kanwar, Sunil, and Robert Evanson. 2001. “Does Intellectual Property Protection Spur Technological Change?” Levine’s Working Paper Archive, no. 122247000000000455.
+
+Kappos, David. 2010. "Reflections on the USPTO Dashboard." Director's Forum: David Kappo's Public Blog, October 13. http://www.uspto .gov/blog/director/entry/reflections_on_the _uspto_dashboard.
+
+Kremer, Michael, and Heidi Williams. 2009. “Incentivizing Innovation: Adding to the Toolkit.” In Innovation Policy and the Economy , Vol. 10, edited by Josh Lerner and Scott Stern, 1–17. Chicago University Press.
+
+Landes, William M., and Richard A. Posner. 2004. Political Economy of Intellectually Property Law. Washington, DC: AEI-Brookings Joint Center for Regulatory Studies.
+
+Lanjouw, Jean O. 1998. "The Introduction of Pharmaceutical Product Patents in India: Heartless Exploitation of the Poor and Suffering?" NBER Working Paper 6366.
+
+Lerner, Josh. 2002. "Patent Protection and Innovation over 150 Years." http://www.epip.eu /papers/20030424/epip/papers/cd/papers _speakers/Lerner_Paper_EPIP_210403.pdf (quotes are from this version). Abridged version published in 2002 as "150 Years of Patent Protection," American Economic Review 92(2): 221-25.
+
+Lerner, Josh. 2009. "The Empirical Impact of Intellectual Property Rights on Innovation: Puzzles and Clues." American Economic Review 99(2): 343-48.
+
+Levin, Richard C., Alvin K. Klevorick, Richard R. Nelson, and Sidney G. Winter. 1987. "Appropriating the Returns from Industrial Research and Development." Brookings Papers on Economic Activity, no. 3, Special Issue on Microeconomics, pp. 783–820.
+
+Llanes, Gastón, and Stefano Trento. 2009. “Patent Policy, Patent Pools, and the Accumulation of Claims in Sequential Innovation.” Economic Theory 50(3): 703–25.
+
+Machlup, Fritz. 1958. “An Economic Review of the Patent System.” Study commissioned by the Senate Judiciary Subcommittee on Patents, Trademarks, and Copyrights, 85th Congress, 2nd session. http://mises.org/document/1182/.
+
+Niccolai, James. 2012. "Android Developers Ignored Sun Patents, Google Exec Testifies." Computerworld, May 9.
+
+Nuvolari, Alessandro. 2004. "Collective Invention during the British Industrial Revolution: The Case of the Cornish Pumping Engine." Cambridge Journal of Economics 28(3): 347–63.
+
+Nuvolari, Alessandro. 2006. "The Making of Steam Power Technology: A Study of Technical Change during the British Industrial Revolution." Journal of Economic History 66(2): 472-76. Cambridge University Press.
+
+Ponce, Carlos J. 2007. "More Secrecy ... More Knowledge Disclosure? On Disclosure Outside of Patents." Levine's Working Paper Archive, no. 122247000000001600.
+
+Quinn, Gene. 2011. "The Cost of Obtaining a Patent in the US." IPWatchdog.com, January 28. http://www.ipwatchdog.com/2011/01/28/the -cost-of-obtaining-patent/id=14668/.
+
+Salmon, Felix. 2012. "Why Patent Trolls Don't Need Valid Patents." March 4. Reuters. http:// blogs.reuters.com/felix-salmon/2012/03/04 /why-patent-trolls-dont-need-valid-patents/.
+
+Sanchez, Julian. 2012. "How Copyright Industries Con Congress." Posted at Cato@Liberty, http://www.cato-at-liberty.org/how-copyright -industries-con-congress/.
+
+---
+
+22 Journal of Economic Perspectives
+
+Scherer, F. M. 1990. Industrial Market Structure and Economic Performance. Houghton Mifflin Company.
+
+Scherer, F. M. 2009. "The Political Economy of Patent Policy Reform in the United States." Journal on Telecommunications and High Technology Law 7(2):167-216.
+
+Shapiro, Cark. 2007. "Patent Reform: Aligning Reward and Contribution." NBER Working Paper 13141.
+
+Shulman, Seth. 2003. Unlocking the Sky: Glenn Hammond Curtiss and the Race to Invent the Airplane. Harper Perennial.
+
+Sinha, Rajiv K., Fernando S. Machado, and Collin Sellman. 2010. "Don't Think Twice, It's All Right: Music Piracy and Pricing in a DRM-Free Environment." Journal of Marketing 74(2): 40–54.
+
+Stigler, George J. 1956. "Industrial Organization and Economic Progress." In The State of the Social Sciences, edited by Leonard D. White, 269-82. University of Chicago Press.
+
+Stigler, George J. 1971. "The Theory of Economic Regulation." Bell Journal of Economics and Management Science 2(1): 3-21.
+
+Svensson, Peter. 2006. “Long-Running Blackberry Patent Battle Ends with $612.5 Million Settlement.” The Daily Reporter (Associated Press), March 4. http://news.google.com/newspapers? nid=1907&dat=20060304&id=5bVGAAAAIBAJ& sjid=0_0MAAAAIBAJ&pg=4540,273762.
+
+US Central Intelligence Agency. n.a. The 1990 CIA World Fact Book. Available from Project Gutenberg: https://ia700401.us.archive.org/32/items /the1990ciaworldf00014gut/world12.txt.
+

processed/00027_W2132240807/document.md

@@ -0,0 +1,378 @@
+Journal of Economic Perspectives-Volume 27, Number 1-Winter 2013-Pages 45-66
+
+# The New Patent Intermediaries: Platforms, Defensive Aggregators, and Super-Aggregators
+
+Andrei Hagiu and David B. Yoffie
+
+Some assets are traded in liquid markets, at transparent prices, with the help of many thriving intermediaries: houses and apartments, stocks and other financial products, books, DVDs, electronics, and all sorts of collectibles. Intellectual property in general and patents in particular—the focus of this paper— are not among those assets (Gans and Stern 2010). The patent market consists mainly of bilateral transactions, either sales or cross-licenses, between large companies. Such deals are privately negotiated and might involve hundreds or thousands of patents. For example, in June 2011, a consortium of Apple, Microsoft, Sony, and several other large tech companies outbid Google to buy Nortel's 6,000 patents and patent applications for $4.5 billion. Google responded first by buying over 1,000 patents from IBM for an undisclosed price, and then by acquiring Motorola Mobile and its more than 17,000 patents for $12.5 billion. In April 2012, Microsoft bought 925 patents from AOL for $1.1 billion, then sold a portion of that portfolio to Facebook for $550 million. And in September 2012, Samsung lost a $1 billion judgment to Apple and faced a potential injunction from a federal judge in a jury trial over patent infringement. The very real threat of adverse jury rulings or injunctions, which might lead to partial or total shutdown of existing businesses, have led to extremely high willingness-to-pay for some intellectual property.
+
+Outside of these bilateral deals, patent buyers and sellers frequently have a hard time finding each other. There is no eBay, Amazon, New York Stock Exchange, or Kelley's Blue Book equivalent for patents, and when buyers and sellers do find
+
+■ Andrei Hagiu is Associate Professor and David B. Yoffie is Max and Doris Starr Professor of International Business Administration, both in the Strategy Unit at the Harvard Business School, Boston, Massachusetts. Their email addresses are ahagiu@hbs.edu and dyoffie @hbs.edu.
+
+http://dx.doi.org/10.1257/jep.27.1.45.
+
+doi=10.1257/jep.27.1.45
+
+---
+
+46 Journal of Economic Perspectives
+
+each other, they usually negotiate under enormous uncertainty: prices of similar patents vary widely from transaction to transaction and the terms of the transactions (including prices) are often secret and confidential.
+
+Inefficient and illiquid markets, such as the one for patents, generally create profit opportunities for intermediaries. In this paper, we begin with an overview of the problems that arise in patent markets, and how traditional institutions like patent brokers, patent pools, and standard-setting organizations have sought to address them. But during the last decade, a variety of novel patent intermediaries has emerged. We will discuss how several online platforms have started services for buying and selling patents but have failed to gain meaningful traction. However, new intermediaries that we call defensive patent aggregators and superaggregators have become quite influential and controversial in the technology industries they touch. In particular, the rising prominence of a new and powerful patent aggregator called Intellectual Ventures has sparked heated debates about the economic role played by intermediaries in the patent market and their effects on innovation. $^1$
+
+One might expect that new intermediaries and competition between them could lead to increased market efficiency. Sometimes, however, intermediaries are able to exploit market inefficiencies without contributing much social value or, worse, they might even exacerbate existing market failures. The goal of this paper is to shed light on the role and efficiency tradeoffs of these new patent intermediaries. In the conclusion, we offer a provisional assessment of how the new patent intermediary institutions affect economic welfare.
+
+## Patent Market Failures and Traditional Patent Intermediaries
+
+Why is the market for patents so illiquid and inefficient? While the root causes are well-known to economists and are a subset of market failures that arise in many markets for ideas, it is useful to summarize them briefly here, highlighting the issues most relevant for patent intermediaries. Gans and Stern (2010) offer a review of market failures in the market for ideas, many of which apply to patents.
+
+First, patents are much more difficult to value than most other goods. This problem arises not simply because patents are intangible assets: after all, intangibles such as brand equity are routinely valued. What sets patents apart is that every patent
+
+1 Because our notion of a patent intermediary is an organization (firm or not-for-profit entity) that directly facilitates the sale or licensing of patents from owners-creators to users, we will not discuss here the patent rating, valuation, and search services that aim to create liquidity indirectly by providing useful patent information. An example of such a service is ArticleOne Partners (http://www.articleonepartners .com/). In addition, we focus specifically on patent intermediaries as opposed to other forms of intellectual property and more general notions of markets for technology (Arora, Fosfuri, and Gambardella 2001) and for ideas (Gans and Stern 2010). Thus, our study does not cover firms like InnoCentive and NineSigma, which connect companies with individuals or institutions that can create pre-patent solutions to science or technology problems.
+
+---
+
+Andrei Hagiu and David B. Yoffie 47
+
+is by definition unique: they lack “comparables,” which are used in many markets to estimate a given asset's value. More importantly, patent value in many modern technologies is subject to strong complementarities and portfolio effects (Gans and Stern 2010; Parchomovsky and Wagner 2005) . The issue of complementarities arises because in industries like semiconductors and smart phones, products are covered by dozens or even hundreds of interdependent patents. As a result, the value of individual patents is heavily discounted. Potential buyers or licensees may not place much value on a given patent sold by itself unless it complements a portfolio that they already own. This greatly reduces the number of buyers and the potential for liquidity. Portfolio effects create asymmetries between large operating firms on one side and individual inventors and small companies on the other side ( Jaffe and Lerner 2004 ) . There is a lower probability for smaller inventors to monetize their patents because they lack a large portfolio and because their owners typically have limited financial resources and legal expertise, which severely undermines their ability to bargain effectively. A well-known example (and the subject of the 2008 movie Flash of Genius ) is that of engineer Robert Kearns, who in 1964 applied for a patent for an intermittent windshield wiper system for automobiles. Manufacturers refused Kearns's requests to sign licensing agreements and began producing cars featuring the wiper system in 1969. Kearns spent decades battling in court for infringement. He eventually earned $ 30 million in settlements from Ford and Chrysler but, in the process, lost his job, divorced, and suffered multiple nervous breakdowns (Schudel 2005) .
+
+Second, both sides of the patent market face high search costs. For patent owners, it is prohibitively costly to find all current users (actual infringers) and all potential applications of their patents. For potential patent buyers or users, it is very costly to find all prior art and patents that “read on” (that is, that might cover the technology within) their products, especially when these products are complex and rely on fast-changing technologies. Indeed, although patent offices around the world as well as private databases provide comprehensive and searchable lists of all patents issued, patent applicants typically seek to disclose only the minimum necessary to obtain the patent, and use language that is oftentimes broad and opaque. This makes it very difficult to figure out their relationship with other patents and prior art, particularly with millions of patents in circulation. To illustrate, consider Apple's “bounce-back” utility patent, which was highly publicized during the recent Apple vs. Samsung trial settled in August 2012 before a California jury (Gallagher 2012). This patent essentially covers a method for allowing users to scroll beyond the edge of an image, webpage, or list and have it bounce back onto the screen. Despite the highly intuitive nature of this functionality, it is quite hard to identify its scope and the way it is meant to be implemented from the language used in the actual patent. $^2$
+
+2 The patent number is 7469381, and its detailed description is available from the US Patent Office at http://patft.uspto.gov/netacgi/nph-Parser?Sect2=PTO1&Sect2=HITOFF&p=1&u=/netahtml/PTO /search-bool.html&r=1&f=G&l=50&d=PALL&RefSch=yes&Query=PN/7469381.
+
+---
+
+48 Journal of Economic Perspectives
+
+Third, patent transactions always happen in the shadow of litigation, which exacerbates valuation problems and creates large transaction costs. Litigation often implies that patents are best viewed as “ probabilistic property rights ” or “ lottery tickets ” (Lemley and Shapiro 2005) : few patents are litigated, but of those that are, approximately half end up being invalidated. Given this risk, many patent owners and users prefer to settle out of court for amounts that have more to do with their opportunity costs of going to trial and their attitude towards risk than with the “ true ” economic value of their patents. Is the plaintiff a small company or individual with limited resources who prefers to settle for a small amount rather than face the possibility of years of litigation? What about a competitor who can be countersued and brought to accept a cross-licensing agreement? Or what about a “ nonpracticing entity ” against which injunctions that they cannot produce the product will not work — because the entity doesn't produce in the first place. Furthermore, some courts have a reputation for bias in favor of small players and against large companies, which makes them attractive patent litigation forums for small players and nonpracticing entities. For example, the Eastern District of Texas received 25 percent of all US patent infringement cases during 2011 and found in favor of patent owners almost 75 percent of the time (Decker 2012) . The prospect of choosing a favorable court setting increases the amount of (inefficient) litigation.
+
+The complexity that arises when valuation is intertwined with litigation has been heightened by the emergence of the US International Trade Commission (ITC) as a new forum for patent battles. The ITC is an independent federal agency with powers to do investigations and fact-finding on international trade issues, including import subsidies, dumping cases, and also issues of whether an imported product infringes on existing intellectual property. The ITC typically decides much faster than federal courts: often 12–15 months vs. several years in federal courts (Analysis Group, n.d.; Fisher 2006) . It also offers the possibility of relatively quick injunctive relief against defendants: it can require that the offending imports be halted, which can be even more effective in extracting monetary settlements. Qualcomm, for example, was forced to negotiate an $ 891 million settlement with Broadcom in 2009, after losing a case at the ITC and facing an import ban (Crothers 2009) . Thus, the ITC has significantly increased the costs of exposure to potential patent infringement lawsuits for firms in traded goods industries such as semiconductors, smartphones, and computers.
+
+These patent market failures are most problematic for individual inventors or small companies, who represent the majority of patent owners. One study, compiling data from a variety of public sources, found that inventors and small businesses contribute 60 percent of all patents in the United States, but only extract 1 percent of total licensing revenues. The remaining 99 percent of licensing revenue goes to large companies (Hagiu, Yoffie, and Wagonfeld 2011, exhibit 11). Of course, it is not shocking that large companies may tend to focus on higher-value patents, but the disjunction is nonetheless striking.
+
+One possible mechanism for small patent owners to address the problems of getting paid for their ideas would be to incorporate them in start-ups and seek either to compete with incumbent companies or to cooperate with them by licensing or
+
+---
+
+The New Patent Intermediaries 49
+
+being acquired (Gans, Hsu, and Stern 2002). In this way, investors, particularly venture capitalists, could mitigate some of these market failures. But many patents are not worth incorporating in a start-up, especially if they are not part of broader portfolios. Furthermore, great inventors are not necessarily great entrepreneurs (Wasserman 2012). In fact, it is arguably more efficient for inventors to specialize in invention rather than to pursue commercialization, a point argued by Lamoreaux and Sokoloff (2003) in the context of late nineteenth century United States and probably equally valid today.
+
+With so many difficulties facing inventors trying to monetize their discoveries, an obvious answer is to create intermediaries that facilitate the sale of patents to users (mainly operating companies), thereby maintaining appropriate incentives for innovation. In the next section, we discuss the three main traditional patent intermediaries studied in the existing economics literature: patent brokers/agents, patent pools, and standard-setting organizations. These traditional patent intermediaries have been around for a long time, but each faces certain limitations which prevent them from solving many of the patent market's problems.
+
+## Three Traditional Patent Intermediaries
+
+Patent brokers help patent owners sell or license their technologies in exchange for a fee contingent on successful transfer. Their activity helps reduce search and transaction costs by investing in specific knowledge and connections on both sides of the market. Brokers often facilitate not just the sale or licensing of patents, but broader technology transfers, which include patents and know-how. They also offer consulting services helping patent owners market and sell their assets. There are a large number of patent brokers, which tend to be small companies with fewer than 10 employees. Some examples include Thinkfire (http://www.thinkfire.com/), IPValue (http://www.ipvalue.com/), Pluritas (http://www.pluritas.com/), and Competitive Technologies (http://www.competitivetech.net/).
+
+Such patent brokers have existed since at least the nineteenth century: for example, Lamoreaux and Sokoloff (2003) document the positive effect of brokers on the US market for patented technology between 1870 and 1920. These brokers were typically patent agents or lawyers who matched inventors looking to sell new technologies with investors or buyers eager to commercialize them. At that time, however, there were few products encompassing hundreds of patented technologies like today: thus, the portfolio effects problem was less prevalent, and patents with fuzzy and overlapping boundaries were relatively rare. The job of modern patent brokers is much harder than those of a century ago. Unlike other markets for assets like stocks or real estate, the existence of many brokers in the patent market does not create sufficient liquidity on its own. Indeed, patent brokers are small in scale and tend to focus on facilitating highend licensing transactions that carry large price tags. Their fees are above 10 percent of the value of the transaction and sometimes reach 20–30 percent (Young 2008), a level high enough to suggest that inefficiencies prevail in the patent market.
+
+---
+
+50 Journal of Economic Perspectives
+
+Patent pools are formal or informal organizations in which for-profit firms come together to license patents to each other or to third parties (Lerner, Strojwas, and Tirole 2007; Shapiro 2001). Some common examples of patent pools include the historical example of the patent pool for sewing machines (see for example Lampe and Moser 2010), along with more recent technology patent pools such as Bluetooth and MPEG-4. Bluetooth is a technology standard for exchanging data over short distances; the corresponding pool brought together patents from 12 companies including Ericsson, IBM, Intel, Motorola, Nokia, and Toshiba (Layne-Farrar and Lerner 2011). MPEG-4 is a method for compressing audio-visual data; this pool contained 29 companies, including Apple, AT&T, Canon, France Telecom, Fujitsu, Hitachi, Microsoft, RealNetworks, and Sharp (Layne-Farrar and Lerner 2011). Patent pools emerged to solve the “multiple marginalization” problem—also known as “royalty-stacking”—which arises when multiple parties hold market power in a chain of production. If all parties attempt to exercise their market power to the fullest, the resulting prices will typically be above the level that would be set by a single party with market power—and the joint profits and social welfare will be lower than in the case of a single party with market power.
+
+While patent pools can create social value by reducing royalty stacking, it is not clear how well they address the traditional problems of patent markets. First, if patents included in a pool are substitutes rather than complements, the pool may turn out to have anticompetitive effects in the form of higher prices: the pool facilitates price collusion at the expense of price competition (Shapiro 2001; Lerner and Tirole 2004) . Second, patent pools can create barriers to entry and innovation, favoring large companies with sizable patent portfolios who are members of the patent pool and discriminating against small companies or individual inventors who find it hard to negotiate their way into the pool. Third, the applicability of patent pools is limited to a small number of markets, where the essential intellectual property to producing a specific product or service is more or less evenly distributed among several large, identifiable players.
+
+Similarly, standard-setting organizations have made it possible for participants in industries where there is an important need for interoperability between many components to come together and voluntarily produce consensus technical standards. Standard-setting organizations create economic value by enabling coordination on (Simcoe 2012) and certification of (Chiao, Lerner, and Tirole 2007; Lerner and Tirole 2006) technical standards. When these organizations endorse a specific technological standard, participants in the relevant industries typically adopt that standard and agree to cross-license or to pay the required royalties to the standard owner(s). The technological standard usually consists of many patents, owned by a patent pool, or on rare occasions by one company or institution. The process of choosing and certifying standards, however, is often subject to conflicts of interest due to interference by large producers (Schmalensee 2009). Furthermore, the scope of standard-setting organizations is limited to a small number of industries and technologies relative to the size of the broad patent market.
+
+---
+
+Andrei Hagiu and David B. Yoffie 51
+
+Thus, while brokers, patent pools, and standard-setting organizations have a role in bridging some of the gaps in the patent market, their effects are limited, and they have not managed to help small inventors get paid for their ideas. Indeed, small patent owners generally do not participate in pools or standardsetting organizations, and most small patent owners are not worth the time of professional brokers.
+
+## Exploiting Market Failures: Nonpracticing Entities
+
+The general lack of depth in patent markets has created a particularly favorable environment for the so-called “nonpracticing entities,” which have become the most controversial patent intermediaries. In essence, nonpracticing entities act as arbitrageurs, first acquiring patents, typically from individual inventors or small companies, and then seeking licensing revenues from operating companies through litigation or the threat of litigation. These entities do not innovate themselves, nor do they produce output. In 2001, nonpracticing entities brought 144 lawsuits targeting over 578 operating companies; by 2011, the numbers had increased to 1,211 lawsuits targeting 5,031 operating companies respectively (according to PatentFreedom research at https://www.patentfreedom.com/research-lot.html).
+
+Two main factors account for the explosion in activity of nonpracticing entities. First, the Internet has greatly reduced transaction costs for inventors to find intermediaries to whom they can sell their patents (Spulber 2011). Although nonpracticing entities appeared in the second half of the 1990s, the way they found undervalued intellectual property assets at that time was largely serendipitous—for example, through personal connections to inventors or sales of distressed assets containing obscure patents. Today, with a quick Internet search, any inventor can locate nonpracticing entities directly or contact brokers who can help one do so (Lohr 2009).
+
+Second, the value and prominence of patents have increased along with the revenues and profits associated with intellectual-property-intensive businesses. This growth was fueled in large part by the explosion of the information and communication technology sectors in areas like software, semiconductors, and mobile communications. Not coincidentally, most of the activity of the nonpracticing entities is concentrated in those sectors. These industries produce complex products and services, which involve many interrelated processes and components. For example, manufacturing an integrated circuit requires hundreds of steps, with literally billions of transistors and thousands of complex algorithms. Consequently, the potential for newly issued patents to have “ fuzzy boundaries ” (in the sense of Besen and Meurer 2008) and to overlap with prior art is very high in these sectors. Furthermore, no firm — even the industry's largest ones — has more than 30 percent of the patents that cover semiconductor design and manufacturing. This fragmented ownership of the relevant intellectual property exacerbates the uncertainty regarding the merits of the many patents involved.
+
+---
+
+52 Journal of Economic Perspectives
+
+Contrast this situation with the pharmaceutical industry, where patents also play a crucial role, but the boundaries of intellectual property are much more clearly defined. Patent claims on new molecules are easily distinguishable from other patented molecules. Not surprisingly then, nonpracticing entities and other patent merchants have been largely absent from the pharmaceutical sector. In contrast, different patents on smartphone user interfaces oftentimes contain closely related claims. For example, the difference between a horizontal and a vertical swiping mechanism for unlocking a touchscreen smartphone leaves lots of room for interpretation. How a jury might construe these claims create big opportunities for nonpracticing entities.
+
+The arbitrage opportunities available to nonpracticing entities are sizable. As of 2010, the median price paid by nonpracticing entities for a patent was approximately $ 100,000 and the mean was $ 400,000 (according to PatentFreedom website, accessed December 2010). On the other side of the market, most patent settlements range between $ 50,000 and a few million dollars (Sharma and Clark 2008). In a few notable cases, however, nonpracticing entities have managed to extract hundreds of millions of dollars. The best-known example is a 2006 settlement in which Research in Motion (maker of the Blackberry smartphones) agreed to pay $ 612.5 million to NTP, a Virginia-based nonpracticing entity, which had sued Research in Motion for infringing on eight wireless email patents (Riordan 2004). While precise data on the distributions of prices paid and settlements received by nonpracticing entities is unavailable, it is useful to consider the following back-of-the-envelope calculation based on the numbers above. If 99/100 settlements are uniformly distributed between $ 50,000 and $ 1 million and 1/100 settlements are for $ 100 million, then the net expected payoff for a risk-neutral nonpracticing entity for purchasing a patent is approximately $ 1.15 million. Even after litigation costs, this offers an attractive arbitrage opportunity.
+
+Nonpracticing entities have attracted financing from investors looking for novel diversification opportunities with high returns. A number of hedge fund, venture capital, and private equity firms either invest in nonpracticing entities or approach small patent-holders directly, offering to finance lawsuits against operating companies in exchange for a cut of any resulting payments (for some examples, see Masnick 2009; Bergelt 2010; McCurdy 2009) .
+
+Nonpracticing entities are sometimes pejoratively known as “patent trolls.” The originator of the patent troll model is generally agreed to be the company TechSearch and its lawyer Raymond Niro. Beginning in the late 1990s, TechSearch originated the practice of buying up patents and suing companies for infringement to demand payments (Bario 2011). In 2001, Intel's in-house lawyer Peter Detkin referred to Niro as a “patent troll” and popularized the term. (Perhaps ironically, Detkin went on to cofound Intellectual Ventures, the largest nonpracticing entity today, which we discuss below.) The meaning of the term “patent troll” has evolved over time, and there is no commonly agreed-on definition. However, trolls are generally viewed as combining the following characteristics: 1) they acquire intellectual property assets, like patents, solely for the purpose of extracting payments
+
+---
+
+The New Patent Intermediaries 53
+
+from alleged infringers; 2) they do not do research or develop any technology or products related to their patents; and 3) they behave opportunistically by waiting until industry participants have made irreversible investments before asserting their claims (Lemley 2008; Schmalensee 2009) .
+
+In itself, buying and reselling patents solely for price arbitrage is not necessarily a harmful practice. One could even argue that it increases market efficiency by creating liquidity and a way for small patent owners to get paid, similar to the function performed by dealers and market-makers in financial markets (McDonough 2006; Schmalensee 2009; Spulber 2011). The main reason that nonpracticing entities can instead create economic harm is that they seek to extract disproportionate payments through two practices. First, they typically engage in “nuisance value” litigation: they sue many companies simultaneously for moderate amounts so that targets are more likely to settle instead of risking a costly and uncertain trial. Second, they attempt to hold up (or “ambush”) practicing companies by bringing the lawsuits at the most vulnerable times for the targets, like just before the release of a new product, when the target can ill afford a risky trial involving its new product shipments. Memory chip companies, for example, accused Rambus of ambushing the industry with litigation just after a new industry standard had been set (Schmalensee 2009). $^3$ These two practices exacerbate patent market inefficiencies. The net effect is to create perverse incentives for some small patent owners to seek out nonpracticing entities to acquire and enforce patents of questionable merit. In addition, the expansion of such lawsuits may well produce a defensive backlash by large operating companies against all small patent owners, even the ones that might have a legitimate and valuable claim.
+
+## Two-sided Patent Platforms: A Failed Solution
+
+In parallel with the increased activity of nonpracticing entities during the 2000s, a number of companies built two-sided platforms in an attempt to create more efficient ways to bring buyers and sellers of patents together. The goal of two-sided patent platforms was to facilitate patent transactions without taking title or ownership of the patents involved. Two main categories of such platforms have been attempted: online marketplaces and live auctions.
+
+Online patent marketplaces appeared as early as 1998, but replicating in the market for patents what eBay has done for collectibles has proven difficult. Some of the online portals dedicated to facilitating patent search and transactions have been shut down or renamed and redirected towards other services. 4 The online platforms
+
+3 The law does try to address this problem through the doctrine of “equitable estoppel,” which can bar enforcement of patents by someone who has deliberately waited until after an investment decision has been locked-in to assert patents. We are grateful to Douglas Melamed for bringing this to our attention.
+
+4 For instance, Patent License and Exchange (pl-x) was created in 1998 as an online intellectual property and licensing marketplace. By 2006 it had been renamed PLX Systems and completely dropped the marketplace idea; instead, it provided software solutions for business and financial management of
+
+---
+
+54 Journal of Economic Perspectives
+
+that are still independent have limited scale, and they function more as brokerage or consulting companies: two representative examples are Yet2 (http://www.yet2 .com) and Tynax (http://www.tynax.com). Both websites contain thousands of listings for both sides of the market. Sellers post detailed information about the patents they want to sell, along with any special conditions (for example, perhaps a license must be granted back to the seller) and without revealing their identity. Buyers can find information about patents that are in the market for sale, search by keywords and patent classes, and post descriptions of specific intellectual property assets in which they have an interest, also without revealing their identity. Both Tynax and Yet2 work with Fortune 500 companies, and for both, keeping the identities of buyers and sellers confidential is a key part of their value proposition. Furthermore, they employ various mechanisms like screening through upfront fees and disclosure requirements to mitigate adverse selection in which only weak patents are offered for sale (Dushnitsky and Klueter 2011)—a potentially serious concern for online trading platforms. Indeed, in the absence of fees, the ratio of low-quality to highquality products is very high on any online marketplace (for example, Craigslist). This clutter significantly raises search costs for buyers, which in turn disincentivizes high-quality product suppliers from participating. The problem is even worse for patents, because search costs are already very high.
+
+Despite the extensive listings on Yet2 and Tynax's online portals, no transactions are completed online. Instead, once a buyer or a seller expresses clear and credible interest in a posting, Tynax or Yet2 manages and facilitates the buyer – seller interaction offline through one of its dealmakers — who is an actual person. The majority of revenues come from commissions on completed transactions: $ 100,000 to $ 10 million for Tynax or 15 percent of licensing fees for Yet2. Thus, both Tynax and Yet2 remain essentially patent and technology brokerage firms.
+
+At first glance, auctions might have seemed like a useful mechanism for eliciting market valuations for patents. The fact that Chicago-based Ocean Tomo managed to organize ten live intellectual property auctions between April 2006 and June 2009 generated significant buzz and optimism regarding the potential for bringing liquidity to the patent market via platforms. These auctions functioned like other live auctions—for example, like art at Sotheby's and Christie's—with an auctioneer taking bids for each lot, which could be a single patent, copyright, trademark, or domain name right, or a bundle of such assets. The lots were sold to the highest bidder on condition that the highest bid exceeded the seller's reserve price.
+
+But the auctions struggled to gain traction. The total value of transactions through Ocean Tomo's ten intellectual property auctions was only $ 114.6 million (Jarosz, Heider, Bazelon, Bieri, and Hess 2010, p. 17) . This total is relatively small, especially when compared to, say, the billions of dollars spent on patent portfolios by
+
+intellectual property for the music and entertainment industry. Other online platforms for matching patent sellers or licensors with buyers or licensees that have disappeared include Open-IP.org, TechEx, PricewaterhouseCoopers' IPEX, and Ocean Tomo's "The Dean's List."
+
+---
+
+Andrei Hagiu and David B. Yoffie 55
+
+Google in 2011–12 alone. The average sales-to-listings ratio over all ten Ocean Tomo auctions was reportedly 38 percent, and the spring 2009 auction only sold six out of 85 lots listed ( Inside IP 2012 ). Part of the reason for the lack of activity in spring 2009 was the financial crisis, but all auctions had been characterized by low participation and little bidding (Jarosz, Heider, Bazelon, Bieri, and Hess 2010, p. 20–22). In June 2009, Ocean Tomo sold its transactions line of business (including auctions and the now-closed “The Dean’s List” online platform) to ICAP, an interdealer broker, for just $ 10 million (ICAP 2009). The live intellectual property auctions were subsequently revived in March 2010 under the joint brand ICAP–Ocean Tomo. The spring 2010 auction (the 11th overall) was reported to have generated $ 14.3 million in transaction value, including buyers' premiums (ICAP 2010).
+
+Thus, while the idea of creating two-sided platforms for matching and facilitating transactions between patent buyers and sellers is appealing in principle, so far none of these platforms has been able to gain significant traction. None is close to creating a sustainable eBay or Sotheby's for intellectual property. One might argue that Tynax and Yet2.com are creating the economic equivalent of Craigslist for patents, but little more. Why is it so hard to establish two-sided platforms for patent transactions?
+
+First, two-sided platforms that attempt to bring together buyers and sellers without ever taking possession of the goods being exchanged face a difficult chicken-and-egg problem. Unlike market-makers who buy and resell, two-sided platforms have to attract a critical mass of both buyers and sellers. Some online platforms managed to attract many listings, but (as pointed out above) they do not facilitate many actual transactions. Ocean Tomo's auctions never achieved sufficient scale to convince buyers and sellers that they would become an important venue for trading patents. Owners of valuable patents did not expect these platforms to offer attractive monetization opportunities for their assets compared to other options like licensing directly, selling to nonprofit entities and splitting the proceeds from litigation, or raising venture capital funding and incorporating. In turn, the lack of valuable patents meant that few large operating companies would participate actively, which confirmed the initial negative expectation of sellers-owners. A broad market was never created; instead a handful of nonprofit entities were very active as buyers in Ocean Tomo's auctions (Malek 2009) .
+
+Second, while online intellectual property platforms like Tynax and Yet2.com have generated some search cost reductions through their thousands of listings, they have been unable to create significant reductions in transaction costs. The sensitivity of intellectual property information and the need for “close-touch” and often in-person due diligence make potential buyers and sellers reluctant to reveal enough details for completing a patent transactions online. Of course, this is why Tynax and Yet2.com still function as offline brokers for the actual transactions. But if personal dealmakers have to be directly involved in each transaction, their business model cannot easily scale up at low marginal cost. Moreover, the final transaction prices and valuations are private information, which cannot be leveraged to create greater transparency and liquidity in the patent market.
+
+---
+
+56 Journal of Economic Perspectives
+
+Will two-sided patent platforms remain limited in scope and scale? Even if they overcome the chicken-and-egg hurdle of how to attract the high-quality and highvalue patents, patent platforms seem unlikely to solve the liquidity problems that plague the market for patents. Indeed, given the heterogeneity and strategic sensitivity of patent transactions, it is hard to see how one could create the equivalent of an eBay for patents. Furthermore, the strong complementarities and portfolio effects across modern patents imply that two-sided platforms are at an inherent disadvantage relative to other types of patent intermediaries who take ownership of patents and are able to exploit those complementarities directly. By definition, two-sided platforms cannot do so. That diagnosis does not rule out the emergence (or growth) of platforms specializing in reducing search costs—similar to Tynax and Yet2.com. There is value in being able to browse through thousands of patents, bundles of patents, and technologies wanted or for sale in one place and in a unified format. The official Patent Office listings—patents granted or under review and searchable patent abstracts—leave significant scope for quasi-brokers to further reduce search costs with better listings and search functionality. As pointed out above, many official patent abstracts are written in such a way as to discourage workarounds and to make the broadest possible claims, which often makes it hard to identify potential applications. In this context, firms such as Yet2 create their own abstracts written in clear language in order to help potential buyers assess the potential benefits of the patented technology they are investigating.
+
+## Defensive Aggregators and Super-aggregators
+
+The rise of nonpracticing entities combined with the failure of patent platforms to bring transparency and liquidity to the patent market (which might have reduced the arbitrage opportunities for nonpracticing entities) have posed a growing threat to operating companies. In response, two new novel patent intermediaries have emerged, which we call defensive aggregators and super-aggregators.
+
+### Defensive Aggregators
+
+There are currently two prominent defensive aggregators: RPX (a for-profit firm, publicly traded since May 2011) and Allied Security Trust (a not-for-profit). In essence, defensive aggregators offer an incomplete insurance policy against patent troll risk to large operating companies. Firms such as Barnes & Noble, Best Buy, Cisco, eBay, HTC, IBM, Intel, McAfee, Microsoft, NEC, Nokia, Panasonic, Research In Motion, Samsung, Sony, and Verizon pay RPX annual subscription fees ranging from $ 65,000 to $ 6.9 million, depending on operating income (as explained at RPX's website: http://www.rpxcorp.com/index.cfm?pageid=85, accessed May 2012). In exchange, RPX identifies patents that might threaten subscribers, acquires those patents (or the right to grant sublicenses) in the open market, and provides all of its subscribers with licenses to those patents. The patents owned by RPX are also made available for use in counterlawsuits against nonmembers who initiate litigation against members.
+
+---
+
+The New Patent Intermediaries 57
+
+Unlike a traditional insurance policy, RPX faces no liability if a subscriber is sued or loses a patent case.
+
+Allied Security Trust, known as AST, offers two main variations on the RPX model. First, RPX decides unilaterally (sometimes in consultation with members) which patents to buy and uses its own capital to do so, while AST identifies patents or portfolios of patents and then solicits acquisition bids from its subscribers, who are also its governing members. Within AST, the bids and the identity of the bidders are kept secret from one another, and each member is required to have sufficient funds in an escrow account in order to support every bid it makes (as explained at the Allied Security Trust website at http://www.alliedsecuritytrust.com/Services /AcquisitionModel.aspx). If the sum of the bids for a particular set of patents is sufficient to close the transaction, then only the members who bid for that particular acquisition receive a license to the relevant intellectual property (as explained at http://www.alliedsecuritytrust.com/Services/LicensingModel.aspx). In the case of RPX, all members receive a license to all patents acquired by RPX. AST's licenses are perpetual from the outset, unlike RPX which introduces vesting periods in its licenses. Members who do not bid in the initial acquisition can still subsequently purchase a license to the patents involved, at a price equal to the highest bid.
+
+Second, after AST acquires a set of patents and licenses its bidding members, it looks to sell those patents. It starts by offering each of the original bidders, starting with the highest one, the opportunity to buy out the entire portfolio by reimbursing the other bidders and AST's related expenses. If none of the bidders is interested, AST places the portfolio for sale with a broker (a divestiture process explained at http://www.alliedsecuritytrust.com/Services/DivestitureProcess.aspx). In contrast, RPX only sells patents occasionally, when it deems that they are no longer useful to its subscribers.
+
+For economists, defensive aggregators raise some interesting issues about contracting. First, the value of RPX to its subscribers seems difficult to verify. Unlike traditional insurers who pay customers when “ accidents ” happen, defensive aggregators get paid to reduce the probability of “ accidents ”— in this case, lawsuits from nonproducing entities. But how can members know that RPX is effectively reducing litigation risk on their behalf? Presumably, part of the answer lies is the number of relevant patents that RPX buys. But perhaps more importantly, subscribers view RPX as offering a more efficient buying service for patents they have already identified as threatening. When patents are critical to their business, operating companies will often buy them on their own. The issue for many firms is what to do about marginally relevant patents: the expected value of the potential damage may not be sufficient to justify the cost of buying the patent unilaterally, but it may be worth the membership fee paid to RPX, who in turn can aggregate payments across multiple subscribers.
+
+Second, defensive aggregators make an intriguing public commitment never to litigate in order to extract revenues (for example, see RPX's website http:// rpxcorp.com/, accessed May 2012). This commitment helps differentiate them from patent trolls and serves to reassure potential subscribers, but at the same time,
+
+---
+
+58 Journal of Economic Perspectives
+
+it creates a significant free-rider problem. When RPX buys a patent (say, for Nokia in smartphones), and eliminates the threat from a troll, then nonsubscribers in the same industries (say, Motorola) equally benefit, so they may be less likely to pay RPX's subscription fees. One way in which RPX mitigates this problem is by adopting a “catch-and-release” approach: it acquires a patent, grants its subscribers a license, and then resells the patent on the open market (preferably to a nonpracticing entity), which means nonsubscribers remain exposed to litigation risk (Hansell 2009) . Still, reselling the patents acquired reduces the value of subscribing to RPX for new members. This approach also complicates the decision for existing members, who have to determine whether to renew their subscriptions.
+
+Third, the defensive aggregator business model faces an inherent limitation by relying exclusively on subscription revenues. RPX has no shot at the huge payoffs that can be achieved by nonpracticing entities (or a super-aggregator like Intellectual Ventures, which we discuss below). In turn, this puts RPX at a disadvantage in acquiring patents. For example, nonpracticing entities can offer payments for patents that are at least partially contingent on what might be received in a later lawsuit—and therefore a much larger potential payoff to owners—whereas RPX can only offer a fixed payment. RPX may also face unreasonable prices from patent owners if the latter interpret an approach by RPX as a sign of interest from its subscribers—who are, after all, large and potentially rich operating companies. This outcome is related to the issue of “awareness-inducing information” in incomplete contract settings studied formally in Tirole (2009). RPX tries to mitigate this problem by forming buying syndicates among its subscribers and then using shell companies to buy patents of interest to the syndicate.
+
+It is still too early to tell whether RPX has managed to address these issues successfully: it was founded in 2008, and most of its members are locked in for a minimum of three years, so there is insufficient data as yet regarding membership renewal rates.
+
+## Super-aggregator(s)
+
+A new type of player, which we call a super-aggregator, has emerged as the largest and most controversial type of intellectual property intermediary. Epitomized by Intellectual Ventures, a super-aggregator is a hybrid between a defensive aggregator, a large nonpracticing entity, and a “weapons dealer,” who can provide intellectual property to litigants on both sides of a battle. At the time of this writing, Intellectual Ventures seems unique because of its size—the company has raised more than $ 5 billion from a variety of investors—but other entities are trying to emulate its model by raising similar amounts of capital.
+
+Intellectual Ventures is a nonpracticing entity. Its first investor, Microsoft, has publicly said that Intellectual Ventures delivers a highly valued service for technology firms (Hagiu, Yoffie, and Wagonfeld 2011). However, critics have described Intellectual Ventures as “the world’s largest patent troll” because it acquires, creates, and seeks to license patents without directly making any products or services itself (Orey 2006). Founded in 2000 by former Microsoft chief
+
+---
+
+Andrei Hagiu and David B. Yoffie 59
+
+technology officer Nathan Myrhvold, as of mid-2012 the firm has spent approximately $ 2 billion building the world's third-largest patent portfolio—roughly 35,000 patents, mostly covering software, semiconductors, communications, and e-commerce. Like a venture capital or private equity firm, Intellectual Ventures is structured as a series of funds. Its two largest funds are dedicated to acquiring existing patents from all possible sources: individual inventors, or small and large companies. Its third fund focuses on developing its own inventions in partnership with scientists; for example, current projects include a new type of nuclear reactor and a laser-based weapon for fighting malaria mosquitoes. A fourth fund is targeted at developing and acquiring pre-filing inventions, mostly from universities in Asia, through a variety of technology transfer deals.
+
+The last two funds distinguish Intellectual Ventures from typical patent trolls, who do not invent. During its first 10 years, Intellectual Ventures also differed from a typical nonpracticing entity in that it had not litigated—at least not directly. The company had instead sought to monetize its patent portfolios through “friendly” licensing deals and, when necessary, by forming shell companies or selling patents to third-party nonpracticing entities who would in turn litigate. This indirect approach changed in December 2010, when Intellectual Ventures started filing direct patent infringement lawsuits against a variety of operating companies. In its first lawsuits, Intellectual Ventures filed three patent infringement suits against nine companies, including McAfee, Symantec, and Hynix Semiconductor. In July 2011, Intellectual Ventures filed its fourth suit against a group of 12 companies, including HP, Dell, Wal-Mart, and Best Buy.
+
+The fundamental feature that sets Intellectual Ventures apart from other nonpracticing entities is that many of its investors are strategic and include prominent technology companies such as Amazon, American Express, Apple, Cisco, eBay, Google, Intel, Microsoft, Nokia, SAP, Sony, Samsung, and Verizon. 5 For these strategic investors, Intellectual Ventures also functions as a defensive patent aggregator. Indeed, firms that invest in Intellectual Ventures automatically receive licenses for subsets of the patents acquired by the firm (earlier investors receive wider coverage), which serves to shield them against lawsuits from trolls or competitors.
+
+The dual structure of Intellectual Ventures as both a nonpracticing entity and a defensive aggregator means that it has a potentially difficult balance to strike between the economic interests of its two types of investors: its strategic investors, who are operating companies, and its financial investors, who include pension funds and university endowments. This conflict was presumably the reason behind the firm's initial reluctance to litigate directly. The “friendly” licensing approach was aligned with the interests of strategic investors-licensees, while financial investors’ interests are conceivably better served by a more aggressive litigation strategy. Suppose, for example, that an operating company is a limited partner in one of Intellectual Ventures' funds, and is only licensed to part of the portfolio. If the
+
+5 The list of investors in Intellectual Ventures has been revealed in the filings for a lawsuit initiated by Intellectual Ventures against Xilinx (XILINX, Inc. v. Intellectual Ventures LLC (N.D. Cal. 2011)).
+
+---
+
+60 Journal of Economic Perspectives
+
+operating company were infringing on new patents bought by Intellectual Ventures, Intellectual Ventures might be reluctant to bring a lawsuit against this company, thus creating an opportunity cost borne by all of its financial investors.
+
+The fundamental premise of the Intellectual Ventures model is that its unprecedented scale helps reduce search and transaction costs, as well as patent valuation uncertainty, on both sides of the market. Because of its size, Intellectual Ventures can single-handedly create liquidity in the patent market. It has become an attractive outlet for a number of small patent owners, including smaller universities, most of whom do not have the necessary legal and technical expertise, resources, and credibility to monetize their intellectual property on their own. On the other side of the market, Intellectual Ventures provides patent buyers and users with a “one-stop shop” for their licensing needs: similar to RPX, the company is more efficient when it comes to search and negotiating with multiple patent owners. Furthermore, the scale of Intellectual Ventures allows it to capitalize on huge portfolio and learning effects in aggregating patents.
+
+Of course, the super-aggregator model also carries large risks. Even after accounting for complementarities and portfolio effects, the inventory risk remains very high: no matter how effectively Intellectual Ventures filters the patents that it buys, many patents turn out to be of low value or poor quality or both (as many as 19 in 20 or 49 in 50, according to the company's own estimates, as described in Hagiu, Yoffie, and Wagonfeld 2011). Furthermore, sorting through and maintaining tens of thousands of patents may actually create diseconomies of scale. After all, patents are rapidly depreciating assets because their value expires after 20 years, and they require payment of maintenance fees to be kept valid (several hundred to a few thousand dollars to be paid at the end of years 3, 7, and 11 (as explained at the USPTO website, http://www.uspto.gov/patents/process/maintain.jsp).
+
+Finally, the time-horizon for Intellectual Ventures investment funds is relatively long at 15–20 years, and one may question whether the firm will ever be able to generate returns for its investors comparable to other investment vehicles with similar time horizons, like venture capital and private equity. The last concern suggests that Intellectual Ventures is under pressure to engage in more lawsuits. Yet the lawsuits raise their own problems: cost escalation and, even more seriously, the risk of having some patents invalidated by the courts, which might cast doubts on the value of Intellectual Ventures' broader patent portfolio.
+
+## Implications and Conclusions
+
+The patent system is inherently a second-best mechanism for trading off the benefits of enhanced future innovation against the costs of temporary distortions of the economic system after innovation has occurred. Furthermore, the practical realities of the patent system have created additional problems: for instance, a substantial number of low-quality, overlapping, and excessively broad patents. Patent intermediaries (including the new ones described in this article) are able
+
+---
+
+The New Patent Intermediaries 61
+
+to profit from the patent system's inherent tension by improving the payoffs to innovators and/or by taxing more heavily the fruits of past innovations. Given the organizational complexity of the new patent intermediaries and the multiplicity of channels through which they affect participants in the patent market, it is very difficult to draw clear conclusions about whether they generate net benefits or costs for society. Nevertheless, it is useful to point out that intermediation mechanisms that move the imperfect patent system in the direction of enhancing rewards for innovation are more likely to be a positive, while mechanisms that move the system in the direction of extracting taxes on prior innovation are likely to be a social negative. The new patent intermediaries clearly do both — and in fact, cannot do one without the other. But their organizational structures and business models may be skewed more heavily on one side or the other, which provides some basis for considering their net social value.
+
+While defensive aggregators are completely aligned with the interests of operating companies in reducing the patent troll threat, this orientation does not mean that they improve the overall efficiency of the patent market. To some extent, they facilitate collusion between large operating companies at the expense of small inventors. By definition, their incentives are to acquire relevant intellectual property at the lowest possible cost to defend their subscribers, not to maximize the value of the patents they acquire. Thus, they are likely to exacerbate the bargaining and information asymmetries between small patent owners and large operating companies (a similar effect to that of traditional cross-licensing practices).
+
+Intellectual Ventures (and other future super-aggregators) are significantly more complicated because of their hybrid nature. Let us consider how a superaggregator affects the incentives of operating companies, financial investors, and small inventors. Operating companies may see their operating costs increase when Intellectual Ventures aggregates and asserts previously “silent” patents against them. But a super-aggregator like Intellectual Ventures may also lower their aggregate search and transaction costs by providing a one-stop group-licensing shop—just like defensive aggregators do for their members. This service is particularly valuable for technology companies in sectors with short innovation cycles. As a consequence, the net effect of Intellectual Ventures on the development and innovation incentives on operating companies is ambiguous. Some operating companies like Microsoft view it as providing a useful patent discovery and licensing service; others view it as a dangerous nonpracticing entity which significantly raises their costs. Small patent owners, individual inventors, and small companies and universities involved in invention unambiguously benefit from the existence of Intellectual Ventures, because it channels more financial rewards to previously undercompensated inventors, which should unambiguously increase their innovation incentives. Similarly, financial (nonstrategic) investors see Intellectual Ventures as a viable vehicle for investing in patents as a new, large, and uncorrelated asset class.
+
+Due to huge economies of scale, it seems most likely that in the long run there will only be a few super-aggregators — or even just one. This concentration raises significant hold-up concerns. A super-aggregator may become nothing more
+
+---
+
+62 Journal of Economic Perspectives
+
+than a super-troll, able to hold up both sides of the market by extracting excessive payments from operating companies (for example, by strategically disaggregating patent portfolios and enforcing the different parts sequentially) while at the same time paying lower compensation to inventors. Perhaps an even greater source of concern is that super-aggregators' incentives may be skewed towards imposing higher fees on operating companies current production activities, rather than facilitating the commercialization of unproven patents (a riskier endeavor).
+
+But, perhaps surprisingly, there could also be significant social efficiency gains from super-aggregator market concentration. Scale leads to significant learning effects in assessing the value of patents, which may create a more reliable mechanism for patent valuation (where patent platforms have failed). Furthermore, in the second-best world created by patent market failures, which lead to excessive patent infringement, it may be efficient to have only a few (or one) market-based enforcer(s). A super-aggregator, in theory, can compensate inventors of a given patent (or portfolio) who otherwise would fall through the cracks. When a superaggregator buys patents in order to assert them against operating companies that attempt to free-ride on the intellectual property, it preserves the incentives for future innovation. Finally, scale and capital structure, and the accompanying large returns promised to financial investors, can act as credible commitments to build valuable patent portfolios and license them broadly to many players in any given industry. In particular, a super-aggregator's ability to sign large numbers of licensees, without the risks of litigation, depends on its reputation. Enforcing even one weak patent for “nuisance value” (like many small nonpracticing entities do) would run the risk of casting doubt over the value of the super-aggregator's broader patent portfolio. This is an instance of the reputation-building mechanism by intermediaries in contexts with goods of uncertain quality, as studied formally by Biglaiser (1993).
+
+The task of empirically measuring the net economic impact of any intellectual property intermediary and deciding whether it is harmful to society is inherently difficult. Such an analysis would require measuring the net effect on operating companies, inventors, universities, and financial investors, both in terms of shortrun payments made or received and in terms of long-run innovation incentives. These effects seem dauntingly complex to measure. For this reason, most recent empirical studies only estimate the effects on one side of the market — and thus are by definition incomplete. Examples include the Bessen, Meurer, and Ford (2011) estimates of the costs imposed by trolls on operating companies between 1990 and 2010, and Tucker (2012) evaluating the effect of trolls on the adoption of medical imaging technology sold by vendors targeted by trolls.
+
+Part of the problem is the difficulty of measuring net transfers to inventors. In many cases, nonpracticing entities make lump-sum payments to inventors in exchange for control of their patents before any litigation occurs; for example, Intellectual Ventures spent over $ 1 billion dollars acquiring patents from various sources before it began suing publicly in late 2010. These transfers are usually not disclosed publicly, unlike the settlements or damages resulting from lawsuits. In the absence of access to such information, empirical research on intellectual property intermediaries might
+
+---
+
+Andrei Hagiu and David B. Yoffie 63
+
+tackle some narrower efficiency questions. For example, an important contributing factor to the effect of nonpracticing entities (including super-aggregators) on innovation incentives is whether they seek to enforce proven patents on existing products or to facilitate the commercialization of unproven patents. Thus, perhaps one could categorize and measure the mix of patents monetized by nonpracticing entities (even without transaction prices) to provide a valuable proxy for their likely effect on innovation.
+
+- ■ We are grateful to the following people for extremely insightful comments and feedback
+on earlier drafts of the article: David Autor, Peter Detkin, Chang-Tai Hseih, Josh Lerner,
+John List, Douglas Melamed, and Timothy Taylor. Britta Kelley provided excellent research
+assistance. The views expressed in this paper are entirely our own.
+## References
+
+Analysis Group. n.d. “Patent Litigation before the International Trade Commission,” p. 2. http:// www.analysisgroup.com/uploadedFiles/Practice _Areas/Patent_Litigation_Before_the_ITC.pdf (accessed May, 2012).
+
+Arora, Ashish, Andrea Fosfuri, and Alfonso Gambardella. 2001. Markets for Technology: The Economics of Innovation and Corporate Strategy. Cambridge, MA: MIT Press.
+
+Bario, David. 2011. “‘Original Patent Troll’ Is Back after Long Hiatus, with a New Name and New Patents.” Am Law Litigation Daily, August 17. http:// amlawdaily.typepad.com/amlawdaily/2011/08 /original-patent-troll-is-back-new-name-new -patents.html.
+
+Bergelt, Keith. 2010. "Patents, Probes and Strength in Unity: Participate in Keeping Open Source Open." Presentation, LinuxCon (Boston, MA), August 10. http://events.linuxfoundation .org/linuxcon2010/bergelt.
+
+Bessen, James, and Michael J. Meurer. 2008. Patent Failure: How Judges, Bureaucrats and Lawyers Put Innovators at Risk. Princeton, NJ: Princeton University Press.
+
+Bessen, James E., Michael J. Meurer, and Jennifer Laurissa Ford. 2011. "The Private and Social Costs of Patent Trolls." Boston University School of Law, Law and Economics Research
+
+Paper No. 11-45. http://papers.ssrn.com/sol3 /papers.cfm?abstract_id=1930272.
+
+Biglaiser, Gary. 1993. "Middlemen as Experts." RAND Journal of Economics 24(2): 212-23.
+
+Chiao, Benjamin, Josh Lerner, and Jean Tirole. 2007. "The Rules of Standard-Setting Organizations: An Empirical Analysis." RAND Journal of Economics 38(4): 905-930.
+
+Crothers, Brooke. 2009. "Qualcomm, Broadcom Reach $891 Million Settlement." CNET News, April 26, 2009. http://news.cnet.com/8301 -13924_3-10227815-64.html (accessed May, 2012).
+
+Decker, Susan. 2012. “A Crackdown on Patently Absurd Lawsuits.” Bloomberg Businessweek , May 10. http://www.businessweek.com/articles/2012-05 -10/a-crackdown-on-patently-absurd-lawsuits.
+
+Dushnitsky, Gary, and Thomas Klueter. 2011. "Is There an eBay for Ideas? Insights from Online Knowledge Marketplaces." European Management Review 8(1): 17–32.
+
+Fisher, Jeffrey M. 2006. "The Pros and Cons of Patent Litigation before the International Trade Commission." Farella, Braun, and Martel LLP, November 15. Available at: http://www.fbm .com/The_Pros_and_Cons_of_Patent_Litigation _Before_the_International_Trade_Commission _11-15-2006.
+
+Gallagher, Billy. 2012. "Apple Awarded $1.049
+
+---
+
+64 Journal of Economic Perspectives
+
+Billion in Damages as Jury Finds Samsung Infringed on Design and Software Patents." Tech Crunch, August 24. http://techcrunch.com/2012/08/24 /apple-wins-patent-ruling-as-jury-finds-samsung -infringes/.
+
+Gans, Joshua S., David H. Hsu, and Scott Stern. 2002. "When Does Start-Up Innovation Spur the Gale of Creative Destruction?" RAND Journal of Economics 33(4): 571-86.
+
+Gans, Joshua S., and Scott Stern. 2010. "Is There a Market for Ideas?" Industrial and Corporate Change 19(3): 805-837.
+
+Hagiu, Andrei, David B. Yoffie, and Alison Wagonfeld. 2011. “Intellectual Ventures.” Harvard Business School Case No. 710-423, Harvard Business School Publishing.
+
+Hansell, Saul. 2009. "Trolling for Patents to Fight Patent Trolls." New York Times, March 30. http:// bits.blogs.nytimes.com/2009/03/30/trolling -for-patents-to-fight-patent-trolls/.
+
+ICAP. 2009. "ICAP Reaches Agreement to Acquire Patent Brokerage Business of Ocean Tomo LLC." June 16. http://www.icap.com /news-events/in-the-news/news/2009/icap -reaches-agreement-to-acquite-patent-brokerage -business-of-ocean-tomo-llc.aspx (accessed May, 2012).
+
+ICAP. 2010. “ ICAP Ocean Tomo Auction Sees Record Bidding. ” March 26. http://www.icap .com/news-events/in-the-news/news/2010/icap -ocean-tomo-auction-sees-record-biding.aspx (accessed May 2012).
+
+Inside IP. 2012. "Going ... Going ... Gone?: The Rises and Falls of Patent Auctions." June 14. http://www.vennershipley.co.uk/show-news-id -249&p=2.html (accessed May 2012).
+
+Jaffe, Adam B., and Josh Lerner. 2004. Innovation and Its Discontents: How Our Broken Patent System is Endangering Innovation and Progress, and What to Do About It. Princeton, NJ: Princeton University Press.
+
+Jarosz, John, Robin Heider, Coleman Bazelon, Christine Bieri, and Peter Hess. 2010. “Patent Auctions: How Far Have We Come?” les Nouvelles, March, pp. 17, 20–22.
+
+Lamoreaux, Naomi R., and Kenneth L. Sokoloff. 2003. "Intermediaries in the U.S. Market for Technology, 1870–1920." In Finance, Intermediaries and Economic Development , edited by Stanley L. Engerman, Philip T. Hoffman, Jean-Laurent Rosenthal, and Kenneth L. Sokoloff, 209–246. Cambridge: Cambridge University Press.
+
+Lampe, Ryan, and Petra Moser. 2010. “Do Patent Pools Encourage Innovation? Evidence from the Nineteenth-Century Sewing Machine Industry.” Journal of Economic History 70(4): 898–920.
+
+Layne-Farrar, Anne, and Josh Lerner. 2011. “To Join or Not to Join: Examining Patent Pool Participation and Rent Sharing Rules.” International Journal of Industrial Organization 29(2): 294–303.
+
+Lemley, Mark A. 2008. "Are Universities Patent Trolls?" Fordham Intellectual Property, Media and Entertainment Law Journal 18(3): Article 2.
+
+Lemley, Mark A., and Carl Shapiro. 2005. "Probabilistic Patents." Journal of Economic Perspectives 19(2): 75-98.
+
+Lerner, Josh, and Jean Tirole. 2004. "Efficient Patent Pools." American Economic Review 94(3): 691-711.
+
+Lerner, Josh, and Jean Tirole. 2006. "A Model of Forum Shopping." American Economic Review 96(4): 1091-1113.
+
+Lerner, Josh, Marcin Strojwas, and Jean Tirole. 2007. "The Design of Patent Pools: The Determinants of Licensing Rules." RAND Journal of Economics 38(3): 610-625.
+
+Lohr, Steve. 2009. “Patent Auctions Offer Protections to Inventors.” New York Times , September 20. http://www.nytimes.com/2009/09/21/technology /21patent.html?_r=0.
+
+Malek, Marcus. 2009. "R.I.P. Ocean Tomo: Complete Auction Analysis." Presentation slides, June 23, http://www.slideshare.net/marcusmalek /complete-ocean-tomo-auction-analysis-marcus -malek-intangitopia (accessed May 2012).
+
+Masnick, Mike. 2009. "Patent Holder Sues McAfee, Gets $25 Million ... But May End up Losing $5 Million due to Everyone It Has to Pay Off." Techdirt, November 4. http://www.techdirt .com/articles/20091103/0333046778.shtml.
+
+McCurdy, Daniel P. 2009. “Patent Trolls Erode the Foundation of the U.S. Patent System.” Science Progress, January 12. http://www.scienceprogress .org/2009/01/patent-trolls-erode-patent-system/.
+
+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(1): 189-228.
+
+Orey, Michael. 2006. “Inside Nathan Myhrvold’s Mysterious New Idea Machine.” With Moria Herbst. Bloomberg Businessweek , July 2, 2006. http://www .businessweek.com/stories/2006-07-02/inside -nathan-myhrvolds-mysterious-new-idea-machine.
+
+Parchomovsky, Gideon, and Richard Polk Wagner. 2005. “Patent Portfolios.” University of Pennsylvania Law Review 154(1): 1–78.
+
+Patent Freedom. n.d. Website. https://www .patentfreedom.com/research-lot.html.
+
+Riordan, Teresa. 2004. "Contest over BlackBerry Patent." New York Times, June 7. http://www .nytimes.com/2004/06/07/technology/07patent .html.
+
+Schmalensee, Richard. 2009. "Standard-Setting
+
+---
+
+The New Patent Intermediaries 65
+
+Innovation Specialists, and Competition Policy." http://papers.ssrn.com/sol3/papers.cfm abstract_id=1219784.
+
+Schudel, Matt. 2005. “Accomplished, Frustrated Inventor Dies.” Washington Post, February 26, 2005. http://www.washingtonpost.com/wp-dyn /articles/A54564-2005Feb25.html.
+
+Shapiro, Carl. 2001. “Navigating the Patent Thicket: Cross Licenses, Patent Pools, and Standard Setting.” In Innovation Policy and the Economy, Vol. 1, edited by Adam Jaffe, Josh Lerner, and Scott Stern, 119–150. Cambridge, MA: MIT Press.
+
+Sharma, Amol, and Don Clark. 2008. "Tech Guru Riles the Industry by Seeking Huge Patent Fees." Wall Street Journal, September 17. http:// online.wsj.com/article/SB122161127802345821 .html.
+
+Simcoe, Timothy. 2012. "Standard Setting Committees: Consensus Governance for Shared
+
+Technology Platforms." American Economic Review 102(1): 305-36.
+
+Spulber, Daniel F. 2011. "Should Business Method Inventions Be Patentable?" Journal of Legal Analysis 3(1): 265-340.
+
+Tirole, Jean. 2009. "Cognition and Incomplete Contracts." American Economic Review 99(1): 265-94.
+
+Tucker, Catherine. 2012. “Patent Trolls and Technology Diffusion.” TILEC Discussion Paper No. 2012-030. http://papers.ssrn.com/sol3/papers .cfm?abstract_id=2136955.
+
+Wasserman, Noam. 2012. The Founder's Dilemmas: Anticipating and Avoiding the Pitfalls That Can Sink a Startup. Princeton, NJ: Princeton University Press.
+
+Young, Eric. 2008. "Patent Marketeers." San Francisco Business Times, March 23. http:// www.bizjournals.com/sanfrancisco/stories /2008/03/24/focus1.html?page=all.
+
+---
+
+66 Journal of Economic Perspectives
+

processed/00046_W2110502388/document.md

@@ -0,0 +1,523 @@
+# HUMAN RIGHTS AND INTELLECTUAL PROPERTY
+
+Mapping the Global Interface
+
+This book analyzes the interface between intellectual property and human rights law and policy. The relationship between these two fields has captured the attention of governments, policymakers, and activist communities in a diverse array of international and domestic venues. These actors often raise human rights arguments as counterweights to the expansion of intellectual property in areas including freedom of expression, public health, education, privacy, agriculture, and the rights of indigenous peoples. At the same time, creators and owners of intellectual property are asserting a human rights justification for the expansion of legal protections.
+
+This book explores the legal, institutional, and political implications of these competing claims in three ways: (1) by offering a framework for exploring the connections and divergences between these subjects; (2) by identifying the pathways along which jurisprudence, policy, and political discourse are likely to evolve; and (3) by serving as a teaching and learning resource for scholars, activists, and students.
+
+Laurence R. Helfer is the Harry R. Chadwick, Sr., Professor of Law at Duke University School of Law, where he codirects the Center for International and Comparative Law and is a member of the faculty steering committee of the Duke Center on Human Rights. He has authored more than 50 publications and has lectured widely on his diverse research interests, which include interdisciplinary analysis of international law and institutions, human rights, and international intellectual property law and policy. He is the coauthor of Human Rights (2d ed., 2009) and the author of Intellectual Property Rights in Plant Varieties: International Legal Regimes and Policy Options for National Governments (2004).
+
+Graeme W. Austin is a Professor of Law at Melbourne University and Victoria University of Wellington. Until 2010, he was the J. Byron McCormick Professor of Law at the University of Arizona. He has lectured on intellectual property law in a variety of settings and is an elected member of the American Law Institute. He has published widely on the topic of intellectual property, including in the Law Quarterly Review and the International Review of Intellectual Property and Competition Law. He is coauthor of International Intellectual Property: Law and Policy (2008).
+
+---
+
+![Figure](figures/figure_000.png)
+
+---
+
+# Human Rights and Intellectual Property Mapping the Global Interface
+
+LAURENCE R. HELFER
+
+Harry R. Chadwick, Sr., Professor of Law
+
+Duke University School of Law
+
+GRAEME W. AUSTIN
+
+
+Professor of Law Melbourne University
+
+
+and Victoria University of Wellington
+
+![Figure](figures/figure_003.png)
+
+CAMBRIDGE UNIVERSITY PRESS
+
+---
+
+CAMBRIDGE UNIVERSITY PRESS
+
+Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Tokyo, Mexico City
+
+Cambridge University Press
+
+32 Avenue of the Americas, New York, NY 10013-2473, USA
+
+www.cambridge.org
+
+Information on this title: www.cambridge.org/9780521711258
+
+c Laurence R. Helfer and Graeme W. Austin 2011
+
+This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press.
+
+First published 2011
+
+Printed in the United States of America
+
+A catalog record for this publication is available from the British Library.
+
+Library of Congress Cataloging in Publication data Austin, Graeme.
+
+Human rights and intellectual property : mapping the global interface / Graeme W. Austin, Laurence R. Helfer.
+
+p. cm.
+
+Includes bibliographical references and index. ISBN 978-0-521-88437-2 (hardback) – ISBN 978-0-521-71125-8 (pbk.)
+
+1. Intellectual property. 2. Intellectual property (International law) 3. Human rights. I. Helfer, Laurence R. II. Title.
+
+K1401.A98 2011
+
+346.04'8-dc22 2010031359
+
+ISBN 978-0-521-88437-2 Hardback
+
+ISBN 978-0-521-71125-8 Paperback
+
+Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party Internet Web sites referred to in this publication and does not guarantee that any content on such Web sites is, or will remain, accurate or appropriate.
+
+---
+
+## Contents
+
+<table><tr><td>Preface</td><td>page xi</td></tr><tr><td>1. Mapping the Interface of Human Rights and Intellectual Property</td><td></td></tr><tr><td>1.1. Thematic Overview and Introduction</td><td>1</td></tr><tr><td>1.2. The International Human Rights System: A Substantive and Institutional Overview</td><td>3</td></tr><tr><td>A. The U.N. Human Rights Treaty System</td><td>5</td></tr><tr><td>B. Mechanisms for Protecting Human Rights under the U.N. Charter</td><td>14</td></tr><tr><td>1.3. The International Intellectual Property System: A Substantive and Institutional Overview</td><td>16</td></tr><tr><td>A. Types of Intellectual Property</td><td>16</td></tr><tr><td>B. Rationales for Intellectual Property</td><td>18</td></tr><tr><td>C. The International Institutional Framework</td><td>24</td></tr><tr><td>1.4. Historical Isolation of the Human Rights and Intellectual Property Regimes</td><td>31</td></tr><tr><td>1.5. Catalysts for the Expanding Intersection of the Human Rights and Intellectual Property Regimes</td><td>34</td></tr><tr><td>A. International Intellectual Property Protection Standards and Enforcement Mechanisms: Reactions and Counterreactions</td><td>34</td></tr><tr><td>1. Pressure by U.S. Intellectual Property Industries to Expand Intellectual Property Protection Standards and Enforcement Mechanisms</td><td>35</td></tr><tr><td>2. The Shift from WIPO to GATT to TRIPS</td><td>36</td></tr><tr><td>3. The Impact of TRIPS and the Rise of TRIPS Plus Treaties</td><td>39</td></tr></table>
+
+
+v
+
+---
+
+vi
+
+Contents
+
+4. Access to Knowledge and the New Politics of Intellectual Property 43 B. New Developments in Human Rights 48 1. The Rights of Indigenous Peoples and Traditional Knowledge 49 2. Conflicts between TRIPS and Economic, Social, and Cultural Rights 52 3. The Human Rights Obligations of Transnational Corporations 57 4. The Human Right of Property and Corporate Intellectual Property Interests 61 1.6. Competing Conceptual Frameworks for Mapping the Interface of Human Rights and Intellectual Property 64 A. Conflict 65 B. Coexistence 73 C. Beyond Conflict and Coexistence 81 2. The Human Right to Health, Access to Patented Medicines, and the Restructuring of Global Innovation Policy 90 2.1. Introduction 90 2.2. Background on the HIV/AIDS Pandemic and Access to Antiretroviral Drugs 92 2.3. The Human Right to Health and the Emerging Right of Access to Medicines 98 A. Justifications for and Critiques of the Human Right to Health 98 B. The Normative Development of the Human Right to Health and of Access to Medicines 105 2.4. Patent Protection for Pharmaceuticals and Revising the TRIPS Agreement to Enhance Access to Medicines 119 A. An Overview of TRIPS Patent Provisions Relating to Access to Medicines 119 B. Recent Examples of Compulsory Licenses to Promote Access to Medicines 127 2.5. Human Rights Approaches to Closing the “Global Drug Gap” Created by Patented Pharmaceuticals 140 A. Are Patents a Barrier to Access to Medicines? 142 B. Human Rights Contributions to Closing the Global Drug Gap 144 C. Implications for Other Intellectual Property Rights 169
+
+---
+
+Contents
+
+vii
+
+3. Creators' Rights as Human Rights and the Human Right of Property 171 3.1. Introduction 171 3.2. Definitional Issues 3.3. Drafting History of UDHR Article 27 and ICESCR Article 15(1)(c) 176 3.4. General Comment No. 17: An Overview 188 3.5. Domestic Law Reform and Creators' Human Rights: Three Case Studies 199 A. Limiting the Reproduction Right for Musical Works B. Restricting the Scope of the Derivative Work Right C. Resurrecting Copyright Formalities 206 3.6. The Human Right of Property 212
+
+4. Rights to Freedom of Expression, to Cultural Participation, and to Benefit from Scientific Advancements 221 4.1. Introduction 4.2. Rationales for the Right to Freedom of Expression 4.3. International, Regional, and Domestic Law Sources 4.4. Rights to Participate in Culture and to Benefit from Scientific Progress 233 4.5. Intersections between Freedom of Expression and Intellectual Property 242 A. Copyright 243 1. The First Amendment to the U.S. Constitution 2. Article 10 of the European Convention on Human Rights 259 B. Trademarks 283 1. The “Essential Function” of a Trademark 286 2. Freedom of Expression and Trademark Doctrine 290 C. Patents 312
+
+5. The Right to Education and Copyright in Learning Materials 316 5.1. Introduction 316 5.2. The Right to Education: Justifications and Rationales 320 5.3. The Right to Education in International Law 326 5.4. The Provision of Learning Materials and the Human Right to Education 332 5.5. Copyright in Textbooks and Learning Materials 335 5.6. Other Impediments to the Provision of Learning Materials
+
+---
+
+viii
+
+Contents
+
+5.7. Mapping the Interface between Copyright and the Right to Education 357 6. The Human Right to Food, Plant Genetic Resources, and Intellectual Property 364 6.1. Introduction 364 6.2. The Evolution of the Right to Food and of Intellectual Property Protection for Plant-Related Innovations 366 A. The Right to Food 366 1. Justifications for and Critiques of the Human Right to Food 366 2. The Normative Evolution of the Human Right to Food 371 B. Intellectual Property Protection for PGRs 379 1. Justifications for and Critiques of Intellectual Property Protection for PGRs 379 2. The Evolution of Intellectual Property Protection for PGRs 381 a. Plant Variety Protection 382 b. Patent Protection for Plant-Related Innovations 386 3. Legal Rules and Policy Objectives in Tension with Intellectual Property Protection for Plant-Related Innovations 390 a. Farmers' Rights 390 b. Regulating Access to PGRs in Nature and in Global Seed Banks 393 c. “Biopriacy” and Intellectual Property Protection of Raw Plant Materials 394 6.3. Specific Controversies Involving the Right to Food and Intellectual Property Protection for PGRs 400 A. The Response to Expanding Intellectual Property Protection for PGRs in the United Nations Human Rights System 400 B. Genetically Modified Seeds and the Right to Food in India 409 1. The Protection of Plant Varieties and Farmers' Rights Act, 2001, and the Seeds Bill, 2004 409 2. Constitutional Protection of the Right to Food and the Review of India's 2008 Report to the Committee on Economic, Social and Cultural Rights 415
+
+---
+
+Contents
+
+ix
+
+7. Indigenous Peoples' Rights and Intellectual Property 432 7.1. Introduction 432 7.2. International Human Rights Law Relating to Indigenous Peoples 438 7.3. Human Rights, Self-Determination, and the Protection of Indigenous Peoples' Intellectual Property 447 7.4. Indigenous Peoples' International Initiatives Relevant to Intellectual Property 457 7.5. Intellectual Property Protections for Traditional Knowledge and Traditional Cultural Expression 461 7.6. Individual and Collective Interests in Indigenous Cultural Productions 483 7.7. Intersections between Indigenous Human Rights and Intellectual Property Issues 496 8. Conclusion 503 8.1. The Unavoidable Intersection of Human Rights and Intellectual Property 504 8.2. Assessing Existing Proposals to Reconcile Human Rights and Intellectual Property 506 8.3. Toward a Human Rights Framework for Intellectual Property 512 A. The Protective Dimension of the Framework 513 B. The Restrictive Dimension of the Framework 516
+
+References 523 Acknowledgments 539 Index 543 Cases Discussed (Selected) 550
+
+---
+
+![Figure](figures/figure_000.png)
+
+---
+
+## Preface
+
+The key terms in this book's subtitle – “mapping,” “global,” and “interface” – reflect our approach to analyzing the relationship between human rights and intellectual property.
+
+Consider first the cartographical trope, “mapping.” It is possible to envision intellectual property law and human rights law as the product of the gradual accretion and spread of international and domestic laws and institutions. The terrain of international intellectual property law was the first to emerge. Initially the subject of discrete bilateral agreements between sovereign nations, its modern form came to be established with the two great multilateral intellectual property treaties from the end of the 19th century: the Paris Convention on industrial property (1883) and the Berne Convention on literary and artistic works (1886). The international human rights regime emerged more recently, with the founding of the United Nations after World War II, and, in particular, the adoption of the Universal Declaration of Human Rights (1948).
+
+From these beginnings, the terrain occupied by both issue areas has expanded significantly in substantive reach, in prescriptive detail, and in geographic scope. In the intellectual property context, the international law relating to patents illustrates this point. At the end of the 19th century, the desirability of domestic – let alone international – patent protection was a matter of sharp debate, even among industrialized nations. For this reason, the Paris Convention contains few substantive rules – although its national treatment and international priority rules for patent registrations were important achievements – and (like the Berne Convention) it has no effective enforcement mechanisms.
+
+Today, in contrast, international intellectual property law imposes a significant and detailed array of substantive and enforcement obligations. The Agreement on Trade Related Aspects of Intellectual Property (TRIPS), which came into force in 1995, obliges member states to recognize patents
+
+xi
+
+---
+
+xii
+
+Preface
+
+in all fields of technology (subject to transitional arrangements for developing nations). TRIPS also dictates the standard by which domestic law deviations from international patent rules are to be tested, and it sets forth detailed requirements in areas such as domestic enforcement procedures. Perhaps most significantly, noncompliance with TRIPS can trigger meaningful sanctions, as a result of the treaty's integration into the international trade regime now administered by the World Trade Organization. That body, through its dispute settlement system, also contributes to the development of international intellectual property norms, along with a number of other key agencies, most notably the World Intellectual Property Organization (WIPO). The expansion of international patent law did not stop with TRIPS. International norms continue to emerge and develop as a result of multilateral, regional, and bilateral agreements. A potentially important new initiative, the AntiCounterfeiting Trade Agreement (ACTA), is currently being negotiated. If adopted, ACTA will shape international intellectual property rules and enforcement mechanisms in a range of different contexts.
+
+The space occupied by the international human rights regime has also grown significantly since its inauguration in the middle of the 20th century. The Universal Declaration gave birth to two foundational treaties that entered into force in 1976 – the International Covenant on Civil and Political Rights and the International Covenant on Economic, Social, and Cultural Rights. The Covenants, together with the general comments, case law, and recommendations of their respective treaty bodies, and the decisions of regional human rights courts and commissions, have significantly bolstered the prescriptive force of human rights law. A particularly noteworthy development has been the widening acceptance of social, economic, and cultural rights that, until the 1990s, remained mostly underdeveloped, particularly in the West. New recognition of the human rights of groups has also emerged – commitments that are especially important to the world's indigenous peoples.
+
+In terms of enforcement, the most important activities are occurring at the regional and domestic levels, especially in Europe but also in the Americas and other regions. National courts increasingly adjudicate human rights treaties directly or draw upon international norms when construing national constitutions and statutes. At all levels, multiple review mechanisms and judicial bodies shape human rights law through their investigative and interpretive activities. Indeed, one critique of the international human rights regime is that it suffers from a surfeit of rules, institutions, and decision makers that risks weakening the system as a whole.
+
+As a result of these and related developments, the respective terrains of both the human rights and intellectual property regimes have grown significantly and the intersections between them have expanded. There now exists a
+
+---
+
+Preface
+
+xiii
+
+broad range of legal, social, political, practical, and philosophical issues that straddle both fields. These intersections are evolving rapidly, requiring a new conceptual cartography to help map the changing landscape.
+
+We explore a number of these intersections in this book. To continue with the patent example introduced earlier, consider the human right to the highest attainable standard of health in the light of the protection of pharmaceutical patents. Many nations once denied patents for new drugs on public health grounds; today, TRIPS obliges member nations to recognize and enforce patents in all fields of technology, including medicines. As a result of these countervailing legal commitments, government agencies, international organizations, and civil society groups must engage with the disciplines of both human rights and intellectual property to develop effective, just, and enduring responses to public health crises and to identify new mechanisms for harnessing private innovation to serve the wider social good. This is already occurring as a growing number of actors typically concerned with human rights issues are becoming engaged in intellectual property issues and (although perhaps to a lesser extent) vice versa.
+
+This discussion also underscores the salience of the term “global” in the book’s subtitle. State and private actors in legal regimes have long recognized the inadequacy of purely domestic responses. In the human rights context, the atrocities of the Second World War engendered a commitment to the idea that sovereign nations cannot be the sole arbiter of the fundamental human entitlements. The founders of the United Nations and the drafters of the Universal Declaration recognized that human rights must be bolstered by international institutions and international legal obligations. In the intellectual property context, both private firms and governments have long recognized that effective responses to piracy and counterfeiting, and, more recently, the protection of genetic resources and indigenous knowledge, cannot be adequately addressed at the domestic level. In addition, there now exist important feedback mechanisms in intellectual property lawmaking, whereby norms developed at the international and domestic levels mutually influence each other.
+
+As we discuss in Chapter 1, the existence of any meaningful engagement between the two areas of law is a relatively recent phenomenon. Scholars and policymakers in each regime are only beginning to recognize areas of mutual concern. Because law is shaped by human agency, the way in which human rights and intellectual property intersect is not an inevitable or predetermined process. The actors who engage with the legal and social policy issues to which both regimes are relevant have a large measure of discretion in determining the character of this interaction. Will there be a seismic clash, a rupturing of tectonic plates, as the two areas move ever closer together and
+
+---
+
+xiv
+
+Preface
+
+finally collide? Or will the engagement be carefully considered, nuanced, and accommodating? Our preference is for the latter kind of engagement, and one of the aims of this book is to provide the substantive materials and original analytical content to help others to explore the intersections between the two regimes in a productive and coherent fashion.
+
+These considerations also explain the use of the term “ interface ” in our subtitle. The most familiar use of the term is in the computing context. It denotes mechanisms for conjoining distinct or contrasting elements and systems: software and hardware, or interfaces between operating systems. Human rights and intellectual property exhibit distinctive systemic characteristics. For the most part they have evolved independently – although, as we discuss in Chapter 3, there is an often-overlooked set of human rights obligations that recognize the rights of creators in their artistic and scientific works – and have been shaped by different sets of actors in distinct institutional contexts and informed by divergent analytical traditions. A key aim of the book, suggested by our use of the term “interface,” is to provide a structure for dialog and engagement between these two – hitherto largely separate – systems.
+
+To that end, Chapter 1 offers a conceptual overview of the relationship between human rights and intellectual property, as well as a brief summary of each area of law. The latter will be useful for readers less familiar with the traditions and substance of one or both areas. Chapter 1 also explores different ways that the relationship between human rights and intellectual property has been understood by scholars and in different legal and policy contexts. The chapters that follow develop the latter theme and present “case studies” of several distinct controversies. Chapter 2 considers the right to health and patented pharmaceuticals; Chapter 3 addresses the human rights associated with certain types of creative activity; Chapter 4 examines the rights of freedom of expression and cultural participation and the right to benefit from scientific progress; Chapter 5 explores the right to education and the potential tensions with copyright protection in learning materials; Chapter 6 examines the human right to food in the context of intellectual property protections in plant genetic materials; Chapter 7 considers the claims that have emerged in the context of indigenous peoples’ struggles for recognition of their rights in respect of traditional knowledge and other forms of cultural production. In a final chapter, we offer a fuller exposition of our own framework for conceptualizing the most productive connections between the human rights and intellectual property regimes.
+
+The decision to defer the exposition of our conceptual framework until the Conclusion in part reflects the genesis of this book. Several years ago, one of us developed a law school course entitled Human Rights and Intellectual
+
+---
+
+Preface
+
+xv
+
+Property. Partly because of the novelty of the topic, no teaching materials existed, a gap that endures today. Teaching the course was a very fulfilling experience. The course brought together students from an array of different backgrounds and with a range of different interests – not only intellectual property and human rights, but also international trade and indigenous peoples’ law and policy issues. The course invited these groups to engage with each other across the intellectual, heuristic, and, sometimes, cultural divides that had informed their thinking about the various issues to which human rights and intellectual property are relevant – issues that we consider at greater length in the case studies in each chapter of this book. The aims of the course included introducing students to the substantive laws, policies, and institutional frameworks of both human rights and intellectual property. But a more ambitious aim was to invite students to develop their own conceptions of how the two areas might interact. Although we have our own views on how the contours of the interface might be mapped, as a pedagogical matter we believe that readers’ engagement with this topic will be richer if they are also encouraged to form their own views as to how this might be achieved. Hence our decision on the placement of the final chapter.
+
+These concerns also reflect the thinking behind our use of the term “mapping” – the present participle form of the verb. Engagement between the two areas of law is a dynamic and evolving process, one to which we hope this book will contribute. But we labor under no pretension that this work is by any measure complete. We look forward to engaging with the responses – including, we imagine, rigorous critiques – that this text might invite.
+
+Our aspirations for the book also extend beyond the classroom context. We hope that it will contribute to the emerging scholarship in the field and to the policy debates that are beginning to occur in both regimes. Here we offer a personal anecdote. When we first entered law teaching in the 1990s, human rights and intellectual property were separate components of our respective research agendas. Our decision to focus our scholarly efforts in these two discrete areas was highly unusual. In fact, a senior colleague counseled one of us to choose one field and abandon the other, warning that there was little benefit – and potentially much risk – in attempting to develop expertise in two such different and unrelated fields. The response offered by the recipient of this well-meaning advice was to acknowledge the lack of substantive connections between the two legal regimes, but to counter that there was much to be learned by interacting with different communities of scholars, government officials, and civil society groups, who rarely, if ever, interacted directly with each other.
+
+More than a decade later, much has changed. When we now explain to colleagues and students that our research explores the intersections between
+
+---
+
+xvi
+
+Preface
+
+intellectual property and human rights, the usual response is a gleam of recognition and a question or two – most often about patented medicines and HIV-AIDS, but increasingly about freedom of expression and online technologies or the moral rights of artists. We are hardly alone in exploring these issues. As we indicated earlier, growing numbers of civil society organizations now include both human rights and intellectual property in their mandates, often specializing in subissues such as patents and the right to health, access to knowledge, or the intersection of human rights, intellectual property, and development. And the global network of commentators and journalists who write about the interface of the two fields is expanding, as revealed by the numerous and diverse entries in this book's extensive References.
+
+For law students, as well as students in cognate disciplines, such as political theory and international relations, much of the value of the book may lie in the extensive Notes and Questions that follow the analysis of each substantive topic. These sections invite the kind of deep engagement and interrogation of substantive issues and conceptual frameworks that characterize universitylevel instruction, at both undergraduate and graduate levels. We also hope that this book will be useful in other contexts and for other actors, including government officials, international organizations, activists, and civil society groups. To that end, discussions of substantive topics often are followed by Issues in Focus. These sections perform a number of functions, including summarizing recent developments and highlighting emerging issues. By deploying a range of different analytical techniques and materials, we hope that the book can be used by, and will be useful for, a wider range of constituencies.
+
+Finally, we would like to acknowledge the many scholars who have contributed to the writing of this book with comments and criticisms. They include Barbara Atwood, Molly Beutz, Jamie Boyle, Audrey Chapman, Graeme Dinwoodie, Maureen Garmon, Toni Massaro, Ruth Okediji, and Peter Yu. We are also grateful for the help of several research assistants, including Laura Duncan, Eric Larson, Lisa Lindemenn, María Méndoza, Casey Mock, Pedro Paranagua, Meryl Thomas, and Amy Zavidow. Erin Daniel provided invaluable assistance in obtaining permissions to reproduce copyrighted materials. Last, but by no means least, are the unswerving dedication and patience of our respective partners, David Boyd and Bryan Patchett, the acknowledgment of whose manifold contributions is itself a reflection of hard-fought human rights struggles.
+
+Laurence R. Helfer Durham, North Carolina, USA December 2010
+
+Graeme W. Austin Wellington, New Zealand
+
+---
+
+## Chapter 8 Conclusion
+
+This book analyzes the interface of human rights and intellectual property from multiple perspectives. Chapter 1 introduces the major legal, institutional, and political aspects of each regime, explains how they came into increasing contact over the past decade, and explores alternative frameworks for conceptualizing their relationship. Each of the remaining chapters adopts a predominantly substantive orientation that examines in depth specific intersections between certain human rights and intellectual property protection rules. In this concluding chapter, we shift focus to elaborate the major transsubstantive themes that are interwoven through the preceding materials. Our aim is twofold. First, and more modestly, we seek to illuminate connections that transcend specific “ hot button ” controversies and to offer deeper insights about the interconnections between the two legal regimes. Second, and more ambitiously, we offer our own analytical framework to assist scholars, policymakers, civil society groups, and students in conceptualizing the relationship between human rights and intellectual property.
+
+We begin in Section 8.1 by reiterating the inevitability of the human rights – intellectual property interface and by rejecting – both as a matter of principle and as a matter of practical politics – arguments for maintaining a firewall between the two regimes and avoiding the difficult work of normative engagement. Section 8.2 evaluates three proposals to demarcate the boundary lines between human rights and intellectual property, proposals whose particularities we describe in greater detail in previous chapters. Section 8.3 offers our own conception of the human rights – intellectual property interface, synthesizing and expanding upon the ideas developed in previous chapters. We distinguish between the protective and restrictive functions of international human rights law in the intellectual property context and propose a framework that identifies when human rights concerns favor revising
+
+503
+
+---
+
+504
+
+Human Rights and Intellectual Property
+
+existing intellectual property protection rules or otherwise restructuring the incentives for human creativity and innovation.
+
+### 8.1. The Unavoidable Intersection of Human Rights and Intellectual Property
+
+The previous chapters of this book describe the growing network of international organizations, government agencies, civil society groups, attorneys, commentators, and journalists whose work focuses on both human rights and intellectual property issues. Many of the individuals who participate in this network view the increasing intersections between the two regimes as beneficial. Some, however, are suspicious of these developments, preferring instead to maintain or even fortify the boundary between the regimes. Others take aim at specific points of intersection, claiming that the overlap of previously unrelated rules and institutions will result in deleterious legal or policy outcomes.
+
+This opposition is partly engendered by a resistance to change among actors who are habituated to the discourse of one complex legal and political system but not the other. But this resistance is animated by more than a reflexive fear of the unfamiliar. The two communities speak very different languages. Intellectual property commentators, especially those working in the Anglo-American tradition, employ the analytical tools of utilitarianism and welfare economics to evaluate the trade-offs between incentives and access and the consequences for the individuals and firms that create, own, and consume intellectual property products. The international human rights movement, by contrast, engages in a discourse of absolutes that seeks to delineate the negative and positive duties of states to respect and promote inalienable individual freedoms. As a result, to label something as a “human right” often invokes – in rhetoric if not always in reality – a language of trumps and unconditional demands. This emphasis on categorical rights and responsibilities appears ill suited to the rapidly changing technological and economic environment in which intellectual property rules operate, an environment that often engenders calls for incremental recalibrations of the balance between incentives and access.
+
+A second basis for resistance to the intersection of human rights and intellectual property stems not from concerns about each regime as it actually exists, but rather from opposition to actors who make rhetorical and, we believe, inflated claims grounded in one regime to support arguments for changing the other. Commentators on both sides have expressed concerns about such overclaiming. Some in the human rights community, for example, fear that
+
+---
+
+Conclusion
+
+505
+
+intellectual property owners – in particular, multinational corporations – will invoke the creators’ rights and property rights provisions of international instruments to lock in maximalist intellectual property rules that will further concentrate wealth in the hands of a few at the expense of the many. Parts of the intellectual property community, by contrast, have expressed the concern that seemingly vague calls for states to “respect, protect, ensure, and fulfill” economic and social rights are code words for more radical campaigns to promote government intervention in private innovation markets and radically scale back or even abolish intellectual property protection. A common factor that unites both sets of fears is the focus on extremist arguments that ignore the actual structure and content of each legal regime.
+
+A third explanation for resistance to the human rights–intellectual property interface stems from a concern with fragmentation of international regimes, overlapping competencies of international institutions, and conflicts among legal obligations. Worries that the international legal system is becoming overly fragmented are widespread. That system, unlike its national counterparts, lacks a single legislative, executive, or judicial body with mandatory, universal powers. It is composed of disaggregated and decentralized rules and institutions that include thousands of multilateral, regional, and bilateral treaties and customary laws; myriad nonbinding declarations and resolutions and other “soft law” norms; intergovernmental organizations with diverse memberships and subject matter competencies; international tribunals, review bodies, and arbitral panels with different jurisdictional mandates; and formal and informal networks of government, private, and hybrid regulators.
+
+Anxieties relating to the fragmentation of international legal regimes are exacerbated by institutional competence concerns. Since the adoption of the TRIPS Agreement in the mid-1990s, many important intellectual property controversies have been litigated within the World Trade Organization (WTO). The merger of trade and intellectual property has provoked a wealth of scholarly debate, much of it unflattering. But calls for the international trading system to give greater consideration to human rights concerns – both those related specifically to intellectual property and more generally – raise difficult issues as to whether WTO decision makers are adequately equipped to mediate these competing values.
+
+At the level of rule conflict, fragmentation concerns run especially high where human rights, intellectual property, and trade intersect. The applicable rules often pull in opposite directions, suggesting to some observers that their interaction is a zero sum game in which the only legal and policy choice is between wider access in the present or more innovation in the future,
+
+---
+
+506
+
+Human Rights and Intellectual Property
+
+never both. These concerns have motivated international review bodies and commentators to propose normative hierarchies that privilege one regime over the other where relevant rules conflict. As the analysis of these hierarchies in Chapter 1 reveals, however, these efforts are insufficiently theorized and highly contested. Government officials, adjudicators, nonstate actors, and scholars are unlikely to accept any wholesale normative prioritization of the two regimes, and they will continue to advance competing claims in the many diverse venues made possible by the international legal system's disaggregated structure. Continued engagement of the two regimes is therefore inevitable. Providing a constructive framework for analyzing and facilitating that engagement is one of the principal motivations for writing this book.
+
+### 8.2. Assessing Existing Proposals to Reconcile Human Rights and Intellectual Property
+
+The analyses in this book and in our previously published articles and essays on which it is based are by no means the only attempts to analyze the relationship between the human rights and intellectual property regimes. A number of scholars and international expert bodies have made thoughtful interventions on these issues over the past decade, and their arguments and proposals have enriched our own ideas about the subject. In previous chapters of this book, we give these commentators and experts pride of place by reproducing and engaging with key extracts of their writings. Here, we paint with a broader brush. We group these contributions into three broad and admittedly simplified categories, highlighting common themes, strengths, and weaknesses and laying the groundwork for our own analysis.
+
+The first group of scholars emphasizes the importance of rediscovering the historical record. $^1$ For these commentators, resolving the normative tensions engendered by the intersection of human rights and intellectual property requires unearthing the original understanding of the long-forgotten creators' rights and cultural benefit clauses in UDHR Article 27 and
+
+1 See, e.g., Audrey Chapman, Approaching Intellectual Property as a Human Right: Obligations Relating to Article 15(1)(c), 35 Copyright Bull. 4 (2001); ECOSOC, Comm. on Econ., Soc. and Cultural Rights, Drafting History of the Article 15(1)(c) of the International Covenant on Economic, Social and Cultural Rights, E/C.12/2000/15 (Oct. 9, 2000) (prepared by Maria Green); Lea Shaver, The Right to Science and Culture, 2010 Wisc. L. Rev. 121 (2010); Lea Shaver & Caterina Spanga, The Right to Take Part in Cultural Life: On Copyright and Human Rights, 27 Wisc. Int'l. L. Rev. 637 (2009); Peter K. Yu, Reconceptualizing Intellectual Property Interests in a Human Rights Framework, 40 U.C. Davis L. Rev. 1039 (2007); Paul Torremans, Copyright as a Human Right, in Paul L. C. Torremans (Ed.), Copyright and Human Rights: Freedom of Expression – Intellectual Property – Privacy 1 (2004).
+
+---
+
+Conclusion
+
+507
+
+ICESCR Article 15. As we explain in greater detail in Chapter 3, these clauses set forth legal obligations and policy objectives closely analogous to those embodied in intellectual property systems. Like the latter systems, the texts of Article 27 and of Article 15, when read together, obligate governments to recognize and reward human creativity and innovation and, at the same time, to ensure public access to the fruits of those endeavors. Striking the appropriate balance between these two goals is the central challenge that both regimes share.
+
+One of the aims of this historical research is to rediscover how the women and men who wrote Articles 27 and 15 understood that this crucial balance would be struck. By carefully parsing the negotiating histories and the wider political and social contexts that gave birth to these clauses, commentators hope to explain why the drafters included the moral and material interests of creators and the public's right to enjoy the benefits of that creativity in universal human rights instruments. For some scholars, however, this enterprise is also a precursor to a second, more ambitious goal: to provide a historical justification for giving greater weight to the public side of the balance between access and innovation and concomitantly reduced protections for the creators and owners of intellectual property products.
+
+We fully support the first objective but are more skeptical of the second. Shedding light on this obscure corner of the human rights regime is undoubtedly a worthy endeavor. However, the historical record provides only limited and ultimately inconclusive guidance. It demonstrates that the drafters strongly endorsed the right to participate in culture and to enjoy the benefits of scientific progress and its applications, rejected proposals to include copyright protection in the UDHR and ICESCR, and divided over the decision to recognize creators' rights as human rights. In the absence of greater specificity, however, the drafting history is too slender a reed on which to ground an alternative framework for how states should balance these competing goals.
+
+This use of history is misguided in another respect. Human rights law and intellectual property law are both famously dynamic, readily adapting to changing circumstances through new rounds of treaty making, interpretations by international tribunals, and revisions of national laws. A framework that privileges the original understanding of Articles 27 and 15 fails to engage with this dynamism and with the evolutions in law, politics, social values, and technology that engendered these adaptations.
+
+A second group of scholars views the increasing attention to intellectual property issues in the human rights regime as an opportunity to reexamine tools that already exist in national intellectual property laws and treaties that help government decision makers to strike a socially optimal balance between
+
+---
+
+508
+
+Human Rights and Intellectual Property
+
+incentivizing private innovation and enriching the cultural, scientific, and information commons. $^2$ For these commentators, gazing at intellectual property through a human rights lens illuminates the fact that governments have rarely used many of these policy levers or have allowed them to fall into desuetude. Once revived or expanded, this argument continues, these tools – which include subject matter exclusions, exceptions and limitations, compulsory licenses, and special and differential treatment of developing countries – can fully achieve the goals that the intellectual property system shares with the human rights regime while avoiding the risks of importing rights claims that are less susceptible to utilitarian balancing.
+
+We endorse calls to revive and expand policy levers that have long been part of the intellectual property regime as a formal matter but that are infrequently utilized in practice, whether because of lack of familiarity or pressure from international organizations, developed countries, or intellectual property industries. And we agree with the conclusions of international expert bodies that the “flexibilities” in intellectual property treaties and statutes are salutary on their own terms as well as essential to maintain compatibility with international human rights law. We part company with these scholars and experts, however, to the extent they assert that bolstering these policy tools is not only necessary but also sufficient to reconcile the human rights and intellectual property regimes.
+
+We reject this conclusion as a matter of both principle and practical politics. As a matter of principle, flexibility mechanisms provide breathing space for governments to promote a wide range of objectives that conflict or are in tension with expansive intellectual property protection rules. Commentators have offered numerous suggestions for manipulating these policy levers to enhance economic development, foster local innovation, and increase technology transfers from developed to developing countries. These are salutary goals, to be sure. But they are insufficiently connected to the protection of fundamental rights and freedoms. Stated another way, intellectual property
+
+2 See, e.g. , P. B. Hugenholtz, Copyright and Freedom of Expression in Europe , in Rochelle Cooper Dreyfuss, Harry First, & Diane Leheer Zimmerman (Eds.), Innovation Policy in an Information Age (2000); P. Bernt Hugenholtz & Ruth L. Okediji, Conceiving an International Instrument on Limitations and Exceptions to Copyright: Final Report (Mar. 2008), available at http://www.ivir.nl/publicaties/hugenholtz/finalreport2008.pdf; Ruth Okediji, Securing Intellectual Property Objectives: New Approaches to Human Rights Considerations , in Margot E. Salomon et al. (Eds.), Casting the Net Wider: Human Rights, Development and New Duty-Bearers 211 (2007); A. Strowel & F. Tulkens, Freedom of Expression and Copyright under Civil Law: Of Balance, Adaptation, and Access , in J. Griffiths & U. Suthersanen (Eds.), Copyright and Free Speech, Comparative and International Analyses (2005).
+
+---
+
+Conclusion
+
+509
+
+flexibility mechanisms expand the regulatory space available to governments. Yet they offer at best only limited guidance for restructuring creativity and innovation policies to promote human rights, including the treaty obligations and customary rules that the vast majority of states have ratified and recognized as legally binding.
+
+As a matter of practical politics, calls to revivify intellectual property flexibility mechanisms face a major structural challenge, one that engagement with the international human rights regime can help to overcome. In the existing intellectual property system, the producers and owners of intellectual property products are the only “rights” holders. All other actors – consumers, future creators, and the public generally – are relegated to an implicitly inferior status. Recognizing this imbalance, commentators have recently introduced proposals for “users’ rights,” “maximum standards” of intellectual property protection, and new international instruments that make exceptions and limitations mandatory rather than permissive.
+
+We believe that many of these initiatives, although beneficial, are misguided in a number of respects. First, the proposals are at odds with more than a century of international intellectual property lawmaking in which treaties establish only basic ground rules (such as national treatment) and minimum standards of protection (such as subject matter eligibility rules and exclusive rights). Expanding this settled approach to embrace user rights and mandatory exceptions and limitations is a challenging and contested enterprise, as demonstrated by recent debates in WIPO on proposals for a treaty on access for the visually impaired. $^3$ In contrast, arguments grounded in human rights reframe the demands of consumers, future creators, and the public as internationally guaranteed entitlements that are conceptually equivalent to those of intellectual property owners and producers.
+
+This linguistic shift is not a semantic trick; nor is it merely a rhetorical move. It also reshapes normative agendas and negotiating strategies. From a normative perspective, such a reframing directs intellectual property reform advocates to work within international human rights venues – in particular the treaty bodies and the special rapporteurs and independent experts of the Human Rights Council whose activities we examine in previous chapters – to clarify ambiguous legal norms and evaluate the human rights consequences of existing intellectual property laws and policies. It would be myopic for
+
+3 See, e.g., James Boyle, Obama's Mixed Record on Tech Policy, Fin. Times (Jan. 25, 2010); Manon Ress, Six Myths about the Treaty for People with Disabilities That Should Be Debunked Next Week?, Knowledge Ecology International (Mar. 5, 2010), available at http://keionline. org/node/795.
+
+---
+
+510
+
+Human Rights and Intellectual Property
+
+these reform advocates to refrain from influencing these human rights actors and, where appropriate, from forging alliances with them, especially now that these actors have devoted significant attention to national and international intellectual property protection rules.
+
+From a negotiating perspective, reform arguments that draw upon the output of these human rights venues have a distinct strategic advantage. They invoke legal rules and norms adopted by institutions whose provenance and legitimacy are well established and that have received the imprimatur of many governments in other international fora. By drawing upon these sources, reform advocates can more credibly claim that a rebalancing of intellectual property protection rules is necessary to harmonize two parallel regimes of internationally recognized “rights.” And they can more easily deflect claims that such rebalancing efforts are merely fig leaves for self-serving legislation by well-resourced user industries or disguised attempts to distort free trade rules or free ride on foreign creators and inventors.
+
+A third approach to reconciling human rights and intellectual property employs the rules of the former regime to bolster arguments for expanding or diminishing the rules of the latter. $^4$ Expansionist arguments are often raised by industries that view their business models and financial viability as tied to the exclusive exploitation rights that intellectual property protection confers. Seizing upon (and often misreading) the creators' rights and property rights clauses of international instruments, these industries seek to lock in maximalist intellectual property protection by invoking the rhetoric of human rights as trumps. A fear of such expansionist claims – and the perceived difficulty of refuting them – explains why some commentators are skeptical of attempts to analyze intellectual property issues in human rights terms.
+
+4 See, e.g., Audrey R. Chapman, Towards an Understanding of the Right to Enjoy the Benefits of Scientific Progress and Its Applications , 8 J. Hum. Rts. 1 (2009); Christophe Geiger, “Constitutionalising” Intellectual Property Law? The Influence of Fundamental Rights on Intellectual Property in the European Union , 37 Int’l Rev. Intell. Prop. & Competition L. 371 (2006); Christophe Geiger, Fundamental Rights, a Safeguard for the Coherence of Intellectual Property Law? , 35 Int’l Rev. Intell. Prop. & Competition L. 268 (2004); Tom Giovanetti & Merrill Matthews, Institute for Policy Innovation, Intellectual Property Rights and Human Rights , IDEAS, Sept. 2005, available at http://www.ipi.org; Burkhart Goebel, Geographical Indications and Trademarks in Europe , 95 Trademark Rep. 1165 (2005); see also Kal Raustiala, Density and Conflict in International Intellectual Property Law , 40 U.C. Davis L. Rev. 1021, 1032 (2007) (stating that “the embrace of [intellectual property] by human rights advocates and entities ... is likely to further entrench some dangerous ideas about property: in particular, that property rights as human rights ought to be inviolable and ought to receive extremely solicitous attention from the international community”).
+
+---
+
+Conclusion
+
+511
+
+These fears are not entirely unfounded, especially in Europe. The recently adopted Charter of Fundamental Rights of the European Union (EU) subsumes intellectual property under the rubric of property, and provides in Article 17(2) that “Intellectual property shall be protected.” References to fundamental rights appear in the recitals of several EU directives on intellectual property. A few national courts in Europe have relied on property guarantees in their respective constitutions when adjudicating intellectual property disputes. And, as discussed in Chapter 3, the European Court of Human Rights has extended the right of property in Protocol No. 1 to the European Convention on Human Rights to copyrights, patents, and trademarks owned by both corporations and individuals.
+
+Viewed in isolation, these trends appear ominous. But they are counterbalanced by a large and growing number of international and domestic decisions, in Europe and elsewhere, that invoke civil and political rights (in particular freedom of expression) and economic and social rights (in particular the right to health) to limit or cabin expansive interpretations of intellectual property protection rules. Commentators and public interest NGOs have endorsed these developments, urging decision makers to reach outside intellectual property's own flexibility mechanisms and safety valves to impose external limits, or maximum standards of protection, upon intellectual property owners.
+
+These two opposing frameworks share a common methodology. Each begins with the existing baseline of intellectual property protection and then invokes selective provisions of international human rights law to bolster arguments for moving that baseline in one direction or the other. The frameworks also share a common flaw. They encourage uncoordinated interventions at the upper and lower boundaries of intellectual property protection, interventions that, over time, would establish both a floor and a ceiling on intellectual property.
+
+This selective use of human rights law to impose upper and lower limits on intellectual property protection standards is worrisome. These efforts have mostly ignored the creators' rights and cultural rights provisions of UDHR Article 27 and ICESCR Article 15. They have instead invoked human rights that are unconcerned with balancing the protection of creators and innovators against the public's right to benefit from the scientific and cultural advances. Lacking a coherent blueprint to undertake the sensitive and policy-laden analysis that such balancing requires, human rights interventions at the upper and lower boundaries of intellectual property law will inevitably be ad hoc. They may also create cycles of underprotection and overprotection, depending on the vagaries of which issues are raised, in which venues, and in what order.
+
+---
+
+512
+
+Human Rights and Intellectual Property
+
+### 8.3. Toward a Human Rights Framework for Intellectual Property
+
+In this section, we offer our own framework for understanding the interface between human rights and intellectual property. As we explain in greater detail below, our framework differs from the approaches reviewed previously in several important respects. First, it is capacious, encompassing the full panoply of human rights and freedoms whose realization is affected by intellectual property protection rules. Second, our framework rejects arguments that invoke human rights to leverage across-the-board expansions or rollbacks of intellectual property protection. Third, our framework is empirically grounded. It urges governments, before revising the status quo, first to determine whether and to what extent intellectual property – as opposed to other factors – impedes or enhances the attainment of desired human rights outcomes. Fourth, our framework is dynamic. It draws inspiration from the drafting history of the creators’ rights and cultural benefit clauses, but recognizes that the human rights and intellectual property regimes are continually evolving in response to changing conceptions of legal entitlements and technological progress.
+
+As an initial matter, we distinguish between the protective and restrictive dimensions of human rights in the intellectual property context. The protective dimension requires states (1) to recognize and respect the rights of individuals and groups to enjoy a modicum of economic and moral benefit from their creative and innovative activities and (2) to refrain from bad faith and arbitrary interferences with intellectual property rights that the state itself has previously granted or recognized. In contrast, the restrictive dimension, which includes both a process component and a substantive standard, identifies the conditions under which the realization of a specific right or freedom requires (1) a diminution of intellectual property protection standards and enforcement measures, (2) a restructuring of incentives for private creativity and innovation, or (3) both.
+
+Our framework also stresses the importance of the process, transparency, and predictability values that are hallmarks of the rule of law. $^5$ The founding documents of the international human rights movement did not emphasize
+
+5 There are many different conceptions of the rule of law. See Note, Thom Ringer, Development, Reform, and the Rule of Law: Some Prescriptions for a Common Understanding of the "Rule of Law" and Its Place in Development Theory and Practice, 10 Yale Hum. Rts. & Develop. L.J. 178 (2007). For a recent application to international intellectual property, see Laurence R. Helfer, Karen J. Alter, & M. Florencia Guerzovich, Islands of Effective International Adjudication: Constructing an Intellectual Property Rule of Law in the Andean Community, 103 Am. J. Int'l L. 1 (2009).
+
+---
+
+Conclusion
+
+513
+
+the rule of law. Today, however, the connection between human rights and the rule of law is well established6 and provides additional arguments for contesting intellectual property initiatives that conflict with rule of law values.
+
+A salient recent example is the Anti-Counterfeiting Trade Agreement (ACTA), a proposed multilateral treaty that would establish more robust obligations to suppress unauthorized uses of intellectual property. For two years, ACTA negotiations occurred in secret and governments refused to disclose a draft text of the treaty. Only after a French civil rights NGO leaked a document revealing “contradictions between the text and public comments by negotiators”? did governments release an official text. 8 Such lack of transparency involving potentially far-reaching changes to intellectual property laws and enforcement mechanisms is disturbing, as is the inability of interested constituencies, in the words of the Committee on Economic, Social and Cultural Rights, to “take part in … any significant decision making processes that have an impact on their rights and legitimate interests.” 9
+
+## A. The Protective Dimension of the Framework
+
+The protective dimension of the human rights framework for intellectual property is grounded in state obligations to respect, protect, and fulfill creators’ rights and the right of property found in several international instruments, most notably UDHR Article 27, ICESCR Article 15(1)(c), and Article 1 of the European Convention’s First Protocol.
+
+Obligations with regard to creators' rights encompass modest economic exploitation and personality guarantees that, taken together, are more circumscribed than those imposed by intellectual property treaties. The limited scope of these guarantees can be deduced from the two principal objectives of recognizing the moral and material interests of creators as human rights.
+
+6 See generally Randall Peerenboom, Human Rights and Rule of Law: What's the Relationship?, 36 Geo. Int'l. L.J. 809 (2005).
+
+7 Monika Ermert, Leaked ACTA Text Shows Possible Contradictions with National Laws, Intellectual Property Watch (Mar. 29, 2010), available at http://www.ip-watch.org/ weblog/2010/03/29/leaked-acta-text-shows-possible-contradictions-with-national-laws/.
+
+8 The Office of the U.S. Trade Representative Releases Statement of ACTA Negotiating Partners on Recent ACTA Negotiations (Apr. 16, 2010), available at http://www.ustr.gov/about-us/ press-office/press-releases/2010/april/office-us-trade-representative-releases-statement-ac. The U.S. Trade Representative stressed that “ACTA will not interfere with a signatory's ability to respect its citizens' fundamental rights and liberties, and will be consistent with the ... TRIPS Agreement and will respect the Declaration on TRIPS and Public Health.” Id.
+
+9 Comm. Econ., Soc. & Cultural Rights, General Comment No. 17: The Right of Everyone to Benefit from the Protection of the Moral and Material Interests Resulting from Any Scientific, Literary or Artistic Production of Which He Is the Author, art. 15(1)(c), U.N. Doc. E/C.12/ GC17, para. 34 (Jan. 12, 2006).
+
+---
+
+514
+
+Human Rights and Intellectual Property
+
+According to the General Comment 17 of the Committee on Economic, Social and Cultural Rights, analyzed in depth in Chapter 3, such rights “safeguard the personal link between authors and their creations and between people or other groups and their collective cultural heritage,” and they protect the “basic material interests which are necessary to enable authors to enjoy an adequate standard of living.”
+
+We interpret these two statements, which recur throughout the General Comment , to imply the existence of a zone of personal autonomy in which individuals can achieve their creative potential, control their productive output, and lead the independent intellectual lives that are essential requisites of any free society. The legal protections required to establish this zone are, however, significantly narrower than those mandated by intellectual property treaties and statutes. As an initial matter, these protections do not apply to corporations or other business entities. But even as to individuals and groups, a state can satisfy its obligation to protect creators' rights in myriad and diverse ways. It may, for example, recognize the same exclusive rights as are found in intellectual property treaties and statutes, but radically reduce terms of protection, expand exceptions and limitations, or both. Alternatively, a state could eschew exclusive rights altogether (except for minimal attribution and integrity guarantees) and substitute a system of liability rules, levies, or government subsidies. $^10$ Under either approach, governments could also designate certain socially valuable uses of knowledge goods as not requiring any remuneration to creators.
+
+The protective dimension of the human rights framework is more expansive than existing intellectual property protection rules in only two respects. First, it encompasses all individuals and groups; the categorical exclusion of a class of creators would be inconsistent with the framework. The absence of protection in some countries for the traditional knowledge of indigenous communities is one example, although, as we discuss at the end of Chapter 7, the potential conflicts between the recognition and assertion of rights in indigenous creativity and other human rights must also be considered.
+
+Second, the protective dimension of the framework imposes a more stringent test for evaluating restrictions within the irreducible core of rights that establishes the zone of autonomy described earlier. Such restrictions must, among other requirements, be “ strictly necessary for the promotion of the general welfare in a democratic society ” and must employ “ the least
+
+10 For a thoughtful discussion of liability rules, see Arti K. Rai, Jerome H. Reichman, Paul F. Uhlir, & Colin Crossman, Pathways across the Valley of Death: Novel Intellectual Property Strategies for Accelerated Drug Discovery , 8 Yale Journal of Health Law, Policy & Ethics 53, 78–80 (2008).
+
+---
+
+Conclusion
+
+515
+
+restrictive measures ... when several types of limitations may be imposed. ” This standard is substantially more constraining than the now ubiquitous “ three-step test ” used to assess the TRIPS-compatibility of exceptions and limitations in national intellectual property laws. 11 For this reason, the rights included in the core must be concomitantly narrow.
+
+We emphasize, however, the limited practical consequences of these additional obligations. For traditional knowledge, for example, a state could eschew exclusive rights altogether in favor of government assistance programs that seek to preserve the creative works of indigenous communities consistently with their religious beliefs and cultural traditions. As for restrictions on creators' human rights, these must be interpreted in light of the narrowness of the zone of autonomy itself and the many permissible approaches to establishing that zone.
+
+The protective dimension of the human rights framework for intellectual property may also justify more expansive legal protections for individuals and groups vis-à-vis other actors involved in the production and distribution of knowledge goods. For example, the framework's emphasis on human creativity rather than economic exploitation may support a more circumscribed approach to work for hire rules that grant authorship and control of copyrighted works to corporate employers at the expense of those who work for them. But nothing requires that any revisions of domestic intellectual property laws adhere to any particular template. To the contrary, the protective dimension of the framework preserves wide latitude for states to regulate innovation and creativity to achieve socially beneficial ends and to tailor regulations to political, economic, and cultural conditions within their borders.
+
+The property rights component of the protective dimension is similarly modest. Government officials included the right of property in the UDHR and in the three regional civil and political rights conventions but excluded it from the ICCPR and the ICESCR. This omission reveals that many countries have no treaty-based obligation to protect private property as a human right, although they may protect property on other legal grounds. And for states that do have such an international obligation, the treaties' drafting histories evidence a clear intent to preserve latitude for governments to adopt economic and social policies that adversely affect property owners while, at the same time, condemning arbitrary deprivations of property by state actors.
+
+11 For example, Article 13 of the TRIPS Agreement provides that “members shall confine limitations or exceptions to exclusive rights to certain special cases which do not conflict with a normal exploitation of the work and do not unreasonably prejudice the legitimate interests of the right holder.”
+
+---
+
+516
+
+Human Rights and Intellectual Property
+
+Among the three regional treaties that include a property rights clause, only the European system has considered whether that provision encompasses ownership of intangible knowledge goods. In Chapter 3, we analyze recent rulings of the European Court of Human Rights that answer this question in the affirmative and also emphasize the limited protection that the Court's case law provides. Here we offer more general guidance to decision makers as to how to interpret the human right of property as applied to intellectual property.
+
+Consistent with the rule of law values that these treaty clauses embody, decision makers should find fault only with arbitrary or ultra vires exercises of state power and bad faith refusals to follow intellectual property protection rules that the state itself has previously recognized as valid. Such actions include, for example, a government ministry that installs copyrighted software on its desktop computers without providing statutorily mandated remuneration to the software's owner, a state-run enterprise that refuses to pay royalties to an inventor whose locally patented process it had previously licensed, and judicial or administrative rules that eschew minimum procedural guarantees such as the ability to present evidence or legal arguments.
+
+As these examples illustrate, the restrictions imposed by treaty-based property rights clauses are minimal and unobtrusive. They allow governments unfettered discretion to fashion their domestic innovation and creativity policies as they see fit, provided only that they adhere to the previously established rules that embody those policies. This narrow focus also justifies the application of these principles to intellectual property owned by corporations and other business entities, since arbitrary and bad faith deprivations of property are not confined to natural persons.
+
+## B. The Restrictive Dimension of the Framework
+
+The restrictive dimension of the framework comes into play where a state expands legal protections for creativity and innovation beyond those required to establish the zone of personal autonomy described in the previous subsection. There are longstanding debates over whether capacious intellectual property protection helps or hinders economic growth, especially in least-developed and developing countries. But even assuming for purposes of argument that advocates for strong intellectual property protection have the better of this debate, they must still contend with the obligations that international human rights law imposes, obligations that may provide an independent legal basis for cabining strong intellectual property protection rules even if they enhance economic development. We part company, however, with commentators who invoke human rights to support an across-the-board
+
+---
+
+Conclusion
+
+517
+
+rollback of intellectual property without regard to context or to the ways in which it can be harnessed to promote the realization of human rights. We advance instead an approach that is both faithful to the diversity of individual rights and freedoms and grounded in empirical reality.
+
+We begin from the uncontroversial premise that the ends of international human rights law – including noninterference with civil and political rights and guaranteeing minimum levels of economic and social well-being in areas such as health, food, and education – can be achieved in a wide variety of ways. Intellectual property protection can help or hinder the attainment of these ends, or it may be entirely irrelevant to their realization. The first component of the framework's restrictive dimension, therefore, is a process inquiry that seeks to determine what role, if any, intellectual property protection actually plays in this regard.
+
+If the institutions, resources, personnel, and other inputs necessary to achieve desired human rights outcomes do not exist or are inadequate, the issue of whether intellectual property also impedes those outcomes may be entirely irrelevant. Stated differently, the barriers to realizing human rights are often overdetermined, with intellectual property functioning as only one among a multiplicity of barriers, and not necessarily the most important one. This analysis harkens back to debates in the early 2000s, discussed in Chapter 2, as to whether pharmaceutical patents hindered access to HIV/AIDS medications in sub-Saharan Africa. Even if antiretroviral drugs were given to these countries free of charge, proponents of strong patent protection claimed, the public health infrastructure needed to distribute them was inadequate and the individuals who received the medicines were incapable of following directions for their ingestion without the assistance of medical professionals.
+
+Nearly a decade later, these arguments have proven to be mostly groundless. But the basic insight underlying these claims – that multiple factors unrelated to intellectual property often act as barriers to human rights outcomes – remains valid. It is difficult to contend, for example, that copyright in educational materials impedes the right to education if there are no school buildings and no teachers. This illustration is, admittedly, an oversimplification. As the discussion in Chapter 5 reveals, however, even in countries with more fully functional educational systems, the adverse consequences of copyright protection are often minimal in comparison to factors such as language barriers, small domestic publishing industries, and tariffs on paper imports. Another relevant variable is the extent to which educational materials are available online without charge. Empirical analyses that consider such availability should not, however, presume that such materials are uniformly available; nor should they ignore the many economic and technological barriers to online access.
+
+---
+
+518
+
+Human Rights and Intellectual Property
+
+For other human rights, by contrast, the concern is not with structural factors that impede access to goods protected by intellectual property but, to the contrary, with the qualities or attributes of goods that are often widely distributed. As explained in Chapter 6 on the right to food, opposition to intellectual property protection for new plant varieties is often bound up with fears about the health and environmental consequences of genetically modified crops or opposition to the marketing practices of agrobiotechnology firms. These are legitimate concerns. But the proper response to them is not – or at least not necessarily – a diminution of intellectual property protection. Rather, what is first required is a careful evaluation of the human rights implications of these claims and the role that intellectual property does or does not play in exacerbating them. If further studies reveal, for example, that certain genetically modified crops are harmful to human health or to farmers who grow traditional plant varieties, an appropriate response by health, environment, or agriculture ministries would be to regulate or prohibit the distribution and sale of such crops. Similarly, if the consolidation of the commercial seed industry enables a few firms with excessive market power to demand artificially high prices for seeds, the remedy lies in national competition laws rather than in restricting intellectual property rules that create incentives for new plant varieties with desirable characteristics. In addition, it may be useful to distinguish between problems caused by the subsistence of intellectual property rights in genetically modified plant varieties as such, and problems engendered by the decisions of public and private actors to adopt, promote, or subsidize such varieties.
+
+The determination of whether and to what extent intellectual property, as opposed to other factors, impedes the attainment of desired human rights outcomes requires careful, objective, and context-specific empirical assessments. Over the last several years, a growing array of international bodies, NGOs, and scholars have turned their attention to the previously understudied issue of how to measure the enjoyment of human rights. The result has been an outpouring of indicators, metrics, benchmarks, impact statements, and other measurement tools that seek to identify with greater precision the levels of rights protections in individual countries and the factors that contribute to or retard their achievement. Most of these tools focus on economic and social rights, whose realization requires identifying aggregate outcomes at the societal as well as the individual level. These quantitative and qualitative indicators and benchmarks have become key elements of the iterative process by which treaty bodies, special rapporteurs, and national courts monitor the progressive realization of rights that the ICESCR protects. $^12$
+
+12 See, e.g., Alana Klein, Judging as Nudging: New Governance Approaches for the Enforcement of Constitutional Social and Economic Rights, 39 COLUM. HUM. RTS. L. REV. 351 (2008); Office of
+
+---
+
+Conclusion
+
+519
+
+These measurement tools do not, as far as we are aware, systematically assess the positive and negative consequences of intellectual property protection on human rights in general or on economic and social rights in particular. But they could easily be revised to include such an evaluative component. A harbinger of this approach is a 2006 report of the National Human Rights Commission of Thailand, which reviewed a draft Thailand–United States Free Trade Agreement. 13 Among other issues, the Commission analyzed the treaty's inclusion of TRIPS Plus provisions from the perspective of the right to health and farmers' rights. It recommended, inter alia , that the Thai government remove from the negotiations stronger intellectual property protection for pharmaceuticals. 14 As other commentators have observed, however, the report also used “emotive language and strong claims about the effects of the FTA without recourse to empirical evidence to support those claims.” 15
+
+The political contestations surrounding the Thai Commission's report highlight the need to develop, in advance of any particular controversy, measurement tools that have been accepted by stakeholders with divergent viewpoints, or at least that reflect their input. These measurement tools should include at least the following components: (1) an evaluation of whether existing or proposed intellectual property protection rules and policies help or hinder the realization of specific human rights outcomes; (2) an assessment, to the greatest extent possible, of the relative causal contributions of intellectual property rules and policies in comparison to other factors; and (3) an identification of the legal and policy measures, whether or not consistent with the existing intellectual property regime, that will facilitate these human rights outcomes.
+
+If the assessment of these issues reveals that non-intellectual property factors are responsible for the lack of progress in realizing human rights ends,
+
+the High Commissioner for Human Rights, Report on Indicators for Monitoring Compliance with Human Rights Instruments, U.N. Doc. HRI/MC/2006/7 (2006); AnnJanette Rosga & Margaret L. Satterthwaite, Trust in Indicators: Measuring Human Rights, 27 Berkeley J. Int'l L. 254 (2009); Lea Shaver, Defining and Measuring Access to Knowledge: Towards an A2K Index, Yale Law School Student Scholarship Papers, available at http://digitalcommons. law.yale.edu/fss_papers/22/ (2007); Sally Engle Merry, Measuring the World: Indicators, Human Rights, and Global Governance (May 2009), available at http://www.iilj.org/research/ documents/I.Merry.MeasuringtheworldASIL.pdf.
+
+13 See National Human Rights Institutions Forum, Human Rights Impact Assessment of the US–Thai Free Trade Agreement (Jan. 22, 2007), available at http://www.nhri.net/news. asp?ID=1115.
+
+14 See James Harrison & Hessa Goller, Trade and Human Rights: What Does "Impact Assessment" Have to Offer?, 8 Hum. Rts. L. Rev. 587, 604 (2008).
+
+15 Id. at 608 (quoting inflammatory statements in the commission's report, including that the treaty will "pave the way for [transnational corporations] to seize power" and "US demands on patents ... clearly reflect greed on [the] part of US pharmaceutical corporations").
+
+---
+
+520
+
+Human Rights and Intellectual Property
+
+as may be the case for the example of genetically modified crops discussed earlier, state and nonstate actors should focus their lawmaking and advocacy strategies on those factors and should not treat intellectual property issues as a proxy for them. Such categorical outcomes are likely to be rare, however. A more frequent result of the assessment process will be a finding that the specific intellectual property rule or policy under scrutiny is one among many factors responsible for deleterious human rights conditions. Such a conclusion implicates policy responses that address both the share of the problem attributable to intellectual property and the type and extent of the harm that it engenders. The structure of domestic institutions, the extent of available resources and their reallocation, the sequencing of policy prescriptions, and the interrelationship among government programs will be important issues in this regard. We hope that the analysis set forth in previous chapters of this book will assist all stakeholders in addressing these issues.
+
+In the final analysis, however, national decision makers will need to decide whether to revise existing intellectual property protection rules and how best to do so. It is here that the second, substantive stage of the framework's restrictive dimension comes into play. In deciding what measures to take, we urge decision makers to begin from the premise that the human rights and intellectual property regimes share the same core objective – to encourage creativity and innovation that benefits society as a whole. It is the different ways that each regime achieves this objective, which create the potential for conflicts between them.
+
+In the intellectual property system, most societal benefits accrue far in the future when knowledge goods enter the public domain and may be freely used by all. Flexibility mechanisms such as exceptions to exclusive rights and compulsory licenses mitigate the costs of this delay. But they can only do so much without harming the incentives to create and innovate in the first instance. In contrast, the human rights regime has much shorter time horizons. The legal entitlements it enshrines are both immediate and urgent. The regime has little tolerance for states that lack the present ability to meet their negative obligation to refrain from repression or their positive commitment to protect and fulfill the minimum essential needs of individuals and groups.
+
+Intellectual property-protected knowledge goods help to satisfy these immediate demands when the owners of these goods sell or license them to consumers. But the monopoly power that accompanies intellectual property rights enables owners to maximize profits by offering knowledge goods at supracompetitive prices that exclude consumers who would have purchased or licensed the goods had they been offered in a competitive market. The
+
+---
+
+Conclusion
+
+521
+
+result is that individuals with greater financial means can afford knowledge goods whereas those with fewer economic resources cannot.
+
+These disparities in affordability apply to intellectual property-protected goods currently being offered for sale or license. But the disparities are exacerbated when incentives to create and innovate are considered from a dynamic perspective. Intellectual property industries respond to existing market signals by fashioning research and development strategies to satisfy the anticipated demands of consumers with financial means. The pernicious consequences of these dynamic innovation incentives are illustrated most starkly in the area of patented medicines. As we analyze in Chapter 2, pharmaceutical companies devote the bulk of their research efforts to identifying new drugs for ailments common in wealthy industrialized nations while eschewing research on diseases that afflict the world's poor, who cannot afford any treatments the companies might have developed.
+
+The intellectual property system is generally agnostic about both the static and the dynamic distributional consequences of monopoly pricing structures. But these distributional consequences are a central concern of human rights law in general and economic and social rights in particular, which prioritize the needs of the most marginalized and disadvantaged individuals and groups above the needs of those with greater financial means. Stated more pointedly, intellectual property protection may help states to satisfy their obligations to protect and fulfill economic and social rights. But its effect is greatest where it is needed least.
+
+There are short-term and long-term responses to this troubling state of affairs. Both responses depend upon the findings of the indicators and impact statements described previously. If these measurement tools reveal that specific intellectual property protection rules are (or, in the case of proposed rules, would be) an immediate obstacle to the realization of specific human rights, governments should revise those rules or, in the case of new rules, reject proposals to adopt them. Impact statements structured according to our recommendations should also indicate which legal and policy measures would help to achieve this result. All other things equal, we think that governments should favor measures compatible with the existing intellectual property regime over measures that are inconsistent with it. But we also believe that governments should be free to choose intellectual propertyinconsistent measures where the indicators and impact statements contain credible evidence that such measures are likely to achieve more extensive human rights benefits. Where the evidence is equivocal or uncertain, measures should be temporary and include sunset clauses to force a revaluation of their merits an appropriate interval after their adoption.
+
+---
+
+522
+
+Human Rights and Intellectual Property
+
+With regard to long-term responses, indicators and impact statements should provide a roadmap for governments to restructure innovation incentives to further human rights ends. Strategies to encourage research relating to neglected diseases have advanced further in this regard than initiatives in other areas. As described in Chapter 2 on the right to health and pharmaceutical patents, these strategies work with intellectual property rather than against it. They redirect incentives and channel market forces to achieve socially valuable ends. Thomas Pogge's research program on human rights and global health is perhaps the most advanced proposal in this regard. We urge states, public interest NGOs, and the staff of international organizations to develop similar proposals for other intersections between the human rights and intellectual property regimes and to tailor incentive structures to the diverse economic and political realities that we describe in previous chapters.
+
+In developing these proposals, actors should also consider whether nonproprietary innovation schemes can help to achieve salutary human rights outcomes. For example, open source systems that require follow-on innovators to share their contribution to collectively produced knowledge goods should be encouraged, provided that the system's policies are fully disclosed to participants. Private contracting and delegation mechanisms, such as Creative Commons, that permit creators to disclaim intellectual property protection in whole or in part deserve similar solicitude. But as with online access to educational materials, the widespread accessibility of nonproprietary alternatives should not be assumed. Where access disparities exist, an overemphasis on nonproprietary mechanisms may have the perverse if unintended effect of disfavoring the less technologically adept or those burdened by economic barriers to online access. Inattention to access disparities thus creates a risk that nonproprietary alternatives will be least available to those who require them most.
+
+Even assuming widespread and equitable access, nonproprietary alternatives may appear contrary to the protective dimension of the human rights framework for intellectual property, which, as described previously, protects a zone of personal autonomy for all creators. In practice, there may be no incompatibility if individuals retain the right to be acknowledged as creators and to receive remuneration for at least some uses. The more fundamental point, however, is that although creators and innovators do indeed possess a narrow class of inalienable economic and personality rights, they can choose how best to exercise those rights so as to construct a zone of personal autonomy that is both self-empowering and conducive to the broader public values that the human rights framework for intellectual property seeks to achieve.
+

processed/00075_W2112085319/document.md

@@ -0,0 +1,514 @@
+THE MODERN LAW REVIEW
+
+ARTICLES
+
+# The Tragedy of the Anticommons: A Concise Introduction and Lexicon
+
+Michael Heller*
+
+This article gives a concise introduction to the ‘tragedy of the anticommons.’ The anticommons thesis is simple: when too many people own pieces of one thing, nobody can use it. Usually, private ownership creates wealth. But too much ownership has the opposite effect – it leads to wasteful underuse. This is a free market paradox that shows up all across the global economy. If too many owners control a single resource, cooperation breaks down, wealth disappears, and everybody loses. Conceptually, underuse in an anticommons mirrors the familiar problem of overuse in a ‘tragedy of the commons.’ The field of anticommons studies is now well-established. Over a thousand scholars have detailed examples from across the innovation frontier, including drug patenting, telecom licensing, climate change, compulsory land purchase, oil field unitisation, music and art copyright, and post-socialist economic transition. Fixing anticommons tragedy is a key challenge for any legal system committed to innovation and economic growth.
+
+Some years ago, a drug company executive presented me with an unsettling puzzle. His scientists had found a potential treatment for Alzheimer's disease, but they couldn't develop it for the market unless the company bought access to dozens of patents. Any single patent owner could demand a huge payoff; some blocked the whole deal.This story does not have a happy ending.The drug sits on the shelf though it might have saved millions of lives and earned billions of dollars.1
+
+Here's a second high stakes puzzle: what's the most underused natural resource in America? The answer may be a surprise. It's the air waves. Over 90 per cent is dead air because ownership of broadcast spectrum is so fragmented. As a result, America's information economy is relatively hobbled; wireless broadband capacity lags far behind that in Japan and Korea. The cost of spectrum underuse may be in the trillions.
+
+And another puzzle: why do we waste weeks of our lives stuck in airports? Bad law for real estate assembly. In America, air travel was deregulated thirty-five years
+
+*Lawrence A.Wien Professor of Real Estate Law, Columbia Law School. This article is adapted from the
+
+41 st Annual Chorley Lecture given at the London School of Economics, 12 June 2012 (finally revised 10 October 2012).The article draws substantially from the Preface and from Chapter 2 of M.A. Heller, The Gridlock Economy: How Too Much Ownership Weeks Markets, Stops Innovation, and Costs Lives (New York: Basic Books, 2008).
+
+1 The fullest account of anticommons theory and solutions appears in The Gridlock Economy ibid. On the anticommons in drug patents, see ibid 49–78 (chapter 3); in telecom, see ibid 79–106 (chapter 4); in land, see ibid 107–142 (chapter 5). See also M. A. Heller (ed), Commons and Anticommons (Cheltenham, UK: Elgar Publishing, 2011) (collecting and reprinting, in two volumes, the key scholarly articles on the theory and economics of commons and anticommons property).
+
+© 2013 The Author.The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6–25 Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
+
+---
+
+Michael Heller
+
+ago.The number of fliers has tripled. So how many new airports have been built since 1975? One. Denver.You can't build new airports, not anywhere, because multiple landowners can block every project. Twenty-five new runways at America's busiest airports would end most routine air travel delays in the country. Imagine that.
+
+All these puzzles share a common cause: when too many people own pieces of one thing, nobody can use it.The anticommons thesis is that simple: when too many people own pieces of one thing, nobody can use it. Usually, private ownership creates wealth. But too much ownership has the opposite effect – it leads to resource underuse in an anticommons. This is a free market paradox I discovered and it shows up all across the global economy. If too many owners control a single resource, cooperation breaks down, wealth disappears, and everybody loses.
+
+There has been an unnoticed revolution in how we create wealth. In the old economy, twenty or thirty years ago, you invented a product and got a patent; you wrote a song and got a copyright; you subdivided land and built houses. Today, the leading edge of wealth creation requires assembly. From drugs to telecom, software to semiconductors, anything high-tech demands the assembly of innumerable patents. And it's not just high tech that's changed — today, cutting edge art and music are about mashing up and remixing many separately-owned bits of culture. Even with land, the most socially-important projects, like new runways, require assembling multiple parcels. Innovation has moved on, but we are stuck with old-style ownership that's easy to fragment and hard to put together.
+
+Fixing anticommons tragedy is a key challenge for our time. Some solutions are entrepreneurial; for example, people can profit from finding creative ways to bundle ownership. Philanthropists can assemble patents for disease cures. Political advocacy and legal reform will be needed to secure solutions. But the first and most important step in solving anticommons tragedy is to name it and make it visible. This article takes that step by briefly introducing the anticommons lexicon. With the right language, anyone can spot links among anticommons puzzles, and all can come together to fix them.
+
+## COMMONS AND ANTICOMMONS
+
+To understand the dilemma of resource underuse in an anticommons, it is helpful to start with overuse in a commons. Aristotle was among the first to note how shared ownership can lead to overuse: `That which is common to the greatest number has the least care bestowed upon it . . . each thinks chiefly of his own, hardly at all of the common interest; and only when he is himself concerned as an individual.' $^2$
+
+2 Aristotle, The Politics and the Constitution of Athens (S. Everson (ed), B. Jowett (trans) Cambridge: Cambridge UP, 1996) 33. Before Aristotle, Thucydides noted that people ‘devote a very small fraction of time to the consideration of any public object, most of it to the prosecution of their own objects. Meanwhile each fancies that no harm will come to his neglect, that it is the business of
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+7
+
+---
+
+The Anticommons
+
+Why do people overuse and destroy things that they value? Perhaps they are shortsighted or dim-witted, in which case reasoned discussion or gentle persuasion may help. But even the clear-headed can overuse a commons, for good reasons. The most intractable overuse tragedy arises when individuals choose rationally to consume a common pool of scarce resources even though each knows that the sum of these decisions destroys the resource for all. In such settings, reason cuts the wrong way and gentle persuasion is ineffective. In other words, I do what's best for me, you do what's best for you, and no one pays heed to the sustainability of the shared resource.
+
+Ecologist Garrett Hardin captured this dynamic well when he coined the phrase tragedy of the commons. $^3$ In 1968 he wrote, `Ruin is the destination toward which all men rush, each pursuing his own best interest in a society that believes in the freedom of the commons. Freedom in a commons brings ruin to all.' $^4$ Since Hardin wrote these lines, thousands have identified additional areas susceptible to overuse and commons tragedy. $^5$
+
+In addition, Hardin's metaphor inspired a search for solutions. Most solutions revolve around two main approaches: regulation or privatisation. Suppose a common lake is being overfished. Regulators can step in and decide who can fish, when, how much, and with what methods. Such direct `command-andcontrol' regulation has dropped from favor, however, partly because it fails so often and partly because of disenchantment with socialist-type regulatory control.
+
+These days, regulators are more likely to look for some way to privatise access to the lake.They know that divvying up ownership can create powerful personal incentives to conserve. Harvest too many fish in your own lake today, starve tomorrow; invest wisely in the lake, profit forever. Extrapolating from such experience, legislators and voters reason – wrongly – that if some private property is a good thing, more must be better. In this view, privatisation can never go too far.
+
+Until now, ownership, competition, and markets – the guts of modern capitalism – have been understood through the opposition suggested by figure 1 . Private property solves the tragedy of the commons. Privatisation beats regulation. Market competition outperforms state control. Capitalism trounces socialism. But these simple oppositions mistake the visible forms of ownership for the whole spectrum. The assumption is fatally incomplete.
+
+somebody else to look after this or that for him; and so, by the same notion being entertained by all separately, the common cause imperceptibly decays' History of the Peloponnesian War (R. Crawley (trans), New York: E.P. Dutton, 1910) bk 1, sec 141.
+
+3 G. Hardin, 'The Tragedy of the Commons' (1968) 162 Science 1243, 1244. The power of Hardin's rhetoric sometimes exceeded the reach of his data. For example, Hardin's work overlooks the important distinction between 'open access' and 'limited access commons'. On the implications of this distinction for common pool resource dilemmas, see below n 15 and accompanying text.
+
+4 Hardin, ibid 1244.
+
+5 See, eg, 'The Digital Library of the Commons' http://dlc.dlib.indiana.edu (last visited 2 October 2012).
+
+8
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+![Figure](figures/figure_001.png)
+
+Figure 1: The standard solution to commons tragedy
+
+Privatising a commons may cure the tragedy of wasteful overuse, but it may inadvertently spark the opposite. English lacks a term to denote wasteful underuse. To describe this type of fragmentation, I coined the phrase tragedy of the anticommons. $^6$ The term covers any setting in which too many people can block each other from creating or using a scarce resource. Rightly understood, the opposite of overuse in a commons is underuse in an anticommons.
+
+This concept makes visible the hidden half of our ownership spectrum, a world of social relations as complex and extensive as any we have previously known (see figure 2 ). Beyond normal private property lies anticommons ownership. As one commentator notes, `To simplify a little, the tragedy of the commons tells us why things are likely to fall apart, and the tragedy of the anticommons helps explain why it is often so hard to get them back together.' $^7$
+
+![Figure](figures/figure_005.png)
+
+Figure 2: Revealing the hidden half of the ownership spectrum
+
+Making anticommons ownership visible is not easy. Let me give you an image that I have found helpful in crystallising the abstract idea. A thousand years ago, the Rhine was one of the world's great trade routes. Boatmen traded under the protection of the Holy Roman Emperor. When the Emperor weakened in the 13 th century, freelance German barons built castles and began collecting their own illegal tolls. The growing gauntlet of `robber baron' castles, over 200 at one point, make for a wonderful bicycling holiday today, but made shipping impracticable back then. The river continued to flow, but for 500 years, boatmen would not bother making the journey. As one boatman's plaintive song went:
+
+6 M.A. Heller, 'The Tragedy of the Anticommons: Property in the Transition from Marx to Markets' (1998) 111 Harvard Law Review 621, 624. In turn, my term builds on earlier conceptual work by Frank Michelman, ibid 667-669 (discussing evolution of the concept); see also F.J. Michelman, 'Ethics, Economics and the Law of Property' (1982) 24 Nomos 3 (discussing what he calls the 'regulatory regime').
+
+7 L. A. Fennell, 'Common Interest Tragedies' (2004) 98 Northwestern University Law Review 907, 936-937.
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+9
+
+---
+
+The Anticommons
+
+The Rhine can count more tolls than miles And knight and priestling grind us down The tollman's heavy hand falls first, Behind him stands the greedy line Master of tolls, assayer, scribe Four man deep they tap the wine. $^8$
+
+Everyone suffered – even the robber barons.The European economic pie shrank. Wealth disappeared. Too many tolls meant too little trade. To understand anticommons tragedy, just update this image. Phantom tollbooths today emerge whenever ownership first arises and property is being created all the time in ways many of us do not realise. Today’s robber barons are often private owners and public regulators, all the people holding vetoes on the path to innovation.Today’s missing river trade takes the form of crushed entrepreneurial energy and forgone investment.When too many public decision makers or private owners can block access to a resource, they harm us all.
+
+Often, we think that governments need only to create clear property rights and then get out of the way. So long as rights are clear, owners can trade in markets, move resources to higher valued uses, and generate wealth. But clear rights and ordinary markets are not enough.The anticommons perspective shows that the content of property rights matters as much as the clarity .Wasteful underuse can arise when ownership rights and regulatory controls are too fragmented.
+
+Making the tragedy of the anticommons visible upends our intuitions about private property. Private property can no longer be seen as the end point of ownership. Privatisation can go too far, to the point where it destroys rather than creates wealth. Too many owners paralyse markets because everyone blocks everyone else. Well-functioning private property is a fragile balance poised between the extremes of overuse and underuse.
+
+## THE MAGICAL PARKING LOT
+
+So far, I've introduced the nutshell version of the commons and anticommons. To understand the concepts more fully, imagine you've discovered an empty paved lot near Leicester Square in London.At first, the parking paradise is free and open to all. No one tickets or tows.You park and go to the theater. No problem. Later, you tell friends, who park there too. No problem. But then others notice, and soon the lot is jammed. Cars are blocked in. Doors are dinged. Fights break out.The lot becomes a scary place.You pay to park elsewhere.
+
+This overused lot is an example of a tragedy of the commons. It's a tragedy because every parker is acting reasonably, but their individual actions quickly sum to collective disaster. Similarly, if a single shepherd has access to a field, the result is well-trimmed grass and fat sheep. But open the field to all shepherds, each of whom may add sheep without regard to the others, and soon there may be nothing left but bare dirt and hungry animals.
+
+8 J.P. Chamberlain, The Regime of the International Rivers: Danube and Rhine (New York: Columbia UP, 1923) 148.
+
+10
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+Overuse tragedies are everywhere: species extinction, ozone depletion, and highway congestion. After Garrett Hardin popularised the `tragedy of the commons' metaphor in 1968, people gained a new language for a phenomenon that was widely experienced, but had been difficult to name.The concept helped people give voice to then-emerging concerns about environmental degradation.
+
+Metaphors can be powerful. The tragedy of the commons concept revealed hidden links among innumerable resource dilemmas, large and small. Spotting this shared structure helped people identify shared solutions. For example, the International Association for the Study of the Commons brings together a global network of scholars, policymakers, and practitioners, while the Digital Library of the Commons hosts an online database that cites about fifty thousand articles related to the commons. 9
+
+How do we solve such tragedies? There are three distinct approaches: privatisation and markets, cooperative engagement, and political advocacy and regulation. Bear in mind that each solution has an analogue on the anticommons side of the property spectrum.
+
+Private property and market transactions can solve overuse tragedy. Recall that in the parking example, you were the first to discover the empty lot.You might claim ownership for yourself based on your original discovery and first possession. Being first is a standard (but not necessarily fair or efficient) way to hand out rights in resources. Another path to private ownership passes through state control. The state might reject your claim of original discovery and instead appropriate the lot and auction it to the highest bidder or transfer it quietly to a crony. However the lot arrives in private hands, it will likely be managed better than if it had remained open to all. Owners can profit if they spruce up the lot, repave it perhaps, paint lines, and keep it clean, safe, and well used.As a parker in a private property regime, you lose the freedom of the commons but gain order and access.
+
+The moral justifications for private ownership are controversial for philosophers, but as a practical matter, moving to private property often does prevent overuse in a commons. Harold Demsetz, author of the leading economic theory of ownership, argues that this ‘conservation effect’ is the main reason private property emerges in, and provides a benefit to, society. $^10$
+
+Because of our private-property focus, we tend to overlook cooperative solutions to overuse dilemmas. Cooperative solutions are often small-scale, context specific, local, and not reliant on legal structures – thus invisible. In the case of our magical parking lot, notes under windshields, gossip on the street, and other neighborly devices can coordinate the parkers. Parkers may figure out how to keep the parking lot running smoothly without state coercion or private ownership. In Governing the Commons , Ostrom demonstrated that close-knit communities around the world have succeeded in managing group property
+
+9 On the IASC, see http://www.iascp.org (last visited 2 October 2012). On the Digital Library of the Commons, see n 6 above.
+
+10 See H. Demsetz, 'Toward a Theory of Property Rights' (1967) 57 American Economic Review, Papers and Proceedings 347, 354-359.
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+11
+
+---
+
+The Anticommons
+
+without tragic outcomes. $^11$ There are thousands of stories of successful cooperation that preserve contested resources and promote overall social welfare.
+
+Finally, state coercion can solve overuse. Cooperative mechanisms may break down if there are too many newcomers coming and going, if people don't really know each other, or if it is otherwise hard to discipline deviants. Then, parkers may move from polite notes under windshields to breaking antennae, purposely scratching cars, slashing tires, and fistfights.The state might assert ownership over the lot, put up a gate, and hand out or sell parking permits. But bureaucracy is costly and often capricious. Political pressure may lead to bizarre uses of the lot. States are rarely nimble or efficient parking-lot operators. Public ownership and management can eliminate the tragedy of the parking-lot commons, but they may create new costs and inconveniences for the parkers.
+
+Privatising a commons may cure the tragedy of wasteful overuse and lead to orderly parking; but it also may inadvertently spark the opposite, a lot that no one can use.The tragedy of the anticommons describes this problem of wasteful underuse. Though the anticommons concept refers at its core to fragmented ownership, the idea extends to fragmented decision-making more generally. Resource use often depends on the outcome of some regulatory process. If the regulatory drama involves too many uncoordinated actors – neighbors and advocacy groups; local, state and federal legislators; agencies and courts – the sheer multiplicity of players may block use of the underlying resource.
+
+How could the parking lot become an anticommons? Recall that underuse in an anticommons is the mirror image of overuse in a commons. Much can go wrong when politicians privatise state-owned resources, when resources are owned for the first time, or when owners divvy up property later on. For example, in privatising the lot, politicians might not want to annoy parkers who are also voters. So they might give free parking spots based on every parker's previous use of the lot. (This is approximately how US regulators have allocated ocean fishing quotas and tradable pollution permits.) If there are thousands of parkers, but say one hundred spots, dozens might have to share each spot. Assembling the fractional shares back into a usable parking lot would require too many deals. Even if each parker behaved reasonably, bargaining is costly. And many of us are not reasonable, especially at seven o'clock in Leicester Square when shows are about to start. So the ‘privatised’ lot may sit empty and unused – an anticommons.
+
+Now substitute sheep in a meadow for the parkers in the lot. If a common field were privatised down to the square inch, no shepherd would be able to graze a single sheep. The same might happen if innumerable heirs separately owned scattered strips of an ancestor's farm. In an anticommons, the grass may be lush and tall and unused; in a commons, it may be picked bare. In both cases, the pasturage can be wasted and the sheep starve.
+
+11 See generally E. Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge: Cambridge UP, 1990) 1–28 (setting out the theoretical framework); E. Ostrom, ‘Coping with Tragedies of the Commons’ (1999) 2 Annual Review of Political Science 493 (discussing solutions).
+
+12
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+The parking lot and shepherd's field show that creating private property can solve the problem of overuse in a commons. But privatisation can go too far. When it does, we can tip into an anticommons, and again everyone loses.Adding the concepts of underuse and anticommons makes visible a new frontier for private bargaining, political debate, and wealth creation. Our goal should be to find the sweet spot for property rights, between commons and anticommons.
+
+## THE NEW WORLD OF USE
+
+My tales of the magical parking lot are a bit of a sleight of hand. They give a succinct overview of overuse and underuse, commons and anticommons. But underuse and anticommons are still squiggly – until recently, my Word spellchecker rejected them by underlining each with red squiggles, and instead suggested undersea and anticommunist as replacements. These squiggles are a signal: the nonexistence of a word can be as telling as its presence. When we lack a term to describe some social condition, it is because the condition does not exist in most people's minds. So, it should be no surprise that we have overlooked the hidden costs of anticommons ownership. We cannot easily fix the problem until we have created a shared lexicon to spot tragedies of the anticommons.
+
+Besides highlighting the language problem, the squiggles prompted me to look around the Internet at overuse and underuse. Googling overuse yielded about ten million hits in late 2012, while underuse generated under one million. ( Commons had 800 million hits and anticommons had 25,000.) To me, the data immediately suggest two possibilities: either overuse is about ten times more important a social problem than underuse, or we are only about ten percent as aware of underuse as we should be.You will not be surprised that I believe the latter to be correct.
+
+To understand the Google results, start with overuse.According to the Oxford English Dictionary (OED), overuse entered the language as a verb in the early 1600s. One of the earliest usages is as apt today as it was centuries ago: ‘When ever we overuse any lower good we abuse it.’ By 1862, the noun form was well recognised: ‘The oyster beds are becoming impoverished, partly by over-use.’ 12
+
+Overuse continues to mean `to use too much' and `to injure by excessive force,' definitions that have been stable for hundreds of years. Many of Google's top links for overuse come from medicine. Doctors diagnose `overuse syndrome' and dozens of `overuse injuries' – injuries from too much tennis, running, violin playing, book reading, whatever. So what is the opposite of overuse?
+
+Ordinary use. The opposite of injuring yourself through too much use and excessive force is staying injury free by using an ordinary amount of force. Instead of abandoning an activity, do it in a reasonable, sustainable way. In medicine as in everyday language, the opposite of overuse is ordinary use (figure 3 ). Since the 1600s, overuse and ordinary use have been an either-or proposition. Either you will feel pain in your elbow, or you will be able to play happily, if not well. When you overuse a resource, bad things happen. It is much better to engage in ordinary use.
+
+12 OED,'overuse, v.' at http://www.oed.com/view/Entry/135291 (last visited 2 October 2012).
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+13
+
+---
+
+The Anticommons
+
+![Figure](figures/figure_001.png)
+
+Figure 3: Ordinary use as the end point
+
+How do we achieve ordinary use? Recall the problem of the magical parking lot. The usual solutions to tragedies of the commons are, as I've mentioned, privatisation, cooperation, and regulation. These three solutions map onto the traditional view that ownership can be organised into three basic types of property: private, commons, and state (figure 4 ). $^13$
+
+![Figure](figures/figure_004.png)
+
+Figure 4: The trilogy of ownership
+
+We all have strong intuitions about private property, but the term is surprisingly hard to pin down. A good starting point is William Blackstone, the foundational eighteenth-century British legal theorist. His oft-quoted definition of private property is `that sole and despotic dominion which one man claims and exercises over the external things of the world, in total exclusion of the right of any other individual in the universe.' $^14$ In this view, private property is about an individual decision maker who directs resource use.
+
+Commons property refers to shared resources, resources for which there is no single decision maker. In turn, the commons can be divided into two distinct categories.The first is open access , a regime in which no one at all can be excluded, like anarchy in the parking lot or on the high seas. Mistakenly, the legal and economics literatures have long conflated the commons with open access, hence reinforcing the link between commons and tragedy. The second type of commons has many names, but for now let's call it group access , a regime in which a limited number of commoners can exclude outsiders but not each other. If the ocean is open access, then a small pond surrounded by a handful of landowners may be group access (or consider the shared mews behind houses in Notting Hill, London's keyholder-only squares, and New York City's Gramercy Park). Group
+
+13 On the property trilogy, see M. A. Heller, 'The Dynamic Analytics of Property Law' (2001) 2 Theoretical Inquiries in Law 79, 82-92.
+
+14 W. Blackstone, Commentaries on the Laws of England: In Four Books, bk 2, *2.
+
+14
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+access is often overlooked even though it is the predominant form of commons ownership and often not tragic at all. 15
+
+State property resembles private property in that there is a single decision maker but differs in that resource use is directed through some process that is, in principle, responsive to the needs of the public as a whole. In recent years, state property has become less central as a theoretical category: the cold war is over, most socialist states have disappeared, intense state regulation of resources has dropped from favor, and privatisation has accelerated. Today, for many observers, the property trilogy can be reduced to an opposition of private and commons property, what one scholar calls simply `all and none' (figure 5 ). $^16$
+
+![Figure](figures/figure_003.png)
+
+Figure 5: The familiar split in ownership
+
+I believe a substantial cause of our cultural blindness to the costs of fragmented ownership arises from the dominance of this too simple image of property. Note how the commons–private opposition tracks the overuse– ordinary use opposition.The former implies that there is nothing beyond private property; the latter suggests that we cannot overshoot ordinary use. Together, these oppositions reinforce the political and economic logic of the global push toward privatisation.We assume, without reflection, that the solution to overuse in a commons is ordinary use in private ownership.This logic makes it difficult to imagine underuse dilemmas and impossible to see the uncharted world beyond private property.
+
+According to the OED, undense is a recent coinage. In its first recorded appearance, in 1960, the word was hedged about with an anxious hyphen and scare quotes: `There might, in some places, be a considerable “under-use” of [parking] meters.' By 1970, copy editors felt sufficiently comfortable to cast aside the quotes: `A country can never recover by persistently under-using its resources, as Britain has done for too long.' The hyphen began to disappear around 1975. $^17$
+
+In the OED, this new word means `to use something below the optimum' and `insufficient use.' The reference to an `optimum' suggests to me how underuse entered English. It was, I think, an unintended consequence of the increasing role of cost-benefit analysis in public policy debates. What happens when we slot
+
+15 On open access versus group access property, see T. Eggertsson, 'Open Access versus Common Property' in T. Anderson and F. McChesney (eds), Property Rights: Cooperation, Conflict, and Law (NewYork: Princeton UP, 2003) 74–85. I advocate that we use the term liberal commons to describe many forms of legally sanctioned group ownership. See generally H. Dagan and M.A. Heller, 'The Liberal Commons' (2001) 110 Yale Law Journal 549.
+
+16 Y. Barzel, Economic Analysis of Property Rights (Cambridge: Cambridge UP, 1989) 71.
+
+17 OED,'under-use, n.' at http://www.oed.com/view/Entry/212195 (last visited 2 October 2012).
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+15
+
+---
+
+The Anticommons
+
+underuse into the opposition in figure 3? Although the result seems simple, it leads to conceptual turmoil (figure 6).
+
+![Figure](figures/figure_002.png)
+
+Figure 6: The new spectrum of use
+
+In the old world of overuse versus ordinary use, our choices were binary and clear-cut: injury or health, waste or efficiency, bad or good. In the new world, we are looking for something more subtle – an ‘optimum’ along a continuum. Looking for an optimum level of use has a surprising twist: it requires a concept of underuse and surreptitiously changes the long-standing meaning of overuse. Like Goldilocks, we are looking for something not too hot, not too cold, not too much or too little – just right. Figure 7 suggests how underuse changes our quest.
+
+![Figure](figures/figure_005.png)
+
+Figure 7: Goldilocks' quest for the optimum
+
+How can we know whether we are underusing, overusing, or optimally using resources? It's not easy, and not just a matter for economic analysis. Searching for an optimum between overuse and underuse sets us on the contested path of modern regulation of risk, an inquiry that starts with economic analysis but quickly implicates our core beliefs. We have to put dollar values on human lives and on the costs of overuse and underuse behaviors – a process filled with moral and political dilemmas. I note this difficult topic to show that finding the optimum requires the idea of underuse and that this new word in turn transforms the meaning of overuse. Overuse no longer just means using a resource more than an ordinary amount. The possibility of underuse reorients policy making from relatively simple either-or choices to the more contentious trade-offs that make up modern regulation of risk.
+
+## THE TRAGEDY OF THE ANTICOMMONS
+
+Adding the idea of `underuse' sets the stage for the anticommons. Looking back at figures 3–7 , you can see there is a gap in our labeling scheme.We have seen the complete spectrum of use , but not the analogous spectrum of ownership . What form of ownership typically coincides with squiggly underuse? The force of symmetry helped reveal a hidden property form. Figure 8 shows my path to the anticommons.
+
+16
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+![Figure](figures/figure_001.png)
+
+Figure 8: An ownership puzzle
+
+I coined the term tragedy of the anticommons to help make visible the dilemma of too fragmented ownership beyond private property. Just as the idea of `underuse’ transforms the continuum of resource use, `anticommons' transforms the continuum of ownership. It shows that the move from commons to private can overshoot the mark (figure 9 ). When privatisation goes too far, resources can end up wasted in an unfamiliar way.
+
+![Figure](figures/figure_004.png)
+
+Figure 9: The full spectrum of ownership
+
+Seeing the full spectrum of ownership has another benefit. Our understanding of commons ownership may help inform solutions to anticommons tragedy. To start, consider the distinction between open access (anarchy open to all) and group access (property that is commons to insiders and private to outsiders).This distinction can do some work on the anticommons side of the spectrum as well. The conventional wisdom has often overlooked group access, but we don’t have to. Under the right conditions, groups of people succeed at conserving a commons resource without regulation or privatisation. $^18$ Cooperation can get us to optimal use. Under what conditions does cooperation work, and what does that teach us about fixing underuse dilemmas?
+
+At the extreme of open access, group norms don't stick. For example, anyone can fish for tuna on the high seas.Tuna fleets work in relative isolation, and their catches can be sold anonymously to diverse buyers. Conservation norms, such as voluntary limits on fishing seasons, may gain little traction. Gossip and other low-cost forms of policing don't work for wide-ranging international fleets. Unless states intervene, overuse is hard to avoid. Whales were saved from extinction more through naval powers enforcing international treaties than through gossip at the harbor bar.
+
+The state can sponsor hybrid solutions. What if the state asserted ownership over lobsters and fish, and then created private rights (such as licenses and tradable quotas) that complement cooperative solutions? Often, such hybrid regimes lead to fairer and higher-yielding results than informal group access can achieve. For
+
+18 See, eg, R. C. Ellickson, Order without Law: How Neighbors Settle Disputes (Cambridge: Harvard UP, 1991) 167-183.
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+17
+
+---
+
+The Anticommons
+
+example, in Australia, the government issues licenses for a sustainable number of lobster traps and enforces strict harvesting limits. Lobstermen can wait to harvest until the lobsters mature, or they can sell their government-created rights, secure in their markets and property. With far less fishing effort, this system yields more and bigger lobsters than US lobstermen catch either in coastal harbor gangs or on the open ocean. 19
+
+Hybrid systems are the cutting edge of natural resource management: examples include not only tradable fishing quotas, but also carbon-emission markets and transferable air-pollution permits. 20 These solutions can work far beyond lobsters and tuna, even beyond natural resources generally.They may reach the edge of high-tech innovation.
+
+Solutions to commons property dilemmas give clues to solving anticommons tragedy. For open access, like the high seas, states must command resource use directly or create hybrid rights, such as fishing quotas.The anticommons parallel to open access is full exclusion in which an unlimited number of people may block each other. With full exclusion, states must expropriate fragmented rights or create hybrid property regimes so people can bundle their ownership. Otherwise, the resource will be wasted through underuse.There is, however, one important respect in which full exclusion differs from open access: an anticommons is often invisible.You have to spot the underused resource before you can respond to the dilemma.
+
+Group access in a commons also has an anticommons parallel: group exclusion in which a limited number of owners can block each other. Recall the multiple owners of our magical parking lot. For both group access and group exclusion, the full array of market-based, cooperative, and regulatory solutions is available. Although self-regulation may be more complex for anticommons resources, 21 close-knit fragment owners can sometimes organise to overcome anticommons tragedy. For group exclusion resources, the regulatory focus should be support for markets to assemble ownership and removal of roadblocks to cooperation.
+
+Group property on the commons or anticommons side of private ownership is exponentially more important than the rare extremes of open access or full exclusion. Much of the modern economy – corporations, partnerships, trusts, condominiums, even marriages – can be understood as legally structured group property forms for resolving access and exclusion dilemmas. 22 We live or die depending on how we manage group ownership. Now, we can see the full spectrum of property, as shown in figure 10 .
+
+19 See generally J. Tierney, 'A Tale of Two Fisheries' New York Times 27 August 2000; J. Acheson, Lobster Gangs of Maine (Hanover, NH: UPNE, 1988).
+
+20 See C. M. Rose, 'Expanding the Choices for the Global Commons: Comparing Newfangled Tradable Allowance Schemes to Old-Fashioned Common Property Regimes' (1999) 10 Duke Environmental Law and Policy Forum 45.
+
+21 See B. Depoorter and S. Vanneste, 'Putting Humpty Dumpty Back Together: Pricing in Anticommons Property Arrangements' (2007) 3 Journal of Law, Economics, and Policy 1.
+
+22 See Dagan and Heller, n 15 above, 552-54; see also H. Dagan and C. J. Frantz, 'Properties of Marriage' (2004) 104 Columbia Law Review 75.
+
+18
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+![Figure](figures/figure_001.png)
+
+Figure 10: The full spectrum of property, revealed $^{24}$
+
+## THE ECONOMICS OF THE ANTICOMMONS
+
+After I proposed the possibility of anticommons tragedy, economist and Nobel laureate James Buchanan and his colleague Yong Yoon undertook to create a formal economic model. They wrote that the anticommons concept helps explain ‘how and why potential economic value may disappear into the “black hole” of resource underutilisation.’23 According to their model, society gets the highest total value from a resource – say, the magical parking lot – when a single decision maker controls its use. As more people can use the lot independently, the value goes down – a tragedy of the commons. And as more people can block each other from the lot, the value also goes down symmetrically – a tragedy of the anticommons. Figure 11 shows their graphic summarising this finding.
+
+![Figure](figures/figure_005.png)
+
+Figure 11: Value symmetry in an anticommons and a commons $^{25}$
+
+After developing their proof and showing how the anticommons construct may apply to a wide range of problems, Buchanan and Yoon concluded that `the anticommons construction offers an analytical means of isolating a central feature of sometimes disparate institutional structures . . . [People] have perhaps
+
+23 J. Buchanan and Y. J. Yoon, 'Symmetric Tragedies: Commons and Anticommons' (2000) 43 Journal of Law and Economics 1, 2.
+
+24 I develop an early version of this spectrum in M. A. Heller, 'The Boundaries of Private Property' (1999) 108 Yale Law Journal 1163, 1194-98.
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+19
+
+---
+
+The Anticommons
+
+concentrated too much attention on the commons side of the ledger to the relative neglect of the anticommons side."
+
+In recent years, economic modeling of the anticommons, including game theory approaches, has become quite sophisticated. 27 At the simplest level, anticommons theory can be understood as a legal twist on the economics of `complements' first described by Antoine-Augustin Cournot in his 1838 Researches into the Mathematical Principles of the Theory of Wealth 28 (and discovered independently by Charles Ellet in 1839 in his work on railway tariffs 29 ). Anticommons theory is a partial corrective for modern economic models that focus on `substitutes' and often neglect the role of `complements'. 30 What's the difference?
+
+In figure 12, Railways A, B, and C are substitute ways to get from here to there. Say the fare is 9. If railway A finds a way to provide service for 8, it will win riders. B and C must become more efficient to keep up. In markets with robust substitutes, competitors have incentives to innovate, lower prices, and thereby indirectly benefit society as a whole. By contrast, Railways D, E, and F are complements. When inputs are complementary, generally you want all or none of them.
+
+![Figure](figures/figure_004.png)
+
+Figure 12: Substitutes versus complements
+
+Again, assume the fare from here to there is 9. D, E, and F each charge 3. Railway D knows that if you want to ride, you must buy its ticket. So why innovate? Instead, D may raise its fare to 5, hoping that E and F lower theirs to 2 each. But why would E and F go along? More likely, they too would raise fares,
+
+25 ibid 8.
+
+26 ibid 12.
+
+27 See F. Parisi, 'Entropy in Property' (2002) 50 American Journal of Comparative Law 595; F. Parisi, N. Schulz and B. Depoorter, 'Fragmentation in Property:Towards a General Model' (2003) 159 Journal of Institutional and Theoretical Economics 594; see also Fennell, n 7 above (arguing that tragedies of the commons are best modeled as prisoner's dilemma games and anticommons as chicken games).
+
+28 A. Cournot, Researches into the Mathematical Principles of the Theory of Wealth (1838) (N.T. Bacon trans, 1897) 99-104 (sections 55-57, introducing theory of complements).
+
+29 C. Ellet, Jr, An Essay on the Laws of Trade in Reference to the Works of Internal Improvement in the United States (1839).
+
+30 On the problem of complements in an information economy, see H. R. Varian, J.V. Farrell, and C. Shapiro, The Economics of Information Technology: An Introduction (Cambridge: Cambridge UP, 2005) 43–45. On the interaction of substitutes and complements in the anticommons context, see G. Dari-Mattiacci and E Parisi, ‘Substituting Complements’ (2006) 2 Journal of Comparative Law and Economics 333.
+
+20
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+so the total exceeds 9, and ridership falls below the optimal level. With complementary competition, incentives to innovate are blunted: if D did lower fares, then E and F just might raise theirs.
+
+It's the same problem if D, E, and F are complementary patents instead of railways. Then innovators face what economist Carl Shapiro calls a `patent thicket', a lot of phantom tollbooths on the route to commercialising new technology. 31 Cournot proved that in markets dominated by complements – whether railways or patents – we can get higher overall social welfare if D, E, and F merge. Here, monopoly trumps competition. Anticommons theory, in turn, moves from railways and patents to ownership and regulation generally.All these concepts describe facets of the same dilemma: too many uncoordinated owners or regulators blocking optimal use of a single resource.
+
+## ANTICOMMONS PUZZLES
+
+Our rhine boaters from a little earlier may seem an esoteric example, but there are a near-infinity of everyday puzzles that share this common structure — one whose solution could jump-start innovation, release trillions in productivity, and help revive the global economy.
+
+Let me return to the drug patent example that opens this article. The Alzheimer’s drug that never came to market is not alone. Numerous potential drugs are lost to anticommons ownership. In the past 30 years, drug R&D has been going steadily up, but discoveries of major new classes of drugs have been declining. Instead, drug companies focus on minor spinoffs of existing drugs for which they have already assembled the necessary property rights. How did this new drug discovery gap happen? Patent anticommons. Paradoxically, more biotech patent owners can mean fewer life-saving innovations. Drugs that should exist, could exist, are not being created.
+
+To date, the most debated application of anticommons theory has built on the above example. The field was sparked in part by my article with Rebecca Eisenberg on the anticommons in biomedical research. 32 Since then, there has been a flurry of follow-on papers, studies, and reports, many of which conclude that patents should be harder to obtain, in part to avoid potential anticommons tragedy effects. 33 In their 2009 book, The Patent Crisis , Dan Burk and Mark Lemley review the most recent literature on patents and biotech innovation and conclude that, `the structure of the biotechnology industry seems likely to run
+
+31 C. Shapiro,‘Navigating the Patent Thicket: Cross Licenses, Patent Pools, and Standard Setting’ in A. B. Jaffe et al (eds), Innovation Policy and the Economy (Cambridge, MA: MIT Press, 2001) vol 1, 119; cf D. L. Burk and M. A. Lemley, The Patent Crisis and How the Courts Can Solve It (Chicago: University of Chicago Press, 2009) (distinguishing anticommons and patent thickets in the patent context) 75–78, 86–92.
+
+32 M.A. Heller and R. S. Eisenberg, 'Can Patents Deter Innovation? The Anticommons in Biomedical Research' (1998) 280 Science 698; see generally Heller, n 1 above 49-78 (Chapter 3, "Where are the Cures?" reviewing ten years of scholarship and debate on anticommons effects in drug development).
+
+33 ibid 65 (discussing three influential policy-oriented reports that argue for patent reform based on anticommons concerns).
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+21
+
+---
+
+The Anticommons
+
+high anticommons risks', particularly when companies are attempting to bring products to market. 34 However, the empirical basis for finding that anticommons effects are stifling innovation remains inconclusive – it remains an important area for empirical work. 35
+
+In addition, the anticommons framework has been used to analyse ownership across the high tech frontier, ranging from Thomas Hazlett's work on broadcast spectrum ownership, what he calls the `tragedy of the telecommons', to Rosemarie Ham Ziedonis' study of anticommons effects in technology patenting. 36 In my Gridlock Economy book, I show that it is not just biotech that is susceptible to anticommons tragedy. Cutting edge art and music are about mashing up and remixing many separately owned bits of culture. 37 And as I earlier said, even with land the most socially important projects often require assembling multiple parcels. 38
+
+Anticommons theory is now well established, but empirical studies have yet to catch up. How hard is it to negotiate around ownership fragmentation? How much does ownership fragmentation slow down technological innovation? Does the effect vary by industry? It is difficult to measure discoveries that should have been made but weren't, industries that could exist but don't. We are just starting to examine these conundrums.
+
+In 2006, a team of law, economics, and psychology researchers reported experimental findings that rejected the presumed symmetry of commons and anticommons. They found that anticommons dilemmas ‘seem to elicit more individualistic behavior than commons dilemmas’ and are ‘more prone to underuse than commons dilemmas are to overuse’. The researchers conclude that ‘if commons leads to “ tragedy ” , anticommons may well lead to “ disaster ” . $^39$
+
+These preliminary findings of bargaining failure around anticommons ownership may help provide some insight into otherwise puzzling economic phenomena. For example, some of the world's biggest energy companies have for years failed to agree on joint management of oil and gas fields they own together.
+
+34 Burk and Lemley, n 31 above 89.
+
+35 Several survey-based studies have questioned whether anticommons tragedy is blocking academic biomedical research. See eg, J. P. Walsh, A. Arora, and W. M. Cohen, ‘Effects of Research Tool Patents and Licensing on Biomedical Innovation’ in W. M. Cohen and S. A. Merrill (eds), Patents in the Knowledge-Based Economy (Washington DC: National Academies Press, 2004) 285–340, 324.
+
+36 On the anticommons in the telecom context, see Heller, n 1 above, 79–106 (Chapter 4, ‘You Can’t Hear Me Now’). See also T.W. Hazlett, ‘Spectrum Tragedies’ (2005) 22 Yale Journal of Regulation 242; R. H. Ziedonis, ‘Don’t Fence Me In: Fragmented Markets for Technology and the Patent Acquisition Strategies of Firms’ (2004) 50 Management Science 804.
+
+37 On the anticommons in the copyright context, see Heller, ibid 9–16 (discussing tragedy in filmmaking, art, history, and music), 27–30 (discussing the anticommons in online search, such as Google Books). See also F. Parisi and B. Depoorter, ‘Fair Use and Copyright Protection: A Price Theory Explanation’ (2003) 21 International Review of Law and Economics 453.
+
+38 On the anticommons in land, see Heller, n 1 above, 107–142 (detailing anticommons tragedies and solutions in land resources).
+
+39 S. Vanneste et al, ‘From “Tragedy” to “Disaster”: Welfare Effects of Commons and Anticommons Dilemmas’ (2006) 26 International Review of Law and Economics 104. Follow-up studies looked at why negotiations fail when presented in anticommons form.They found more failure as the number and complementarity of fragment owners increase. Also, as uncertainty increases, losses become even more pronounced. Depoorter and Vanneste, n 21 above, 21–23.
+
+22
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+If one company pumps the oil too fast, it can wreck the pressure in the gas field; if the other extracts gas too fast, it traps the oil. American law has offered them an effective legal tool, called `unitisation', to overcome anticommons tragedy and smooth joint management of divided oil and gas interests. Yet firms block each other year after year. 40
+
+How can this be? Oil units aren't a case of two spiteful neighbors arguing over a broken backyard fence. They involve arm's-length business negotiations between savvy corporations. Everyone has good information about the underlying geologic and technical issues. The gains from cooperation are in the billions of dollars. Why doesn't one firm sell its interest to the other? Why don't the firms merge? What's going on? The experimental studies are beginning to give us explanations rooted in the psychology of the anticommons. Even the most sophisticated businesspeople can fail to reach agreement when a negotiation is framed in anticommons terms.
+
+## ROUNDING OUT THE LEXICON - CAVEATS AND CAUTIONS
+
+In rounding out the anticommons lexicon, there are some caveats: first, my term focuses on one form of underuse, the tragedy that arises when ownership is too fragmented. Here, multiple owners block each other from using a scarce resource. Underuse can also arise in the monopoly context, when a single owner blocks access to a resource. This situation may be tragic, but it is not an anticommons tragedy in my sense of the term.
+
+In the old economy, many companies held monopolies – Ma Bell (the American telephone monopoly), railways, local water utilities, and others. Society gained the economic benefits of scale and scope from allowing these sectors to be monopolised. The state policed against abuse of monopoly power through complex rate regulation and oversight. Phone lines were cheaper and more available than in many other countries.The costs of these monopolies were often invisible, costs such as deferred and dampened innovation.
+
+In an information economy, any piece of intangible property, such as a patent, is also a monopoly.We award patents because monopoly profits create incentives to invent and because patents give inventors incentives to disclose their discoveries (without patents people might prefer to invent things they could keep secret). On the other hand, drugs would be cheaper and lives could be saved if competitors could make generic copies at will.To balance the values of innovation, disclosure, and competition, the US Congress keeps shifting the bundle of rights that a patent confers.
+
+The dilemmas of any individual monopoly in the old or new economy are a great topic, but are not relevant here. For better or worse, these quandaries are familiar.We do not, however, have much experience dealing with the interaction
+
+40 On the costs of 'excessive anarchy' in the oil industry, see G. D. Libecap and J. L. Smith, 'The Economic Evolution of Petroleum Property Rights in the United States' (2002) 31 Journal of Legal Studies S589.The same tragedy affects excessive pumping of groundwater. See B. H. Thompson Jr, 'Tragically Difficult:The Obstacles to Governing the Commons' (2000) 30 Environmental Lawyer 241, 250.
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+23
+
+---
+
+The Anticommons
+
+of ownership fragments or an array of blocking patents. The anticommons lexicon addresses not monopoly per se, but ownership multiplicity.
+
+Next, here's a caution: when talking about an anticommons, stay away from absolutes. First, you shouldn't assume that anticommons ownership is inevitably tragic. 41 If we lived in a world where people had perfect information and could bargain with each other at no cost, they could avoid anticommons tragedy every time (just as, in a perfect world, there would be no commons tragedy, or for that matter, tragedy of any sort). 42 In practice, however, bargaining is never free, people shirk duties and hold out for payoffs, and there are cognitive limits that shape owners' decisions. In the real world, anticommons ownership is not necessarily tragic, but it does tend that way.
+
+Second, it's theoretically possible that an anticommons may face overuse instead of underuse. 43 For example, consider real estate development along the California coast. It's a mess. Multiple community groups, environmentalists, neighbors, and government agencies may each prefer different versions of a project. Even in that regulatory morass, though, overbuilding may occur if it is sufficiently costly to exercise each right to veto development. Every opponent of development may prefer to go surfing and hope the others sit through the boring public hearings. If enough people opt for a free ride, a project might face too little opposition, not too much. It's an empirical question whether the California coast tips toward over- or under-building. That said, in most cases I've seen, anticommons regulation tends to be associated with too little economic development, not too much. 44
+
+Finally, there is a caveat that comes from the legal theorist Carol Rose. Certain resources, such as roads and waterways, are sometimes owned most efficiently in common. As Rose points out, creating and enforcing private-property rights is itself costly; sometimes these costs exceed the gains, not just economically but also socially. Village greens and town halls may strengthen communities in ways that are socially valuable but hard to quantify in monetary terms. Rose coined the term `the comedy of the commons' to describe these social and economic benefits that can flow from group access. 45
+
+Rose's insight is equally true on the anticommons side – there are both economic and social reasons that we may prefer group exclusion to sole ownership. For example, it's possible that creating multiple vetoes may help preserve a treasured resource against transient political pressures for development – for instance, Central Park in New York City or Indian burial grounds in Arizona. 46
+
+41 These points are developed in Heller, n 6 above at 676.
+
+42 See generally R. H. Coase, 'The Problem of Social Cost' (1960) 1 Journal of Law and Economics 1.
+
+43 Fennell develops this insight in her theory of common interest tragedies, see n 7 above, 934-937.
+
+44 Just as an anticommons theoretically may lead to overuse, it is possible for a commons to be associated with underuse, ibid. For a real-world example, see W. W. Buzbee, 'The Regulatory Fragmentation Continuum, Westway, and the Challenges of Regional Growth' (2005) 21 Journal of Law and Politics 323.
+
+45 C. M. Rose, 'The Comedy of the Commons: Custom, Commerce, and Inherently Public Property' (1986) 53 University of Chicago Law Review 711; see also R. C. Ellickson, 'Property in Land' (1993) 102 Yale Law Journal 1315, 1336-1338.
+
+46 On the potential use of an anticommons to preserve Central Park, see A. Bell and G. Parchonovsky, 'Of Property and Antiproperty' (2003) 102 Michigan Law Review 1, 3-4, 31-36, 60-61.
+
+24
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR 6-25
+
+---
+
+Michael Heller
+
+Similarly, `conservation easements' intentionally use anticommons ownership to foster environmental goals. 47 (With a conservation easement, the owner sells or gives away the right to develop land, gets a tax break, and retains the right to continue a current use such as farming.) The underuse created by split ownership may be justifiable if the environmental gains exceed fragmentation costs. On balance, though, I'm sceptical. What happens a generation from now when communities want to reduce sprawl but face a patchwork of easements that make `in-fill' development prohibitively difficult? Many conservation easements look to me like potential anticommons tragedies. 48
+
+The `comedy of the anticommons' insight suggests that sometimes, for some resources, we should promote little or no use. Most of the time, for most resources, however, some positive level of use will be socially most valuable. Underuse is rarely the optimum.
+
+## TOWARD A NON-SQUIGGLY LANGUAGE
+
+We have millennia of practice in spotting tragedies of the commons. When too many people fish, fisheries are depleted. When too many people pollute, we choke on dirty air. Then, we spring into action with market-based, cooperative, and legislative solutions. Similarly, we have a lot of experience spotting underuse caused by a particular monopoly owner. We have created regulatory bodies that know (more or less) what to do with such dilemmas.
+
+But underuse caused by multiple owners is unfamiliar. The affected resource is hard to spot. Our language is new. A tragedy of the anticommons may be as costly to society as the more familiar forms of resource misuse, but we have never noticed, named, debated, or learned how to fix underuse. How do we stumble into the problem of too many owners? How do we get out? As a first step, underuse in a tragedy of the anticommons should be squiggly no more.
+
+Fixing anticommons tragedies is a key challenge for our time. Individual entrepreneurship and legal reform will be important. But I want to stress that the first and most important step to solving an anticommons is to name it and make it visible.With the right language, we can all spot links among ownership puzzles, and we can all come together to fix them. Nothing is inevitable about an anticommons. Every ownership puzzle results from choices we make, and can change, about how to control the resources we value most.
+
+47 See J. D. Mahoney, 'Perpetual Restrictions on Land and the Problem of the Future' (2002) 88 Virginia Law Review 739, 785.
+
+48 ibid 785-786 (noting potential anticommons tragedy in conservation easements).
+
+© 2013 The Author. The Modern Law Review © 2013 The Modern Law Review Limited. (2013) 76(1) MLR. 6-25
+
+25
+

processed/00079_W2039252480/document.md

@@ -0,0 +1,563 @@
+International Studies Quarterly (2007) 51, 559-581
+
+# The Political Economy of AIDS Treatment: Intellectual Property and the Transformation of Generic Supply
+
+Kenneth C. Shadden London School of Economics
+
+This article examines the relationship between intellectual property (IP) and public health, with a focus on the extension of AIDS treatment in the developing world. While most of the literature on IP and health examines the conditions affecting poor countries' capacities to acquire essential medicines, I show the distinct—and more complicated—political economy of production and supply. IP regulations alter the structure of generic pharmaceutical sectors in the countries capable of supplying essential medicines, and changes in market structure affect actors' economic and political interests and capacities. These new constellations of interests and capacities have profound implications for the creation and maintenance of political coalitions in support of on-going drug supply. The result is that the global AIDS treatment campaign becomes marked by mismatches of interests and capacities: those actors capable of taking the economic, legal, and political steps necessary to increase the supply and availability of essential drugs have diminished interest in doing so, and those actors with an interest in expanding treatment may lack the capacities to address the problem of undersupply. By focusing centrally on actors' interests in and capacities for economic and political action, the article restores political economy to analysis of an issue-area that has been dominated by attention to international law. And by examining the fragility of the coalitions supporting the production and supply of generic drugs, the article points to the limits of transnational activist networks as enduring agents of change.
+
+Since the 1990s, the relationship between changing international regulations on intellectual property (IP) and access to medicines in the developing world has been the subject of an ever-growing body of literature. The World Trade Organization’s (WTO) Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) requires that countries offer patents on pharmaceutical products. Traditionally many developing (and developed) countries did not allow drugs to
+
+Author's note: I presented earlier versions of this paper at the First Global International Studies Conference at Bilgi University (Istanbul, Turkey) in August 2005, and the conference on “ Frontiers of Regulation: Assessing Scholarly Debates and Policy Challenges ” at the University of Bath (UK), in September 2006. I wish to thank the following people for sharing their insights and providing helpful comments and suggestions on earlier drafts: Brook Baker, Tony Barnett, Shannad Basheer, Jim Bjorkman, Catherine Boone, Angela Burger, Nick Corby, Graham Dutfield, Tim Dyson, Kevin Gallagher, Christopher Garrison, Padmashree Gehl Sampath, John Gerring, Cheri Grace, Homi Katrak, Marcus Kurtz, Evan Lieberman, Susan Martin, Duncan Matthews, Christopher May, Andrew Schrank, Hakan Seckinelgin, and Susan Sell. I am also grateful to ISQs editor and the three anonymous reviewers for their encouraging and constructive feedback.
+
+c 2007 International Studies Association. Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford OX4 2DQ, UK
+
+---
+
+560
+
+The Political Economy of AIDS Treatment
+
+be patented, but as of 2005 all but the poorest “ Least Developed Countries ” must do so. This monumental change in the global governance of IP has prompted a wave of research into the question of how the transition to near-universal pharmaceutical patenting will affect drug prices and, subsequently, developing countries' access to essential medicines. $^1$ Because patent protection can raise the price of drugs, $^2$ concern has almost universally been directed toward understanding how to make IP regulations more flexible (so to increase access to more affordable generic drugs and constrain the prices of patented drugs) and how to mobilize funding to increase poor people's and poor countries' abilities to purchase essential drugs.
+
+In this article, I build on the contemporary scholarship by problematizing what most analyses of access take as a constant—the existence and availability of generic versions of essential drugs. If generic drugs exist, then analysts, activists, and policymakers can focus on the steps needed to increase consumers' ability to acquire them, such as how to overcome legal, IP-related barriers and how to increase purchasing power via the development of funding mechanisms. But production and supply precede access, and the constellations of political actors and strategies that facilitate the latter may be less effective with regard to the former.
+
+To highlight the distinct — and more complicated — political economy of production and supply, I focus on drugs for treatment of HIV/AIDS. TRIPS — and in particular the aspects of TRIPS related to pharmaceutical patents — sets new incentives for drug producers. Because these new incentives inspire new patterns of investment and specialization, they can have significant effects on the structure of generic pharmaceutical sectors in the countries capable of supplying essential medicines. The ensuing changes in market structure affect actors' economic and political interests and capacities, and new sets of interests and capacities have profound implications for the creation and maintenance of political coalitions in support of on-going drug supply. The result, then, is that the global AIDS treatment campaign becomes marked by mismatches of interests and capacities: those actors capable of taking the steps necessary to increase the supply and availability of high-quality, affordable drugs have diminished interest in doing so, and those actors with an interest in expanding treatment may lack the capacities to address the problem of undersupply.
+
+The present analysis has profound — and worrying — implications for public health and the global campaign to extend AIDS treatment. Since the early 2000s an increasing number of people living with HIV/AIDS in the developing world have begun to receive treatment. Although the World Health Organization (WHO) fell short of its goal of three million people receiving treatment by the end of 2005, the trajectory is clearly one of increased treatment, and most international health experts expect further scaling up in the years to come. $^3$ Indeed, the emerging consensus is that expanding treatment is not just essential but
+
+1 These questions have been addressed by social scientists from the fields of politics, economics, and law, along with health professionals and activists. A sampling of the literature includes, among others, Abbott (2005), Attaran (2004), Attaran and Gillespie-White (2001), Baker (2005), Chaudhuri (2005), CIPR (2002), Correa (2000a), Fink (2001), Granville (2002), Heins (2008, forthcoming), Katrak (2004a), Lanjouw (1998), Matthews (2004, 2006), May (2002), Moatti, Coriat, Souteyrand, Barnett, Dumoulin, and Flori (2003) MSF-WHO-UNAIDS (2003), Nogués (1993), Roffe, Tansey, and Vivas-Eugui (2006), Sell (2002, 2006), Shadlen (2004a,b), Weinberg (2005), and WHO (1998, 2002, 2005). See also the abundant studies (and links to further research) at the portals of Médecins Sans Frontières, available at http://www.accessmed-msf.org (Accessed April 7, 2007), and the Consumer Project on Technology, available at http://www.cptech.org/ip/health/ (Accessed April 7, 2007). WHO (2001) provides an extensive (though dated) annotated bibliography of the literature.
+
+2 Patents confer owners with exclusive rights over the use of their knowledge. When a firm has a patent on a given drug, no other actor may sell the drug without the authorization of the patent-holder. In contrast, when a drug is unpatented, the absence of this legal barrier to entry means that consumers can typically select from a range of suppliers. All else being equal, then, patents introduce higher prices.
+
+$^{3}$ In 2005 the G8 set as a goal universal access to AIDS treatment by 2010.
+
+---
+
+KENNETH C. SHADLEN
+
+561
+
+feasible, and that with the mobilization of sufficient resources an ever-increasing number of people can benefit from AIDS treatment (WHO/UNAIDS 2006). But the analysis in this article suggests otherwise: what has worked so far in expanding treatment may not continue to work as the challenge becomes one of supply.
+
+In addition to the obviously important implications for global public health, the analysis also advances our thinking on international political economy in three significant and closely related ways. First, by focusing centrally on actors' interests in and capacities for economic and political action, this article should reorient scholars' attention away from legal provisions per se and toward politics. Much of the debate about IP and access to medicines has emphasized the legal dimensions; even the political scientists who have addressed the topic have focused on the making and changing of laws (Sell 2002, 2003; Shadlen 2004a) . But legal mechanisms only define the realm of the possible; they establish the parameters of what activities are acceptable. Legal mechanisms must be activated, and any efforts to do so put private and public officials in the crosswinds of competing interests. Understanding the interests and capacities of actors to take advantage of legally sanctioned opportunities and, critically, to secure complementary actions on the part of public officials, brings us to the realm of politics. Beyond the questions of what scenarios and activities are possible, political analysis demands that attention be paid to changing constellations of interests in and capacities for pursuing different possible scenarios and activities. That is, in addition to considering what actors can do according to prevailing legal arrangements, we also need to analyze what they do do in response to the incentives embedded in laws and in accordance with their interests and capacities.
+
+Second, by focusing on the fragility of the political arrangements that support and facilitate generic drug supply, the article points to the potential limits of transnational activist movements as agents of change. The global treatment access movement has played a vital role in expanding treatment throughout the developing world since the late 1990s (‘t Hoen 2002; Sell 2002, 2006; d'Adesky 2004; Smith and Siplon 2006) . Without the intense mobilization and political pressure applied by the activist movement to relax legal barriers to acquiring generic drugs and to increase global funding for treatment and care (i.e., to facilitate access ), it is unlikely that we would have witnessed such remarkable increases in AIDS treatment since 2000. But as the challenges of access are supplemented by the challenges of supply, the political strategies that have worked in the past may be less effective. Indeed, one of the staples of the literature on transnational activism (and contentious politics more generally) is that the development and effectiveness of such movements is conditioned by the “ opportunity structure ” (Tarrow 1994; Keck and Sikkink 1998; Khagram, Riker, and Sikkink 2002; Schurman 2004) . As I show, the opportunity structure is changing: broad changes in international political economy shift the political terrain on which activists must operate and fundamentally alter the nature of available alliance partners.
+
+Third, the article shows how the temporal dimensions of social phenomena pose challenges for political action. Political scientists increasingly value the analytic importance of time, and in particular the notion that the effect of given events can vary widely depending on when the events occur (Pierson 2004) . Treating AIDS, however, introduces a different perspective on time, for it is a temporally unbounded process. Because AIDS cannot be cured, only treated, people who receive treatment now will need to receive essential drugs for the rest of their lives — and because they will develop resistance they will eventually need new medicines. Politically, this means that the institutional arrangements and social coalitions that facilitate treatment (including both supply of and access to essential medicines) need to be continuously reproduced. AIDS treatment is certainly not unique in its temporal dimensions, and analysts should be
+
+---
+
+562
+
+The Political Economy of AIDS Treatment
+
+alert to the distinct issues involved in policy areas marked by a need for continuous political reproduction.
+
+The remainder of the article consists of five sections. In section one I provide background on the HIV/AIDS epidemic and treatment, and I explain the importance of price stability for scaling up AIDS treatment. In doing so I draw attention to the key role that generic drugs play in global treatment campaigns. In section two I review, briefly, how developing countries' strategies for obtaining reliable supplies of affordable drugs are affected by the new international political economy of IP. I explain how new international regulations can pose barriers to access, but how many of those potential barriers have been relaxed. In section three I shift attention from drug access to drug production and supply (specifically undersupply ). To do so I disaggregate the generic pharmaceutical sector, distinguishing three basic segments according to price:cost ratios. I explain how new IP regulations transform market structure by turning the incentives dramatically against investing in production of generic versions of new drugs for AIDS treatment, and I consider the effects that the emerging market structure has on political coalitions for expanding treatment to meet the goals established by the international community. In section four I examine these dynamics—the effects of IP on the generic pharmaceutical sector and the political consequences of the emerging market structure—in the country that is the most important supplier of generic AIDS drugs, India. In the conclusion, I summarize the main findings and assess a range of mechanisms and strategies for restoring stability to the supply of essential medicines for AIDS treatment.
+
+## HIV/AIDS and ARV Treatment
+
+HIV causes AIDS, and AIDS is fatal. By the end of 2005, approximately 39 million people worldwide were living with HIV, and each year nearly three million people die of AIDS (UNAIDS 2006: 7–50). Though HIV/AIDS is universal, 4 the epidemic has had the greatest impact in the developing world. As Table 1 indicates, roughly ninety-five percent of the adults and children living with HIV live outside of North America, Western, and Central Europe. Sub-Saharan Africa alone accounts for three-fifths of the total.
+
+Antiretroviral drugs (ARVs) do not cure HIV/AIDS, but they allow people inflicted with the illness 5 to live normal lives. Although the introduction of combination therapy (where multiple drugs are taken together to slow reproduction of the virus) in the late 1990s converted AIDS into a chronic illness in much of the OECD, it remained a death sentence in most of the developing world, where the high cost of drugs made treatment generally unaffordable for all but the wealthiest. Since the early 2000s, however, an increasing number of people living with HIV/AIDS in the developing world have begun to receive ARV treatment. Whereas fewer than five percent of the people with HIV/AIDS in the developing world who needed treatment were estimated to be receiving ARV therapy in 2001, the share increased to roughly twenty percent by the end of 2005 (WHO/UNAIDS 2006). Although the World Health Organization fell short of its goal of three million people receiving treatment by the end of 2005, the trajectory is clearly one of increased treatment. From 2003 alone, when the 3 x 5 plan was launched, the number increased threefold, and in 2005 the G8 set as a goal universal access to AIDS treatment by 2010 (WHO/UNAIDS 2006). Not everyone with HIV needs ARV treatment.
+
+4 “HIV/AIDS has been reported from every inhabited continent and from every country” (Barnett and Whiteside 2006: 8).
+
+$^{5}$ I use the term "illness" to refer both to people who are HIV-infected and people suffering from AIDS.
+
+---
+
+KENNETH C. SHADLEN
+
+563
+
+TABLE 1. HIV/AIDS Prevalence and Deaths, by Region (2005)
+
+<table><tr><td>Region</td><td>Adults and Children Living with HIV</td><td>Prevalence</td><td>Adult and Child Deaths due to AIDS</td><td>Adults and Children Living with HIV as Share of Global Total</td></tr><tr><td>Sub-Saharan Africa</td><td>24,500,000</td><td>6.1\North Africa and Middle East</td><td>440,000</td><td>0.2\Asia</td><td>8,300,000</td><td>0.4\Oceania</td><td>78,000</td><td>0.3\Latin America</td><td>1,600,000</td><td>0.5\Caribbean</td><td>330,000</td><td>1.6\East Europe and Central Asia</td><td>1,500,000</td><td>0.8\North America, West</td><td>2,000,000</td><td>0.5\and Central Europe</td><td></td></tr></table>
+
+
+Source: WHO/UNAIDS (2006: 13, Figure 2.3). WHO/UNAIDS presents the figures as ranges and as estimates within ranges. This table is constructed using the estimates, which generally fall in the middle of each range.
+
+"Universal" treatment by 2010 means roughly 10 million people, a six-seven times increase over current levels.
+
+The challenges to extending AIDS treatment in the developing world are daunting (WHO 2003; Tayler 2004; World Bank 2004) . Drugs are expensive, as is diagnostic equipment. Trained healthcare professionals are needed for diagnosing patients, delivering ARVs, monitoring patients, responding to medical emergencies and dealing with the emergence of opportunistic infections. Some second-line treatments have refrigeration requirements, which add further complexities to supply management and to the already-immense challenges of procurement and patient care. The list goes on; no one would deny the extent of the challenges. Yet we know that ARV treatment is possible, even in resourcepoor settings in the developing world. $^6$
+
+The cornerstone to an effective treatment program is that affordable medicines be available. Drugs are the key input, not because they are necessarily the most expensive component of treatment, but because they are irreplaceable. Functional—if not optimal—substitutes can be found to address inadequacies in other components of treatment: different sorts of infrastructure can be made suitable, alternative healthcare providers can be deployed, and so on. But no amount of managerial and political creativity can create functional substitutes for ARVs: if drugs are not available, treatment is impossible, full stop.
+
+Drug prices, of course, are not the only relevant issue, as the previous discussion indicates. But low (high) prices can encourage (discourage) resource mobilization necessary to address some of the complementary challenges. For public health ministries operating with scarce resources, the high price of drugs can serve as a disincentive to invest in the development of the healthcare infrastructure that is essential for treatment. When drugs are affordable, in contrast, improving healthcare infrastructure may appear as a more worthwhile task. Thus, lower drug prices can create incentives (and free resources) to build necessary infrastructure (Schwartländer, Stover, Walker, Bollinger, Gutierrez, McGreevey, Opuni, Forsythe, Kumaranayake, Watts, and Bertozzi 2001; Berwick 2002: 214;
+
+6 The WHO strategy for scaling up ARV treatment in the developing world is based on a protocol-driven public health approach, rather than individual treatment regimens. The idea is that a protocol-driven approach, based on a formulary of a small number of drugs, is simpler to implement (administratively and medically) and thus more appropriate for resource-poor settings. See WHO (2003). Mukherjee, Farmer, Niyizonkiza, McCorkle, Vandenwarker, Teixeira, and Kim (2003) and Sever, Leger, Charles, Noel, Bonhomme, Bois, George, Kenel-Pierre, Wright, Gulick, Johnson, William Pape, and Fitzgerald (2005) discuss the challenges to treatment in resource-poor settings.
+
+---
+
+564
+
+The Political Economy of AIDS Treatment
+
+TABLE 2. Providers of ARVs to Global Fund, by Type of Manufacturer (2004)
+
+<table><tr><td></td><td>Units</td><td>Expenditures</td></tr><tr><td>Brand-name</td><td>40.7%</td><td>53.0%</td></tr><tr><td>Generic</td><td>59.3%</td><td>47.0%</td></tr></table>
+
+
+Source: GFATM (2005).
+
+Nattrass and Geffen 2005). $^7$ The relationship between ARV prices and AIDS treatment is supported by recent experience. Since 2000 the price of key drugs declined to less than one dollar per day per patient. These price reductions, combined with significant increases in funding from governments, have provided the foundations for the expansion of ARV treatment.
+
+A critical factor in lowering drug prices has been the existence of a vibrant market for high-quality generic ARVs. The key actors in this regard are pharmaceutical firms from India: it is estimated that more than half of those receiving treatment in the developing world are treated with generic ARVs produced in India. $^8$ Table 2 provides data on ARV purchases supported by the Global Fund Against AIDS, TB, and Malaria (GFATM) in 2004. Both the importance of generics and the existence of price differentials are obvious: although nearly sixty percent of the drugs purchased were generics, these purchases accounted for just less than half of the expenditures.
+
+The data in Table 2 greatly understate the effect of generic medicines on prices. Not only are generic drugs themselves ordinarily less expensive than their brand-name equivalents, but competition introduced by generics yields price reductions across the board (Lucchini, Cisse, Duran, de Cenival, Comiti, Gaudry, and Moatti 2003; Kovsted 2005; MSF 2005; Wainberg 2005) . Thus, the “ brandname ” prices reflected in Table 2 are already significantly reduced from the prices observed in the absence of generic competition.
+
+It is important to consider the complex relationship between generic and brand-name sources of supply. After all, the introduction of generic competition has coincided with the proliferation and extension of discount, tiered-pricing (and also voluntary licensing) programs on the part of brand-name producers, and by December 2005 more than 716,000 people in low- and middle-income countries were receiving treatment with brand-name drugs (WHO/UNAIDS 2006: 21). $^9$ A critical point to underscore, however, is that non-binding, concessionary schemes such as these, in addition to often having onerous conditions attached and lacking transparency, lack stability in the absence of generic competition (Lucchini et al. 2003; Shadlen 2004b) . Discount prices and licensing agreements are restricted to specific drugs, for example, with terms, quantities, and other details largely set by the supplier and subject to change according to suppliers' policy. Or take the case of voluntary licensing, where terms are formally established in negotiations between licensor and licensee: the relative bargaining power of the two parties depends on whether or not the licensor is threatened by generic competition. The point here is not that brand-name pharmaceutical firms' tiered pricing and voluntary license programs cannot exist independently of generics. After all, such schemes, provided they allow the firms to retain control over distribution, present firms with minimal opportunity costs, since they sell few drugs in poor countries without making significant price
+
+$^{7}$ Shadlen (2004b) discusses this in terms of a "price-infrastructure trap."
+
+$^{8}$ This estimate was provided to me by an official from the WHO. Precise figures of this sort do not exist.
+
+9 Note that someone using brand-name ARVs may also be using ARVs provided by generic suppliers, which is one of the reasons why precise tallying is so difficult. In any case, increased treatment has been possible because of increased availability of bath generic and brand-name medicines.
+
+---
+
+KENNETH C. SHADLEN
+
+565
+
+reductions. Furthermore, these schemes can bring much-desired positive reporting and public relations and, more strategically, serve as loss-leaders, help increase sales of other products, and keep competitors out of markets. The point, rather, is that these schemes are insufficient, on their own, because they tend to lack stability and transparency. There are sound reasons to question whether the benevolence of brand-name firms is a sufficiently reliable and stable mechanism for securing long-term supplies of ARVs (Shadlen 2004b; MSF 2005). Because of the competition it introduces, the existence of generic alternatives makes the global AIDS treatment campaign less dependent on and vulnerable to changes in brand-name firms' marketing and pricing decisions. $^10$
+
+In sum, significant extension of AIDS treatment in the developing world has been made possible by the widespread availability of affordable and high-quality ARVs. Extending treatment (e.g., to approach the ambitious goals set out by the G8, and beyond) or even just sustaining treatment at current levels depend on high-quality ARVs remaining available and affordable. Yet here is the catch: because treatment cannot be terminated, the need for drugs to be available is never-ending; and because patients develop immunity and treatment regimens need adjustment, affordable and high-quality versions of new ARVs must be available as well. The drugs that work today will be ineffective tomorrow, meaning that the political, economic, and legal conditions that facilitate the availability of today's drugs must be continuously reproduced. The remainder of the article examines the challenges to the durability of generic competition, and thus the stability of supply of affordable ARVs.
+
+## TRIPS and Generic Supply
+
+The TRIPS agreement sets new and universal standards to which IP regimes in all countries that are members of the WTO must conform. $^11$ For the purposes of this article, the most important aspects of TRIPS are the requirement that countries grant patents in pharmaceutical products and the greater degree of exclusivity countries must give to patent holders. As indicated, prior to the introduction of TRIPS many countries did not issue patents on pharmaceutical products, and had wide latitude on how strongly they protected the rights of patent holders. Where drugs are not patented or where the exclusive rights of patent-owners are weak, patients and healthcare providers typically face fewer obstacles obtaining generic versions of new drugs. Thus, the new obligations established by TRIPS led to extensive concern about the availability of affordable medicines in the developing world (see note 1).
+
+It is important to note that many ARVS are not patented in many developing countries. The reasons for this are multiple. First, the poorest ( “ least developed countries ” ) are not required to issue pharmaceutical patents until 2016. Second, those countries that did not grant patents to pharmaceuticals prior to 1995 did not have to begin doing so until 2005. $^12$ Third, even where countries issue pharmaceutical patents, drugs that were already on the market prior to a country changing its patent laws typically cannot be patented (i.e., only new drugs are eligible). Fourth, notwithstanding the formal availability of patents, originator firms
+
+10 As the prominent international IP lawyer Fred Abbott notes, “developing countries on the whole have shared interests in assuring that there will be alternative production of medicines not under the control of patent holders and that they will have access to newer products, wherever produced” (Abbott 2005: 323–324). See also Katrak (2004a).
+
+$^{11}$ This is necessarily a very brief and schematic overview. For more detailed analysis, see, for example, Correa (2000b), Watal (2000), and Matthews (2004).
+
+$^{12}$ Note that many developing countries with important generic pharmaceutical sectors (e.g., Argentina, Brazil, Mexico, and Thailand) did not take full advantage of this transition period and began offering pharmaceutical patents prior to 2005.
+
+---
+
+566
+
+The Political Economy of AIDS Treatment
+
+often choose not to patent each of their drugs in all countries, especially smaller markets. $^13$ Thus, many drugs that are patented in the OECD are still not patented in many developing countries.
+
+The limited extent of pharmaceutical patents in the developing world has led some observers to conclude that patents are unimportant and that the concerns expressed by many are unwarranted (e.g., Attaran and Gillespie-White 2001; Attaran 2004) . But such analyses, based on unweighted tallying of patents, are misleading. After all, few developing countries have the economic and technological capacity to produce their own ARVs, regardless of the patent situation. Most countries import their ARVs; and where ARVs are produced locally, the active principal ingredients (APIs) are generally imported. From a treatment perspective, for example, the fact that patenting is low in Malawi and Zambia is irrelevant, since neither country has the capacity to produce ARVs. Instead, they might depend on South African production, where the situation is unquestionably complicated by patents. $^14$
+
+Rather than examining patents across the board, our attention must be directed toward the more advanced developing (and also developed) countries with sophisticated pharmaceutical sectors, that is, the potential exporters. All larger countries—developed and developing—with the capacity to produce and export ARVs are WTO members (or wish to join, e.g., Iran, Russia) and are therefore bound by TRIPS to make pharmaceutical patents available. What this means, quite simply, is that eventually all new drugs are likely to be patented in all countries with the capacity to produce them . Hence, the key question: how will ARV treatment be affected by the on-going transition toward a world where all new drugs can—and almost certainly will—be patented in all countries with production capacity?
+
+For developing countries with more advanced pharmaceutical sectors, the challenges to securing stable supplies of affordable ARVs are manageable. Provided that the country in question has a patent regime with adequate public health provisions, countries can make drugs affordable by authorizing domestic generic manufacture, importing generics from abroad, and negotiating price reductions with brand-name producers. The key tool in a country's patent regime is the compulsory license, whereby the government allows a local entity (a private firm and/or government agency) to produce and distribute a good under patent, without the consent of the patentee. $^15$ Countries meeting their TRIPS obligations can design compulsory licensing systems to encourage price competition (Correa 2000a; CIPR 2002). For example, a country's patent regime might permit the government to classify high prices or limited supply as violations of public interest or forms of patent abuse, and subsequently threaten to issue a license to an alternative firm if the patent holder fails to lower the price or increase supply. $^16$ Of course, the threat to issue a compulsory license may be sufficient to encourage firms with patents to lower the prices at which they provide the goods over which they have control. But whether or not countries can make credible threats depends in part on their patent regimes — the ability to lower prices in these ways depends on the grounds under which a government can issue a
+
+15 Because patents are national, firms obtain and defend their patents in each country. Where markets are small, firms may decide that the costs of obtaining and maintaining a patent outweigh the benefits.
+
+14 This point was made by a letter to the editor of the Journal of American Medical Association in response to Attaran and Gillespie-White (2001). See Boelaert, Lynen, Van Damme, Colebunders, Goemaere, Kaninda, Ciafli, Mulemba, 't Hoen, Pécoul, Sans Frontières, Selgelid, Schuklenk, Attaran, and Gillespie-White (2002) and also Grace (2005: 18–19).
+
+15 Compulsory licenses can be issued for importation as well, not just production. The point is that the government compels the patent owner to license the exclusive rights of production, importation, and distribution. Compulsory licenses are often referred to as “non-voluntary licenses.” See Correa (2000a,b), Reichman and Hasenzahl (2003), and Tayler (2004).
+
+16 In 2001, following concern of anthrax attacks, the United States used a threat of compulsory license to induce Bayer to reduce the price of the drug Ciprofloxacine (“Cipro”) to the Department of Health and Human Services. The Canadian government acted similarly at this time.
+
+---
+
+KENNETH C. SHADLEN
+
+567
+
+compulsory license, which ministries have the power to do so, whether the ruling could be appealed, and so on.
+
+Brazil offers the most prominent case of a developing country obtaining a stable and affordable supply of ARVs by threatening compulsory licenses (Galvão 2002; Granville 2002; Orsi, Hasenclever, Fialho, Tigre, and Coriat 2003; Cohen and Lybecker 2005) . Brazil did not issue patents on pharmaceuticals prior to April 1997, and because patents are not offered retroactively to existing drugs that were already on the market prior to 1995, local suppliers in Brazil are free to produce generic versions of many ARVs. The problem the Brazilian government has faced is that newer, patented ARVs included in its national HIV/AIDS treatment program — the ones that came on the market after 1995 — consume a disproportionate amount of state resources allocated to purchasing medicines. In response, the health ministry has used the facilities under its TRIPS-compliant IP law to negotiate price reductions. For example, in August 2001 Brazil announced its intention to issue a compulsory license on an ARV to which the patent was held by the Swiss firm Roche. Roche responded to the threat by reducing the price of the drug in Brazil, and subsequently no license was issued. A similar episode occurred with Roche, Abbott, and Merck in 2003, and then, again with Abbott in 2005. What makes these threats effective negotiating tools is that they are credible, and one reason why the threats are credible is because the Brazilian government exploited the flexibilities under the TRIPS agreement and integrated public health provisions into the national patent regime. $^17$
+
+It is important to emphasize that Brazil's measures to reduce the price of patented ARVs are entirely legal: TRIPS permits countries to design and use their patent regimes as the Brazilians have. $^18$ However, the uncertainty regarding countries' rights and the external pressures to exceed TRIPS obligations, a scenario faced by many developing countries, have inhibited other countries from emulating the Brazilian model. Thus, the Doha Declaration on the TRIPS Agreement and Public Health, adopted in November 2001, which confirms the rights set out in TRIPS, marks an important advance for this reduced sub-set of middle-income developing countries with well-developed pharmaceutical sectors. Although the Declaration does not reform TRIPS or release countries from their new obligations regarding the establishment and protection of intellectual property, it provides the political space for more countries to design and use their patent systems to secure stable access to affordable ARVs (Shadlen 2004a).
+
+But what about poorer, less developed countries that lack the ability to produce ARVs locally? For these countries, the ability to emulate the Brazilian strategy for lowering the price of drugs depends on the ability to import generic versions. Yet doing so is difficult, on account of the restrictions that TRIPS places on compulsory licenses in exporting countries. Quite simply, importing a good presupposes someone else exporting that good, and TRIPS requires that goods produced in one country under a compulsory license be “predominantly” for domestic use (Article 31.f). To the extent that new ARVs become patented in export-capable countries, this requirement could hamper provision of generic versions of such drugs to developing countries. To illustrate this, return to the example of Brazil's threats to issue compulsory licenses. The threats are credible because of Brazil's patent regime, as discussed, and also because Brazil has the capacity to produce the drugs locally. Patent-holding pharmaceutical firms know that Brazil has not just
+
+17 As the article was going to press, Brazil issued a compulsory license on Efivarenz, an ARV whose patent was owned by Merck.
+
+18 In 2001, the US filed a complaint in the WTO that Brazil's IP regime violated TRIPS, but the complaint was directed toward another aspect of Brazilian law.
+
+---
+
+568
+
+The Political Economy of AIDS Treatment
+
+the legal but also the industrial capacity to introduce generic competition, and they respond by lowering their own prices. But Brazil, though not unique, is special. 19 Most developing countries lack the capacity to produce drugs locally, which makes threats to issue compulsory licenses for local production empty threats—but so too are threats to issue compulsory licenses for import, if potential exporters in more industrialized developing and developed countries are hamstrung by TRIPS.
+
+Much of the contemporary politics of IP in the WTO has revolved around addressing this issue of how the supply of generic ARVs may be affected by the restrictions that TRIPS places on producers in export-capable countries. In October 2001, on the eve of the Doha Ministerial meeting, the developing countries proposed a reading of TRIPS that would have simplified the export of generic drugs to poor countries. $^20$ The proposal was based on an interpretation of Article 30 of the TRIPS Agreement, which addresses the conditions under which actors can use patented knowledge without obtaining any permissions from the state or the owner of the exclusive rights. The developing countries sought to make foreign public health emergencies one such condition. According to their proposal, a drug firm in one country would be able to supply a generic drug to a country that was experiencing a public health crisis but unable to produce the drug locally. The exporting firm would be able to do so even if the drug were under patent in the exporting country. For example, pharmaceutical firms in Canada or India (post-2005) would be allowed to produce generic versions of patented drugs to export to Malawi or Nicaragua, even if these firms were not the patent holders in Canada or India — and they would be able to do so without needing any steps to be taken by the Canadian or Indian governments.
+
+Under strong pressure from the transnational pharmaceutical industry, developed country governments, led by the US, EU, and Switzerland, rejected this proposal (Matthews 2004; Abbott 2005) . Allowing automatic exceptions, they maintained, would undermine one of the basic objectives of TRIPS, which was to universalize patent protection in pharmaceutical products. Instead, most developed countries preferred a solution based on Article 31 of TRIPS, which addresses compulsory licensing. The developed countries' proposals would condition production and trade in generic medicines on compulsory licenses being issued in all jurisdictions where required. Again, compulsory licenses would be necessary only in countries where the drug in question is under patent. If a drug were not patented in the developing country that lacked pharmaceutical manufacturing capacity and sought to import, no such license would be required. But a license would be required in the exporting country, where, at some point after 2005, the drug is almost certainly to be patented.
+
+The issue was unresolved at the WTO's 2001 Ministerial meeting. Paragraph six of the Doha Declaration simply recognized the special problems that TRIPS poses for developing countries that lack local manufacturing capacity and called on the TRIPS Council to address the problem. In August 2003, however, after nearly 2 years of debate and on the eve of the WTO's Fifth Ministerial Meeting in Cancún, Mexico, a temporary resolution was finally agreed (WTO 2003a). The settlement, which partially waives Art. 31.f, included increased regulations for issuing compulsory licenses for export to poor
+
+19 Even for Brazil this is a problem on account of diminished domestic manufacturing capacity in APIs and the inability of local producers to supply enough drugs quickly enough to lower prices (Orsi et al. 2003: 132; Wall Street Journal 2003). As a result Brazil threatens to issue compulsory licenses not just for local production but also for importation from China and India. In short, even larger and more industrialized developing countries rely on the existence of a vibrant international market for generic ARVs.
+
+20 See paragraph 9 of developing country proposal to TRIPS Council, WTO, IP/C/W/312, WT/GC/W/450, October 4, 2001.
+
+---
+
+KENNETH C. SHADLEN
+
+569
+
+countries (and extensive safeguards against inexpensive generic drugs being redirected back into wealthier markets). The August 2003 agreement, along with a supplementary statement from the Chair of the WTO's General Council (WTO 2003b), also included a list of developing and developed countries that pledged not to use the system as importers. After 2 years of non-action, both the agreement and the supplementary statement were formalized as a permanent amendment to TRIPS in December 2005.
+
+Since 2003, a number of export-capable countries have revised their patent laws to incorporate the waiver of Article 31.f (Matthews 2006). Canada was the first to do so, followed by Norway. Reforms are on course in China, the European Union, South Korea, and a number of countries with more advanced pharmaceutical sectors. In India, after significant activism by civic organizations and opposition parties, the final version of the amended Patent Act, passed in March 2005, also permits compulsory licensing for export.
+
+The new international regulations and the subsequent changes to national IP laws create new political challenges. In Canada, for example, the requirement that drugs eligible for export under CL be included on an official list of authorized drugs means that some actor—in industry or civil society—must petition relevant public officials for inclusion. Similarly, under the Indian regulations, firms must receive approval by the Controller General of Patents to export drugs produced under CL. To understand how changes in the global IP system may fundamentally transform the supply of subsequent generations of generic ARVs, then, we need to move analytic attention beyond the actual legality of producing and exporting such drugs and examine the interests and capacities of economic and political actors that might seek to engage in such activities and to elicit necessary measures by public officials. The following sections undertake such analysis.
+
+## The Political Economy of Undersupply
+
+The global treatment regime is marked by mismatches between interests and capacities. On the one hand, firms capable of taking advantage of legal opportunities to produce and export generic ARVs may lose interest in doing so, and if they are less vested in such operations they may also be less prepared to devote resources to get public officials to invoke the legal provisions that facilitate generic production. On the other hand, firms interested in taking advantage of legal opportunities to produce and export generic ARVs may lack capacity to do so, and the coalition of actors seeking to expand generic ARV supply may lack the resources to secure necessary public action.
+
+To make sense of these mismatches and to understand the implications for ARV supply, it is useful to consider the generic pharmaceutical industry according to three segments: commodity generics, specialty generics, and hybrid generics. Commodity generics (CGs) are drugs whose patents have expired (or perhaps never existed). Aspirin is a version of a CG: there are no longer patents anywhere in the world on aspirin, so any pharmaceutical firm in any country can produce aspirin (distribution, however, depends on approval from national health authorities). Most of the thousands of generic firms in the world produce CGs, and low barriers to entry make this an intensely competitive segment of the market. $^21$
+
+Specialty generics (SGs) refer to new drug delivery systems, novel combinations of existing drugs, and non-infringing processes of patented drugs. Included as SGs are drugs whose patents in the US and EU have recently expired or are about to expire, the most lucrative end of the generic market. In the US, for
+
+$^{21}$ Commodity generics are also referred to as "worldwide generics" and "plain vanilla generics."
+
+---
+
+570
+
+The Political Economy of AIDS Treatment
+
+TABLE 3. Disaggregating the Generic Drug Sector
+
+<table><tr><td>Market Segment</td><td>Prices</td><td>Costs</td><td>Principal Export Markets</td></tr><tr><td>Commodity Generics (CGs)</td><td>Low</td><td>Low</td><td>Developing and Least Developed ("less regulated")</td></tr><tr><td>Specialty Generics (SGs)</td><td>High</td><td>High</td><td>Developed ("regulated")</td></tr><tr><td>Hybrid Generics (HGs)</td><td>Low</td><td>High</td><td>Developing and Least Developed</td></tr></table>
+
+
+example, the first generic producer to obtain authorization from the Food and Drug Administration (FDA) can receive six months of market exclusivity, in which its product competes only against the brand-name product. $^22$
+
+Hybrid generics (HGs) are drugs that are under patent in some countries but not in the country where being produced (or if produced locally done so under compulsory license). HGs generally cannot be exported to countries where the drugs are patented, but they can be sold in any country that either does not have pharmaceutical patents or where the originator firm did not obtain a patent or where the importing country has issued a compulsory license. The generic ARVs that are used in HIV/AIDS treatment throughout the developing world fall into the HG category. As patents expire on some of the older ARVs (e.g., those with priority dates from he mid-late 1980s), they will become commodities, but ARVs with later priority dates will remain HGs. Most importantly, all subsequent generations of ARVs will fall into the HG category as well.
+
+Table 3 presents a basic overview of the variable price:cost ratios facing producers in the three distinct segments of the industry. CGs have low prices, because minimal barriers to entry and intense competition drive prices down, and low costs, for production processes are well-known and there are no legal obstacles. Firms that specialize in CGs require high volumes to compensate for tight margins.
+
+SGs have relatively high prices, as the drugs are sold in more regulated markets, and high costs, for more research and innovation are required and obtaining regulatory approval is more demanding. When a drug is no longer patented—that is, when the knowledge is in the public domain—everyone with sufficient scientific and technical capacity can engage in production and sales, but for SGs considerable resources must be invested in being the first to reverse engineer the drug. Indeed, in the case of firms seeking to exploit the opportunities presented by US legislation, if the firm is not the first to receive FDA authorization the advantages of doing so dissipate. Moreover, administrative and legal factors create fairly high barriers to entry, as considerable resources must be invested in obtaining approval in more regulated markets, and becoming “first to file” also exposes the firm to being sued for patent infringement. Thus, while only an option for those firms with sufficient technical capacity and financial and legal resources, the SG segment is the most profitable—and desirable—segment of the industry. $^23$
+
+Now consider generic versions of new ARVs, the example of an HG of principal concern in this article. Here the price:cost ratios are least favorable, combining low prices with the high costs. Prices of generic ARVs must be low, for it is only legal to sell them in countries without pharmaceutical patents, and these
+
+22 Specialty generics are also referred to as “ value-added generics. ” Many analysts use the SG label to refer exclusively to novel delivery systems (e.g., developing versions of drugs that can be injected directly into the blood). My usage is broader, including all generics that are technologically, legally, and administratively more complex, for these are the entry barriers that make SG more upmarket.
+
+25 Indeed, the SG segment is at the top of the value chain in the generic pharmaceutical sector, a step below new product development.
+
+---
+
+KENNETH C. SHADLEN
+
+571
+
+are without exception poorer countries with low purchasing power. $^24$ But the costs of producing and marketing generic versions of new ARVs are quite high. Obviously, the costs of HGs (and also SGs) are less than the costs of developing new patentable products, but they are significantly more than those presented by commodities. The reason for this is that reverse-engineering of newer drugs is more difficult and expensive. Firms need to dedicate resources toward researching the compound and learning how to reverse engineer the drugs; $^25$ they need to invest in developing production capacity; they must meet the higher standards of good manufacturing practices that are conditions for participating in publicly funded access programs; they need to comply with packaging requirements and find reliable distribution networks (WHO 2004). And, of course, firms engaging in the production of HGs face transaction costs that derive from the legal environment, such as the necessity of requesting compulsory licenses from the government and defending themselves against complaints of patent infringement.
+
+These observations allow us to make a set of hypotheses regarding the political economy of ARV production and supply. Because larger firms that are capable of bearing the scientific, administrative, and legal costs of HGs are also capable of bearing the costs of SGs, they are less likely to be attracted to the low-price, high-cost HG segment of the market. At the same time, smaller firms that seek to fill the niche may not be repelled by the low prices, but they are unlikely to be able to bear the costs of the HG segment. From a treatment perspective, the problem is that the current suppliers of ARVs tend to be more sophisticated and capable firms (Gehl Sampath 2005; Grace 2005; MSF 2005; WHO 2006). Not only are very few firms involved in this segment of the market, those that are involved tend to be among the largest firms in any given country. Thus, to the extent that larger pharmaceutical firms' interests in participating in HG markets decline, the risk of ARV undersupply increases.
+
+The emerging industrial structure poses critical challenges for the global campaign to increase ARV treatment. The movement to secure public action for scaling up global treatment loses a potentially powerful actor: to the extent that larger firms lose interest in producing and exporting generic ARVs, they are also likely to lose interest in lobbying to secure necessary actions on the part of public officials. But at the same time, those firms with an interest in doing so, in addition to lacking scientific and legal capacity, are also likely to lack the political resources to secure public action. As a result, the principal advocates of such government action in export-capable countries are poor people in need of treatment in far-away lands, and more generally the transnational network of treatment access activists. Hence the mismatch: actors capable of securing public action for scaling up global treatment (i.e., the ability to demand compulsory licenses for export to poor people in poor countries) may have few incentives to act, while actors needing action (i.e., people with HIV/AIDS in developing countries) may lack the necessary political resources.
+
+This is not to suggest that poor people with HIV/AIDS who depend on the existence of high-quality, generic ARVs lack representation altogether, but they face significant hurdles in securing necessary public action. Most obviously, poor people with HIV/AIDS in sub-Saharan Africa are not powerful or important constituents in export-capable countries such as Canada and India. Instead, they are
+
+24 There are exceptions to this broad characterization of the HG market. As explained above, because patents are typically not available retroactively to products already on the market, some ARVs remain unpatented in some middle-income developing countries. Yet as the transition to full pharmaceutical patentability continues in the post2005 environment, the exceptions to the statement in the text will become fewer and fewer (i.e., the only countries where ARVs will be unpatented will be the poorest LDCs).
+
+25 It is important to emphasize that the costs of learning how to make generic versions of ARVs applies to future generations of drugs, not the current crop. The reverse engineering costs of today's drugs have already been borne.
+
+---
+
+572
+
+The Political Economy of AIDS Treatment
+
+represented indirectly, through the transnational network of treatment activists. And while this activist network has been enormously successful in expanding access to generic ARVs by pressing for relaxation of legal impediments and for increases in global funding (Sell 2002; `t Hoen 2002; d'Adesky 2004; Friedman and Mottiar 2005; Seckinelgin 2006; Smith and Siplon 2006) , the new challenges in the realm of production and supply may stretch the capacities of the global treatment movement.
+
+Indeed, it is important to emphasize how the emerging global political economy of ARV supply presents the transnational treatment activist network with new, formidable challenges. In addition to working to relax legal barriers and increase funding, treatment activists now also face the fight of getting foreign governments and, importantly, private firms in export-capable countries that already participate in intensely competitive market segments, to take politically risky and economically costly measures for poor people in other countries. Yet the capacity of activist movements to affect outcomes when private investors are involved depends on the structure of the relevant industries (Schurman 2004) . Producing generic versions of new ARVs for export to poor countries is an activity with exceptionally thin margins, a characteristic that becomes exacerbated as the costs — not just legal, but economic and political — to reverse engineering, producing, and exporting generic drugs increase. But at the same time that the new IP rules make this segment less attractive to larger, capable firms, the challenge confronting treatment activists is to get more firms involved and politically vested in this activity.
+
+The subsequent challenges appear even more daunting when we consider the temporal dimensions of ARV treatment. Treating AIDS is not like rescuing victims of natural disasters, such as earthquakes or floods. In the latter instances, at some point the interventions are complete. Yet AIDS treatment is never complete. Patients infected with HIV need treatment for the rest of their lives (pending discovery of a cure). But patients and populations develop immunity to particular ARVs, so regimens need to be altered and updated with new medicines. Just as yesterday's ARVs have been replaced by a newer generation, the ARVs that are most effective today will need to be replaced by a subsequent generation, and so on. Politically, what this means is that actors and coalitions face the never-ending tasks of continuously reproducing conditions that are supportive and facilitative of generic ARV production and supply. Whether or not political arrangements that depend so heavily on transnational activists have sufficient endurance and stability is an open question, but the transformation in industrial structure and the subsequent thinning-effect this has on the treatment coalition certainly increases the challenges to continuous political reproduction.
+
+## TRIPS, India, and the Sustainability of ARV Supply
+
+For the present analysis, the most important country is India, with its extensive pharmaceutical sector that has acted as a principal supplier of generic drugs to the developing world. 26 India delayed the availability of product patents on pharmaceuticals until 2005, taking full advantage of the transition period allowed under TRIPS. Indeed, prior to 2005 India was the last country with an advanced pharmaceutical sector not to offer drug patents; and many Indian firms produce drugs whose quality is recognized by the World Health Organization (WHO 2006). In sum, India's IP system contributed to the growth of a pharmaceutical sector whose active presence in the global ARV market greatly increased the feasibility of extending AIDS treatment in poor countries, directly, through the supply of affordable ARVs, and indirectly, by placing pressure on brand-name firms.
+
+$^{26}$ A good argument can be made that China is equally important (Grace 2004, 2005).
+
+---
+
+KENNETH C. SHADLEN
+
+573
+
+Thus, the fate of treatment in the developing world depends to a large extent on the effects that the introduction of product patents has on the Indian pharmaceutical sector. As indicated, India’s Patent Act (Article 92A) establishes conditions under which compulsory licensing for export is acceptable.
+
+To understand the new political economy of ARV supply in India, we need to make a few quick observations about the broader changes that full implementation of TRIPS is likely to engender. The introduction of patent protection will bring about a period of consolidation and intense change in the Indian pharmaceutical sector (Lanjouw 1998; Fink 2001; Katrak 2004b; Chaudhuri 2005; Gehl Sampath 2005) . Full implementation of TRIPS introduces new competitive pressures and compels firms to adopt new strategies to exploit new market opportunities. As the substantial skills that Indian firms have developed should allow them to find niche areas and take advantage of new opportunities, the introduction of product patents need not spell denationalization and a return to the pre-1970 industrial structure. Indeed, a number of India's largest firms have responded to the pending (and now actual) reintroduction of product patents by increasing their investments in research and development, and thereby seeking to engage in more innovative activities. Ranbaxy, for example, which is India's largest drug firm, has massively increased its R & D spending and begun a patenting strategy; other large firms have similar aspirations (Chaudhuri 2005; Gehl Sampath 2005) . By combining the skills developed under the previous IP regime with the country’s impressive scientific infrastructure and low costs, many firms in India aim to develop new drugs, independently or in collaboration with OECD-based pharmaceutical firms.
+
+Two points should be made with reference to Indian firms' efforts to move into new product development. First, even the largest firms are probably too small to make significant in-roads into one of the world's most capital-intensive and oligopolistic sectors. Indian firms do not invest nearly enough in R & D, notwithstanding their lower cost structures, and they lack the skills and resources that are necessary to complete the full course of drug development, from discovery of new chemical entities to formulation to clinical trials to marketing and distribution. 27 Instead, we should expect manufacturing and sales of generics to remain as a critical source of revenue for most local firms. The most likely scenario is for the leading Indian firms to increase their presence in global markets for generic medicines, including competition against large OECD-based generic firms for shares of more regulated markets (i.e., SGs).
+
+But second, and more importantly, for the purposes of this analysis the extent to which firms succeed in new product development and/or upgrade to SGs is of little consequence, for both scenarios have the same implications for ARV production and supply. As the most sophisticated local firms upgrade to SGs (and some focus on new product development), hybrid generics, with high costs and low prices, will constitute a smaller part of their operations. Even though the new Indian Patent Act permits firms to export drugs produced under compulsory license, the incentives these firms face may discourage them. The existence of legal provisions, for example, Article 92 (A), make this possible, but, again, the key issues are not simply what actors can do but what they have the interests and capacities to do. It may not be worthwhile for Indian firms to undertake the considerable scientific, legal, and political steps necessary to acquire the capacities and rights to produce and export drugs that they will then have to supply at marginal cost. They have few incentives to do so now, but they will likely have
+
+27 Despite recent increases, Indian firms spend considerably less on R & D than transnational brand-name firms, whether measured in terms of absolute expenditures or as percent of sales. According to Chaudhuri (2005: 167), the top twelve Indian firms' combined R & D expenditures amount to just nine percent of the amount spent by the tenth largest brand-name firm. See also Grace (2004, 2005) and Gehl Sampath (2005).
+
+---
+
+574
+
+The Political Economy of AIDS Treatment
+
+even fewer incentives in the future—both because the costs of doing so will be greater and they will be less vested in those sorts of operations.
+
+To be sure, engendering diversification away from commodity and hybrid generics and toward specialized generics (and, ideally, new product development) are among the central objectives of India's amended Patent Act. If successful, a scenario that features national firms gaining ever-larger shares of the lucrative SG market in the US and making in-roads into currently underprovided European and East Asian markets will be highly developmental: we will be witnessing indigenous firms using the skills developed under the previous IP regime to consolidate their positions in market segments with significantly higher barriers to entry and, subsequently, higher returns (Schrank 2005) .
+
+But such changes in the structure and interests of India's pharmaceutical industry may have worrying effects on the global supply of ARVs. At a time when the world may be looking to private firms to increase their capacity to produce generic ARVs, the firms capable of doing this are likely to have diminished interest in doing so. The reason is that the firms that export ARVs to poor countries, the larger ones, are precisely those firms that will come under the most intense pressure to upgrade into speciality generics in order to survive in the new competitive environment. Even those firms most intent on diversifying exports and targeting more regulated markets (i.e., upgrading to SGs) may continue to export CGs to developing countries, as this remains an easy and reliable source of revenue with low costs and little risk (Grace 2004). $^28$ But this logic is less convincing when applied to the HG segment: firms are unlikely to dedicate additional resources toward the production and exports of new ARVs, with low prices and significantly higher costs and risks. That is, as new generations of ARVs emerge, they will be patented in India — unlike the current generation of drugs. And although the Patent Act will allow Indian firms to obtain permission to reverse-engineer and produce generic versions for export under compulsory licenses, the costs that more diversified firms will face in doing so from scratch are likely to outweigh the benefits.
+
+The behavior of India's leading pharmaceutical firms provides evidence in support of these hypothesized changes. Along with dedicating more resources toward new product innovation, since the late 1990s India's leading firms have focused on developing new drug delivery systems and non-infringing processes for export to regulated markets, that is, specialized generics (Chaudhuri 2005; Gehl Sampath 2005) . $^29$ Much of their R & D effort is directed toward reverse engineering drugs nearing the end of their patent terms, with an eye toward receiving market exclusivity as “first to file” in the US (Bhandari 2005) . $^30$ FDA approval for export to the US market has been obtained for roughly 150 generic drugs, and more generally, exports to regulated markets have been the fastestgrowing component of Indian firms' portfolios (Chaudhuri 2005) . $^31$ A detailed survey of emerging business strategies in the Indian pharmaceutical sector (Gehl Sampath 2005) shows unambiguously that the highest priority for India's largest firms is to increase their share of generic exports to regulated markets. More than three-quarters of the firms indicated that they did not regard Article 92(A)
+
+28 Gehl Sampath's findings suggest otherwise, that although many firms have supplied sub-Saharan Africa and other poor countries, “there is sufficient evidence to presume that this will not continue post-2005” (2005: 59).
+
+29 Some firms (most notably Ranbaxy), have also adopted aggressive strategies of challenging pharmaceutical patents owned by brand-name firms.
+
+30 See also, the comments of India's Minister of Commerce and Industry (Express Pharma Pulse 2005) and (Singh and Chatterjee 2004).
+
+31 Indeed, the competitive pressures that Indian firms now face and the subsequent concern with upgrading export markets coincides with immense pressures in the OECD to reduce the costs of medicines, thus making India's high-quality yet inexpensive generic drug exports more welcome.
+
+---
+
+KENNETH C. SHADLEN
+
+575
+
+as a useful option. $^32$ And of the minority that regard Article 92(A) more positively, sixty percent were small firms that lack the technical capacity and political resources necessary to take advantage of the legal opportunities in any case.
+
+These firm-level strategies are not surprising. After all, India's generic exports mean much more to the world than they do to Indian pharmaceutical firms. As indicated, Indian generics account for roughly fifty percent of ARVs used in developing countries, but one estimate puts that figure at roughly four to ten times the importance of these exports to the firms themselves, in terms of their total sales. 33 To put it simply, Indian firms do lots of things besides produce generic ARVs for the developing world, and the asymmetry in this relationship is bound to intensify over time as the firms diversify in response to the new IP environment. ARVs already constitute a small part of the portfolio for the firms engaged in this activity, and these are the ones that are most integrated internationally and best prepared to take advantage of new opportunities — and they will be in competition against the largest and best-resourced firms in the world. In such an environment, their strategies are likely to “ be dictated by survival needs and not by issues related to access to medicines of the general public, whether in India or other least developed countries ” (Gehl Sampath 2005: 67, emphasis added). Translated, to have a strategy “ dictated by survival needs ” means they will have acute concern with price:cost ratios — they are likely to avoid sectors where prices are decreasing and costs increasing. In such an environment producing generic ARVs for export to poor countries becomes less attractive.
+
+Nor do the interests, preferences, and intentions revealed by political lobbying indicate that producing and exporting generic ARVs is a high priority. When we examine what India's pharmaceutical firms and their trade associations lobbied for during the course of the legislative debate on amending the Patent Act in 2005, we see a concern with securing clauses to facilitate upgrading into SGs. For example, the trade organization dominated by the largest domestic drug firms, the Indian Pharmaceutical Association (IPA), worked doggedly to assure that patent rights granted to applications in the “mailbox” would have little effect on local producers already engaged in generic production, avoid “new use” patents, limit the scope of patentability so as to favor the sorts of incremental innovations that local firms do best, and allow for pre-grant opposition on patent applications. 34 In doing so, the local pharmaceutical industry had the support of domestic activists (e.g., the Affordable Medicines and Treatment Campaign, Alternative Law Forum, Indian Network for People Living with HIV/AIDS, Lawyers Collective on AIDS, and National Working Group on Patent Laws) that feared the introduction of a product patent regime would raise the cost of medicines locally. With regard to securing the right to export drugs produced under compulsory licenses, however, there is little evidence that this was a high priority for the national pharmaceutical sector. That is, in contrast to the pharmaceutical sector's focus on revising clauses of the Patent Act that would facilitate upgrading into SGs, it was principally NGOs and health activists—and particularly international NGOs such as MSF and the Consumer Project on Technology—that led the charge for revision of Article 92(A) to retain the possibility of generic drugs
+
+52 “The common reason was that it increased the procedural hassles associated with such exports enormously, and this was not considered worthwhile, especially since the economic returns from such exports were very low” (Gehl Sampath 2005: 65).
+
+$^{33}$ These data were provided by a market analyst researching the Indian pharmaceutical sector.
+
+54 The “mailbox” refers to India's obligation to accept patent applications to be examined post-2005, once product patents became available. Denying “new use” patents means that new forms of pre-existing products cannot be patented without demonstration of increased efficacy. Pre-grant opposition, as the phrase suggests, allows those objecting to a patent application to express their objections prior to legal rights being granted. See Basheer (2005).
+
+---
+
+576
+
+The Political Economy of AIDS Treatment
+
+(including, but not exclusively, ARVs) being exported to poorer developing countries lacking pharmaceutical manufacturing capacity. $^{35}$
+
+Again, the emerging industrial structure has political implications, in that the new constellations of interests pose immense challenges for continuous expansion of ARV treatment. To the extent that the global treatment campaign depends on the supply of generic ARVs from India, it depends on Indian firms to remain vested in the HG segment, and it depends on the Indian government to use the new Patent act in such a way as to keep the HG segment alive. The two are related, in that the likelihood of the government standing down the pressures that it faces from global brand-name pharmaceutical firms and OECD governments not to facilitate generic production is greater when a powerful domestic constituent is demanding that it do so. Yet when new ARVs are introduced in coming years, drugs that will be essential as resistance develops to current treatments, if there are few firms that are legally, financially, and technically capable of producing high-quality generic versions for export, then there will be few firms that are also interested and capable of securing public action to activate the legal provisions in India's new patent law that would clear the way for large-scale production of generic ARVs. National and transnational treatment activists in India will continue to push for such action, but clearly the coalition supporting and facilitating generic ARV production will have been significantly weakened by the new orientation of India's leading pharmaceutical firms. Quite simply, the emerging industrial structure in the Indian pharmaceutical sector has profound implications for the ability to continuously reproduce the political arrangements that are necessary to sustain and expand generic ARV supply.
+
+## Conclusion: Overcoming the Challenges to Stabilizing Supply
+
+Expanding and sustaining AIDS treatment in the developing world—and sustaining expanded treatment—depend on the existence of generic ARVs. While the absence of generic competition does necessarily imply unaffordability, it removes a disciplining lever that can force brand-name firms to make their own products more readily available. This is recognized nearly universally. When, for example, the CEO of Abbott Laboratories was asked why his firm was reducing prices on drugs to developing countries, he explained the pressures his firm faces: “Frankly, because it is required. If I don’t provide our products in Africa, governments will license our intellectual property to others who can. Governments will intervene. Make no mistake, they will do that.” 36 The important thing to remember, however, is that the ability of most developing countries to introduce generic competition (i.e., to “intervene” and “license” IP) depends not just on having patent regimes that take advantage of the flexibilities in the TRIPS Agreement, but their ability to secure foreign suppliers.
+
+The emerging international regulations on IP engender changes that threaten the supply of high quality, generic ARVs needed for AIDS treatment. I have examined how IP institutions alter the interests of private firms, and how these transformed interests affect political coalitions needed to facilitate expanded ARV supply. One need not be apocalyptic to appreciate the concern and recognize the danger that looms over the horizon: at a time of projected increase in global demand for generic ARVs, supply is unlikely to keep apace. Poorer devel-
+
+35 Because many of the concerns and objectives of the various actors overlapped, it is extraordinarily difficult to parse the interests of IPA lobbyists from NGOs, and then again the interests of Indian vs. transnational NGOs. But when it comes to securing the right to export generic drugs produced under compulsory licenses, there is little evidence that this was a high priority for national pharmaceutical firms.
+
+$^{36}$ As quoted by London (2004).
+
+---
+
+KENNETH C. SHADLEN
+
+577
+
+oping countries, international organizations, and non-governmental healthcare providers may demand these drugs, but the critical sources of supply—upon which the world currently depends—may have been fundamentally transformed.
+
+How might the pending problem of undersupply be overcome? It is worth considering two strategies for reinvigorating generic ARV production. The first, and most obvious, strategy is to reform global IP rules to facilitate continued supply of generic ARVs. That is essentially what the developing countries' 2001 proposed interpretation of TRIPS Article 30 would have done, by removing the legal and political obstacles that pharmaceutical firms face in producing and exporting generic ARVs (see discussion above). Of course, for this very reason such proposals have been rejected by governments of countries where the researchbased pharmaceutical industry is based. These firms and their governments regard generic firms in less favorable light, essentially as “free riders” on the global public good of drug innovation. Indeed, one of the central objectives of TRIPS (from the OECD perspective) was to reduce “free riding” and thus provide the research-based pharmaceutical sector with protection from generic competition (Drahos 1995; Sell 2003) . Thus, however sensible this strategy may be from a public health perspective, it has been shown to have little political viability (Matthews 2004) .
+
+Alternatively, then, a second strategy might be to shift attention away from larger, export-capable countries and focus instead on pharmaceutical manufacturing in poorer countries. That is, since the least developed countries are not obligated to offer pharmaceutical patents until 2016, there should be no legal barriers to local production and export of generic ARVs — the constraints of Article 31.f and the subsequent waiver do not yet apply. 37 Yet the challenges to creating generic pharmaceutical sectors in such countries are extensive. Putting IP laws in place that are appropriate for such purposes is more difficult than one might imagine. Many of the poorest WTO members have already fully implemented their TRIPS obligations — transition periods notwithstanding — and offer product patents on pharmaceuticals. They are not obliged to have done so, but re-revising their laws to, for example, make pharmaceuticals unpatentable would expose them to intense international pressures. Furthermore, even if such legal changes were made, the technical and economic obstacles to developing pharmaceutical sectors are immense (WHO 2004; Kaplan and Laing 2005). In India, for example, the growth of the pharmaceutical sector was driven not just by the 1970 Patent Act, which made pharmaceutical products unpatentable, but also by an active state role in sponsoring research and transferring technology, knowledge, and skills from government labs to private firms (Kumar 2003; Chaudhuri 2005: 30 – 60). The advanced chemistry skills that proved so effective in the emergence of India's indigenous pharmaceutical sector developed only with the assistance of concerted government policies that are largely absent in the poor countries that have until 2016. These countries would almost certainly need beyond 2016 to develop indigenous pharmaceutical sectors. Yet it may be difficult to inspire investment knowing that after 10 years the IP regime will change. Or to put it differently, the pre-TRIPS legal environment that spurred high levels of domestic investment in India, for example, cannot be replicated because that legal environment did not have a known terminal point. 38 In any case, focusing exclusively on productive capacity without attention to scale and cost is problematic. National self-sufficiency in ARVs is not just difficult to achieve legally and technically, but even if achieved would not provide producers with sufficient
+
+$^{57}$ Bangladesh and Kenya are frequently cited as potential production sites.
+
+58 Formally, the 2016 deadline is flexible and countries can receive additional extensions, but only by petition and approval—and it is a pretty certain bet that developed countries would not approve an extension in a country where the local generics sector is growing.
+
+---
+
+578
+
+The Political Economy of AIDS Treatment
+
+scale as to lower costs. $^39$ Quite simply, there are strong public health grounds for having foreign suppliers available.
+
+This brings us back to where we started. Demand for ARVs is increasing and will continue to increase as international funding for ARV treatment increases. The expectation is that supply will keep apace and high-quality, affordable versions of drugs will be available. But assuring access to medicines requires attention not just to demand but supply as well. In this article I have shown that the political economy of supply is distinct from—and more complicated than—that of demand. The challenges may not be insurmountable, but the analysis should serve as a warning bell and, it is hoped, spark more inquiry into how to create and sustain political conditions supportive of large-scale generic supply.
+
+## References
+
+Abbott, Frederick M. (2005) The WTO Medicines Decision: World Pharmaceutical Trade and the Protection of Public Health. The American Journal of International Law 99(2):317-358.
+
+D'ADESKY, ANNE-CHRISTINE. (2004) Moving Mountains: The Race to Treat Global AIDS. New York: Verso.
+
+Attaran, Amir. (2004) How Do Patents and Economic Policies Affect Access to Essential Medicines in Developing Countries? Health Affairs 23(3):155-166.
+
+Attaran, Amir, and Lee Gillespie-White. (2001) Do Patents for Antiretroviral Drugs Constrain Access to AIDS Treatment in Africa? Journal of the American Medical Association 286(15):18861892.
+
+BAKER, BROOK K. (2005) India's 2005 Patent Act: Death by Patent or Universal Access to Second- and FutureGeneration ARVs? Background Paper, Health GAP. Available at http://www.healthgap.org/press_ releases/05/091905_HGAP_BP_India_patent_baker.html. (Accessed April 5, 2007).
+
+BARNETT, TONY, AND Alan WHITESIDE. (2006) AIDS in the Twenty-First Century: Disease and Globalization. 2nd edition. Basingstoke: Palgrave.
+
+Basheer, Shammaad. (2005) India's Tryst with TRIPS: The Patents (amendment) Act 2005. Indian Journal of Law and Technology. Vol. 1. Available at http://ssrn.com/abstract=764066. (Accessed April 5, 2007).
+
+BERNICK, DONALD. (2002) 'We all have AIDS': Case for Reducing the Cost of HIV Drugs to Zero. British Medical Journal 324(7331):214-216.
+
+BHANDARI, BHUPESH. (2005) Patent regime? Not the end of the road. Business Standard. 2 February. Available at http://www.rediff.com/money/2005/feb/02guestl.htm. (Accessed April 5, 2007).
+
+Boelaert, Marleen, Lut Lynen, Wim Van Damme, Robert Colebunders, Eric Goemaere, AnneValerie Kaninda, Laura Ciaffi, Maryline Mulemba, Ellen 't Hoen, Bernard Pécoul, Médecins Sans Frontières, Michael J. Seligalid, Udo Schuklenk, Amir Attaran, and Lee Gillespie-White. (2002) Do Patents Prevent Access to Drugs for HIV in Developing Countries? Journal of American Medical Association 287(7):809–938.
+
+Chaudhuri, Sudip. (2005) The WTO and India's Pharmaceuticals Industry: Patent Protection, TRIPS, and Developing Countries. New Delhi: Oxford University Press. (India).
+
+CIPR. (2002) Integrating Intellectual Property Rights and Development Policy. London: Commission on Intellectual Property Rights.
+
+Cohen, Jillian Clare, and Kristina M. Lybecker. (2005) AIDS Policy and Pharmaceutical Patents: Brazil's Strategy to Safeguard Public Health. The World Economy 28(2):211-230.
+
+CORREA, CARLOS. (2000a) Integrating Public Health Concerns into Patent Legislation in the Developing Countries. Geneva: South Centre.
+
+CORREA, CARLOS. (2000b) Intellectual Property Rights, the WTO and Developing Countries: The TRIPS Agreement and Policy Options. London and New York: Zed Books.
+
+Drahos, Peter. (1995) Global Property Rights in Information: The Story of TRIPS at the GATT. Prometheus 13(1):6-19.
+
+EXPRESS PHARMA PULSE. (2005) Patent Act Will Not Affect Drugs Export: Kamal Nath, January 31. Available at http://www.expresspharmapulse.com/story.php?&idno=687. (Accessed April 20, 2007).
+
+39 This is why the inclusion of a list of middle-income developing countries that have agreed not to import generic drugs under the December 2005 waiver of Article 31.f may be problematic as well. Ultimately, that developing national self-sufficiency in drugs is being encouraged (if only implicitly) by the WTO is ironic, to say the least.
+
+---
+
+KENNETH C. SHADLEN
+
+579
+
+Fink, Carsten. (2001) Patent Protection, Transnational Corporations and Market Structure: A Simulation Study of the Indian Pharmaceutical Industry. Journal of Industry, Company and Trade 1(1):101-121.
+
+Friedman, Steven, and Shauna Mottiar. (2005) A Rewarding Engagement? The Treatment Action Campaign and the Politics of HIV/AIDS. Politics & Society 33(4):511-565.
+
+GALVĀO, JANE. (2002) Access to Antiretroviral Drugs in Brazil. The Lancet 360(9348):1862-1865.
+
+Gehl. Sampath, Padmashree. (2005) Economic Aspects of Access to Medicines after 2005: Product Patent Protection and Emerging Firm Strategies in the Indian Pharmaceutical Industry. United Nations University-INTECH. Available at http://www.who.int/intellectualproperty/studies/ PadmashreeGehlSampathFinal.pdf. (Accessed April 5, 2007).
+
+GFATM. (2005) Purchase Price Report. Geneva: Global Fund Against Aids, Tuberculosis, and Malaria. Available at http://web.theglobalfund.org/prm/. (Accessed April 5, 2007).
+
+GRACE, CHERI. (2004) The Effect of Changing Intellectual Property on Pharmaceutical Industry Prospects in India and China: Considerations for Access to Medicines. DFID Health Systems Resource Centre. Available at http://www.dfid.gov.uk/pubs/files/indiachinadomproduce.pdf.
+
+GRACE, CHERI. (2005) A Briefing Paper for DFID: Update on China and India and Access to Medicines. DFID Health Systems Resource Centre. Available at http://www.dfidhealthrc.org/what_new/ Final%20India%20China%20Report.pdf.
+
+Granville, Brigitte, ed. (2002). The Economics of Essential Medicines. London: Royal Institute of International Affairs.
+
+Heins, Volker. (2008, forthcoming) Human Rights, Intellectual Property, and the Struggles for Recognition. Human Rights Review 9(2).
+
+‘T Hoen, Ellen. (2002) TRIPS, Pharmaceutical Patents, and Access to Essential Medicines: A Long Way from Seattle to Doha. Chicago Journal of International Law 3(1):27-46.
+
+Kaplan, Warren, and Richard Laing. (2005) Local Production of Pharmaceuticals: Industrial Policy and Access to Medicines. HNP Discussion Paper. Washington, DC: The World Bank.
+
+KATRAK, HOMI. (2004a) Affordable Prices for Essential Medicines for Developing Countries: Some Economic Issues. Development Policy Review 22(3):309-320.
+
+KATRAK, HOMI. (2004b) The Indian Pharmaceutical Industry and Consumer Welfare: Prospects under the TRIPS Regime. In The WTO and Developing Countries, edited by H. Katrak and R. Strange. New York: Palgrave/Macmillan.
+
+KECK, M., AND K. SIKKINK. (1998) Activists Beyond Borders: Advocacy Networks in International Politics. Ithaca: Cornell University Press.
+
+Khagram Sanjeev, James Riker, and Kathryn Sikkink, eds. (2002) Reconstructing World Politics: Transnational Social Movements, Networks and Norms. Minneapolis, MN: University of Minnesota Press.
+
+Kovsted, Jens. (2005) Scaling Up AIDS Treatment in Developing Countries: A Review of Current and Future Arguments. Development Policy Review 23(4):465-482.
+
+KUMAR, NAGESH. (2003) Intellectual Property Rights, Technology and Economic Development: Experiences of Asian Countries. Economic and Political Weekly 38(3):209-225.
+
+LANJOUW, JEAN O. (1998) The Introduction of Pharmaceutical Product Patents in India: Heartless Exploitation of the Poor and Suffering? NBER Working Paper 6366.
+
+London, Simon. (2004) Benevolence and the Bottom Line. Financial Times, July 14. Available at http://www.ft.com. (Accessed April 20, 2007).
+
+Lucchini, Stéphane, Boubou Cisse, Ségolène Duran, Marie de Cenival, Caroline Comiti, Marion Gaudry, and Jean-Paul Moatti. (2003) Decrease in Prices of Antiretroviral Drugs for Developing Countries: From Political ‘Philanthropy’ to Regulated Markets. In Economics of AIDS and Access to HIV/AIDS Care in Developing Countries: Issues and Challenges , edited by J. Moatti, B. Coriat, Y. Souteyrand, T. Barnett, J. Dumoulin, and Y. Flori. Paris: ANRS.
+
+MATTHEWS, DUNCAN. (2004) 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(1):73-107.
+
+MATTHEWS, DUNCAN. (2006) From the August 30, 2003 WTO Decision to the December 6, 2005 Agreement on an Amendment to TRIPS: Improving Access to Medicines in Developing Countries? Intellectual Property Quarterly 10(1):91-130.
+
+MAY, CHRISTOPHER. (2002) Unacceptable Costs: The Consequences of Making Knowledge Property in a Global Society. Global Society 16(2):123-144.
+
+Moatti, Jean-Paul, Benjamin Coriat, Yves Souteyrand, Tony Barnett, Jérôme Dumoulin, and Yves-Antoine Flori, ed. (2003) Economics of AIDS and Access to HIV/AIDS Care in Developing Countries: Issues and Challenges. Paris: ANRS.
+
+---
+
+580
+
+The Political Economy of AIDS Treatment
+
+MSF. (2005) Untangling the Web of Price Reductions: A Pricing Guide for the Purchase of ARVs for Developing Countries. 8th edition. Available at http://www.doctorswithoutborders.org/publications/reports/ 2005/untanglingthewebv8.pdf. (Accessed April 5, 2007).
+
+MSF-WHO-UNAIDS. (2003) Surmounting Challenges: Procurement of Antiretroviral Medicines in Low- and Middle-Income Countries. The Experience of Médecins Sans Frontières. Available at http:// www.accessmed-msf.org/documents/procurementreport.pdf. (Accessed April 5, 2007).
+
+Mukherjee, J.S., P. E. Farmer, D. Niyizonkiza, L. McCorkle, C. Vanderwarker, P. Teixeira, and J. Y. Kim. (2003) Tackling HIV in Resource Poor Countries. British Medical Journal 327(7423):1104-1106.
+
+Nattrass, Nicoli, and Nathan Geffen. (2005) The Impact of Reduced Drug Prices on the CostEffectiveness of HAART in South Africa. African Journal of AIDS Research 4(1):65-67.
+
+NOGUÉs, JULIO. (1993) Cost and Benefits of Introducing Patent Protection for Pharmaceutical Drugs in Developing Countries. The Developing Economies 31(1):24-53.
+
+Orsi, Fabienne, Lia Hasenclever, Beatriz Fialho, Paulo Tigre, and Benjamin Coriat. (2003) Intellectual Property Rights, Anti-AIDS Policy and Generic Drugs: Lessons from the Brazilian Public Health Program. In Economics of AIDS and Access to HIV/AIDS Care in Developing Countries: Issues and Challenges, edited by J. Moatti, B. Coriat, Y. Souteyrand, T. Barnett, J. Dumoulin, and Y. Flori. Paris: ANRS.
+
+PIERSON, PAUL. (2004) Politics in Time: History, Institutions, and Social Analysis. Princeton: Princeton University Press.
+
+REICHMAN, JEROME H., AND CATHERINE HASENZAHL. (2003) Non-voluntary Licensing of Patented Inventions: Historical Perspective, Legal Framework under TRIPS, and an Overview of the Practice in Canada and the USA. UNCTAD-ICTSD Project on IPRs and Sustainable Development, Issue Paper No. 5.
+
+ROFFE, PEDRO, GEOFF TANSEY, AND DAVID VIVAS-EUGUI, ed. (2006) Negotiating Health: Intellectual Property and Access to Medicines. London: Earthscan.
+
+SCHRANK, ANDREW. (2005) Conquering, Comprador, or Competitive: The National Bourgeoisie in the Developing World. In New Directions in the Sociology of Global Development, edited by F. BUTTEL and P. MCMICHAEL. Amsterdam: JAI/Elsevier.
+
+SCHURMAN, RACHEL. (2004) Fighting 'Frankenfoods': Industry Opportunity Structures and the Efficacy of the Anti-Biotech Movement in Western Europe. Social Problems 51(2):243-268.
+
+Schwartländer, B., J. Stover, N. Walker, L. Bollinger, J. P. Gutierrez, W. McGreevey, M. Opuni, S. Forsythe, L. Kumaranayake, C. Watts, and S. Bertozzi. (2001) Resource Needs for HIV/AIDS. Science 292(5526):2434–2436.
+
+SECKINELGIN, Hakan. (2006) The Multiple Worlds of NGOs and HIV/AIDS: Rethinking NGOs and their Agency. Journal of International Development 18(5):715-727.
+
+SELL, SUSAN K. (2002) TRIPS and the Access to Medicines Campaign. Wisconsin International Law Journal 20(3):481-522.
+
+SELL, SUSAN K. (2003) Private Power, Public Law: The Globalization of Intellectual Property Rights. London: Cambridge University Press.
+
+SELL, SUSAN K. (2006) Books, Drugs and Seeds: The Politics of Access. Paper presented at the annual meeting of the International Studies Association, San Diego, CA (March).
+
+Sever, Patrice, Paul Leger, Macarthur Charles, Francine Noel, Gerry Bonhomme, Gyrlande Bois, Erik George, Stefan Kenel-Pierre, Peter F. Wright, Roy Gulick, Warren D. Johnson, Jean William Pape, and Daniel W. Fitzgerald. (2005) Antiretroviral Therapy in a Thousand Patients with AIDS in Haiti. New England Journal of Medicine 353(22):2325-2334.
+
+SHADLEN, KENNETH C. (2004a) Patents and Pills, Power and Procedure: The North-South Politics of Public Health in the WTO. Studies in Comparative International Development 39(3):76-108.
+
+SHADLEN, KENNETH C. (2004b) Challenges to Treatment: The Price-Infrastructure Trap and Access to Medicines. Journal of International Development 16(8):1169-1180.
+
+Singh, Pragya, and Dev Chatterjee. (2004) Pharma's in the Den. The Sunday Express, December 19. Available at http://www.indianexpress.com/full_story.php?content_id=61096. (Accessed April 20, 2007).
+
+SMITH, RAYMOND, AND PATRICIA SiPLON. (2006) Drugs into Bodies: Global AIDS Treatment Activism. Westport: Praeger.
+
+TARROW, Sidney. (1994) Power in Motion: Social Movements and Contentious Politics. Cambridge: Cambridge University Press.
+
+Tayler Yolanda, ed. (2004) Battling HIV/AIDS: A Decision Maker's Guide to the Procurement of Medicines and Related Supplies. Washington, DC: The World Bank.
+
+---
+
+KENNETH C. SHADLEN
+
+581
+
+UNAIDS. (2006) Report on the Global AIDS Epidemic: A UNAIDS 10 th Anniversary Special Edition. Geneva: UNAIDS.
+
+Wainberg, Mark A. (2005) Generic HIV Drugs: Enlightened Policy for Global Health. New England Journal of Medicine 352(8):747-750.
+
+Wall Street Journal. (2003) Brazil to Stir Up AIDS-Drug Battle: Nation to Authorize Imports of Generics, Citing the Cost of Big Companies' Drugs. 5 September.
+
+Watal, Jayashree. (2000) Intellectual Property Rights in the World Trade Organization: The Way Forward for Developing Countries. New Delhi: Oxford University Press (India).
+
+WHO. (1998) Globalization and Access to Drugs: Perspectives on the WTO/TRIPS Agreement. Health Economics and Drugs, DAP Series No. 7 (Revised), WHO/DAP.98.9. Available at http://whqlibdoc. who.int/hq/1998/WHO_DAP_98.9_revised.pdf. (Accessed April 5, 2007).
+
+WHO. (2001) Globalization, Patents and Drugs: An Annotated Bibliography. Health Economics and Drugs, EDM Series No. 10. WHO/EDM/PAR/2001.1.
+
+WHO. (2002) Network for Monitoring the Impact of Globalization and TRIPS on Access to Medicines. Health Economics and Drugs, EDM Series No. 11, WHO/EDM/PAR/2002.1.
+
+WHO. (2003) Scaling Up Antiretroviral Therapy in Resource-Limited Settings: Treatment Guidelines for a Public Health Approach. 2003 Revision. Geneva: World Health Organization. Available at http://www. who.int/hiv/pub/prev_care/en/arvrevision2003en.pdf. (Accessed April 5, 2007).
+
+WHO. (2004) Manufacture of Antiretrovirals in Developing Countries and Challenges for the Future. EB114/15. Available at http://www.who.int/gb/ebwha/pdf_files/EB114/B114_15-en.pdf. (Accessed April 5, 2007).
+
+WHO. (2005) Intellectual Property Protection: Impact on Public Health. WHO Drug Information 19(3):236-241.
+
+WHO. (2006) Access to HIV/AIDS Drugs and Diagnostics of Acceptable Quality. Procurement, Quality and Sourcing Project. http://mednet3.who.int/prepqual. (Accessed April 5, 2007).
+
+WHO/UNAIDS. (2006) Progress on Global Access to HIV Antiretroviral Therapy: A Report on 3 by 5 and Beyond. Geneva: World Health Organization.
+
+World Bank. (2004) HIV/AIDS Medicines and Related Supplies: Contemporary Context and Procurement. Technical Guide. Washington, DC: The World Bank.
+
+WTO. (2003a) Implementation of Paragraph 6 of the Doha Declaration on the TRIPS Agreement and Public Health: Decision of the General Council of 30 August 2003. WT/L/540. Available at http:// www.wto.org/english/tratop_e/trips_e/implem_para6_e.htm. (Accessed April 5, 2007).
+
+WTO. (2003b) The General Council Chairperson's Statement. Available at http://www.wto.org/english/ news_e/news03_e/trips_stat_28aug03_e.htm. (Accessed April 5, 2007).
+

processed/00084_W2262800383/document.md

@@ -0,0 +1,536 @@
+# Exposing the Probabilistic Causal Structure of Discrimination
+
+Francesco Bonchi$^{1,2}$ Sara Hajian$^{2}$ Bud Mishra$^{3}$ Daniele Ramazzotti$^{4}$
+
+1Algorithmic Data Analytics Lab ISI Foundation, Turin, Italy francesco.bonchi@isi.it
+
+2EURECAT Barcelona, Spain sara.hajian@eurecat.org
+
+3New York University New York, USA mishra@nyu.edu
+
+4Milano-Bicocca University Milan, Italy daniele.ramazzotti@disco.unimib.it
+
+## ABSTRACT
+
+Discrimination discovery from data is an important task aiming at identifying patterns of illegal and unethical discriminatory activities against protected-by-law groups, e.g., ethnic minorities. While any legally-valid proof of discrimination requires evidence of causality, the state-of-theart methods are essentially correlation-based, albeit, as it is well known, correlation does not imply causation.
+
+In this paper we take a principled causal approach to the data mining problem of discrimination detection in databases. Following Suppes' probabilistic causation theory , we define a method to extract, from a dataset of historical decision records, the causal structures existing among the attributes in the data. The result is a type of constrained Bayesian network, which we dub Suppes-Bayes Causal Network ( SBCN ). Next, we develop a toolkit of methods based on random walks on top of the SBCN , addressing different anti-discrimination legal concepts, such as direct and indirect discrimination, group and individual discrimination, genuine requirement, and favoritism. Our experiments on real-world datasets confirm the inferential power of our approach in all these different tasks.
+
+## 1. Introduction
+
+The importance of discrimination discovery. At the beginning of 2014, as an answer to the growing concerns about the role played by data mining algorithms in decisionmaking, USA President Obama called for a 90-day review of big data collecting and analysing practices. The resulting report 1 concluded that “ big data technologies can cause societal harms beyond damages to privacy ” . In particular, it expressed concerns about the possibility that decisions informed by big data could have discriminatory effects, even in the absence of discriminatory intent, further imposing less favorable treatment to already disadvantaged groups.
+
+Discrimination refers to an unjustified distinction of individuals based on their membership, or perceived membership, in a certain group or category. Human rights laws prohibit discrimination on several grounds, such as gender, age, marital status, sexual orientation, race, religion or belief, membership in a national minority, disability or illness. Anti-discrimination authorities (such as equality enforcement bodies, regulation boards, consumer advisory councils) monitor, provide advice, and report on discrimination compliances based on investigations and inquiries. A fundamental role in this context is played by discrimination discovery in databases , i.e., the data mining problem of unveiling discriminatory practices by analyzing a dataset of historical decision records.
+
+Discrimination is causal. According to current legislation, discrimination occurs when a group is treated “ less favorably ” [1] than others, or when “ a higher proportion of people not in the group is able to comply ” with a qualifying criterion [2] . Although these definitions do not directly imply causation, as stated in [3] all discrimination claims require plaintiffs to demonstrate a causal connection between the challenged outcome and a protected status characteristic. In other words, in order to prove discrimination, authorities must answer the counterfactual question: what would have happened to a member of a specific group (e.g., nonwhite), if he or she had been part of another group (e.g., white)?
+
+“ The Sneetches ” , the popular satiric tale $^2$ against discrimination published in 1961 by Dr. Seuss, describes a society of yellow creatures divided in two races: the ones with a green star on their bellies, and the ones without. The StarBelly Sneetches have some privileges that are instead denied to Plain-Belly Sneetches. There are, however, Star-On and Star-Off machines that can make a Plain-Belly into a StarBelly, and viceversa. Thanks to these machines, the causal relationship between race and privileges can be clearly measured, because stars can be placed on or removed from any belly, and multiple outcomes can be observed for an individual. Therefore, we could readily answer the counterfactual question, saying with certainty what would have happened to a Plain-Belly Sneetch had he or she been a Star-Belly Sneetch.
+
+In the real world however, proving discrimination episodes is much harder, as we cannot manipulate race, gender, or sexual orientation of an individual. This highlights the need to assess discrimination as a causal inference problem [4]
+
+1http://www.whitehouse.gov/sites/default/files/ docs/big_data_privacy_report_may_1_2014.pdf
+
+Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
+
+Copyright 20XX ACM X-XXXXX-XX-X/XX/XX ...$10.00.
+
+2http://en.wikipedia.org/wiki/The_Sneetches_and_ Other_Stories
+
+arXiv:1510.00552v3 [cs.DB] 8 Mar 2017
+
+---
+
+from a database of past decisions, where causality can be inferred probabilistically. Unfortunately, the state of the art of data mining methods for discrimination discovery in databases does not properly address the causal question , as it is mainly based on correlation-based methods (surveyed in Section 2 ).
+
+Correlation is not causation. It is well known that correlation between two variables does not necessarily imply that one causes the other. Consider a unique cause $X$ of two effects, $Y$ and $Z$ : if we do not take into account $X$ , we might derive wrong conclusions because of the observable correlation between $Y$ and $Z$ . In this situation, $X$ is said to act as a confounding factor for the relationship between $Y$ and $Z$ .
+
+Variants of the complex relation just discussed can arise even if, in the example, $X$ is not the actual cause of either $Y$ or $Z$ , but it is only correlated to them, for instance, because of how the data were collected. Consider for instance a credit dataset where there exists high correlation between a variable representing low income and another variable representing loan denial and let us assume that this is due to an actual legitimate causal relationship in the sense that, legitimately, a loan is denied if the applicant has low income. Let us now assume that we also observe high correlation between low income and being female, which, for instance, can be due to the fact that the women represented in the specific dataset in analysis, tend to be underpaid. Given these settings, in the data we would also observe high correlation between the variable gender being female and the variable representing loan denial, due to the fact that we do not account for the presence of the variable low income. Following common terminologies, we will say that such situations are due to spurious correlations .
+
+However, the picture is even more complicated: it could be the case, in fact, that being female is the actual cause of the low income and, hence, be the indirect cause of loan denial through low income. This would represent a causal relationship between the gender and the loan denial, that we would like to detect as discrimination.
+
+Disentangling these two different cases, i.e., female is only correlated to low income in a spurious way, or being female is the actual cause of low income, is at the same time important and challenging. This highlights the need for a principled causal approach to discrimination detection.
+
+Another typical pitfall of correlation-based reasoning is expressed by what is known as Simpson's paradox $^3$ according to which, correlations observed in different groups might disappear when these heterogeneous groups are aggregated, leading to false positives cases of discrimination discovery. One of the most famous false-positive examples due to Simpson’s paradox occurred when in 1973 the University of California, Berkeley was sued for discrimination against women who had applied for admission to graduate schools. In fact, by looking at the admissions of 1973, it first appeared that men applying were significantly more likely to be admitted than women. But later, by examining the individual departments carefully, it was discovered that none of them was significantly discriminating against women. On the contrary, most departments had exercised a small bias in favor of women. The apparent discrimination was due to the fact that women tended to apply to departments with lower
+
+rates of admission, while men tended to apply to departments with higher rates [5] . Later in Section 5.5 we will use the dataset from this episode to highlight the differences between correlation-based and causation-based methods.
+
+Spurious correlations can also lead to false negatives (i.e., discrimination existing but not being detected) as is commonly seen in “ reverse-discrimination ” . The typical case is when authorities take affirmative actions, e.g., with compensatory quota systems, in order to protect a minority group from a potential discrimination. Such actions, while trying to erase the supposed discrimination (i.e., the spurious correlation), fail to address the real underlying causes for discrimination, potentially ending up denying individual members of a privileged group from access to their rightful shares of social goods. In the early 70's, a case involving the University of California at Davis Medical School highlighted one such incident as the school's admissions program reserved 16 of the 100 places in its entering class for “ disadvantaged ” applicants, thus unintentionally reducing the chances of admission for a qualified applicant. $^4$
+
+These are just few typical examples of the pitfalls of correlation-based reasoning in the discovery of discrimination. Later in Section 5.5 we show concrete examples from real-world datasets where correlation-based methods to discrimination discovery are not satisfactory.
+
+Our proposal and contributions. In this paper we take a principled causal approach to the data mining problem of discrimination detection in databases. Following Suppes' probabilistic causation theory [6, 7] we define a method to extract, from a dataset of historical decision records, the causal structures existing among the attributes in the data.
+
+In particular, we define the Suppes-Bayes Causal Network ( SBCN ), i.e., a directed acyclic graph ( DAG ) where we have a node representing a Bernoulli variable of the type $\langle  attribute = value \rangle$ for each pair attribute-value present in the database. In this DAG an arc $(A,B)$ represents the existence of a causal relation between $A$ and $B$ (i.e., $A$ causes $B$ ). Moreover, each arc is labeled with a score, representing the strength of the causal relation.
+
+Our SBCN is a constrained Bayesian network reconstructed by means of maximum likelihood estimation (MLE) from the given database, where we force the conditional probability distributions induced by the reconstructed graph to obey Suppes' constraints: i.e., temporal priority and probability rising . Imposing Suppes' temporal priority and probability raising we obtain what we call the prima facie causes graph [7] , which might still contain spurious causes (false positives). In order to remove these spurious case we add a bias term to the likelihood score, favoring sparser causal networks: in practice we sparsify the prima facie causes graph by extracting a minimal set of edges which best explain the data. This regularization is done by means of the Bayesian Information Criterion (BIC) [8] .
+
+The obtained SBCN provides a clear summary, amenable to visualization, of the probabilistic causal structures found in the data. Such structures can be used to reason about different types of discrimination. In particular, we show how using several random-walk-based methods, where the next step in the walk is chosen proportionally to the edge weights, we can address different anti-discrimination legal concepts.
+
+3http://en.wikipedia.org/wiki/Simpson's_paradox
+
+4http://en.wikipedia.org/wiki/Regents_of_the_ University_of_California_v._Bakke
+
+---
+
+Our experiments show that the measures of discrimination produced by our methods are very strong, almost binary, signals: our measures are very clearly separating the discrimination and the non-discrimination cases.
+
+To the best of our knowledge this is the first proposal of discrimination detection in databases grounded in probabilistic causal theory.
+
+Roadmap. The rest of the paper is organized as follows. Section 2 discusses the state of the art in discrimination detection in databases. In Section 3 we formally introduce the SBCN and we present the method for extracting such causal network from the input dataset. Once extracted our SBCN , in Section 4 we show how to exploit it for different concepts of discrimination detection, by means of randomwalk methods. Finally Section 5 presents our experimental assessment and comparison with correlation-based methods on two real-world datasets.
+
+## 2. Related Work
+
+Discrimination analysis is a multi-disciplinary problem, involving sociological causes, legal reasoning, economic models, statistical techniques [9, 10] . Some authors [11, 12] study how to prevent data mining from becoming itself a source of discrimination. In this paper instead we focus on the data mining problem of detecting discrimination in a dataset of historical decision records, and in the rest of this section we present the most related literature.
+
+Pedreschi et al. [13, 14, 15] propose a technique based on extracting classification rules (inductive part) and ranking the rules according to some legally grounded measures of discrimination (deductive part). The result is a (possibly large) set of classification rules, providing local and overlapping niches of possible discrimination. This model only deals with group discrimination.
+
+Luong et al. [16] exploit the idea of situation-testing [17] to detect individual discrimination. For each member of the protected group with a negative decision outcome, testers with similar characteristics ( $k$ -nearest neighbors) are considered. If there are significantly different decision outcomes between the testers of the protected group and the testers of the unprotected group, the negative decision can be ascribed to discrimination.
+
+Zliobaite et al. [18] focus on the concept of genuine requirement to detect that part of discrimination which may be explained by other, legally grounded, attributes. In [19] Dwork et al. address the problem of fair classification that achieves both group fairness, i.e., the proportion of members in a protected group receiving positive classification is identical to the proportion in the population as a whole, and individual fairness, i.e., similar individuals should be treated similarly.
+
+The above approaches assume that the dataset under analysis contains attributes that denote protected groups (i.e., direct discrimination). This may not be the case when such attributes are not available, or not even collectable at a micro-data level as in the case of the loan applicant's race. In these cases we talk about indirect discrimination discovery. Ruggieri et al. [20, 21] adopt a form of rule inference to cope with the indirect discovery of discrimination. The correlation information is called background knowledge, and is itself coded as an association rule.
+
+Mancuhan and Clifton [22] propose Bayesian networks
+
+as a tool for discrimination discovery. Bayesian networks consider the dependence between all the attributes and use these dependencies in estimating the joint probability distribution without any strong assumption, since a Bayesian network graphically represents a factorization of the joint distribution in terms of conditional probabilities encoded in the edges. Although Bayesian networks are often used to represent causal relationships, this needs not be the case, in fact a directed edge from two nodes of the network does not imply any causal relation between them. As an example, let us observe that the two graphs $A \to B \to C$ and $C \to B \to A$ impose exactly the same conditional independence requirements and, hence, any Bayesian network would not be able to disentangle the direction of any causal relationship among these events.
+
+Our work departs from this literature as:
+
+- 1. it is grounded in probabilistic causal theory instead of
+being based on correlation;
+2. it proposes a holistic approach able to deal with differ-
+ent types of discrimination in a single unifying frame-
+work, while the methods in the state of the art usually
+deal with one and only one specific type of discrimina-
+tion;
+3. it is the first work to adopt graph theory and social
+network analysis concepts, such as random-walk-based
+centrality measures and community detection, for dis-
+crimination detection;
+Our proposal has also lower computational cost than methods such as [13, 14, 15] which require to compute a potentially exponential number of association/classification rules.
+
+## 3. SUPPES-BAYES CAUSAL NETWORK
+
+In order to study discrimination as a causal inference problem, we exploit the criteria defined in the theories of probabilistic causation [6] . In particular, we follow [7] , where Suppes proposed the notion of prima facie causation that is at the core of probabilistic causation. Suppes' definition is based on two pillars: ( i ) any cause must happen before its effect ( temporal priority ) and ( ii ) it must raise the probability of observing the effect ( probability raising ).
+
+Definition 1 (Probabilistic causation [7]). For any two events $h$ and $e$ , occurring respectively at times $t_h$ and $t_e$ , under the mild assumptions that $0 < \mathcal{P}(h), \mathcal{P}(e) < 1$ , the event $h$ is called a prima facie cause of the event $e$ if it occurs before the effect and the cause raises the probability of the effect, i.e. $t_h < t_e$ and $\mathcal{P}(e \mid h) > \mathcal{P}(e \mid \neg h)$ .
+
+In the rest of this section we introduce our method to construct, from a given relational table $\mathcal{D}$ , a type of causal Bayesian network constrained to satisfy the conditions dictated by Suppes' theory, which we dub Suppes-Bayes Causal Network ( SBCN ).
+
+In the literature many algorithms exist to carry out structural learning of general Bayesian networks and they usually fall into two families [23] . The first family, constraint based learning , explicitly tests for pairwise independence of variables conditioned on the power set of the rest of the variables in the network. These algorithms exploit structural conditions defined in various approaches to causality [6, 24, 25] . The second family, score based learning , constructs
+
+---
+
+a network which maximizes the likelihood of the observed data with some regularization constraints to avoid overfitting. Several hybrid approaches have also been recently proposed [26, 27, 28] .
+
+Our framework can be considered a hybrid approach exploiting constrained maximum likelihood estimation (MLE) as follows: (i) we first define all the possible causal relationship among the variables in $\mathcal{D}$ by considering only the oriented edges between events that are consistent with Suppes’ notion of probabilistic causation and, subsequently, (ii) we perform the reconstruction of the SBCN by a score-based approach (using BIC), which considers only the valid edges.
+
+We next present in details the whole learning process.
+
+### 3.1 Suppes' constraints
+
+We start with an input relational table $\mathcal{D}$ defined over a set $A$ of $h$ categorical attributes and $s$ samples. In case continuous numerical attributes exists in $\mathcal{D}$ , we assume they have been discretized to become categorical. From $\mathcal{D}$ , we derive $\mathcal{D}'$ , an $m \times s$ binary matrix representing $m$ Bernoulli variables of the type $\langle  attribute = value \rangle$ , where an entry is 1 if we have an observation for the specific variable and 0 otherwise.
+
+Temporal priority. The first constraint, temporal priority, cannot be simply checked in the data as we have no timing information for the events. In particular, in our context the events for which we want to reason about temporal priority are the Bernoulli variables $\langle  attribute = value \rangle$ .
+
+The idea here is that, e.g., income = low cannot be a cause of gender = female , because the time when the gender of an individual is determined is antecedent to that of when the income is determined. This intuition is implemented by simply letting the data analyst provide as input to our framework a partial temporal order $r: A \to \mathbb{N}$ for the $h$ attributes, which is then inherited from the $m$ Bernoulli variables $^5$ .
+
+Based on the input dataset $\mathcal{D}$ and the partial order $r$ we produce the first graph $G=(V,E)$ where we have a node for each of the Bernoulli variables, so $|V|=m$ , and we have an arc $(u,v)\in E$ whenever $r(u)\leq  r(v)$ . This way we will immediately rule out causal relations that do not satisfy the temporal priority constraint.
+
+Probability raising. Given the graph $G=(V,E)$ built as described above the next step requires to prune the arcs which do not satisfy the second constraint, probability raising, thus building $G'=(V,E')$ , where $E'\subseteq E$ . In particular we remove from $E$ each arc $(u,v)$ such that $\mathcal{P}(v\mid u)\leq  \mathcal{P}(v\mid \neg u)$ . The graph $G'$ so obtained is called prima facie graph.
+
+We recall that the probability raising condition is equivalent to constraining for positive statistical dependence [27] : in the prima facie graph we model all and only the positive correlated relations among the nodes already partially ordered by temporal priority, consistently with Suppes' characterization of causality in terms of relevance.
+
+### 3.2 Network simplification
+
+Suppes' conditions are necessary but not sufficient to evaluate causation [28] : especially when the sample size is small, the model may have false positives (spurious causes), even after constraining for Suppes' temporal priority and probability raising criteria (which aim at removing false negatives). Consequently, although we expect all the statistically relevant causal relations to be modelled in $G'$ , we also expect some spurious ones in it.
+
+In our proposal, in place of other structural conditions used in various approaches to causality, (see e.g., [6, 24, 25] ), we perform a network simplification (i.e., we sparsify the network by removing arcs) with a score based approach, specifically by relying on the Bayesian Information Criterion (BIC) as the regularized likelihood score [8] .
+
+We consider as inputs for this score the graph $G'$ and the dataset $\mathcal{D}'$ . Given these, we select the set of arcs $E^* \subseteq E'$ that maximizes the score:
+
+$$score_{BIC}\left(\mathcal{D}^{\prime}, G^{\prime}\right)=L L\left(\mathcal{D}^{\prime} \mid G^{\prime}\right)-\frac{\log s}{2} dim\left(G^{\prime}\right) .$$
+
+In the equation, $G'$ denotes the graph, $\mathcal{D}'$ denotes the data, $s$ denotes the number of samples, and $\dim(G')$ denotes the number of parameters in $G'$ . Thus, the regularization term $-\dim(G')$ favors graphs with fewer arcs. The coefficient $\log s/2$ weighs the regularization term, such that the higher the weight, the more sparsity will be favored over “ explaining ” the data through maximum likelihood. Note that the likelihood is implicitly weighted by the number of data points, since each point contributes to the score.
+
+Assume that there is one true (but unknown) probability distribution that generates the observed data, which is, eventually, uniformly randomly corrupted by false positives and negatives rates (in $[0,1)$ ). Let us call correct model , the statistical model which best approximate this distribution. The use of BIC on $G'$ results in removing the false positives and, asymptotically (as the sample size increases), converges to the correct model. In particular, BIC is attempting to select the candidate model corresponding to the highest Bayesian Posterior probability, which can be proved to be equivalent to the presented score and its $log(s)$ penalization factor.
+
+We denote with $G^* = (V, E^*)$ the graph that we obtain after this step. We note that, as for general Bayesian network, $G^*$ is a DAG by construction.
+
+### 3.3 Confidence score
+
+Using the reconstructed $\sf{SBCN}$ , we can represent the probabilistic relationships between any set of events (nodes). As an example, suppose to consider the nodes representing respectively $income = low$ and $gender = female$ being the only two direct causes (i.e., with arcs toward) of $loan = denial$ . Given $\sf{SBCN}$ , we can estimate the conditional probabilities for each node in the graph, i.e., probability $loan = denial$ given $income = low$ AND $gender = female$ in the example, by computing the conditional probability of only the pair of nodes directly connected by an arc. For an overview of state-of-the-art methods for doing this, see [23] . However, we expect to be mostly dealing with full data, i.e., for every directly connected node in the $\sf{SBCN}$ , we expect to have several observations of any possible combination $attribute = value$ . For this reason, we can simply estimate the node probabilities by counting the observations
+
+5 Note that our learning technique requires the input order $r$ to be correct and complete in order to guarantee its convergence. Nevertheless, if this is not the case, it is still capable of providing valuable insights about the underlying causal model, although with the possibility of false positive or false negative causal claims.
+
+---
+
+![Figure](figures/figure_000.png)
+
+Figure 1: One portion of the SBCN extracted from the Adult dataset. This subgraph corresponds to the $C_2$ community reported later in Table 3 (Section 5 ) extracted by a community detection algorithm.
+
+in the data. Moreover, we will exploit such conditional probabilities to define the confidence score of each arc in terms of their causal relationship.
+
+In particular, for each arc $(v,u) \in E^*$ involving the causal relationship between two nodes $u,v \in V$ , we define a confidence score $W(v,u) = \mathcal{P}(u \mid v) - \mathcal{P}(u \mid \neg v)$ , which, intuitively, aims at estimating the observations where the cause $v$ is followed by its effect $u$ , that is $\mathcal{P}(u \mid v)$ , and the ones where this is not observed, i.e., $\mathcal{P}(u \mid \neg v)$ , because of imperfect causal regularities. We also note that, by the constraints discussed above, we require $\mathcal{P}(u \mid v) \gg \mathcal{P}(u \mid \neg v)$ and, for this reason, each weight is positive and no larger than 1, i.e., $W: E^* \rightarrow (0,1]$ .
+
+Combining all of the concepts discussed above, we conclude with the following definition.
+
+## Definition 2 (Suppes-Bayes Causal Network).
+
+Given an input dataset $\mathcal{D}'$ of $m$ Bernoulli variables and $s$ samples, and given a partial order $r$ of the variables, the Suppose-Bayes Causal Network $SBCN = (V, E^*, W)$ subsumed by $\mathcal{D}'$ is a weighted DAG such that the following requirements hold:
+
+• [Suppes' constraints] for each arc $(v,u) \in E^*$ involving the causal relationship between nodes $u,v \in V$ , under the mild assumptions that $0 < \mathcal{P}(u), \mathcal{P}(v) < 1$ :
+
+$$r(v) \leq  r(u)  and  \mathcal{P}(u \mid v)>\mathcal{P}(u \mid \neg v) .$$
+
+• [Simplification] let $E'$ be the set of arcs satisfying the Suppes' constraints as before; among all the subsets of $E'$ , the set of arcs $E^*$ is the one whose corresponding graph maximizes BIC:
+
+$$E^{*}=\underset{E \subseteq E^{\prime}, G=(V, E)}{\arg \max }\left(L L\left(\mathcal{D}^{\prime} \mid G\right)-\frac{\log s}{2} dim(G)\right) .$$
+
+$W(v,u)=\mathcal{P}(u\mid v)-\mathcal{P}(u\mid\lnot v),\ \forall (v,u)\in E^{*}$
+
+An example of a portion of a SBCN extracted from a realworld dataset is reported in Figure 1 .
+
+Algorithm 1 summarizes the learning approach adopted for the inference of the SBCN . Given $\mathcal{D}'$ an input dataset over $m$ Bernoulli variables and $s$ samples, and $r$ a partial order of the variables, Suppes' constraints are verified (Lines 4-9) to construct a DAG as described in Section 3.1 .
+
+The likelihood fit is performed by hill climbing (Lines 1221), an iterative optimization technique that starts with an arbitrary solution to a problem (in our case an empty graph) and then attempts to find a better solution by incrementally visiting the neighbourhood of the current one. If the new candidate solution is better than the previous one it is considered in place of it. The procedure is repeated until the stopping criterion is matched.
+
+The ! StoppingCriterion occurs (Line 12) in two situations: $(i)$ the procedure stops when we have performed a large enough number of iterations or, $(ii)$ it stops when none of the solutions in $G_{neighbors}$ is better than the current $G_{fit}$ . Note that $G_{neighbors}$ denotes all the solutions that are derivable from $G_{fit}$ by removing or adding at most one edge.
+
+Time and space complexity. The computation of the valid DAG according to Suppes' constraints (Lines 4-10) requires a pairwise calculation among the $m$ Bernoulli variables. After that, the likelihood fit by hill climbing (Lines 11-21) is performed. Being an heuristic, the computational cost of hill climbing depends on the sopping criterion. However, constraining by Suppes' criteria tends to regularize the problem leading on average to a quick convergence to a good solution. The time complexity of Algorithm 1 is $\mathcal{O}(sm)$ and the space required is $\mathcal{O}(m^2)$ , where $m$ however is usually not too large, being the number of attribute-value pairs, and not the number of examples.
+
+---
+
+```bash
+#1 Learning the SBCN
+#1 Inputs: $\mathcal{D}'$ an input dataset of $m$ Bernoulli variables and
+#1 samples, and $r$ a partial order of the variables
+#1 Output: $SBCN(V, E^*, W)$ as in Definition 2
+#1 #1 [Suppes' constraints]
+#1 for all pairs $(v, u)$ among the $m$ Bernoulli variables do
+#1 if $r(v) \leq r(u)$ and $\mathcal{P}(u \mid v) > \mathcal{P}(u \mid \neg v)$ then
+#1 add the arc $(v, u)$ to $SBCN$ .
+#1 end if
+#1 end for
+#1 #1 [Likelihood fit by hill climbing]
+#1 #1 Consider $G(V, E^*, W)_{fit} = ø$ .
+#1 while !StoppingCriterion() do
+#1 #1 Let $G(V, E^*, W)_{neighbors}$ be the neighbor solutions of
+$G(V, E^*, W)_{fit}$ .
+#1 #1 Remove from $G(V, E^*, W)_{neighbors}$ any solution
+#1 whose arcs are not included in $SBCN$ .
+#1 #1 Consider a random solution $G_{current}$ in
+#1 #1 $G(V, E^*, W)_{neighbors}$ .
+#1 #1 $score_{BIC}(\mathcal{D}', G_{current}) > score_{BIC}(\mathcal{D}', G_{fit})$
+#1 then
+#1 #1 $G_{fit} = G_{current}$ .
+#1 #1 $\forall$ arc $(v, u)$ of $G_{fit}$ , $W(v, u) = \mathcal{P}(u \mid v) - \mathcal{P}(u \mid \neg v)$ .
+#1 #1 end if
+#1 #1 end while
+#1 #1 $SBCN = G_{fit}$ .
+#1 #1 return $SBCN$ .
+```
+
+### 3.4 Expressivity of a SBCN
+
+We conclude this Section with a discussion on the causal relations that we model by a $SBCN$ .
+
+Let us assume that there is one true (but unknown) probability distribution that generates the observed data whose structure can be modelled by a DAG . Furthermore, let us consider the causal structure of such a DAG and let us also assume each node with more then one cause to have conjunctive parents: any observation of the child node is preceded by the occurrence of all its parents. As before we call correct model, the statistical model which best approximate the distribution. On these settings, we can prove the following theorem.
+
+Theorem 1. Let the sample size $s \to \infty$ , the provided partial temporal order $r$ be correct and complete and the data be uniformly randomly corrupted by false positives and negatives rates (in $[0,1)$ ), then the $SBCN$ inferred from the data is the correct model.
+
+Proof. [Sketch] Let us first consider the case where the observed data have no noise. On such an input, we observe that the prima facie graph has no false negatives: in fact $\mathcal{P}(e \wedge c) = \mathcal{P}(e)$ $\forall [c \rightarrow e]$ modelling a genuine causal relation, , thus the probability raising constraint is satisfied, so it is the temporal priority given that we assumed $r$ to be correct and complete.
+
+Furthermore, it is know that the likelihood fit performed by $BIC$ converges to a class of structures equivalent in terms of likelihood among which there is the correct model: all these topologies are the same unless the directionality of some edges. But, since we started with the prima facie graph which is already ordered by temporal priority, we can conclude that in this case the $SBCN$ coincides with the correct model.
+
+To extend the proof to the case of data uniformly randomly corrupted by false positives and negatives rates (in $[0,1)$ ), we note that the marginal and joint probabilities change monotonically as a consequence of the assumption that the noise is uniform. Thus, all inequalities used in the preceding proof still hold, which concludes the proof.
+
+In the more general case of causal topologies where any cause of a common effect is independent from any other cause (i.e., we relax the assumption of conjunctive parents), the $SBCN$ is not guaranteed to converge to the correct model but it coincides with a subset of it modeling all the edges representing statistically relevant causal relations (i.e., where the probability raising condition is verified).
+
+## 4. Discrimination Discovery by Random Walks
+
+In this section we propose several random-walk-based methods over the reconstructed SBCN , to deal with different discrimination-detection tasks.
+
+### 4.1 Group discrimination and favoritism
+
+The basic problem in the analysis of direct discrimination is precisely to quantify the degree of discrimination suffered by a given protected group (e.g., an ethnic group) with respect to a decision (e.g., loan denial). In contrast to discrimination, favoritism refers to the case of an individual treated better than others for reasons not related to individual merit or business necessity: for instance, favoritism in the workplace might result in a person being promoted faster than others unfairly. In the following we denote favoritism as positive discrimination in contrast with negative discrimination.
+
+Given an SBCN we define a measure of group discrimination (either negative or positive) for each node $v \in V$ . Recall that each node represents a pair $\langle  attribute = value \rangle$ , so it is $\langle  gender = female \rangle$ essentially what we refer to as a group, e.g., . Our task is to assign a score of discrimination $ds^-: V \rightarrow [0,1]$ to each node, so that the closer $ds^-(v)$ is to 1 the more discriminated is the group represented by $v$ .
+
+We compute this score by means of a number $n$ of random walks that start from $v$ and reaches either the node representing the positive decision or the one representing the negative decision. In these random walks the next step is chosen proportionally to the weights of the out-going arcs. Suppose a random walk has reached a node $u$ , and let $deg_{out}(u)$ denote the set of outgoing arcs from $u$ . Then the arc $(u,z)$ is chosen with probability
+
+$$p(u, z)=\frac{W(u, z)}{\sum_{\varepsilon \in \operatorname{degout}(u)} W(e)} .$$
+
+When a random walk ends in a node with no outgoing arc before reaching either the negative or the positive decision, it is restarted from the source node $v$ .
+
+Definition 3 (Group discrimination score). Given an $SBCN = (V, E^*, W)$ , let $\delta^- \in V$ and $\delta^+ \in V$ denote the nodes indicating the negative and positive decision, respectively. Given a node $v \in V$ , and a number $n \in \mathbb{N}$ of random walks to be performed, we denote as $rw_{v \to \delta^-}$ the number of random walks started at node $v$ that reach $\delta^-$ earlier than $\delta^+$ . The discrimination score for
+
+---
+
+the group corresponding to node $v$ is then defined as
+
+$$d s^{-}(v)=\frac{r w_{v \to \delta^{-}}}{n} .$$
+
+This implicitly also defines a score of positive discrimination (or favoritism): $ds^+(v) = 1 - ds^-(v)$ .
+
+Taking advantage of the SBCN we also propose two additional measures capturing how far a node representing a group is from the positive and negative decision respectively. This is done by computing the average number of steps that the random walks take to reach the two decisions: we denote these scores as $as^-(v)$ and $as^+(v)$ .
+
+### 4.2 Indirect discrimination
+
+The European Union Legislation [2] provides a broad definition of indirect discrimination as occurring “where an apparently neutral provision, criterion or practice would put persons of a racial or ethnic origin at a particular disadvantage compared with other persons”. In other words, the actual result of the apparently neutral provision is the same as an explicitly discriminatory one. A typical legal case study of indirect discrimination is concerned with redlining : e.g., denying a loan because of ZIP code, which in some areas is an attribute highly correlated to race. Therefore, even if the attribute race cannot be required at loan-application time (thus would not be present in the data), still race discrimination is perpetrated. Indirect discrimination discovery refers to the data mining task of discovering the attributes values that can act as a proxy to the protected groups and lead to discriminatory decisions indirectly [13, 14, 11] .
+
+In our setting, indirect discrimination can be detected by applying the same method described in Section 4.1 .
+
+### 4.3 Genuine requirement
+
+The legal concept of genuine requirement refers to detecting that part of the discrimination which may be explained by other, legally-grounded, attributes; e.g., denying credit to women may be explainable by the fact that most of them have low salary or delay in returning previous credits. A typical example in the literature is the one of the “ genuine occupational requirement ” , also called “ business necessity ” in [29, 30] . In the state of the art of data mining methods for discrimination discovery, it is also known as explainable discrimination [31] and conditional discrimination [18] .
+
+The task here is to evaluate to which extent the discrimination apparent for a group is “ explainable ” on a legal ground. Let $v \in V$ be the node representing the group which is suspected of being discriminated, and $u_l \in V$ be a node whose causal relation with a negative or positive decision is legally grounded. As before, $\delta^-$ and $\delta^+$ denote the negative and positive decision, respectively. Following the same random-walk process described in Section 4.1 , we define the fraction of explainable discrimination for the group $v$ :
+
+$$f e d^{-}(v)=\frac{r w_{v \to u_{l} \to \delta^{-}}}{r w_{v \to \delta^{-}}},$$
+
+i.e., the fraction of random walks passing trough $u_1$ among the ones started in $v$ and reaching $\delta^-$ earlier than $\delta^+$ . Similarly we define $fed^+(v)$ , i.e., the fraction of explainable positive discrimination.
+
+### 4.4 Individual and subgroup discrimination
+
+Individual discrimination requires to measure the amount of discrimination for a specific individual, i.e., an entire record in the database. Similarly, subgroup discrimination refers to discrimination against a subgroup described by a combination of multiple protected and non-protected attributes: personal data, demographics, social, economic and cultural indicators, etc. For example, consider the case of gender discrimination in credit approval: although an analyst may observe that no discrimination occurs in general, it may turn out that older women obtain car loans only rarely.
+
+Both problems can be handled by generalizing the technique introduced in Section 4.1 to deal with a set of starting nodes, instead of only one. Given an $SBCN = (V, E^*, W)$ let $v_1, \ldots, v_n$ be the nodes of interest. In order to define a discrimination score for $v_1, \ldots, v_n$ , we perform a personalized PageRank computation with respect to $v_1, \ldots, v_n$ . In personalized PageRank, the probability of jumping to a node when abandoning the random walk is not uniform, but it is given by a vector of probabilities for each node. In our case the vector will have the value $\frac{1}{n}$ for each of the nodes $v_1, \ldots, v_n \in V$ and zero for all the others. The output of personalized PageRank is a score $ppr(u|v_1, ..., v_n)$ of proximity/relevance to $\{v_1, ..., v_n\}$ for each other node $u$ in the network. In particular, we are interested in the score of the nodes representing the negative and positive decision: i.e., $ppr(\delta^-|v_1, ..., v_n)$ and $ppr(\delta^+|v_1, ..., v_n)$ respectively.
+
+Definition 4 (Generalized discrimination score). Given an $SBCN = (V, E^*, W)$ , let $\delta^- \in V$ and $\delta^+ \in V$ denote the nodes indicating the negative and positive decision, respectively. Given a set of nodes $v_1, ..., v_n \in V$ , we define the generalized (negative) discrimination score for the subgroup or the individual represented by $\{v_1, ..., v_n\}$ as
+
+$$g d s^{-}\left(v_{1}, ..., v_{n}\right)=\frac{p p r\left(\delta^{-} \mid v_{1}, ..., v_{n}\right)}{p p r\left(\delta^{-} \mid v_{1}, ..., v_{n}\right)+p p r\left(\delta^{+} \mid v_{1}, ..., v_{n}\right)} .$$
+
+This implicitly also defines a generalized score of positive discrimination: $gds^+(v_1,...,v_n)=1-gds^-(v_1,...,v_n)$ .
+
+## 5. Experimental Evaluation
+
+This section reports the experimental evaluation of our approach on four datasets, Adult , German credit and census-income from the UCI Repository of Machine Learning Databases 6 , and Berkeley Admissions Data from [34] . These are well-known real-life datasets typically used in discrimination-detection literature.
+
+Adult: consists of 48,842 tuples and 10 attributes, where each tuple correspond to an individual and it is described by personal attributes such as age, race, sex, relationship, education, employment, etc. Following the literature, in order to define the decision attribute we use the income levels, $\leq $ 50K (negative decision) or $>$ 50K (positive decision). We use four levels in the partial order for temporal priority: age, race, sex, and native country are defined in the first level; education, marital status, and relationship are defined in the second level; occupation and work class are defined in the third class, and the decision attribute (derived from income) is the last level.
+
+6http://archive.ics.uci.edu/ml
+
+---
+
+Table 1: SBCN main characteristics.
+
+<table><tr><td>\textit{Dataset}</td><td>|V|</td><td>|A|</td><td>\textit{avgDeg}</td><td>\textit{maxInDeg}</td><td>\textit{maxOutDeg}</td></tr><tr><td>Adult</td><td>92</td><td>230</td><td>2.5</td><td>7</td><td>19</td></tr><tr><td>German credit</td><td>73</td><td>102</td><td>1.39</td><td>3</td><td>7</td></tr><tr><td>Census-income</td><td>386</td><td>1426</td><td>3.69</td><td>8</td><td>54</td></tr></table>
+
+
+![Figure](figures/figure_002.png)
+
+Figure 2: Distribution of edge scores.
+
+German credit : consists of 1000 tuples with 21 attributes on bank account holders applying for credit. The decision attribute is based on repayment history, i.e., whether the customer is labeled with good or bad credit risk. Also for this dataset the partial order for temporal priority has four orders. Personal attributes such as gender, age, foreign worker are defined in the first level. Personal attributes such as employment status and job status are defined in the second level. Personal properties such as savings status and credit history are defined in the third level, and finally the decision attribute is the last level.
+
+Census-income : consists of 299,285 tuples and 40 attributes, where each tuple correspond to an individual and it is described by demographic and employment attributes such as age, sex, relationship, education, employment, ext. Similar to Adult dataset, the decision attribute is the income levels and we define four levels in the partial order for temporal priority.
+
+Building the SBCN just take a handful of seconds in German credit and Adult , and few minutes in Census-income on a commodity laptop. The main characteristics of the extracted SBCN are reported in Table 1 , while the distribution of the edges scores $W(e)$ is plotted in Figure 2 .
+
+As discussed in the Introduction we also use the dataset from the famous 1973 episode at University of California at Berkeley, in order to highlight the differences between correlation-based and causation-based methods.
+
+Berkeley Admissions Data : consists of 4,486 tuples and three attributes, where each tuple correspond to an individual and it is described by the gender of applicants and the department that they apply for it. For this dataset the partial order for temporal priority has three orders. Gender is defined in the first level, department in the second level, and finally the decision attribute is the last level. Table 2 is a three-way table that presents admissions data at the University of California, Berkeley in 1973 according to the variables department (A, B, C, D, E), gender (male, female), and outcome (admitted, denied). The table is adapted from data in the text by Freedman, et al. [34] .
+
+<table><tr><td colspan="2">Male</td><td colspan="2">Female</td><td rowspan="2">Department</td></tr><tr><td>Admitted</td><td>Denied</td><td>Admitted</td><td>Denied</td></tr><tr><td>512</td><td>313</td><td>89</td><td>19</td><td>A</td></tr><tr><td>313</td><td>207</td><td>17</td><td>8</td><td>B</td></tr><tr><td>120</td><td>205</td><td>202</td><td>391</td><td>C</td></tr><tr><td>138</td><td>279</td><td>131</td><td>244</td><td>D</td></tr><tr><td>53</td><td>138</td><td>94</td><td>299</td><td>E</td></tr><tr><td>22</td><td>351</td><td>24</td><td>317</td><td>F</td></tr></table>
+
+
+Table 2: Berkeley Admission Data
+
+### 5.1 Community detection on the SBCN
+
+Given that our SBCN is a directed graph with edge weight, as a first characterization we try to partition it using a random-walks-based community detection algorithm, called Walktrap and proposed in [33] , whose unique parameter is the maximum number of steps in a random walk (we set it to 8), and which automatically identifies the right number of communities. The idea is that short random walks tend to stay in the same community (densely connected area of the graph). Using this algorithm over the reconstructed SBCN from Adult dataset, we obtain 5 communities: two larger ones and three smaller ones (reported in Table 3 ). Interestingly, the two larger communities seem built around the negative ( $C_1$ ) and the positive ( $C_2$ ) decisions.
+
+Figure 1 in Section 3 shows the subgraph of the SBCN corresponding to $C_2$ (that we can call, the favoritism cluster): we note that such cluster also contains nodes such as sex _ Male , age _ old , relationship _ Husband . The other large community $C_1$ , can be considered the discrimination cluster: beside the negative decision it contains other nodes representing disadvantaged groups such as sex _ Female , age _ young , race _ Black , marital _ status _ Never _ married . This good separability of the SBCN in the two main clusters of discrimination and favoritism, highlights the goodness of the causal structure captured by the SBCN .
+
+### 5.2 Group discrimination and favoritism
+
+We next focus on assessing the discrimination score $ds^-$ we defined in Section 4.1 , as well as the average number of steps that the random walks take to reach the negative and positive decisions, denote $as^-(v)$ and $as^+(v)$ respectively.
+
+Tables 4 , 5 and 6 report the top-5 and bottom-5 nodes w.r.t. the discrimination score $ds^-$ , for datasets Adult , German and Census-income , respectively. The first and most important observation is that our discrimination score provides a very clear signal, with some disadvantaged groups having very high discrimination score (equal to 1 or very close), and similarly clear signals of favoritism, with groups having $ds^-(v) = 0$ , or equivalently $ds^+(v) = 1$ . This is more clear in the Adult dataset, where the positive and negative decisions are artificially derived from the income attribute. In the German credit dataset, which is more realistic as the decision attribute is truly about credit, both discrimination and favoritism are less palpable. This is also due to the fact that German credit contains less proper causal relations, as reflected in the higher sparsity of the SBCN . A consequence of this sparsity is also that the random walks generally need more steps to reach one of the two decisions. In Censusincome dataset, we observe favoritism with respect to married and asian-pacific individuals.
+
+---
+
+Table 3: Communities found in the SBCN extracted from the Adult dataset by Walktrap [33] . In the table the attributes are shortened as in parenthesis: age (ag), education (ed), marital status (ms), native country (nc), occupation (oc), race(ra), relationship (re), sex (sx), workclass (wc).
+
+<table><tr><td>\textbf{$C_1$}</td></tr><tr><td>\textbf{negative\_dec}, wc:Private, ed:Some\_college, ed:Assoc\_acdm,</td></tr><tr><td>\textbf{ms:Never\_married}, ms:Divorced, \textbf{ms:Widowed},</td></tr><tr><td>\textbf{ms:Married\_AF\_spouse}, oc:Sales, oc:Other\_service,</td></tr><tr><td>oc:Priv\_house\_serv, re:Own\_child, re:Not\_in\_family, re:Wife,</td></tr><tr><td>\textbf{re:Unmarried}, re:Other\_relative, \textbf{ra:Black}, oc:Armed\_Forces,</td></tr><tr><td>oc:Handlers\_cleaners, oc:Tech\_support, oc:Transport\_moving,</td></tr><tr><td>ed:7th\_8th, ed:10th, ed:12th, ms:Separated,</td></tr><tr><td>ed:HS\_grad,ed:11th, nc:Outlying\_US\_Guam\_USVLetc,</td></tr><tr><td>nc:Haiti, \textbf{ag:young}, \textbf{sx:Female}, ra:Amer\_Indian\_Eskimo,</td></tr><tr><td>nc:Trinadad\_Tobago, nc:Jamaica, oc:Machine\_op\_inspect,</td></tr><tr><td>ms:Married\_spouse\_absent, oc:Adm\_clerical,</td></tr><tr><td>\textbf{$C_2$}</td></tr><tr><td>\textbf{positive\_dec}, oc:Prof\_specialty, wc:Self\_emp\_not\_inc,</td></tr><tr><td>ms:Married\_civ\_spouse, oc:Craft\_repair,oc:Protective\_serv,</td></tr><tr><td>\textbf{re:Husband}, ed:Prof\_school, wc:Self\_emp\_inc,</td></tr><tr><td>\textbf{ag:old} , wc:Local\_gov, \textbf{oc:Exec\_managerial},</td></tr><tr><td>ed:Bachelors, ed:Assoc\_voc, ed:Masters, wc:Never\_worked,</td></tr><tr><td>wc:State\_gov, ed:Doctorate, \textbf{sx:Male}, nc:India, nc:Cuba</td></tr><tr><td>\textbf{$C_3$}</td></tr><tr><td>oc:Farming\_fishing, wc:Without\_pay, nc:Mexico, nc:Canada,</td></tr><tr><td>nc:Italy, nc:Guatemala, nc:El\_Salvador, ra:White,</td></tr><tr><td>nc:Poland, ed:1st\_4th, ed:9th\_ed:Preschool, ed:5th\_6th</td></tr><tr><td>\textbf{$C_4$}</td></tr><tr><td>nc:Iran, nc:Puerto\_Rico, nc:Dominican\_Republic,</td></tr><tr><td>nc:Columbia, nc:Peru, nc:Nicaragua, ra:Other</td></tr><tr><td>\textbf{$C_5$}</td></tr><tr><td>nc:Philippines, nc:Cambodia, nc:China, nc:South,</td></tr><tr><td>nc:Japan, nc:Taiwan, nc:Hong, nc:Laos, nc:Thailand,</td></tr><tr><td>nc:Vietnam, ra:Asian\_Pac\_Islander</td></tr></table>
+
+
+### 5.3 Genuine requirement
+
+We next focus on genuine requirement (or explainable discrimination). Table 7 reports some examples of fraction of explainable discrimination (both positive and negative) on the Adult dataset. We can see how some fractions of discrimination against protected groups, can be “ explained ” by intermediate nodes such as having a low education profile, or a simple job. In the case these intermediate nodes are considered legally grounded, then one cannot easily file a discrimination claim.
+
+Similarly, we can observe that the favoritism towards groups such as married men, is explainable, to a large extent, by higher education and good working position, such as managerial or executive roles.
+
+### 5.4 Subgroup and Individual Discrimination
+
+We next turn our attention to subgroup and individual discrimination discovery. Here the problem is to assign a score of discrimination not to a single node (a group), but to multiple nodes (representing the attributes of an individual or a subgroup of citizens). In Section 4.4 we have introduced based on the PageRank of the positive and negative decision, $ppr(\delta^+)$ and $ppr(\delta^-)$ respectively, personalized on the nodes of interest. Figure 3 presents a scatter plot of $ppr(\delta^+)$ versus $ppr(\delta^-)$ for each individual in the German credit dataset. We can observe the perfect separation between individuals corresponding to a high personalized PageRank with respect to the positive decision, and those associated with a high personalized PageRank relative to the negative decision.
+
+Table 4: Top-5 and bottom-5 groups by discrimination score $ds^-(v)$ in Adult dataset.
+
+<table><tr><td></td><td>$ds^-(v)$</td><td>$as^-(v)$</td><td>$as^+(v)$</td></tr><tr><td>relationship\_Unmarried</td><td>1</td><td>1.164</td><td>-</td></tr><tr><td>marital\_status\_Never\_married</td><td>0.996</td><td>1.21</td><td>2.14</td></tr><tr><td>age\_Young</td><td>0.995</td><td>2.407</td><td>3.857</td></tr><tr><td>race\_Black</td><td>0.994</td><td>2.46</td><td>4.4</td></tr><tr><td>sex\_Female</td><td>0.98</td><td>2.60</td><td>3.76</td></tr><tr><td></td><td>$ds^-(v)$</td><td>$as^-(v)$</td><td>$as^+(v)$</td></tr><tr><td>relationship\_Husband</td><td>0</td><td>-</td><td>2</td></tr><tr><td>marital\_status\_Married\_civ\_spouse</td><td>0</td><td>-</td><td>2.06</td></tr><tr><td>sex\_Male</td><td>0</td><td>-</td><td>3.002</td></tr><tr><td>native\_country\_India</td><td>0.002</td><td>4.0</td><td>3.25</td></tr><tr><td>age\_Old</td><td>0.018</td><td>2.062</td><td>2.14</td></tr></table>
+
+
+Table 5: Top-5 and bottom-4 groups by discrimination score $ds^-(v)$ in German credit. We report only the bottom-4, because there are only 4 nodes in which $ds^+(v) > ds^-(v)$ .
+
+<table><tr><td>residence\_since\_le\_1d6</td><td>1</td><td>6.0</td><td>-</td></tr><tr><td>residence\_since\_gt\_2d8</td><td>1</td><td>2.23</td><td>-</td></tr><tr><td>residence\_since\_from\_1d6\_le\_2d2</td><td>1</td><td>6.0</td><td>-</td></tr><tr><td>age\_gt\_52d6</td><td>0.86</td><td>3.68</td><td>4.0</td></tr><tr><td>personal\_status\_male\_single</td><td>0.791</td><td>5.15</td><td>5.0</td></tr><tr><td></td><td>$ds^-(v)$</td><td>$as^-(v)$</td><td>$as^+(v)$</td></tr><tr><td>job\_unskilled\_resident</td><td>0</td><td>-</td><td>2.39</td></tr><tr><td>personal\_status\_male\_mar\_or\_wid</td><td>0.12</td><td>8.0</td><td>4.4</td></tr><tr><td>age\_le\_30d2</td><td>0.186</td><td>7.0</td><td>3.34</td></tr><tr><td>personal\_status\_female\_div\_or\_sep\_or\_mar</td><td>0.294</td><td>6.48</td><td>4.4</td></tr></table>
+
+
+Table 6: Top-5 and bottom-5 groups by discrimination score $ds^-(v)$ in Census-income dataset.
+
+<table><tr><td></td><td>$ds^-(v)$</td><td>$as^-(v)$</td><td>$as^+(v)$</td></tr><tr><td>MIGSAME\_Not\_in\_universe\_under\_1\_year\_old</td><td>0.71</td><td>4.09</td><td>8.82</td></tr><tr><td>WKSWORK\_94\_5\_inf</td><td>0.625</td><td>3.0</td><td>6.76</td></tr><tr><td>AWKSTAT\_Not\_in\_labor\_force</td><td>0.59</td><td>2.0</td><td>6.16</td></tr><tr><td>VETYN\_0.5\_20.5</td><td>0.58</td><td>1.01</td><td>5.17</td></tr><tr><td>MARSUPWT\_3188\_455\_4277\_98</td><td>0.55</td><td>5.0</td><td>9.25</td></tr><tr><td></td><td>$ds^-(v)$</td><td>$as^-(v)$</td><td>$as^+(v)$</td></tr><tr><td>AHGA\_Doctorate\_degreePhD\_EdD</td><td>0</td><td>-</td><td>3.07</td></tr><tr><td>AMARITL\_Married\_A.F\_spouse\_present</td><td>0</td><td>-</td><td>4.49</td></tr><tr><td>AMJOCC\_Sales</td><td>0</td><td>-</td><td>2.0</td></tr><tr><td>ARACE\_Asian\_or\_Pacific\_Islander</td><td>0</td><td>-</td><td>6.47</td></tr><tr><td>VETYN\_20\_5\_32\_5</td><td>0</td><td>-</td><td>5.89</td></tr></table>
+
+
+Table 7: Fraction of explainable discrimination for some exemplar pair of nodes in the Adult dataset.
+
+<table><tr><td>Source node</td><td>Intermediate</td><td> $fed^{-}(v)$ </td></tr><tr><td>race.Amer.Indian.Eskimo</td><td>education.HS_grad</td><td>0.481</td></tr><tr><td>sex.Female</td><td>occupation.Other_service</td><td>0.310</td></tr><tr><td>age_Young</td><td>occupation.Other_service</td><td>0.193</td></tr><tr><td>relationship_Unmarried</td><td>education.HS_grad</td><td>0.107</td></tr><tr><td>race_Black</td><td>education.11th</td><td>0.083</td></tr><tr><td>Source node</td><td>Intermediate</td><td> $fed^{+}(v)$ </td></tr><tr><td>relationship_Husband</td><td>occupation.Exec.managerial</td><td>0.806</td></tr><tr><td>sex_Male</td><td>occupation.Exec.managerial</td><td>0.587</td></tr><tr><td>native_country_Iran</td><td>education.Bachelors</td><td>0.480</td></tr><tr><td>native_country_India</td><td>education.Prof.school</td><td>0.415</td></tr><tr><td>age_Old</td><td>occupation.Exec.managerial</td><td>0.39</td></tr></table>
+
+
+Such good separation is also reflected in the generalized discrimination score (Definition 4 ) that we obtain by combining $ppr(\delta^+)$ versus $ppr(\delta^-)$ .
+
+---
+
+![Figure](figures/figure_000.png)
+
+Figure 3: Scatter plot of $ppr(\delta^+)$ versus $ppr(\delta^-)$ for each individual in the German credit dataset.
+
+![Figure](figures/figure_002.png)
+
+Figure 4: Individual discrimination: histogram representing the distribution of the values of the generalized discrimination score $gds^-$ for the population of the German credit dataset.
+
+In Figure 4 we report the distribution of the generalized discrimination score $gds^-$ for the population of the German credit dataset: we can make a note of the clear separation between the two subgroups of the population.
+
+In the Adult dataset (Figure 5 ) we do not observe the same neat separation in two subgroups as in the German credit dataset, also due to the much larger number of points. Nevertheless, as expected, $ppr(\delta^+)$ and $ppr(\delta^-)$ still exhibit anticorrelation. In Figure 5 we also use colors to show two different groups: red dots are for age Young and blue dots are for age Old individuals. As expected we can see that the red dots are distributed more in the area of higher $ppr(\delta^-)$ .
+
+The plots in Figure 6 have a threshold $t \in [0,1]$ on the X-axis, and the fraction of tuples having $gds^-() \geq t$ on the Y-axis, and they show this for different subgroups. The first plot, from the Adult dataset, shows the group female, young, and young female. As we can see the individuals that are both young and female have a higher generalized discrimination score. Similarly, the second plot shows the groups old, single male, and old single male from the German credit dataset. Here we can observe much lower rates of discrimination with only 1/5 of the corresponding populations having $gds^-() \geq 0.5$ , while in the previous plot it was more than 85 % .
+
+![Figure](figures/figure_007.png)
+
+Figure 5: Individual discrimination: scatter plot of $ppr(\delta^+)$ versus $ppr(\delta^-)$ for each individual in the Adult dataset. Red dots are for age Young and blue dots are for age Old .
+
+![Figure](figures/figure_009.png)
+
+Figure 6: Subgroup discrimination: plots reporting a threshold $t \in [0,1]$ on the X-axis, and the fraction of tuples having $gds^-() \geq t$ on the Y-axis. The top plot is from Adult, while the bottom is from German credit.
+
+### 5.5 Comparison with prior art
+
+In this section, we discuss examples in which our causation-based method draws different conclusions from the correlation-based methods presented in [13, 14, 15] using the same datasets and the same protected groups $^7$ .
+
+The first example involves the foreign _ worker group from German Credit dataset, whose contingency table is reported in Figure 7 . Following the approaches of [13, 14, 15] the foreign _ worker group results strongly discriminated. In fact Figure 7 shows an $RD$ value ( risk difference ) of 0.244 which is considered a strong signal: in fact $RD>0$ is already considered discrimination [15] .
+
+However, we can observe that the foreign_worker group is per se not very significant, as it contains 963 tuples out of
+
+$^{7}$ We could not compare with [22] due to repeatability issues.
+
+---
+
+![Figure](figures/figure_000.png)
+
+Figure 7: Contingency table for foreign _ worker in the German credit dataset.
+
+![Figure](figures/figure_002.png)
+
+Figure 8: Contingency table for race_black in the Adult dataset.
+
+1000 total. In fact our causal approach does not detect any discrimination with respect to foreign _ worker which appears as a disconnected node in the SBCN.
+
+The second example is in the opposite direction. Consider the race _ black group from Adult dataset whose contingency table is shown in Figure 8 . Our causality-based approach detects a very strong signal of discrimination ( $ds^-()=0.994$ ), while the approaches of [13, 14, 15] fail to discover discrimination against black minority when the value of minimum support threshold used for extracting classification rules is more than 10 % . On the other hand, when such minimum support threshold is kept lower, the number of extracted rules might be overwhelming.
+
+Finally, we turn our attention to the famous example of false-positive discrimination case happened at Berkeley in 1973, that we discussed in Section 1 . Figure 9 presents the SBCN extracted by our approach from Berkeley Admission Data . Interestingly, we observe that there is no direct edge between node sex _ Female and Admission _ No . And sex _ Female is connected to node Admission _ No through nodes of Dep _ C , Dep _ D , Dep _ E , and Dep _ F , which are exactly the departments that have lower admission rate. By running our random walk-based methods over SBCN we obtain the value of 1 for the score of explainable discrimination confirming that apparent discrimination in this dataset is due the fact that women tended to apply to departments with lower rates of admission.
+
+Similarly, we observe that there is no direct edge between node sex _ Male and Admission _ Yes . And sex _ Male is connected to node Admission _ Yes through nodes of Dep _ A , and Dep _ B , which are exactly the departments that have higher admission rate. By running our random walk-based methods over SBCN we obtain the value of 1 for the score of explainable discrimination confirming that apparent favoritism towards men is due to the fact that men tended to apply to departments with higher rates of admission.
+
+However, following the approaches of [13, 14, 15] , the contingency table shown in Figure 10 can be extracted from Berkeley Admission Data . As shown in Figure 10 , the value of $RD$ suggests a signal of discrimination versus women.
+
+![Figure](figures/figure_009.png)
+
+Figure 9: The SBCN constructed from Berkeley Admission Data dataset.
+
+![Figure](figures/figure_011.png)
+
+Figure 10: Contingency table for female in the Berkeley Admission Data dataset.
+
+This highlights once more the pitfalls of correlation-based approaches to discrimination detection and the need for a principled causal approach, as the one we propose in this paper.
+
+## 6. CONCLUSIONS
+
+Discrimination discovery from databases is a fundamental task in understanding past and current trends of discrimination, in judicial dispute resolution in legal trials, in the validation of micro-data before they are publicly released. While discrimination is a causal phenomenon, and any discrimination claim requires to prove a causal relationship, the bulk of the literature on data mining methods for discrimination detection is based on correlation reasoning.
+
+In this paper, we propose a new discrimination discovery approach that is able to deal with different types of discrimination in a single unifying framework. It is the first discrimination detection method grounded in probabilistic causal theory. We define a method to extract a graph representing the causal structures found in the database, and then we propose several random-walk-based methods over the causal structures, addressing a range of different discrimination problems.
+
+Our experimental assessment confirmed the great flexibility of our proposal in tackling different aspects of the discrimination detection task, and doing so with very clean signals, clearly separating discrimination cases.
+
+Repeatability. Our software together with the datasets used in the experiments are available at http://bit.ly/ 1GizSIG .
+
+## 7. REFERENCES
+
+[1] Australian Legislation, "Equal opportunity act-victoria state,(b) anti-discrimination act-queensland state, 2008."
+
+---
+
+[2] EU Legislation, "(a) race equality directive,2000/43/ec, 2000; (b) employment equality directive, 2000/78/ec, 2000; (c) equal treatment of persons, european parliament legislative resolution, 2009."
+
+[3] S. R. Foster, "Causation in antidiscrimination law: Beyond intent versus impact," Hous. L. Rev., vol. 41, p. 1469, 2004.
+
+[4] M. Dabady, R. M. Blank, C. F. Citro et al. , Measuring racial discrimination . National Academies Press, 2004.
+
+[5] P. J. Bickel, E. A. Hammel, and J. W. O'Connell, "Sex bias in graduate admissions: Data from berkeley," Science , vol. 187, no. 4175, 1975.
+
+[6] C. Hitchcock, “Probabilistic causation,” in The Stanford Encyclopedia of Philosophy , 2012.
+
+[7] P. Suppes, A probabilistic theory of causality. North-Holland Publishing Company, 1970.
+
+[8] G. Schwarz, "Estimating the dimension of a model," Annals of Statistics, vol. 6, no. 2, pp. 461-464, 1978.
+
+[9] B. Custers, T. Calders, B. Schermer, and T. Zarsky, Discrimination and Privacy in the Information Society: Studies in Applied Philosophy, Epistemology and Rational Ethics . Springer, 2012, vol. 3.
+
+[10] A. Romei and S. Ruggieri, “A multidisciplinary survey on discrimination analysis,” The Knowledge Engineering Review , pp. 1–57, 2013.
+
+[11] S. Hajian and J. Domingo-Ferrer, “A methodology for direct and indirect discrimination prevention in data mining,” TKDE , 25(7), 2013.
+
+[12] F. Kamiran and T. Calders, “Data preprocessing techniques for classification without discrimination,” KAIS , 33(1), 2012.
+
+[13] D. Pedreshi, S. Ruggieri, and F. Turini, “Discrimination-aware data mining,” in KDD , 2008.
+
+[14] D. Pedreschi, S. Ruggieri, and F. Turini, “Measuring discrimination in socially-sensitive decision records.” in SDM , 2009.
+
+[15] S. Ruggieri, D. Pedreschi, and F. Turini, “Data mining for discrimination discovery,” TKDD , 4(2), 2010.
+
+[16] B. T. Luong, S. Ruggieri, and F. Turini, “k-nn as an implementation of situation testing for discrimination discovery and prevention,” in KDD , 2011.
+
+[17] I. Rorive, “Proving discrimination cases: The role of situation testing,” 2009.
+
+[18] I. Zliobaite, F. Kamiran, and T. Calders, “Handling conditional discrimination,” in ICDM , 2011.
+
+[19] C. Dwork, M. Hardt, T. Pitassi, O. Reingold, and R. Zemel, "Fairness through awareness," in ITCS , 2012.
+
+[20] D. Pedreschi, S. Ruggieri, and F. Turini, “Integrating induction and deduction for finding evidence of discrimination,” in ICAIL , 2009.
+
+[21] S. Ruggieri, S. Hajian, F. Kamiran and X. Zhang, “Anti-discrimination Analysis Using Privacy Attack Strategies,” in PKDD , 2014.
+
+[22] K. Mancuhan and C. Clifton, “Combating discrimination using bayesian networks,” Artificial Intelligence and Law , 22(2), 2014.
+
+[23] D. Koller and N. Friedman, Probabilistic graphical models: principles and techniques . MIT press, 2009.
+
+[24] P. Menzies, “Counterfactual theories of causation,” in The Stanford Encyclopedia of Philosophy , spring 2014 ed., E. N. Zalta, Ed., 2014.
+
+[25] J. Woodward, “Causation and manipulability,” in The Stanford Encyclopedia of Philosophy , winter 2013 ed., E. N. Zalta, Ed., 2013.
+
+[26] E. Brenner and D. Sontag, “Sparsityboost: A new scoring function for learning bayesian network structure,” arXiv:1309.6820 , 2013.
+
+[27] L. O. Loohuis, G. Caravagna, A. Graudenzi, D. Ramazzotti, G. Mauri, M. Antoniotti, and B. Mishra, “Inferring tree causal models of cancer progression with probability raising,” PLoS ONE , 9(12), 2014.
+
+[28] D. Ramazzotti et al. “Capri: Efficient inference of cancer progression models from cross-sectional data,” Bioinformatics , 31(18), 2015.
+
+[29] US Federal Legislation, “(a) equal credit opportunity act, (b) fair housing act, (c) intentional employment discrimination, (d) equal pay act, (e) pregnancy discrimination act, (f) civil right act., 2010.”
+
+[30] E. Ellis and P. Watson, EU anti-discrimination law . Oxford University Press, 2012.
+
+[31] S. Hajian, J. Domingo-Ferrer, A. Monreale, D. Pedreschi, and F. Giannotti, “Discrimination-and privacy-aware patterns,” DAMI , 29(6), 2015.
+
+[32] G. Jeh and J. Widom, "Scaling personalized web search," in WWW 2003 .
+
+[33] P. Pons and M. Latapy, “Computing communities in large networks using random walks,” in ISCIS , 2005.
+
+[34] D. Feedman, R. Pisani, and R. Purves, "Statistics," 3rd Edition, WW Norton & Company, New York (1998).
+

processed/00087_W2596828501/document.md

@@ -0,0 +1,737 @@
+理想研織究智
+
+Management and Organization Review
+
+Management and Organization Review 13:1, March 2017, 15-38 doi: 10.1017/mor.2016.53
+
+理組研織究管
+
+# History and the Debate Over Intellectual Property
+
+Mike W. Peng, 1 David Ahlstrom, 2 Shawn M. Carraher, 1 and Weilei (Stone) Shi3
+
+$^{1}$ University of Texas at Dallas, USA, $^{2}$ The Chinese University of Hong Kong, Hong Kong, and
+
+$^{3}$ City University of New York, USA
+
+ABSTRACT This article responds to recent calls for organizational research to address larger, more globally relevant questions and to pay attention to history, by analyzing the crucial debate over intellectual property rights (IPR) between the United States and China. Despite the recent US position, the United States has not always been a leading IPR advocate. Rather, it was a leading IPR violator during the nineteenth century. An institution-based view of IPR history suggests that both the US refusal to protect foreign IPR in the nineteenth century and the current Chinese lack of enthusiasm to meet US IPR demands represent rational choices. However, as cost-benefit considerations change institutional transitions are possible. We predict that to the same extent the United States voluntarily agreed to strengthen IPR protection when its economy became sufficiently innovation-driven, China will similarly improve its IPR protection.
+
+KEYWORDS China, history, institution-based view, intellectual property rights (IPR), United States
+
+## INTRODUCTION
+
+A leading debate in contemporary US-China relations centers on intellectual property rights (IPR). Given the scope and scale of potential IPR violations in China, [1] the United States has frequently sought improved IPR protection in China and has been frustrated by the lack of progress, in spite of recent improvements. What does the future of IPR protection in China hold? Despite the obvious importance of the IPR debate, it has attracted little attention from organizational scholars – and little coverage in the pages of Management and Organization Review (MOR). Do organizational scholars such as those in the MOR/International Association for Chinese Management Research (IACMR) community have anything to contribute to the debate over IPR?
+
+Responding to recent calls for organizational research to address larger, more globally relevant questions (Davis, 2015 ; Lewin, 2015 ) and to pay attention to history (Ahlstrom, Lamond, & Ding, 2009 ; Barley, 2016 ) , we argue that a key to predicting the future development of IPR lies in a deeper understanding of
+
+c 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+16
+
+M. W. Peng et al.
+
+history, which can help to clear up misconceptions. For example, many take for granted that the United States is a leading IPR champion and China a leading IPR violator. Yet ironically, as recently as in the nineteenth century, the United States was a leading IPR violator (Raustiala & Sprigman, 2012 , 2013 ) . Between the founding of the United States in 1776 and the enactment of the Chace Act in 1891, pirating British publications (such as books) and entertainment (such as stage plays) was widely practiced by numerous Americans to make a living (and a `killing') (Lohr, 2002 ) . This article endeavors to demonstrate how history can help to identify patterns and further resonates with the recent call for international comparisons to better understand how history matters (Friedman & Jones, 2011 ) .
+
+Leveraging an institution-based view of IPR protection (Peng, 2013 ; Peng, Ahlstrom, Carraher, & Shi, 2017 ) , we focus on why US institutions supported IPR violation initially. Next, we investigate why US institutions later changed to ban such piracy. How the United States went from being a leading IPR violator to a leading IPR champion is a fascinating chapter in history that, we argue, can help inform the modern debate over IPR. In doing so, we respond to recent calls for organizational researchers to address larger, more globally relevant questions with societal ramifications (Davis, 2015 ; 186; Lewin, 2015 : 2) via `significant historical research' as a way forward (Ahlstrom & Wang, 2009 ; Barley, 2016 : 6) .
+
+## THE INSTITUTION-BASED VIEW
+
+The institution-based view is part of a broader intellectual movement centered on the new institutionalism throughout the social sciences. Organizational scholars have increasingly realized that institutions are more than background conditions (Meyer & Peng, 2016; Peng, 2003) . Instead, `institutions directly determine what arrows a firm has in its quiver as it struggles to formulate and implement strategy' (Ingram & Silverman, 2002: 20) . Consequently, an institution-based view has emerged (Ahuja & Yayavaram, 2011; Marquis & Raynard, 2015; Peng, Wang, & Jiang, 2008) . Further development can benefit from following well-accepted research designs of history research – `probing over long periods of time and finding commonalities among institutions' (Wren, 1987 : 341) .
+
+The institution-based view in the organizational literature (Ahuja & Yaravaram, 2011 ; Marquis & Raynard, 2015 ; Peng, Sun, Pinkham, & Chen, 2009 ) has roots in sociological institutionalism (DiMaggio & Powell, 1983 ; Scott, 2014 ) and economic institutionalism (North, 1990 ; Williamson, 1985 ) as well as allied fields such as political science (Hall & Taylor, 1996 ; Pierson, 2004 ). Authors in the organizational field often take an integrative approach, drawing from various disciplines (Ahuja & Yayavaram, 2011 : 1633; Peng, 2003 : 276; Peng et al., 2009 : 64), which is the approach followed here.
+
+The `rules of the game' (North, 1990 ) , an institution's most fundamental role is to reduce uncertainty and provide meaning (Scott, 2014 ) . The institution-based view `focuses on the dynamic interaction between institutions and organizations
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+17
+
+and considers strategic choices as the outcome of such an interaction' (Peng et al., 2009: 66) . In other words, institutions are `both constraints and facilitators' on the decisions and behavior of organizations (Ahuja & Yayavaram, 2011: 1633) .
+
+A leading proposition of the institution-based view is that individuals (such as managers, policymakers, and consumers) and organizations (such as government departments and firms) `rationally pursue their interests and make strategic choices within the formal and informal constraints in a given institutional framework' (Peng et al., 2009 : 67) . Of course, the rationality discussed here is bounded rationality (Williamson, 1985 ) . Individuals and organizations are assumed to be `intendedly rational, but only limitedly so' (Simon, 1961 : xxiv). Often associated with institutional economics, this rational choice school is one of the three major schools of thought within the institutionalism literature – the other two are sociological and historical (Hall & Taylor, 1996 ; Pierson, 2004 ) .
+
+Sociological institutionalism focuses more on the mimetic pressures from isomorphism and the meaning provided by institutions (DiMaggio & Powell, 1983 ; Scott, 2014 ) . Historical institutionalism argues that historical contexts shape institutions, which create path dependencies that are difficult to break (Arthur, 1994 ; David, 1985 ; Steinmo, Thelen, & Longstreth, 1992 ) . While there is theoretical tension among these three schools of thought, they may `be more complementary and less competitive than we realize' (Pierson, 2004 : 131; see also Peng et al., 2009 : 64; Ahuja & Yayavaram, 2011 : 1633) . In an effort to propel organizational research further, the institution-based view aspires to draw from the best insights from these three forms of institutionalism – rational choice, sociological, and historical (Hall & Taylor, 1996 ) . Before examining the institutions underpinning IPR protection, it may be useful to first discuss two `popular' explanations concerning IPR in China: culture and politics.
+
+## THE CULTURE AND POLITICS BEHIND IPR IN CHINA
+
+Today, China's IPR laws are generally well-crafted and have pushed a lot of IPR violation under ground (Chow, 2003 ; IP Commission, 2013 ; Yu, 2014b ) . However, enforcement continues to be weak. Despite some progress, repeated negotiations between the United States and China have not produced satisfactory headway. In the context of the contemporary US-China IPR debate, a fundamental puzzle is why China does not pay sufficient attention to IPR protection and why so many Chinese firms and individuals are willingly engaged in piracy, counterfeiting, and other acts of IPR violation (Yu, 2008 ) . Various debate points can boil down to two popular explanations: culture and politics.
+
+The cultural argument fingers Chinese culture, dating back 2500 years to Confucianism, for a major weakness: a lack of respect for IPR. For example, the title of Alford's (1995) influential book – To Steal a Book is an Elegant Offense: Intellectual Property Law in Chinese Civilization – sums it well. Suttmeier and Yao (2011: 19) further add that `Chinese culture still seems to have trouble
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+18
+
+M. W. Peng et al.
+
+valuing intangible assets.' As culture generally changes slowly (Hofstede, 2007 ) , though there is some debate about this (McCloskey, 2016 ) , if it were the main driver of behavior concerning IPR, then it would be difficult to envision better IPR protection in China in the absence of significant changes to its culture.
+
+However, is China the only country that has shown disrespect for IPR? The answer is no (Robinson, 2016 ) . Enough examples exist to refute the notion that IPR violation is driven by certain deeply held, largely invariant cultural values (Peng, 2013 ) . In China's pre-Qin period (before 221 BCE), traditional Chinese scholars despised the act of plagiarism (Feng, 2007 ) . Numerous swings in IPR protection also occurred in Europe since the Roman Empire where plagiarism (such as appropriating poems and plays) was common (Bugbee, 1967 ; Jaffe & Lerner, 2007 ) . Significant violation and theft of IP were regularly noted in the Middle Ages as well (Jaffe & Lerner, 2007 ) . [ 2 ] For example, in the fifteenth century, the Dutch busied themselves making counterfeit Chinese porcelain (Johns, 2009 ) .
+
+In the twentieth century, 1960s Japan was a global leader for counterfeit goods. In the 1970s and 1980s, Taiwan, South Korea, and Hong Kong grabbed this dubious distinction (Kumar, 2003 ; Peng, 2013 ) , and recent music piracy has involved people from practically every country (Hill, 2007 ) . For example, Russia is now home to some of the most notorious piracy websites in the world (Galtsova, 2008 : 21) . Clearly, IPR violation is not unique to China. The more abstract argument about a Confucian value system driving IPR violation does not stand up well to scrutiny.
+
+A political school of thought in this debate suggests that `Chinese political culture does not lend itself to the concept of IP ownership' (Zimmerman, 2013 : 142). During the Mao era in China, the emphasis on public (state) ownership was directly at odds with the emphasis on private rights inherent in IPR (Lubman, 1999 ). During the subsequent reform period, China's decentralized political structure encouraged local protectionism. This made IPR protection more problematic (Chow, 2003 ; Mertha, 2007 ; Poncet, 2005 ). Although the IPR laws on the books now look impressive and are (mostly) compatible with WTO (2009) mandates, it is the weak enforcement that is often fingered as contributing to the unsatisfactory IPR protection.
+
+However, in 2008 during the Beijing Olympics, the sale of fake Olympic merchandise completely disappeared (Yu, 2014a ) . Similar successes were achieved during the Asian Games in Guangzhou in 2010 and the Summer Youth Olympics in Nanjing in 2014. Did the counterfeiters become more patriotic? Or did the government demonstrate that effective protection of IPR was possible, once it was given political priority? Regardless of the answer, clearly the Chinese can be politically motivated and organizationally capable of eradicating IPR violation if so inclined. Thus, the political argument – just like the cultural argument – can also be rejected. Instead, when there is a will, there is a way.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+19
+
+## THE INSTITUTIONS BEHIND IPR IN CHINA
+
+As `rules of the game,' IPR institutions provide the incentive structures that affect the costs of doing business – in this case the costs of engaging via-a-vis the costs of not engaging in piracy and counterfeiting (Peng, 2013 ; Peng et al., 2017 ; Yu, 2014a ) . Given weak IPR protection (and weak enforcement if caught), thousands of firms and tens of thousands of individuals have made rational decisions (from their standpoint at least) to engage in piracy and counterfeiting (Hill, 2007 ; Li, 2004 ) . Stronger IP protection such as criminalizing all counterfeiting activities – instead of only criminalizing large scale counterfeiting above a certain threshold – is likely to significantly reduce counterfeiters' incentives. Between 2007 and 2009, the WTO adjudicated the dispute between the United States and China (Yu, 2008 , 2011 ) . The United States advocated a `zero tolerance' approach. China insisted on a `threshold' approach, which would only penalize violations above the threshold of `large scale' (or using the WTO vocabulary, `commercial scale'). While the WTO (2009) respected China's threshold approach and the United States did not win, China's reluctance to criminalize all IPR violations was evident. A lingering question is: Why?
+
+The institution-based view first highlights the path dependent nature of IPR development (North, 1990 ) . Path dependence maintains that the choices one faces for any given circumstance are limited by related decisions made in the past, even though past circumstances may no longer be relevant (David, 1985 ) . Earlier practices and initial conditions of potentially not paying enough attention to IPR thus cast a long shadow.
+
+Second, when viewed from a relatively long span of three decades, China's IPR system `has indeed come a very long way' (Yu, 2013 : 88). It essentially started from scratch in 1985 (Lubman, 1999 ). Handling over one million patent applications annually since 2010 (Suttmeier & Yao, 2011 ), China already has the world's largest volume of domestic patent applications ( The Economist , 2014 : 73) and it is now one of the world's leaders in filing international (ICT) patent applications (Yu, 2013 : 88). While China's IPR regulations are fundamentally compatible with the letter of the expectations of the WTO and its sister organization, World Intellectual Property Organization (WIPO), critics note that their enforcement, however, does not appear to meet the spirit of such expectations.
+
+Finally, institutional transitions moving toward a more effective IPR protection and enforcement regime have both costs and benefits (Mazzoleni & Nelson, 1998 ; Moser , 2013 ; Peng , 2003 , 2013 ) . Given widespread scholarly diffusion, explicit international pressures, and initial legal reforms, it is unlikely that Chinese policymakers, managers, and scholars are not aware of the innovation-boosting benefits of better IPR protection (Abrami, Kirby, & McFarlan, 2014 ; Lewin, Kenney, & Murmann, 2016 ; McKinsey, 2015 ; Yu, 2008 ) . However, even for proven violations, current law generally imposes a maximum fine of one million RMB ( $ 160,000) – only in a small number of cases when the loss to the plaintiff and
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+20
+
+M. W. Peng et al.
+
+income of the defendant are difficult to assess can the judge award damages greater than one million RMB. However, the average award is only 190,000 RMB ( $ 30,000), which is hardly enough to deter violations and that may not even cover legal costs (McKinsey, 2015: 115). Why does China not significantly increase the maximum fine for IPR violations? The reluctance seems to stem from the concern that at this early stage of China's development, satisfying US IPR demands may result in foreign (and primarily US) IP rights holders benefiting more (Liang & Xue, 2010: 490; Yu, 2013: 97). In short, costs may exceed benefits.
+
+How does the future of IPR protection look in China? We argue that the history of IPR development in the United States identifies interesting historical parallels and offers a great deal of insights to help inform today's debate.
+
+## THE UNITED STATES AS AN IPR VIOLATOR[3]
+
+Although widely known to scholars (Chaudhry & Zimmerman, 2009 : 175; Khan, 2013 : 67; Lopes & Casson, 2012 : 308; Raustiala & Sprigman, 2013 : 29) , the US experience as a leading IPR violator during the nineteenth century is seldom mentioned as a part of the modern IPR discourse. It is important to note that we are not offering a radical revision or reinterpretation of the US IPR history in the nineteenth century. Our characterization of the United States as an IPR violator follows the mainstream views of historians who specialize in this era (Gordon, 2005 ; Khan, 2005 ) .
+
+Similar to our earlier rejection of the cultural and political arguments in the contemporary IPR debate (such as the Chinese culture was not aware of the importance of IPR and the Chinese polity was not capable of offering IPR protection), we can also reject the cultural and political arguments concerning nineteenth-century America. From its founding, the United States had a conceptualization of IPR and a formal system of IPR protection. George Washington, in his first inaugural speech, encouraged a system of protection for inventors. Abraham Lincoln, the only US president to hold a patent, was a strong proponent of IPR. Adopted in 1787, the Constitution of the United States included a provision for protecting IPR in Article I, Section 8:
+
+The Congress shall have Power . . . To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.
+
+However, such IPR protection only protected US authors and inventors. Section 5 of the 1790 Copyright Act explicitly stated (Yu, 2014b: 9):
+
+Nothing in this Act shall be construed to extend to prohibit the importation or vending, reprinting or publishing within the United States, of any map, chart, book or books, written, printed, or published by any person not a citizen of the United States, in foreign parts or place without the jurisdiction of the United States.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+21
+
+In short, foreign intellectual property was up for grabs. [ 4 ] During much of the nineteenth century, American pirates were busy knocking off secret British designs for looms and mills, while producers and publishers extensively pirated foreign publications, art, and drama (Khan & Sokoloff, 2001: 237). The US government `stood ready to help them' (Raustiala & Sprigman, 2013: 29).
+
+Even American author Edgar Allan Poe got into the piracy game. The only book Poe published that was successful enough to be reprinted during his lifetime was not House of Usher or Rue Morgue , but The Conchologist's First Book (or A System of Testaceous Malacology, Arranged Expressly for the Use of Schools – a Textbook on Molluscs first published in 1839) (Gould, 1993 ; Quinn, 1998 ) . This was because Poe and his coauthor plagiarized works by British authors and such plagiarism was quickly discovered. Poe admitted as much in his own writings, but with no international copyright agreement, the original British publisher had no recourse (Gould, 1993 ) .
+
+During Charles Dickens' visit to the United States in 1842, he was appalled by the ubiquitous pirating of his work. He called for better protection of IPR. However, the US media, which regularly pirated British works, serialized novels, and other content, contended that Dickens should be appreciative for his popularity and that he was greedy to complain about his work being pirated (Tomalin, 2011 : 128 – 132).
+
+There was no shortage of British critics such as Dickens who lamented that the American culture – historically centered on lawlessness and rebellion (from a British standpoint) – might have been hopeless in improving IPR protection (Tomalin, 2011 ) . Moreover, painful and fruitless IPR negotiations between Britain and the United States led many British commentators to conclude that Americans were politically unwilling to protect IPR. These cultural and political arguments were similar to those invoked to explain the apparent `hopelessness' of the recent IPR situation in China.
+
+From an institution-based view, we argue that the US refusal to protect foreign IPR (prior to 1891) was rational. Given the lower levels of literary and economic development, benefits from the protection of foreign IPR would simply accrue to foreign inventors, authors, and firms (such as British publishers), while domestic consumers had to shoulder higher costs for books, media products, and innovative goods. In path dependence terms, institutions such as those supporting the violation of foreign IPR, once in place, would be hard to change (Arthur, 1994 ). However, institutions do change – as illustrated in the next section.
+
+## INSTITUTIONAL TRANSITIONS IN THE UNITED STATES
+
+Institutions can change (Campbell, 2004 ) . Institutional isomorphism asserts pressures on individuals, organizations, and even nation-states to conform, thus triggering the long-term processes for institutional transitions (DiMaggio & Powell, 1983 ) . In the long run, development of informal institutions such as clear norms on what are legitimate behaviors can be identified (Scott, 2014 ) . Sometimes,
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+22
+
+M. W. Peng et al.
+
+`nation-states, subject to only modest coercion or control, adopt standard identities and structural forms' (Meyer, Boli, Thomas, & Ramirez, 1997 : 174) . [ 3 ] In the IPR arena, nineteenth-century America and twentieth-first-century China did eventually switch from violating to respecting IPR – at least in terms of formal law-making. In 1891, to the pleasant surprise of British critics, the United States voluntarily strengthened its IPR laws with the passing of the International Copyright Act (commonly known as the Chace Act after its sponsor, Senator Jonathan Chace from Rhode Island), which extended IPR protection to foreign works. What was the cause for such a major institutional transition in the US IPR regime?
+
+Foreign (primarily British) pressures played a role behind the enactment of the Chace Act. However, given the repeated US refusals to British demands prior to 1891, such coercion was likely not the main driver of change. A more significant driving force behind such a transformation was likely the pressures from numerous indigenous inventors, authors, and organizations within the United States. As the close of the nineteenth century approached, the United States had turned from being a net consumer to a net producer of intellectual products (Fisher, 1999 ; Khan & Sokoloff, 2001 , 2004 ) . As more American authors, publishers, and producers were eager to market their products overseas (a leading market was Britain and the British Empire), they naturally desired better IPR protection from foreign governments (Thompson, 1986 ) . However, in the absence of reciprocity, foreign governments were unwilling to grant US rights holders IPR protection. They would only offer IPR protection for US rights holders if the United States offered equivalent protection to non-US IP rights holders in the United States.
+
+Further, in the late nineteenth century, American IP was being increasingly pirated elsewhere – especially in Canada. Similar to the Americans, the Canadians offered IPR protection to domestic (Canadian and British Empire) inventors, authors, and publishers, but did not offer IPR protection to foreigners (essentially Americans). Thus, pirating US publications was widespread in Canada, causing an uproar among American publishers and writers such as Mark Twain. [ 6 ] Given these changing winds, it was rational to start offering IPR protection in the United States to foreign authors and inventors. Essentially, the costs of offering local protection to foreign IPR were thought to be outweighed by the benefits of having American IPR protected overseas.
+
+In other words, piracy ` promoted domestic publishing output' in nineteenthcentury America (Khan, 2013 : 67, emphasis added) . For nineteenth-century America and many developing economies today, strong IPR `to primarily protect the rights of foreigners hindered their ability to imitate the technologies of the advanced countries and to learn by doing' (Khan, 2013 : 68) . A cost-benefit analysis suggests that only when the US economy (especially its IP production) was strong enough could the United States afford to offer IPR protection to foreigners (Raustiala & Sprigman, 2013 : 29) . Charles Dickens would probably be turning in his grave if he heard that the leading pirating nation of his time, the United States, had become a leading IPR advocate recently.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+23
+
+## THE SWEET AND SOUR HISTORY OF US IPR PROTECTION[7]
+
+Institutional transitions do not take place over night and do not necessarily unfold in a linear fashion (North, 1990 ; Scott, 2014 ) . Twists, turns, and setbacks are likely. Even after major transitions such as the 1891 Chace Act, two episodes stand out: (1) the complaint that US patent law was `over protecting' German chemical firms and (2) the US refusal to protect fashion design as IPR.
+
+### US-German Rivalry in the Chemical Industry
+
+As two rapidly developing economies, the United States and Germany engaged in strong rivalry prior to 1914. Competition in the innovation-driven chemical industry was especially intense (Chandler, 1990 ) . Between 1900 and 1910, the United States granted 4,068 patents to chemical and related inventions. Twothirds of them went to four leading German firms – Badische Anilin und SodaFabrik (BASF), Bayer (Farbenfabriken), Hoechst, and IG Farben (Khan, 2013 : 64) . German firms dominated the market for innovations in dyes and chemicals, and used patent rights to foreclose on US rivals. Many US managers and policymakers complained that when the US chemical industry was still an `infant industry', such (overly) strong IPR protection of globally competitive German rivals – thanks to the 1891 Chace Act – prevented US firms from producing knock-off products. In short, such `over protection' made German rivals stronger and conversely the fledgling US firms weaker.
+
+Only the confiscation (without compensation) of German IPR in 1917 consisting of several thousand patents, trademarks, and copyrights as well as German corporate assets – thanks to the US entry into World War I (WWI) and an amendment of the Trading with Enemy Act – did the competitiveness of the US chemical industry improve significantly (Khan, 2013 ) . Specifically, such IPR violation resulted in a 20 % increase in US domestic inventions during and in the aftermath of WWI (Moser & Voena, 2012 ) . However, despite such a `windfall stock' of German IPR, the absorptive capacity of US firms was still limited (Khan, 2013 : 64) . After WWI, German firms were able to regain their former competitive advantage in dyes and chemicals.
+
+### US-French Competition in the Fashion Industry
+
+From the beginning of the fashion industry, France has developed a number of leading firms and the United States has had numerous imitators, copycats, and pirates (Pouillard, 2011 ) . While one can debate the creativity differences between fashion industry practitioners in the two countries, from an important but often overlooked institution-based view, we need to realize the crucial institutional differences governing this industry. French copyright law protected fashion design, while US copyright law did not and still does not offer such protection (Pouillard, 2011 ) . Prior to WWII, making sketches at the Paris fashion openings was forbidden
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+24
+
+M. W. Peng et al.
+
+by French law. Nevertheless, numerous Americans attended such fashion shows and sketched the designs from memory right after the shows. They then illegally sold such unauthorized copies to American designers in New York, who would start copying the designs immediately.
+
+The efforts made by French fashion houses and their legal US importers (typically high-end retailers) to stop such copying, which was illegal in France but legal in the United States, were difficult. In 1930, a bill outlawing such copying passed the House of Representatives but failed to pass the Senate. Viewed as counter to the American ideal of social equality, the bill, according to its critics, would `take away from the poor working girl the right to wear the same pattern of goods that the wealthy people do' (Pouillard, 2011 : 331) . Further, opponents argued that copying pushed American designers to innovate and that such piracy boosted economic growth (Pouillard, 2011 : 343 – 344) . Today the debate continues to rage. A recent effort was the Design Piracy Prohibition Act sponsored by Representative Robert Goodlatte from Virginia, which was debated by the US Congress in 2011. But it failed to pass.
+
+### Analysis of the Two US Episodes
+
+Overall, the sweet and sour history of IPR protection in the United States – both before and after the enactment of the 1891 Chace Act – has been neither linear nor comprehensive. From an institution-based view, the path dependence of not respecting (certain) foreign IPR is evident in the efforts to expropriate German IPR during WWI and in the continued refusal to protect fashion design at the expense of the French fashion industry. The processes are indeed long term – several decades after 1891 for the chemical industry and over a century (and still ongoing) for the fashion industry. But despite such industry-specific exceptions, generally speaking, post-1891 institutional transitions in IPR protection have certainly resulted in significantly improved IPR protection in the United States.
+
+Informed by the institution-based view, our historical examination has revealed that (1) the United States engaged in significant IPR violation when its economy was taking off, and (2) the development of admittedly better US IPR protection in recent times has been neither linear nor comprehensive. Next, we use the same institution-based lens to shed light on two recent IPR developments in China.
+
+## TWO RECENT DEVELOPMENTS IN CHINA
+
+### The Surge of IPR Proclamations, Patenting, and Litigation
+
+Far from being an IPR `backwater', since the beginning of the 2000s, China has become one of the most eager countries in promoting innovation and IPR (Abrami et al., 2014 ; Cao, Li, Li, & Liu, 2013 ; Cheng & Huang, 2016 ; Hu & Mathews,
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+25
+
+Table 1. IPR civil cases in China and the United States
+
+<table><tr><td></td><td>Copyright cases</td><td>Trademark cases</td><td>Patent cases</td><td>Total IPR civil cases</td></tr><tr><td>China (2010)</td><td>24,700</td><td>8,460</td><td>5,785</td><td>38,945</td></tr><tr><td>United States (2010)</td><td>2,192</td><td>2,792</td><td>1,674</td><td>6,658</td></tr></table>
+
+
+Source: Extracted from data reported in Suttmeier, R. P., & Yao, X. 2011. China's IP transition: Rethinking intellectual property rights in a rising China (13). NBR Special Report 29. Seattle: National Bureau of Asian Research.
+
+2008 ; Lewin et al., 2016 ; Li, 2012 ; Liu, Simon, Sun, & Cao, 2011 ; Williamson & Yin, 2014 ) . According to the National Patent Development Strategy (SIPO, 2010 ):
+
+By 2020, China will become a country with a comparatively high level in terms of the creation, utilization, protection, and administration of patents... The quantity of patents for inventions for every 1 million people, and the quantity of patent applications in foreign countries will quadruple (3)... By 2015, the annual quantity of applications for patents for inventions, utility models and designs will reach 2 million. China will rank among the top two in the world in terms of the annual number of patents on inventions granted to domestic applicants and the quality of patents filed will further improve (4).
+
+Despite the overwhelmingly negative Western media coverage of IPR violation in China, it is difficult to argue that the Chinese government is not serious about IPR (Cao et al., 2013 ; Hu & Mathews, 2008 ; Lewin et al., 2016 ; Yu, 2008 ) . Total patent applications in China jumped from 476,000 in 2005 to over 1.2 million in 2010 (Suttmeier & Yao, 2011 : 13) . Today China is the world champion in the number of patent applications ( The Economist , 2014 : 73) . While the quality of Chinese patents (such as numerous alleged `junk patents') may be questioned (Cheng & Huang, 2016 ; The Economist , 2014 ) , some scholars suggest that it `may not be a serious concern' (Li, 2012 : 236; see Boeing, Mueller, & Sander, 2016 ) . $^[8]$ At least in terms of volume, China's patent system `has arguably accomplished what no other patent system in the world has ever achieved' (Yu, 2013 : 88) .
+
+Given the domestic and international complaints about the lack of IPR expertise among judges, China has recently set up a system of specialized IP courts led by judges specifically trained in IPR enforcement. As a result, IPR litigation has skyrocketed. Table 1 shows 38,945 IPR civil cases in 2010 (most of which involving Chinese plaintiffs suing other alleged Chinese IPR violators). [ 9 ] In comparison, the United States had 6,658 such cases. In 2013, the number of IPR case adjudicated in China reached over 100,000 (McKinsey, 2015: 115). China has become the world's most litigious country with respect to IPR – measured by the number of IPR litigation cases (Suttmeier & Yao, 2011: 13).
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+26
+
+M. W. Peng et al.
+
+![Figure](figures/figure_002.png)
+
+Figure 1. The costs and benefits of IPR protection and enforcement institutions
+
+## Analysis of China's Surge of IPR Proclamations, Patenting, and Litigation
+
+What can explain China's transformation from long-time reluctance to recent eagerness in promoting IPR? From an institution-based view, this development deviates from path dependence. This is why it is labeled `counterintuitive' by Suttmeier and Yao ( 2011 : 5) . Historical forces generally constrain contemporary laws and practices to be similar to past ones (Klochikhin, 2012 ) . But exogenous influences and endogenous changes to institutions are possible and can modify a seemingly entrenched path (Greif & Laitin, 2004 ) .
+
+As the self-reinforcing aspects of the informal institutional regime are gradually undermined by changes in the environment, both endogenous and exogenous changes are enabled. One endogenous element of this change in the United States was paradoxically a push by the railroads in the 1870s to make patenting more difficult and narrow (Usselman, 1999 ) . Some inventions were thought to be small incremental changes and should fall under the purview of trade secrets within firms and not IP. This debate eventually did lead to the tightening of IP laws, in spite of the railroads' protests. China has started to experience similar debates about IP and its optimal scope given its current level of development (Suttmeier & Yao, 2011 ) .
+
+The long-term institutional processes of enhancing legitimacy have asserted their influence – thanks to foreign pressures as well as an increase in invention by firms and citizens and gradual recognition from the public of its value (Sokoloff & Khan, 1990 ) . This increasingly accumulated momentum and court cases have led to a gradual appreciation of the wider benefits of strengthened IPR protection.
+
+Over time, the cost-benefit analysis underpinning institutional transitions becomes evident (Greif & Laitin, 2004 ; Peng, 2003 ) . Figure 1 illustrates the costs and benefits of IPR protection and enforcement institutions. Setting up
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+27
+
+such institutions is costly. During time 1, the costs of better IPR protection and enforcement outweigh the benefits – at least viewed from the perspective of domestic Chinese firms (a situation almost identical to nineteenth-century America). Over time, both the Chinese government and firms have realized more sustained economic growth will increasingly need to draw on IPR (Cao et al., 2013 ; Lewin et al., 2016 ; Liu et al., 2011 ). Past time 2 (the proverbial `crossover point' [Yu, 2013 ] or `point of inflection' [Peng, 2003 ] ), enhanced IPR protection and enforcement will be more beneficial, thus motivating the emergence of stronger IPR institutions (such as specialized IPR courts), which in turn fuel the more extensive use of these institutions (as evidenced by the rising number of IPR litigation cases). In sum, institutional transitions fostering stronger IPR in a country known for IPR violation is possible – as long as stakeholders such as policymakers, managers, consumers, lawyers, and judges believe they can `do better' in a new environment (North, 1990 : 8).
+
+## The Promotion of ‘Indigenous Innovation’
+
+If the first new development of improved and widely accepted IPR institutions is encouraging, a second development – the Chinese government's promotion of ‘indigenous innovation’ – has attracted significant criticism, especially from the international business community (IP Commission, 2013 ; USTR, 2010 ). China's first ever National Patent Development Strategy stated:
+
+The level of self-relied intellectual property will be higher by a large margin and the quantity of intellectual property will be greater. China will rank among the advanced countries of the world in terms of the annual number of patents for inventions granted to the domestic applicants, while the number of overseas patent applications filed by Chinese applicants should greatly increase. A number of world-famous brands will emerge. The proportion of the GDP accounted for by the value of core copyright industries will greatly increase (SIPO, 2010 : 2).
+
+Such an explicit emphasis on indigenous innovation has been criticized by US and foreign rights holders as well as their governments. These sentiments are summarized by the US Trade Representative in the 2010 National Trade Estimate Report on Foreign Trade Barriers:
+
+A troubling trend that has emerged... is China's willingness to encourage domestic or `indigenous' innovation at the cost of foreign innovation and technologies... In order to qualify as `indigenous' innovation under the accreditation system, and therefore be entitled to procurement preferences, a product's intellectual property must originally be registered in China (USTR, 2010 : 69).
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+28
+
+M. W. Peng et al.
+
+### Analysis of China's Promotion of 'Indigenous Innovation'
+
+The institution-based view can again contribute to a better understanding of this development. First, just like the case of US chemical and fashion industries, the path dependent nature of protecting domestic firms in certain industries is hard to break off from in China – even after significant commitments are made to enhance IPR protection (Howell, 2015) .
+
+Second, being aware of the long-term processes, we caution against a `knee-jerk' reaction of harsh criticism (as reflected in USTR [ 2010 ] ). Instead, we advocate a more nuanced reading of history. Specifically, history suggests that US policies regarding foreign IPR during much of the nineteenth century (prior to the 1891 Chace Act) can be regarded as US `indigenous innovation policies'. One lesson from history is that such policies `may help encourage the development of local stakeholders' as advocates of stronger IPR protection and enforcement (Yu, 2013 : 91; see also Jandhyala, 2015 : 287). Recall that it was the emergence of a critical mass of indigenous stakeholders such as US publishers, authors (such as Mark Twain), and inventors (such as Eli Whitney), who decisively pushed the US government to promote stronger IPR protection. Encouragingly, a new generation of indigenous stakeholders interested in better IPR protection is emerging in China. Jack Ma, founder of Alibaba, has called counterfeit goods a `cancer' and vowed to mobilize both Alibaba and other stakeholders to fight counterfeit goods ( The Economist , 2016 : 52).
+
+Finally, an interest in fostering institutional transitions offers a counterintuitive suggestion that the US government, US rights holders, as well as the international business community assist China to strengthen its indigenous innovation policies. This is because fostering China's indigenous innovation policies is compatible with initiating cognitive beliefs about the value of IPR and the need to protect it. It is also compatible with the long-run goal of improving IPR protection for all firms in China – both Chinese and foreign. To do this, foreign governments can help the Chinese government to create more uniform IPR laws and encourage their consistent enforcement around the country (Poncet, 2005 ) .
+
+## DISCUSSION
+
+### Contributions
+
+This article endeavors to respond to calls for organizational researchers to address broader, bigger issues of our times (Davis, 2015 ; Lewin, 2015 ) . Overall, at least two contributions emerge. First, we join Peng et al. ( 2017 ) in demonstrating how history can be directly relevant in informing a crucial, ongoing debate. History can help to identify patterns and parallels that can inform present day explanations and provide potential propositions for future research, practice, and policy (Ferguson, 1999 ) . For example, Weber studied the history of the Roman Legions, the Prussian Army, and the Roman Catholic Church to develop his theory about effective
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+29
+
+organizing (Weber, Baehr, & Wells, 2002 ) . Chandler ( 1990 ) advocated research on firm strategy and structure by investigating major industrial firms in Britain, Germany, and the United States. Military history has long influenced the analysis and formulation of strategy (Ahlstrom et al., 2009 ) . Extending such traditions of history research, our article suggests that historical variation is an important complement to contemporary cross-sectional variation in articulating conceptual issues (Jones & Khanna, 2006 ) . Our approach also resonates with the recent call for an international comparison in better understanding how history matters (Friedman & Jones, 2011 ) .
+
+How generalizable is the American experience in the nineteenth century to today's China? Although the United States and China differ on a number of dimensions, the international exchange and diffusion of ideas and practices have scaled new heights recently (Paulson, 2015 ) . Neither of these two countries – nor any other country for that matter – can evolve its own IPR institutions in total isolation. Institutional systems, especially formal ones, can have considerable influence on each other. [ 10 ] If we embrace a more global and longer view of history, we see that documented IPR violation started at least during the Roman times (Chaudhry & Zimmerman, 2009 ) , and that new laws and their enforcement contributed to improvements of IPR (Raustiala & Sprigman, 2012 ; Yu, 2014a ) .
+
+Second, by addressing and leveraging a specific episode of history, this article also enriches the institution-based view (Ahuja & Yayavaram, 2011 ; Meyer & Peng, 2016 ; Peng et al., 2008 , 2009 ) . Our historical excursion identifies the distinction and similarities in the course of events across time and location (Kieser, 1994 ) . Some have argued that culture, and particularly Confucianism, is linked to weak IPR protection (Alford, 1995 ) . Our historical examination has revealed similar culture-based arguments made about a US `frontier culture' of lawlessness in the appropriation of IP in the nineteenth century. Few would make the argument about a lawless frontier culture in US IPR today – except in the area of fashion where some French fashion houses continue to complain (Pouillard, 2011 ) . In spite of the concerns about IPR, China has clearly made strides in recent years.
+
+Specifically, we argue that economic development in China will facilitate improved IPR protection as economic development in the United States did in the nineteenth century. The poorest countries allocate few resources to innovation and thus have little IPR to protect. As incomes grow to middle income levels, some inventive capacity tends to emerge, but competition remains based on imitation. As a result, the majority of economic and political interests prefer weak IPR protection. As an economy develops further, additional inventive capacity and demands for high-quality products emerge, and commercial lobbies form to demand effective protection – a domestic interest that coincides with the foreign interest in better IPR protection.
+
+In summary, organizational research that addresses historical context is beneficial to theory development. This is because it does not allow complex phenomena such as the development IPR protection to be automatically attributed
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+30
+
+M. W. Peng et al.
+
+to culture or some other latent variable with little further explanation (Singh, 2007 ) , while avoiding limited, episodic understandings of such phenomena. By studying the history of IPR development in the United States, we can better identify and explain the potential parallels to China's current and future IPR development.
+
+## Policy and Practical Implications
+
+Several public and business policy implications emerge (Khoury & Peng, 2011; Paulson, 2015) . In terms of public policy, Raustiala and Sprigman suggest that `the United States should consider its own history as a pirate nation – and relax' (2013: 30). With an historical mindset, it is possible to appreciate that the differences between the United States and China in the IPR debate are not as significant as certain media and trade groups may portray. A practical lesson from a greater appreciation of history is to look for similarities that can help overcome the differences (Yu, 2011) .
+
+However, it is important to point out that certain groups benefit from an environment where the differences between the United States and China are magnified. `The more differences there are, the more valuable their expertise will become, and the more they can influence the policy and business debate' (Yu, 2011 : 1128) . When these differences are sharpened, important areas of potential cooperation – such as those dealing with China's indigenous innovation policies – become difficult. `The two countries will be worse off as a result' (Yu, 2011 : 1128; see also Paulson, 2015 ).
+
+Policymakers in China and other developing countries need to better arm themselves with a good command of the historical knowledge and to be aware of the crossover point – the point of inflection beyond which benefits will outweigh the costs of world-class IPR protection. The WTO requires that developing countries raise their IPR levels to the standard in force in more developed countries at the time of negotiation. However, by examining the welfare implications of protecting IPR and bringing technology transfer and tariffs into the equation, Naghavi ( 2007 ) argues that prior to the crossover point , following the highest standards in IPR protection is not necessarily ideal.
+
+US managers who complain about IPR violation in China today can learn how British authors, composers, and firms dealt with counterfeiters in the United States and elsewhere over 100 years ago. `Litigation was normally a last resort. Private agreements with offenders were cheaper and quicker, both because they could be kept confidential and because they avoided unduly alarming consumers about quality or alerting them to the possibility of purchasing a substitute at a cheaper price' (Lopes & Casson, 2012 : 303) . For example, nineteenth-century British composers Gilbert and Sullivan brought their entire troupe to New York to perform The Pirates of Penzane , and thus upstaged the theatrical pirates who had made so much money (without paying any royalties) from pirating Gilbert and Sullivan's previous smash hit, HMS Pinafore (Gould, 1993 ) . In contemporary China,
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+31
+
+numerous multinationals have been able to adapt to the Chinese IPR system and maximize their performance, leveraging a Gilbert-and-Sullivan-style pre-emptive strategy by bringing the innovation to China very early in its development (Liang & Xue, 2010 ) . Specifically, they file patents and trademarks – including Chineselanguage trademarks – as soon as possible, and/or set up strategic alliances with dependable Chinese partners right away (Shi, Sun, Pinkham, & Peng, 2014 ) . Such aggressive activities in China `makes major patent infringement very difficult' (Liang & Xue, 2010 : 491) .
+
+Managers in China who are currently involved in IPR violation need to be aware of the rapidly evolving institutional transitions toward better IPR protection. Although a `copycat' strategy centered on IPR violation has been viable (Shenkar, 2010 ) , institutional transitions may make this strategy obsolete. The best managers expect strategy to shift over time by constantly deciphering changes in the `big picture' (Peng, 2003 ) . That is why a good command of the historical knowledge of IPR evolution elsewhere in the world can be helpful. As institutional transitions unfold, the highest-performing firms seem to be those that convert the gains from an earlier strategy centered on IPR violation into value-adding capabilities that thrive on innovation. For example, Huawei, which was sued by Cisco for alleged IPR violation in 2003, has emerged as a leading innovator. It now routinely files one of the largest numbers of international patents among all firms in the world. On the other hand, `firms that fail to realize the passing of their time are likely to fall behind or go out of business' (Peng, 2003 : 292) .
+
+## Limitations and Future Research Directions
+
+We have barely scratched the surface of IPR development in the United States, China, and elsewhere around the world. We have neither covered IPR development in other parts of today's developed economies (those in Europe such as Great Britain and Germany and those in Asia such as Japan and Korea), nor that in other parts of developing economies (such as Brazil and India). Following Peng et al. ( 2017 ) , new insights will certainly emerge when future historically oriented IPR research probes these areas.
+
+Similarly, future research should also address abstract arguments about culture and its impact on important institutional variables. For example, the more abstract argument about a Confucian value system driving a unique form of Chinese IPR as some commentators have argued does not stand up well to either historical or analytical scrutiny. In particular, it fails to distinguish between Confucianism and a range of ethical teachings that have also influenced and continue to influence China's legal system and cultural milieu such as Taoism and Buddhism. The teachings in these systems are not in harmony with one another when it comes to IPR and other aspects of organizing (Ahlstrom & Wang, 2010 ) . This is not to say that culture does not matter, but scholars must be careful not to rush to cultural
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+32
+
+M. W. Peng et al.
+
+explanations before examining historical parallels, institutional developments, and other key processes that may be largely distinct from cultural values (Singh, 2007).
+
+## CONCLUSION
+
+How can history inform the debate on IPR? How can organizational scholars add to such an important debate? In short, history can be used both to understand the past and the present better and to make difficult decisions concerning the future in a more informed way (Neustadt & May, 1986 ) . Instead of strict path dependence, cultural determinism, or basic episodic assessment history helps to explain the past and the present, and suggests possible future directions given institutional constraints by assessing present-day events historically (Draper, 1996 ; North, 1990 ) . Joining Peng et al. ( 2017 ) , this article has extended the institutionbased view by demonstrating how the history of the United States transforming itself from a leading IPR violator to a leading IPR champion can be used to inform the contemporary US-China debate on IPR. Leveraging this period of US history, we predict that to the same extent that the United States voluntarily agreed to strengthen IPR protection when the US economy became sufficiently innovationdriven, China will similarly enhance its IPR protection. We further predict that when Chinese IPR are significantly violated abroad, China will become more serious about IPR protection. Therefore, we suggest that China's indigenous innovation policies need to be strengthened instead of discouraged by foreign IP rights holders, governments, and other stakeholders.
+
+## NOTES
+
+We thank Arie Lewin (Editor-in-Chief) for his encouragement and guidance. Earlier versions were presented at AAOM (Hong Kong, 2015); Copenhagen Business School; Hong Kong Baptist; HKUST; IACMR (Beijing, 2014; and Hangzhou, 2016); Renmin; Seattle; Shanghai Jiao Tong; Simon Fraser; Stockholm; Tsinghua; Universities of Newcastle, San Francisco, Sydney, and Texas at Dallas; Western Washington; and Xi'an Jiaotong. We thank Kevin Au, Michael Bond, Xiaoping Chen, Ping Deng, Tony Fang, Sam Garg, Steve Globerman, Syd Gray, Siggi Gudergan, Michael Jacobsen, Kwok Leung, J. T. Li, Jing Li, John Li, Peter Li, John Lin, Klaus Meyer, Bo Nielson, David Reid, Tom Roehl, Danny Shapiro, Pek-Hooi Soh, Hao Tan, Alain Verbeke, Cristina Vlas, David Whetton, Zhengzheng Xie, Xiaohua Yang, Jingtao Yi, Michael Young, Wu Zhan, and Jing Zhao for helpful discussions. This research has been supported by the Jindal Chair at UT Dallas.
+
+- [1] While we focus on China, numerous other countries have also been implicated by the United
+States Trade Representative (USTR, 2010 ) as IPR violators, such as Argentina, Brazil, Chile,
+Egypt, India, Israel, Mexico, Paraguay, Russia, Thailand, Turkey, Ukraine, and Venezuela
+(Chaudhry & Zimmerman, 2009) .
+[2] For example, the Roman writer Vitruvius (257–180 BCE) revealed IP theft during a literary
+contest in Alexandria. While serving as a judge in the contest, Vitruvius determined that some
+poets in the contest had stolen the words and phrases of other writers. After being found out,
+the false poets were then tried, convicted, and disgraced (Bugbee, 1967 ) . Yet in spite of such
+occasional prosecutions in Rome and later in medieval Europe, there were few laws codifying
+IP protection in Roman and medieval times, and the prosecution of offenses tended to be ad
+hoc (Bugbee, 1967 ) .
+[3] While we leverage the US experience, it is important to note that Germany, Japan, and
+numerous other developed countries were also significant IPR violators when their economies
+were taking off. However, ‘all the major imitators subsequently became leading economic
+powers, indicating an apparent connection between learning by imitation and economic
+growth’ (Lopes & Casson, 2012 : 308) .
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+33
+
+[4] American inventors sometimes found their inventions violated by their countrymen as well. For example, Eli Whitney invented the Cotton Gin in 1793 and patented it in 1794 under America's new Patent Law. Although this invention revolutionized the US cotton industry, it proved impossible to enforce as competent carpenters were able to rapidly produce and sell knock-off Cotton Gins. Whitney spent years pursuing patent lawsuits, and not until 1807 was the patent properly enforced. Whitney would realize only about $100,000 ($2,000,000 today). Much of the money was donated from southern state governments thankful for his invention (Gordon, 2005).
+
+[5] In addition to IPR, another visible area of long-term institutional isomorphism is the adoption of democracy, which has spread from 69 countries in the 1980s to 120 in the 2000s.
+
+[6] Mark Twain had to establish residency in Canada in order to protect the copyright of his novel The Prince and the Pauper in Canada. Given the costs involved, few authors could possibly afford to do that.
+
+[7] This title is inspired by Yu's ( 2014b ) title: `The sweet and sour story of Chinese intellectual property rights'.
+
+[8] The quality of many US patents can be similarly questioned. For example, BusinessWeek (2006: 60) reports that the United States is ‘awash in a sea of junk patents’.
+
+[9] In China, about 3% of all recent civil IPR litigation involved a foreign party (Suttmeier & Yao, 2011: 34).
+
+[10] For example, California contract law in terms of investment term sheets and related contractual agreements has impacted the practice and rules regarding venture capital and start-up firms around the world – including those in China (Ahlstrom, Bruton, & Yeh, 2007; Bruton & Ahlstrom, 2003).
+
+## REFERENCES
+
+Abrami, R. M., Kirby, W. C., & McFarlan, F. W. 2014. Why China can't innovate: And what it's doing about it. Harvard Business Review, 92(3): 107-111.
+
+Ahlstrom, D., Bruton, G. D., & Yeh, K. S. 2007. Venture capital in China: Past, present, and future. Asia Pacific Journal of Management, 24(3): 247-268.
+
+Ahlstrom, D., Lamond, D., & Ding, Z. 2009. Reexamining some management lessons from military history. Asia Pacific Journal of Management, 26(4): 617–642.
+
+Ahlstrom, D., & Wang, L. C. 2009. Groupthink and France's defeat in the 1940 campaign. Journal of Management History, 15(2): 159-177.
+
+Ahlstrom, D., & Wang, L. C. 2010. Entrepreneurial capitalism in East Asia: How history matters. In H. Landström & F. Lohrke (Eds.), The historical foundations of entrepreneurial research : 406–428. Cheltenham, UK: Edward Elgar.
+
+Ahuja, G., & Yayavaram, S. 2011. Explaining influence rents: The case for an institution-based view of strategy. Organization Science, 22(6): 1631–1652.
+
+Alford, W. P. 1995. To steal a book is an elegant offense: Intellectual property law in Chinese civilization. Stanford, CA: Stanford University Press.
+
+Arthur, W. B. 1994. Increasing returns and path dependence in the economy. Ann Arbor, MI: University of Michigan Press.
+
+Barley, S. R. 2016. Ruminations on how we became a mystery house and how we might get out. Administrative Science Quarterly, 61: 1–8.
+
+Boeing, P., Mueller, E., & Sander, P. 2016. China's R&D explosion: Analyzing productivity effects across ownership types and over time. Research Policy, 45(1): 159-176.
+
+Bruton, G. D., & Ahlstrom, D. 2003. An institutional view of China's venture capital industry: Explaining the differences between China and the West. Journal of Business Venturing , 18(2): 233–259.
+
+Bugbee, B. 1967. The genesis of American patent and copyright law. New York: Public Affairs Press.
+
+BusinessWeek. 2006. The patent epidemic. January 9: 60-62.
+
+Campbell, J.L. 2004. Institutional change and globalization. Princeton, NJ: Princeton University Press.
+
+Cao, C., Li, N., Li, X., & Liu, L. 2013. Reforming China's S&T system. Science, 341(August 2): 460-462.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+34
+
+M. W. Peng et al.
+
+Chandler, A. D. 1990. Strategy and structure: Chapters in the history of the industrial enterprise. Boston: MIT press.
+
+Chaudhry, P. E., & Zimmerman, A. 2009. The economics of counterfeit trade. Heidelberg, Germany: Springer.
+
+Cheng, M. L., & Huang, C. 2016. Transforming China's IP system to stimulate innovation. In A. Y. Lewin, M. Kenney & J. P. Murmann (Eds.), China's innovation challenge: Overcoming the middle-income trap: 152–188. New York: Cambridge University Press.
+
+Chow, D. C. K. 2003. The legal system of the People’s Republic of China in a nutshell. St. Paul, MN: West.
+
+David, P. 1985. Clio and the economics of QWERTY. American Economic Review, 75(2): 227332.
+
+Davis, G. F. 2015. What is organizational research for? Administrative Science Quarterly, 60(2): 179-188.
+
+DiMaggio, P.J., & Powell, W.W. 1983. The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2): 147160.
+
+Draper, T. 1996. A struggle for power: The American Revolution. New York: Times Books.
+
+Feng, N. H. 2007. To steal a book is just an elegant offense? Discussion on books' copyright protection in imperial China. Library and Information Service, 51(11): 58-61.
+
+Ferguson, N. (Ed.). 1999. Virtual history: Alternatives and counterfactuals. New York: Basic Books.
+
+Fisher, W. W. 1999. The growth of intellectual property: A history of the ownership of ideas in the United States. Working paper, Harvard Law School. [Cited 15 July 2016]. Available from URL: www.cyber.law.harvard.edu
+
+Friedman, W. A., & Jones, G. 2011. Business history: Time for debate. Business History Review, 85(1): 1-8.
+
+Galtsova, P. 2008. Intellectual property reform in Russia: Analysis of part four of the Russian Civil Code. Master’s thesis, Faculty of Law, Lund University, Sweden.
+
+Gordon, J. S. 2005. Empire of wealth: The epic history of American economic power. New York: Harper Perennial.
+
+Gould, S. J. 1993. Poe's greatest hit. Natural History, 102(7): 10-19.
+
+Greif, A., & Laitin, D. 2004. A theory of endogenous institutional change. American Political Science Review, 98(4): 633-652.
+
+Hall, P. A., & Taylor, R. 1996. Political science and the three new institutionalisms. Political Studies, 44: 936-957.
+
+Hill, C. W. L. 2007. Digital piracy: Causes, consequences, and strategic responses. Asia Pacific Journal of Management, 24(1): 9-25.
+
+Hofstede, G. 2007. Asian management in the 21st century. Asia Pacific Journal of Management, 24(4): 411-420.
+
+Howell, A. 2015. Indigenous innovation with heterogeneous risk and new firm survival in a transitioning Chinese economy. Research Policy, 44(10): 1866–1876.
+
+Hu, M.-C., & Mathews, J. A. 2008. China's national innovation capacity. Research Policy, 37(9): 1465-1479.
+
+Ingram, P., & Silverman, B. 2002. Introduction. In P. Ingram & B. Silverman (Eds.), The new institutionalism in strategic management: 1-30. Amsterdam: Elsevier.
+
+IP Commission, The (The Commission on the Theft of American Intellectual Property). 2013. The IP commission report. Seattle: National Bureau of Asian Research.
+
+Jaffe, A. B., & Lerner, J. 2007. Innovation and its discontents. Princeton, NJ: Princeton University Press.
+
+Jandhyala, S. 2015. International and domestic dynamics of intellectual property protection. Journal of World Business, 50(2): 284–293.
+
+Johns, A. 2009. Piracy: The intellectual property wars from Gutenberg to Gates. Chicago: University of Chicago Press.
+
+Jones, G., & Khanna, T. 2006. Bringing history (back) into international business. Journal of International Business Studies, 37(4): 453-468.
+
+Khan, B. Z. 2005. The democratization of invention: Patents and copyrights in American economic development, 1790-1920. New York: Cambridge University Press.
+
+Khan, B.Z. 2013. Selling ideas: An international perspective on patenting and markets for technological innovations, 1790–1930. Business History Review, 87(1): 39–68.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+35
+
+Khan, B. Z., & Sokoloff, K. L. 2001. The early development of intellectual property institutions in the United States. Journal of Economic Perspectives, 15(3): 233-246.
+
+Khan, B. Z., & Sokoloff, K. L. 2004. Institutions and democratic invention in nineteenth-century America. American Economic Review, 94(2): 395-401.
+
+Khoury, T., & Peng, M. W. 2011. Does institutional reform of intellectual property rights lead to more inbound FDI? Evidence from Latin America and the Caribbean. Journal of World Business, 46(3): 337-345.
+
+Kieser, A. 1994. Why organization theory needs historical analyses – and how this should be performed. Organization Science, 5(4): 608–620.
+
+Klochikhin, E. A. 2012. Russia's innovation policy: Stubborn path-dependencies and new approaches. Research Policy, 41(9): 1620–1630.
+
+Kumar, N. 2003. Intellectual property rights, technology, and economic development: Experiences of Asian countries. Economic and Political Weekly, January 18: 209-226.
+
+Lewin, A. Y. 2015. Letter from the editor. Management and Organization Review, 11(1): 1-3.
+
+Lewin, A. Y., Kenney, M., & Murmann, J. P. (Eds.). 2016. China's innovation challenge: Overcoming the middle-income trap. New York: Cambridge University Press.
+
+Li, S. 2004. Why is property right protection lacking in China? An institutional explanation. California Management Review, 46(3): 100-115.
+
+Li, X. 2012. Behind the recent surge of Chinese patenting: An institutional view. Research Policy, 41(1): 236-249.
+
+Liang, Z., & Xue, L. 2010. The evolution of China's IPR system and its impact on the patenting behaviors and strategies of multinationals in China. International Journal of Technology Management, 51(5): 469–496.
+
+Liu, F., Simon, D. F., Sun, Y., & Cao, C. 2011. China's innovation policies: Evolution, institutional structure, and trajectory. Research Policy, 40(7): 917-937.
+
+Lohr, S. 2002. The intellectual property debate takes a page from 19th-century America. New York Times, October 14.
+
+Lopes, T. D. S., & Casson, M. 2012. Brand protection and the globalization of British business. Business History Review, 86(2): 287-310.
+
+Lubman, S. 1999. Bird in a cage: Legal reform in China after Mao. Stanford, CA: Stanford University Press.
+
+Marquis, C., & Raynard, M. 2015. Institutional strategies in emerging markets. Academy of Management Annals, 9(1): 291-335.
+
+Mazzoleni, R., & Nelson, R. R. 1998. The benefits and costs of strong patent protection: A contribution to the current debate. Research Policy, 27(3): 273-284.
+
+McCloskey, D. N. 2016. Bourgeois equality: How ideas, not capital or institutions, enriched the world. Chicago: University of Chicago Press.
+
+McKinsey Global Institute (McKinsey). 2015. The China effect on global innovation. Report.
+
+Mertha, A. C. 2007. The politics of piracy: Intellectual property in contemporary China. Ithaca, NY: Cornell University Press.
+
+Meyer, J. W., Boli, J., Thomas, G. M., & Ramirez, F. O. 1997. World society and the nation-state. American Journal of Sociology, 103(1): 144-181.
+
+Meyer, K. E., & Peng, M. W. 2016. Theoretical foundations of emerging economy business research. Journal of International Business Studies, 47(1): 3–22.
+
+Moser, P. 2013. Patents and innovation: Evidence from economic history. Journal of Economic Perspectives, 27(1): 23–44.
+
+Moser, P., & Voena, A. 2012. Compulsory licensing: Evidence from the Trading with the Enemy Act. American Economic Review, 102(2): 396-427.
+
+Naghavi, A. 2007. Strategic intellectual property rights policy and North-South technology transfer. Review of World Economics, 143(1): 55-78.
+
+Neustadt, R. E., & May, E. R. 1986. Thinking in time: The uses of history for decisionmakers. New York: Free Press.
+
+North, D. C. 1990. Institutions, institutional change, and economic performance. New York: Norton.
+
+Paulson, H. 2015. Dealing with China: An insider unmasks the new economic superpower. New York: Grand Central.
+
+Peng, M. W. 2003. Institutional transitions and strategic choices. Academy of Management Review, 28(2): 275-296.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+36
+
+M. W. Peng et al.
+
+Peng, M. W. 2013. An institution-based view of IPR protection. Business Horizons, 56(2): 135139.
+
+Peng, M. W., Ahlstrom, D., Carraher, S. M., & Shi, W. 2017. An institution-based view of global IPR history. Journal of International Business Studies, forthcoming.
+
+Peng, M. W., Sun, S. L., Pinkham, B., & Chen, H. 2009. The institution-based view as a third leg for a strategy tripod. Academy of Management Perspectives, 23(4): 63-81.
+
+Peng, M. W., Wang, D. Y. L., & Jiang, Y. 2008. An institution-based view of international business strategy: A focus on emerging economies. Journal of International Business Studies, 39(5): 920–936.
+
+Pierson, P. 2004. Politics in time: History, institutions, and social analysis. Princeton, NJ: Princeton University Press.
+
+Poncet, S. 2005. A fragmented China: Measure and determinants of Chinese domestic market disintegration. Review of International Economics, 13(3): 409-430.
+
+Pouillard, V. 2011. Design piracy in the fashion industries of Paris and New Yorkin the interwar years. Business History Review, 85(2): 319–344.
+
+Quinn, A. H. 1998. Edgar Allan Poe: A critical biography. Baltimore, MD: Johns Hopkins University Press.
+
+Raustiala, K., & Sprigman, C. 2012. The knockoff economy: How imitation sparks innovation. New York: Oxford University Press.
+
+Raustiala, K., & Sprigman, C. 2013. Fake it till you make it: The good news about China's knockoff economy. Foreign Affairs, 92(4): 25-30.
+
+Robinson, R. 2016. How Google stole the work of millions of authors. Wall Street Journal, February 8.
+
+Scott, W. R. 2014. Institutions and organizations: Ideas, interests, and identities. Thousand Oaks, CA: Sage.
+
+Shenkar, O. 2010. Copycats: How smart companies use imitation to gain a strategic edge. Boston: Harvard Business School Press.
+
+Shi, W., Sun, S. L., Pinkham, B. C., & Peng, M. W. 2014. Domestic alliance network to attract foreign partners: Evidence from international joint ventures in China. Journal of International Business Studies, 45(3): 338-362.
+
+Simon, H. A. 1961. Administrative behavior (2nd ed.). New York: Macmillan.
+
+Singh, K. 2007. The limited relevance of culture to strategy. Asia Pacific Journal of Management, 24(4): 421-428.
+
+Sokoloff, K. L., & Khan, B. Z. 1990. The democratization of invention during early industrialization: Evidence from the United States, 1790–1846. Journal of Economic History, 50(2): 363– 378.
+
+State Intellectual Property Office (SIPO). 2010. National patent development strategy (2011– 2020). Beijing: SIPO.
+
+Steinmo, S., Thelen, K., & Longstreth, F. (Eds.) 1992. Structuring politics: Historical institutionalism in comparative analysis. Cambridge: Cambridge University Press.
+
+Suttmeier, R. P., & Yao, X. 2011. China's IP transition: Rethinking intellectual property rights in a rising China . NBR Special Report 29. Seattle: National Bureau of Asian Research.
+
+The Economist. 2014. Intellectual property in China: Patent fiction. December 13: 73-74.
+
+The Economist. 2016. Counterfeiting and piracy: Stamping it out. April 23: 51-52.
+
+Thompson, K. 1986. Exporting entertainment: America in the world film market, 1907– 1934. London: British Film Institute.
+
+Tomalin, C. 2011. Charles Dickens: A life. New York: Viking.
+
+US Trade Representative (USTR). 2010. National trade estimate report on foreign trade barriers. Washington, DC: USTR.
+
+Usselman, S. 1999. Patents, engineering professionals, and the pipelines of innovation: The internalization of technical discovery by nineteenth-century American railroads. In N. R. Lamoreaux, D. M. G. Raff, & P. Temin (Eds.), Learning by doing in markets, firms, and countries : 61–102. Chicago: University of Chicago Press.
+
+Weber, M., Baehr, P. R., & Wells, G. C. 2002. The Protestant ethic and the ‘spirit’ of capitalism and other writings. New York: Penguin Books.
+
+Williamson, O. E. 1985. The economic institutions of capitalism. New York: Free Press.
+
+Williamson, P.J., & Yin, E. 2014. Accelerated innovation: The new challenge from China. MIT Sloan Management Review, 55(4): 27–34.
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+History and the Debate Over Intellectual Property
+
+37
+
+Wren, D. 1987. Management history: Issues and ideas of teaching and research. Journal of Management, 13(2): 339-350.
+
+Yu, P.K. 2008. Intellectual property, economic development, and the China puzzle. In H. C. Hansen (Ed.), Intellectual property law and policy : 990–1023. Portland, OR: Hart Publishing.
+
+Yu, P. K. 2011. The TRIPS enforcement dispute. Nebraska Law Review, 89: 1046-1131.
+
+Yu, P. K. 2013. Five oft-repeated questions about China's recent rise as a patent power. Cardozo Law Review de Novo, 78-114.
+
+Yu, P. K. 2014a. The curious case of fake Beijing Olympics merchandise. In I. Calboli & E. Lee (Eds.),
+
+Trademark protection and territoriality challenges in a global economy: 259–282.
+
+Cheltenham, UK: Edward Elgar.
+
+Yu, P. K. 2014b. The sweet and sour story of Chinese intellectual property rights. In G. Dutfeld & U. Suthersanen (Eds.), Technology, progress, and prosperity. London: Palgrave Macmillan.
+
+Zimmerman, A. 2013. Contending with Chinese counterfeits: Culture, growth, and management responses. Business Horizons, 56(2): 141-148.
+
+Mike W. Peng ( mikepeng@utdallas.edu ), PhD, University of Washington, is the Jindal Chair of Global Strategy at the University of Texas at Dallas. He is a National Science Foundation CAREER Award winner, a Fellow of the Academy of International Business, and an IACMR Scholarly Contribution Award winner. The Asia Pacific Journal of Management has named its best paper award the Mike Peng Best Paper Award. He has authored Global Strategy, Global Business , and Global , which are available in Chinese, Portuguese, and Spanish, and used in over 30 countries.
+
+David Ahlstrom (ahlstrom@baf.cuhk.edu.hk), PhD, New York University, is a professor at The Chinese University of Hong Kong. He has published numerous articles in journals such as the Strategic Management Journal, Academy of Management Review, Journal of International Business Studies, and Entrepreneurship: Theory & Practice. His work has also appeared multiple times in The Wall Street Journal. He co-authored the textbook International Management: Strategy and Culture in the Emerging World. David has also guest edited four special issues of Asia Pacific Journal of Management and two special issues of Entrepreneurship: Theory & Practice. Shawn M. Carraher (shawn.carraher@utdallas.edu), PhD, University of Oklahoma, has served as Division Chair of Management History and Technology & Innovation Management (TIM) within the Academy of Management, President of the Small Business Institute and the Southwestern Academy of Management, and Dean of the Endowed Chairs and Past Division Chairs Committee within the AOM TIM Division. He is a clinical professor at the University of Texas at Dallas. In 2015 at the Säid Business School at Oxford University, Shawn was named one of the top 50 Outstanding Global Business Educators in the world.
+
+Weilei (Stone) Shi (weilei.shi@baruch.cuny.edu), PhD, University of Pittsburgh, is an associate professor of strategy at Zicklin School of Business, Baruch College – City University of New York. He serves on the editorial board of Journal of Management, Academy of Management Perspectives, and Asia Pacific Journal
+
+© 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+
+---
+
+38
+
+M. W. Peng et al.
+
+of Management. His research interests center on the interaction between strategy and international management. Specifically, Stone is interested in examining M & A and alliances from both temporal and network perspectives. Before joining academia Stone worked for Roland Berger Strategy Consultancy, a global consulting firm headquartered in Munich, Germany.
+
+Manuscript received: August 22, 2016
+
+Final version accepted: September 27, 2016 (number of revisions – 1)
+
+Accepted by: Editor-in-Chief Arie Y. Lewin
+
+c 2017 The International Association for Chinese Management Research
+
+https://doi.org/10.1017/mor.2016.53 Published online by Cambridge University Press
+

processed/00106_W2584475832/document.md

@@ -0,0 +1,609 @@
+![Figure](figures/figure_000.png)
+
+Research Policy 46 (2017) 693-707
+
+Contents lists available at ScienceDirect
+
+Research Policy
+
+journal homepage: www.elsevier.com/locate/respol
+
+![Figure](figures/figure_005.png)
+
+# Secondary pharmaceutical patenting: A global perspective
+
+![Figure](figures/figure_007.png)
+
+Bhaven N. Sampat a, * , Kenneth C. Shadlen b
+
+a Department of Health Policy and Management, Columbia University, United States
+
+b Department of International Development, London School of Economics, United Kingdom
+
+## A RTICLE INFO
+
+### Article history:
+
+Received 7 January 2016 Received in revised form 21 December 2016 Accepted 9 January 2017 Available online 13 February 2017
+
+### Keywords:
+
+Pharmaceuticals Patents TRIPS Developing countries Patent quality Patent examination
+
+## ABSTRAC
+
+Pharmaceutical firms’ use of secondary patents to extend periods of exclusivity generates concerns among policymakers worldwide. In response, some developing countries have introduced measures to curb the grant of these patents. While these measures have received considerable attention, there is limited evidence on their effectiveness. We follow a large sample of international patent applications in the US, Japan, the European Patent Office, and corresponding filings in three developing countries with restrictions on secondary patents, India, Brazil, and Argentina. We compare primary vs. secondary grant rates across countries, consider the differential fates of “twin” applications filed in multiple countries, and undertake detailed analyses of patent prosecution in the three developing countries. Our analyses indicate that measures to restrict secondary patents in developing countries are having limited impact. In none of these three countries are specific policies toward secondary patents the principal determinant of grant rates. Our analyses also suggest the importance of other procedural aspects of patent systems, beyond the formal policies targeting secondary applications, that affect outcomes for these applications in developing countries.
+
+© 2017 Elsevier B.V. All rights reserved.
+
+## 1. Introduction
+
+Taking out multiple patents on different aspects of a drug in order to cordon off competitors is standard practice in pharmaceuticals. In addition to primary patents, firms commonly attempt to acquire secondary patents on alternative forms of molecules, different formulations, dosages, and compositions, and new uses. Devising patenting strategies to extend periods of protection is an essential aspect of “life cycle management” in the pharmaceutical industry ( Burdon and Sloper , 2003 ; Howard , 2007 ; European Commission , 2009 ; Sternitzke , 2010 ; Ellery and Hansen , 2012 ; Kapczynski et al. , 2012 ) . This paper discusses policy challenges raised by secondary patenting, provides comparative data on secondary patent grant rates, and evaluates the effectiveness of restrictions on secondary patents in developing countries.
+
+While firms increasingly attempt to obtain secondary patents, policymakers have grown concerned about their effects, since they can extend periods of exclusivity beyond the dates in which protection would otherwise lapse if the only protection came from the primary patent on the molecule. Some have argued that patents on alternative molecular forms, formulations, or uses are of lower
+
+"quality" than primary patents too, in that they are less likely to be novel or manifest inventive step ( Correa , 2007 ; Kesselheim , 2007 ; Eisenberg , 2008 ) . And as with more general debates over patent quality ( Jaffe and Lerner , 2004 ; de Rassenfosse et al. , 2016 ; GAO , 2016 ) , there are concerns that patent offices worldwide may erroneously grant secondary applications that don't warrant patentability, but once granted restrict competition.
+
+Secondary patents are a particular source of concern in developing countries, where pharmaceutical patenting is new. The World Trade Organization's (WTO) 1995 Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) required developing countries to change their patent laws to be more like those in developed countries. Prior to TRIPS few developing countries allowed pharmaceutical products to be patented. Doing so is now obligatory for nearly all WTO members.1
+
+While TRIPS universalizes pharmaceutical patenting, some developing countries have exploited flexibilities built into the agreement to try to limit the grant of secondary patents. Three prominent examples of countries doing so are India, Brazil, and Argentina. Fearing the effects that secondary patents might have on pharmaceutical markets and access to medicines, and worried by
+
+* Corresponding author. E-mail address: bns3@columbia.edu (B.N. Sampat).
+
+1 Thirty-four WTO members classified as "Least Developed Countries" are exempted from this obligation until 2033. Before TRIPS many developing countries allowed for process patents in pharmaceuticals, but not product patents.
+
+http://dx.doi.org/10.1016/j.respol.2017.01.005 0048-7333/© 2017 Elsevier B.V. All rights reserved.
+
+---
+
+694
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+the difficulties of circumventing or removing patents once granted, each of these countries introduced provisions to restrict secondary patenting.
+
+These countries' approaches toward secondary patents have been championed by academics, civil society groups and nongovernmental organizations, and cited as models to be emulated ( Reichman , 2009 ; Kapczynski , 2013 ; South Centre , 2011 ; UNAIDS , 2011 ) . They have also been criticized by the pharmaceutical industry as unfair limitations on their abilities to obtain patents ( PhRMA , 2016 ) . $^2$ Though there are policy concerns, and some research indicating that many secondary patents are issued in developing countries ( Abud et al. , 2015 ; Correa et al. , 2011 ) , there is little evidence on the share of secondary applications granted and the effectiveness of countries' specific restrictions.
+
+This paper evaluates the effects of measles to limit secondary patents in India, Brazil, and Argentina on patent office outcomes in these countries. We do so in three ways. First, we compare differences between primary and secondary patent grant rates in these countries to differences in three patent offices (the U.S., EPO, and Japan) that do not have measures toward secondary patents. If the developing country policies are functioning effectively, we should observe differences across countries in the differential grant rates between primary and secondary patents. Second, we compare grant rates for secondary patents in developing countries for "twins," the same applications filed in different jurisdictions. Exploiting the twins nature of international patent applications is increasingly common for developed countries ( Jensen et al. , 2006 ; Hopkins et al. , 2007 ; Lemley and Sampat , 2012 ; Webster et al. , 2014 ; de Rassenfosse et al. , 2016 ; Christie et al. , 2016 ) , but few analyses have done so for developing countries ( Sampat and Amin , 2013 ; Sampat and Shadlen , 2015a ) . Third, since grant rates may be a blunt indicator of policy effectiveness, we provide data on the details of patent prosecution for secondary patent applications filed in the developing countries. This allows us to examine the role that the specific policies are having, in relation to other influences on secondary patenting grant rates. These final analyses build on and extend recent work in the U.S. that uses prosecution history data to get “inside the black box” of patent examination, to provide insights on the functioning of patent systems, beyond what can be learned from grant rates alone ( Drahos , 2010 ; Lemley and Sampat , 2012 ; Carley et al. , 2015 ; Frakes and Wasserman forthcoming ) .
+
+We find that developing countries' measures to restrict secondary patents are having less impact than one might expect from the considerable attention (positive and negative) they attract. Neither India nor Brazil exhibit lower grant rates for secondary patents than for primary patents, which is a differential that we would expect to observe if these countries' measures were having their intended impact. These results are robust across the overall sample, and the sets of twin applications. Though we do observe this differential in Argentina, detailed analyses of prosecution suggest that in none of these three countries are specific policies toward secondary patents the principal determinant of grant rates. In investigating this, we find suggestive evidence that long patent office backlogs in the developing countries give applicants time to learn about the importance and quality of their applications, leading them to abandon applications deemed not worth pursuing.
+
+In the following section we provide a general overview of the challenges posed by secondary patents globally, discuss why secondary patenting is a particularly salient policy issue in developing countries where pharmaceutical patents are new, and describe the policies that India, Brazil, and Argentina have enacted to limit the grant of such patents. In Section 3 we discuss the data sources we
+
+use to provide comparative evidence on secondary patent grant rates and to assess the roles played by the developing countries' restrictions. In Section 4 we present our empirical results, examining cross-national grant rates, grant rates for "twin" applications, and detailed analyses of secondary patent prosecution. Section 5 discusses these results. Section 6 synthesizes the main findings of the paper, addresses the limitations of the study, and points to avenues for future research.
+
+## 2. Secondary patents and public policy
+
+Secondary patents can restrict competition, deny consumers the benefit of generic entry, and thus allow for supra-competitive prices. While this is true of all patents, the grant of secondary patents draws particular criticism from those who believe they represent less research investment than novel molecules, and thus do not warrant patent protection ( Correa, 2007, 2014 ) . Related to this, because applications for secondary patents are typically filed after applications for primary patents, and patents last twenty years from the date of application, secondary patents, if granted, can potentially extend periods of market exclusivity. Pharmaceutical firms use secondary patents to retain exclusive rights to valuable, revenue-generating drugs for as long as possible, a strategy that has been attributed to the high costs of research and development, the low success rate in creating products that work in the lab and clinic and can gain regulatory approval, and the fact that significant portions of available patent periods will ordinarily have lapsed before successful products ever get on the market ( European Commission, 2009 ) . While in industry the use of secondary patents to extend periods of market exclusivity is referred to as “life cycle management” (e.g. Burdon and Sloper, 2003; Ellery and Hansen, 2012 ) , critics use the more pejorative term “evergreening” ( Rathod, 2010; Correa, 2014 ) .
+
+Even in the absence of specific policies targeting secondary patents, legal scholars believe that conventional patent standards, that an invention must be novel and demonstrate inventive step (in the USA, be “non-obvious”) ought to make secondary patent applications more difficult to obtain ( Eisenberg , 2008 ) . But there is also concern that resource-constrained patent offices commonly grant low “quality” patents ( Jaffe and Lerner , 2004 ) , i.e. patents that do not satisfy conventional patent standards and that, with more rigorous scrutiny, would have been rejected. Some have argued that the U.S. patent system is particularly permissive, on account of the incentives facing examiners to grant patents and its unique continuation practice that can reward applicants who are persistent ( Lemley and Moore , 2004 ; Amin and Kesselheim , 2012 ) , though the US is not alone in being criticized in this regard ( Moir , 2013 ) . The perception that the lax application of traditional patent standards can contribute to excessive granting of low-quality secondary patents in developing countries is widespread too ( Drahos , 2008 , 2010 ; Correa , 2007 , 2014 ; Reichman , 2009 ; Löfgren and Williams , 2013 ) .
+
+One way to address the problems that may be created by the granting of secondary patents is to invalidate them via litigation, as is common in the U.S. and many developed countries ( Hemphill and Sampat , 2011 ) . In developing countries, however, smaller markets and greater resource and information asymmetries between patent holders and potential challengers make this a less attractive solution ( Sampat and Shadlen , 2015a ) . Rather than relying on litigation to invalidate low-quality secondary patents after they have been issued, countries implementing new patent laws under TRIPS have been encouraged to introduce measures to address secondary patents at the point of examination. Such measures try to limit the grant of secondary patents in the first place, reflecting a belief that, in the language of Drahos (2008) , prevention is better
+
+2 One of the controversial aspects of the proposed Trans-Pacific Partnership had been to place limitations on countries' abilities to deny some secondary patents.
+
+---
+
+B.N. Sampat, K.C. Shodlen / Research Policy 46 (2017) 693-707
+
+695
+
+than treatment. $^3$ Although secondary patents in pharmaceuticals (and other fields) may be more vulnerable to rejection or invalidation under traditional patentability standards (e.g. novelty and inventive step) as noted above, this approach reflects a concern that ordinary examination practices are not enough, and additional restrictions specifically targeting secondary patents are needed.
+
+The three most prominent attempts to limit secondary patents at the point of examination are in India, Brazil, and Argentina. India took full advantage of the transitional provisions in TRIPS, which allowed countries to wait until 2005 to begin granting pharmaceutical patents. Accordingly, India began accepting applications in 1995, when TRIPS formally came into effect, though these were held waiting in a “mailbox” until 2005. In the final amendments to its new TRIPS-compliant patent law in early 2005, as the country began to examine pharmaceutical patent applications, India introduced Section 3(d), a provision that was explicitly designed to minimize the grant of secondary patents ( Basheer and Reddy, 2008; Kapczynski, 2009). According to Section 3(d), new forms of known substances are not eligible for protection unless the applicant can demonstrate increased “efficacy.”4
+
+Brazil introduced pharmaceutical patents in 1997, and in 2001 the new patent law was reformed to condition the grant of drug patents not only on the approval of the Brazilian patent office (INPI), but also on the “prior consent” of the Ministry of Health’s surveillance agency (ANVISA). When the system was put in place there was confusion about how exactly the health agency would carry out its new task. ANVISA subsequently decided to use its authority to try to limit the grant of secondary patents. The health agency created its own intellectual property division, and developed its own examination guidelines, more restrictive than INPI’s, targeting secondary patents ( Basso, 2006; Silva, 2008; Shadlen, 2011; Correa, 2014).
+
+Under the arrangements that were in place until 2012, the examination process began at the patent office. When INPI determined that the patent should not be granted, then the application was rejected and the process ended. However, when INPI determined that the patent should be granted, the application was passed to ANVISA for the health agency's examination team to make its own judgement. If ANVISA issued its consent, INPI then granted the patent; if ANVISA withheld its consent, INPI was prohibited from granting the patent (notwithstanding its original favorable evaluation). The Prior Consent system was revised in 2012, reversing the order of operation, but the core feature of Brazil's system – pharmaceutical patents need to be approved by both the INPI and ANVISA to be granted – remains in place.5
+
+In Argentina, pharmaceutical products became patentable in 2000. In 2002, the patent office issued new examination guidelines that barred patents on one type of secondary patent, those claiming
+
+second medical uses. Then, in 2012, the patent office issued new and significantly more restrictive guidelines that instruct examiners to reject most forms of secondary pharmaceutical patents. Argentina's approach toward secondary patents differs from that of India and Brazil in terms of coverage. For most of the period in our analysis, the prohibition on second medical uses is the sole measure designed to limit secondary patents. The more restrictive guidelines are only in place for applications examined as of May 2012.
+
+As discussed above, these provisions have received considerable attention, but little research has been conducted on their effects. Berndt and Cockburn (2014), for example, assume that Section 3(d) is restricting patent grants in India but their research design does not allow them to not provide direct evidence on this. Basso (2006), Shadlen (2011) and Barbosa (2013) analyze the conflict between INPI's and ANVISA's approaches to secondary patents in Brazil, but without systematically assessing the effects of the dual examination arrangements on patent grant outcomes. Other academics have expressed concerns that patent examiners in developing countries, tied into global examination networks, working under severe resource constraints, and facing pressures to clear the large inflow of applications, may not effectively enforce laws and guidelines and there may be significant gaps between laws on the books and patent prosecution in practice (Kapczynski, 2009; Drahos, 2010). Yet there is little direct evidence of this either. Drahos (2010) shows how the transnational integration of patent offices generates a common "interpretive examination culture," but the analysis does not provide data on convergence in terms of patent office outcomes.
+
+A previous study examining Indian and Brazilian patent applications on about 160 drugs launched between 1996 and 2004 with at least one U.S. patent, found these provisions were rarely used (Sampat and Shadlen, 2015a). However, that work focused on a small number of applications with various special characteristics (including that they tended to be older applications, and they were associated with “successful” drugs already on the market that had U.S. patents). Moreover, that study was only able to ensure similarity of applications for a small number of cases, making comparing grant rates on secondary patents difficult. Most importantly, by focusing only on India and Brazil there was no baseline against which to assess grant rates for secondary patents. Nor do we know much about the Argentinean approach. One earlier study has examined secondary patenting in Argentina (Correa et al., 2011), but because that paper focuses only on granted patents it cannot provide insight into how the patent office's approach to secondary applications functions.
+
+In this paper we use a novel dataset of international patent applications, code each application as primary or secondary, and identify outcomes in each of the developing countries, and, as a basis for comparison, in three developed country patent offices. In addition to providing descriptive results on cross-national grant rates for secondary patents, we examine the effectiveness of the restrictions in Argentina, Brazil, and India in several different ways. The following section describes the dataset and research strategy.
+
+## 3. Data
+
+### 3.1. PCT applications and coding
+
+The majority of global pharmaceutical patent applications are filed through the Patent Cooperation Treaty (PCT), which allows single applications to be deposited in multiple jurisdictions after undergoing preliminary analysis by an International Searching Authority. Our analysis focuses on grant rates for "national phase"
+
+3 Another reason why secondary patenting has particular salience in developing countries relates to the timing of TRIPS implementation. Because of the transition provisions in TRIPS, countries were not required to grant patents with priority dates (first global filing date) before 1995. This means that for many drugs that were launched in the 1990s and 2000s, primary patents are not eligible, despite the global spread of pharmaceutical patenting, and the only patent protection available in developing countries would be via secondary patents ( Sampat and Shadlen, 2015b ) . This transitional feature heightened the stakes of policy choices about secondary patents.
+
+4 Section 3(d) was a surprise to most observers, including the pharmaceutical industry. It has since been the source of much controversy. The provision was (unsuccessfully) challenged in the Indian Supreme Court by Novartis, following the Indian Patent Office's rejection of a secondary patent on a cancer drug "Gleevec" (imatinib mesylate). The Novartis case galvanized opposition to 3(d) from the transnational pharmaceutical industry, and also vigorous defense of the provision from civil society organizations, health activists, and international organizations.
+
+5 Under the new workflow, pharmaceutical patent applications are first sent to ANVISA: if ANVISA rejects, the examination process is to end there; if ANVISA approves, the application is then examined by INPI.
+
+---
+
+696
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+applications in each country – the U.S., Europe,6 Japan, Brazil, and India – that emanate from PCT applications. In the case of the one non-PCT member, Argentina, we use the Argentinian equivalents of "national phase" filings.7
+
+Using the World Intellectual Property Organization's Patent Statistics database, we identified all PCT applications filed between 2000 and 2002 that had at least one International Patent Classification of A61K or C07D, the main classes associated with drugs. From these, we focused on applications that were filed (either as original filings or as national phase filings through the PCT) at the US, European and Japanese patent offices. Applying these criteria left 15,815 applications. We focused on the years 2000–2002, to allow a long window to observe outcomes. (Otherwise, many applications would remain pending, especially in the developing countries with large patent office backlogs.) Another benefit of focusing on this time period is that the Indian and Brazil national phase applications from PCT applications filed during these years would have been submitted before Section 3(d) was introduced or ANVISA's secondary patent restrictions were fully implemented, and before the most restrictive Argentinian guidelines were enacted, limiting the effect of selective filing on our results.
+
+Since our research involved working with individual applications, to keep the analyses manageable we further restricted the set to those filed between January and July. Patent classifications are known to be noisy. Scanning this set revealed it contained many applications that were not actually for pharmaceuticals; we consulted the Thomson Reuters “Chemical Patent Index” (CPI) code for each application and used these codes to narrow to true pharmaceuticals, resulting in a final set of 5193 applications.8
+
+In some of our analyses, we will examine matched “twin” applications, i.e. the same PCT applications that have national phase entries in each of the PCT countries (and equivalent national applications in the non-PCT country, Argentina). While the specific claims filed in individual jurisdictions vary slightly, by and large these applications are substantively similar, if not always “identical” twins.
+
+For our analyses we will need to know which of the PCT applications (and by extension, the national phase filings that result) include primary claims or only secondary claims. We had a pharmaceutical patent attorney code each of the applications, using a coding guide adapted from Hemphill and Sampat (2011). The Appendix A includes the first pages of the coding guide. About 8 % of the applications contained only process claims. We dropped these, since our focus is on product patenting. Of the remaining 4765 applications, roughly 38 % were coded as including a novel active ingredient claim and thus were classified as "primary," and 62 % had no novel active ingredient claims and were classified as "secondary."
+
+### 3.2. National outcomes
+
+We collected information on whether each PCT application was filed and granted in each country. For the 5 PCT members, we col-
+
+lected information on any national phase applications filed in these patent offices via the PCT. Since Argentina is not a member of the PCT, there we focus on any national applications that are equivalents of the PCT applications. For expositional convenience, though with a slight abuse of terminology, we refer to all of these, including those filed in Argentina, as “national phase” applications. We determined whether each of the national phase applications was granted in each country, as of 2015. In cases where a given PCT spawned more than one application (including, for example, through “continuation” and “divisional” applications) we determined whether any of the applications were granted.9
+
+For the three developing countries we recorded outcomes beyond overall grant rates, to distinguish between applications that were rejected, abandoned or withdrawn, and those still pending. In India and Brazil, we also collected detailed information on prosecution histories, including, for those that were not granted, the specific reasons for non-grant (India) and role of ANVISA in the process (Brazil). In Argentina we examined the fate of use patents and distinguished between those applications reviewed before and after the patent office's new examination guidelines came into effect. This additional information helps us better understand the role of the countries' approaches toward secondary patents in determining outcomes for the national phase applications.
+
+It is important to emphasize what this sample excludes. In focusing on national phase applications, we are not including the underlying priority applications on which the PCT was based. This is not an issue for developing countries, since almost all applications are filed there based on the PCT, but for the US, EPO, and Japan only applications filed via the PCT are in our sample. 10
+
+### 3.3. Application importance
+
+In addition to any laws or policies regarding secondary patents, applicants' levels of effort in pursuing patent protection also may influence national outcomes. This varies with the importance of the invention, based on information at the time of filing and information that accumulates over time.
+
+One measure of importance is family size: the number of countries in which an application is filed, based on data from to the Derwent World Patents Index (DWPI). Family size is a commonly used measure of invention importance, based on the expectation firms will file their more important inventions more broadly ( Lanjouw et al. , 1998 ) .
+
+Given long patent office backlogs and slower prosecution in developing countries ( Schultz and Madigan , 2016 ) , and that most applications may not be examined until well after they were filed and even granted in many developed countries, we can also look at a measure of importance further into the lifecycle, whether the patent was granted in the U.S., and, if so, if it was renewed. For national phase applications filed and granted at the USPTO, we collected information from the USPTO on they were renewed (as of October 2015) or allowed to lapse ( USPTO, 2015 ). This allows us to distinguish, among the applications in our sample that were filed at the USPTO, between those never granted, those granted but allowed to lapse (reflecting diminished importance and effort over time), and those granted and maintained (reflecting the greatest degree of importance and effort).
+
+6 Because of a shared European patent office, the EPO, we refer to "Europe" as if it were a country.
+
+7 In the Appendix A we describe the steps taken to map PCT applications to their national versions, and our approach for determining the outcome of each national application.
+
+8 The CPI codes are based on expert coding of the applications. Each application can have many CPI codes. We restricted the set of patent applications to those with at least one "B" (Pharmaceutical) coding, dropping 826 applications. Among the remaining applications, we also determined which were likely biotechnologyrelated (those coded as BO4-E, F, G or D05-H). We also dropped these applications (about a third of the total) since our focus here is on small molecule drugs where secondary patents are more prevalent and regarded as potentially important impediments to competition among multiple suppliers.
+
+9 These "child" applications contain subject matter from previously-filed "parent" applications.
+
+10 In practice, almost all PCT applications filed at the EPO “go national” in the EPO through the PCT. This is because applications from member countries that are filed at the EPO automatically receive new Euro-PCTapplication numbers. About a quarter of PCT applications filed through the US go national in the US via the PCT (typically these are based on provisional priority applications in the U.S. that are then abandoned), and the analogous figure is 55% for Japan.
+
+---
+
+B.N. Sampat, K.C. Shodlen / Research Policy 46 (2017) 693-707
+
+697
+
+Table 1
+
+Descriptive statistics for the 4765 PCT applications in the sample.
+
+<table><tr><td colspan="5">Descriptive Statistics</td></tr><tr><td></td><td>Mean</td><td>Min</td><td>Max</td><td>N</td></tr><tr><td>Secondary application</td><td>0.62</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>US filed</td><td>0.67</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>EP filed</td><td>1.00</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>JP filed</td><td>0.93</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>BR filed</td><td>0.36</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>IN filed</td><td>0.24</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>AR filed</td><td>0.20</td><td>0</td><td>1</td><td>4765</td></tr><tr><td>US granted</td><td>0.61</td><td>0</td><td>1</td><td>3184</td></tr><tr><td>EP granted</td><td>0.51</td><td>0</td><td>1</td><td>4756</td></tr><tr><td>JP granted</td><td>0.29</td><td>0</td><td>1</td><td>4450</td></tr><tr><td>BR granted</td><td>0.05</td><td>0</td><td>1</td><td>1698</td></tr><tr><td>IN granted</td><td>0.41</td><td>0</td><td>1</td><td>1155</td></tr><tr><td>AR granted</td><td>0.12</td><td>0</td><td>1</td><td>931</td></tr><tr><td>PCT family size</td><td>8.95</td><td>2</td><td>31</td><td>4765</td></tr><tr><td>Granted and Maintained in US</td><td>0.29</td><td>0</td><td>1</td><td>3184</td></tr></table>
+
+
+Table 1 provides basic descriptive data for the full sample. Of the 4765 PCT applications, 62% are secondary. Recall that our sample is based on PCT applications with original (priority) or national phase filings in the U.S., EPO, and Japan. About two thirds of the PCTs had national phase applications in the U.S., over 90% in Japan, and nearly all in Europe. The share with national phase applications is much lower in the developing countries (24% in India, 35% in Brazil, and 20% in Argentina). On average the applications in our set were filed in nine countries. And for the 3184 PCT with US national applications, 29% were granted and renewed to date.
+
+## 4. Secondary pharmaceutical patenting
+
+We use these data in four ways. First, we consider how secondary patent grant rates vary across countries, conditional on filing and by the two measures of invention importance. Doing so provides us with a baseline for understanding these countries' grant rates, and a point of departure for our subsequent focus on the effectiveness of specific instruments to restrict secondary patents. Second, we examine cross-country differentials between grant rates for primary vs. secondary patents. If national restrictions on secondary patents are functioning effectively, we should see a larger “difference in differences” between primary and secondary grant rates in countries with such restrictions than in countries without. Third, we examine the fate of the same applications filed in different countries. This analysis of “twins” allows us to control for applicants' propensity to file different applications in different countries, and thus make sure that the results we observe reflect differences in how countries treat secondary applications rather than the characteristics of the applications filed nationally. Fourth, we provide detailed analyses of secondary patent prosecution outcomes in the three developing countries, extending our focus beyond grant rates. Doing so provides a complementary “bottom up” view on the specific roles played by these countries' measures to restrict secondary patents.
+
+### 4.1. Cross-national comparisons of grant rates
+
+Of the 2964 secondary applications, 971 had national phase filings in Brazil through the PCT, 641 in India, 2047 in the US, 2956 in the EPO, and 2735 in Japan, and 444 had equivalents filed in Argentina. Fig. 1(a) shows grant rates for secondary applications in each country, conditional on filing. We count a PCT application as "granted" in a country if any national phase application (including continuations or divisionals) is granted. The U.S. grants about half of the secondary applications, roughly the same as in Europe. Surprisingly in view of the controversy surrounding Indian restric-
+
+tions on secondary patenting, India is the median country in terms of granting secondary patents; Japan has a much lower grant rate than India. The lowest grant rates are in Argentina and Brazil.
+
+Grant rates are the result of several factors, including the types of applications filed in a country, patent laws and guidelines, how laws and guidelines are enforced, and patent office processing speed. We will consider the role these factors may have in explaining overall trends in more detail below. For now, a simpler explanation for differences may be that applicants try harder on some applications than others. To examine whether the differences across countries reflect only applicant effort, we examined how grant rates vary across the family size distribution.
+
+Fig. 2 shows that grant rates increase with family size in all countries. However, the gradient is notably less steep in the South American countries. If we focus on the top decile of the family size distribution (applications filed in 25 or more countries), several things stand out. The EPO has the highest grant rate for these important applications (90%), followed by the US (granting 81%). India is still the median, though the grant rate at the top of the distribution is nearly 60% (compared to the 42% that we saw above, for all Indian applications). Even at the top of the distribution, only 15% of secondary applications filed in Argentina, and 8% filed in Brazil, were granted.
+
+Family size is a measure of application importance at the time the application is filed. A complementary measure is U.S. maintenance status, which can shine a light on importance as applicants obtain additional information after original filing decisions are made. Applications that are granted in the U.S. and maintained there are likely to be ones that drug companies most care about, for these applications are likely to be associated with drugs that have passed product development and perhaps even market entry tests. Fig. 3 shows grant rate in the five other countries based on U.S. maintenance status. Again we see sharp gradients, with applications granted in the U.S. and maintained having much higher grant rates in other countries than applications that are not granted, or granted and allowed to lapse. As with the family size, the basic rank ordering of countries by grant rate of secondary applications is similar at the top of the distribution to what we saw in Fig. 1 . Effort appears to matter in Argentina and Brazil, as elsewhere, but even for these applications that are granted and maintained in the U.S., grant rates in the two South American countries are low.
+
+The data presented so far suggest that secondary patents are easiest to obtain in the U.S. and Europe and hardest in Argentina and Brazil, with India and Japan in between, and that in all countries more important inventions, by either measure of importance, fare better. However, the fact that differences across countries in grant rates for secondary patents persist, even for the most important inventions, calls for further explanation.
+
+### 4.2. Understanding the differences in developing countries
+
+Differences in national grant rates, overall and conditioned on importance, may not necessarily tell us about the permissiveness of each country towards secondary applications, or, for the countries with restrictions, their effectiveness. There are other reasons that grant rates may vary, including the speed by which applications are examined and other attributes of national patent offices, and the efforts that applicants make (on account of the economic importance of particular markets, for example, or new information obtained about a drug's promise).
+
+One way to account for country-specific factors that affect outcomes is to compare differences in secondary grant rates to differences in primary grant rates. Even if we saw differential grant rates by type of application, it would be important to compare these to differential grant rates in countries without specific restrictions on secondary patents, since secondary patents may be more vulner-
+
+---
+
+698
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+![Figure](figures/figure_002.png)
+
+Fig. 1. Grant rates by country. Note: Grant rates conditional on filing. AR = Argentina, BR = Brazil, IN = India, US = USA, EP = European Patent Office, JP = Japan.
+
+![Figure](figures/figure_004.png)
+
+Fig. 2. Grant rates for secondary applications by family size.
+
+Note: Based on the outcomes for the 2047 secondary applications filed in the US.
+
+able to rejection anyhow (through novelty and inventive step), even absent specific policies targeting them. That is, it is useful to look at the difference in differences: comparing the difference between secondary and primary grant rates between countries that do and
+
+do not have specific restrictions. Comparing Fig. 1(a) and (b) allows us to do this. Here we see that neither India nor Brazil has a lower secondary grant rate than a primary grant rate, an observation that is not what we would expect if restrictions targeting secondary
+
+---
+
+B.N. Sampat, K.C. Shodlen / Research Policy 46 (2017) 693-707
+
+699
+
+![Figure](figures/figure_002.png)
+
+Fig. 3. Grant rates for secondary applications by US status.
+
+Note: Based on the outcomes for the 2047 secondary applications filed in the US.
+
+patents were effective. By contrast in Argentina there is a 7 percentage point differential, which is greater than that in two of the developed country patent offices (EPO and JPO) but much smaller than that in the U.S.11
+
+Beyond differences in national characteristics of patent systems, which the comparison to primary patents aims to account for, another issue that may bias cross-country comparisons is that not all the PCT applications are filed in all countries. As a result, the outcomes may be reflecting as much about characteristics of the applications filed in each country as the countries' examination practices. To provide better comparability, Fig. 1(c) and (d) show analogous data for "twin" applications where the same PCT is filed in all six countries. For secondary twins, in 1(c) we see the same patterns: grant rates are highest in the EPO and US and lowest in Argentina and Brazil, with India and Japan in the middle. What about the difference in differences? Among this set of twins, primary applications fare better than secondary applications in each of the developed countries, with the largest differential in the U.S.
+
+More strikingly, in two of the three countries which have explicit restrictions on secondary patents, there is little evidence that secondary applications fare worse than primary applications do: in India the difference is much smaller than in the developed countries, while in Brazil the primary and secondary grant rates are identical. In Argentina, however, the grant rate for secondary patents remains much lower than that of primaries in this twins sample too, suggesting this country's policies towards secondary patent applications may be achieving their objectives.
+
+This twins analysis focuses on the smaller number of applications that went national in all PCT countries using the PCT and had an equivalent Argentinian filing. This may also introduce bias, since for applications originating from the U.S. in particular, firms often do not use the PCT to go national, and as result the set of twins in
+
+Fig. 1(c) and (d) does not include many of the US-origin applications (see Table 1 and related discussion). 12
+
+To eliminate this bias, we can consider only twins filed via PCT in the three developing countries, regardless of how they went national in the developed countries. In India and Brazil almost all applications filed by multinationals go national via PCT, and in Argentina as national equivalents filed after the main application is filed globally. We identified 501 twin applications that were filed in each of these countries. Fig. 1 (e) and 1 (f) show outcomes. Argentina and Brazil have low grant rates for both types of patents, with India considerably higher, and among the three countries only Argentina shows a sharp difference between primary and secondary grant rates. As in all of the analyses above, neither Brazil nor India seem to discriminate much between primary and secondary patent applications, suggesting their policies may be having little direct effect.
+
+Limiting the sample to twins filed in all six jurisdictions, or the broader set filed in each of the three developing countries, facilitates cross-country comparison. But there is a trade-off: the applications filed in multiple countries are a select set, so results from this set may be less generalizable. 13 While this would be a
+
+11 We find similar results in a regression framework, as reported in the Appendix.
+
+12 This is not an issue in the difference-in-difference analyses over the whole sample reported immediately above.
+
+13 To examine how select, we compared the twins to other secondary applications on several dimensions: patent application family size, (the number of countries where the application was filed, from DWPI), the number of claims in the PCT application, PCT filing year, the number of forward citations to the WO application (from DWPI), and whether granted and maintained in the U.S. For the 151 secondary twinsfiled in all six countries, family size was the only variable significantly different than those for all other secondary applications (17.4 countries for the twins, 8.2 for the others; p < 0.001). For the 223 twins filed in the three developing countries, the results were similar, with only family size significantly different (17.1 vs. 8.02; p < 0.001). Unsurprisingly the twin applications were thus filed more broadly, but do not appear that different based other observable characteristics. However, for some of the measures (forward cites, and share granted and maintained in US) the twins do appear slightly more "valuable" (even if the difference is not significant at
+
+---
+
+700
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+problem if we saw conflicting results, the main results are robust to whether we look at all applications (the most general set, with the largest sample size) and the two sets of twin applications (the most comparable applications, but a select set, with smaller sample sizes). In neither India nor Brazil is there evidence that policies regarding secondary patent matter much (based on raw differences, or differences-in-differences when comparing to other countries without these policies). In Argentina, however, there is evidence more suggestive of effectiveness.
+
+### 4.3. Detailed analyses of secondary patent prosecution
+
+Grant rates on secondary patents may be problematic as indicators of the effectiveness of developing country policies. They are driven by other features of national patent systems and based on different sets of national filings. We tried to account for these issues using the difference-in-difference and twins approaches, respectively. A different “bottom up” way to assess the effects of the three developing countries' approaches to secondary patents is to examine the details of patent prosecution. This approach has the advantage of allowing us to see exactly what roles the policy provisions may have had in the prosecution of the applications, and to better understand other influences on national grant rates for secondary patents.
+
+We determined whether applications filed in these three countries were granted, pending, rejected, or abandoned or withdrawn by the applicant. For the rejected Indian applications we determined the grounds cited by the patent office. In particular, we noted whether the specific provision targeting secondary patents, Section 3(d), was cited by examiners, and if so whether it was cited alone or along with other substantive grounds for rejection. For the Brazilian filings, in addition to outcomes at the patent office, we collected information on any role ANVISA had in the examination process. For applications filed in Argentina, we looked specifically at applications for "second uses," prohibited as of 2002, and distinguished between those examined before and after the patent office's more restrictive guidelines were introduced in 2012. 14
+
+Fig. 4(a) shows detailed outcomes for secondary applications in India. The grant rate is 40%, similar to what we saw above. A small share (5%) remains pending in India. Roughly 27% were withdrawn or abandoned before examination. For these applications, prosecution could not directly have been affected by Section 3(d). Of the 194 applications that the Indian Patent Office rejected, roughly 28% of the full set, nearly three quarters of these did not make any mention of 3(d). Instead, these applications were rejected typically for failing to satisfy conventional patentability grounds, such as novelty and inventive step. Forty-eight applications (7% of the full set, one quarter of the rejected applications) were rejected on grounds including Section 3(d), though the reasons cited for rejection in nearly all of these cases included conventional patentability grounds too. Thus there is reason to believe that these applications may have been
+
+rejected even in the absence of Section 3(d). Only two applications for secondary patents were rejected solely on 3(d) grounds. These data do more than reinforce our previous findings that 3(d)'s contribution to the overall outcomes for secondary patents filed in India is less substantial than many expect; they also suggest that, even for rejected applications, 3(d) plays a small independent role.
+
+Fig. 4 (b) shows detailed outcomes in Brazil. As all of the crossnational analyses have revealed, Brazil has a low grant rate: only 5 % of applications for secondary patents in Brazil were granted. Eleven percent of the applications remain pending in Brazil, reflecting the country's substantial backlog. The modal outcome in Brazil is withdrawal before the completion of examination, an outcome experienced by nearly 60 % of the applications. Applications in Brazil are withdrawn overwhelmingly because the applicant stopped paying fees, before the application underwent substantive examination. Withdrawal, and not rejection, is the main explanation for Brazil's low grant rate.
+
+Prior Consent rejections are rare: only 19 of the secondary applications (10% of rejected applications, and 2% of the full set) were rejected after being handled by ANVISA. Moreover, since we code Prior Consent rejections liberally, including in this category any application that was seen by ANVISA that was later rejected by INPI or withdrawn, these figures actually overstate the role of the patent examiners in the health agency. Of the 19 Prior Consent rejections, only four were under the old workflow, where the applications were first approved by INPI before being reviewed by ANVISA. Of the remaining 15 under the new workflow, where applications go to ANVISA first, only one was denied consent by ANVISA.15 In sum, looking at the entirety of secondary applications filed in Brazil in our dataset, in only one half of one percent of these (5 of 972) did an application end up either rejected or withdrawn following ANVISA's denial of consent.
+
+Fig. 4(c) shows detailed outcomes for Argentina. As in Brazil the grant rate is low, at just 7%. The rejection rate is low, too, at roughly 10%. Here too, in fact more than in the other countries, withdrawal before substantive examination is the dominant outcome.
+
+Unlike 3(d) and Prior Consent, which target broad classes of secondary patents, the Argentinian policy during most of the period under study focused on one type of secondary patent: second uses. Unfortunately we do not have information on specific grounds for rejection in Argentina, so our ability to evaluate the role of this specific provision is limited. However, our application codings (see Appendix A ) allow us to identify applications that only have use claims. None of the 73 pure use applications filed in Argentina were granted, suggesting that the 2002 restriction was effective. However, since pure use applications account for only 15% of the secondary applications in Argentina the specific provisions cannot be the main explanation for the low grant rate. 16
+
+What about the new, more restrictive examination guidelines introduced in 2012, which target a broader range of secondary patents? Of the 533 secondary applications filed in Argentina, 137 were examined under the new guidelines. The grant rate for secondary patent applications fell from 9% to less than 1% after the new guidelines. While in part this reflects censoring (newer applications are more likely to be pending) it is notable that the grant rate for primary applications fell by much less, from 14 to 10%. This suggests the new guidelines may be having an effect as well. But since
+
+conventional levels). Of course it is also possible that they are different on unobservables, which is why it is important to interpret results from these analyses together with those from the other complementary approaches.
+
+14 Our analyses are based on all national phase applications filed in each country that emanate from our original set of PCT applications, and that some PCTs have multiple national phase applications. From the 2964 secondary PCT applications in our sample, this approach yields 697 applications in India, 972 in Brazil, and 533 in Argentina. While we calculated grant rates above based on whether any of the national applications were granted, here we examine outcomes for all national phase applications, since we are calculating a range of outcomes (beyond whether “any national phase granted”) that may vary within individual PCT applications. For example, if one PCT was linked to four national phase applications in a country, in the earlier analyses we classified the PCT as granted in the country if any of the four applications were granted, while here we treat each of the four as discrete applications with their own outcomes.
+
+15 Of the remaining applications, some received ANVISA's consent but were later rejected by INPI or withdrawn, and some were withdrawn before ANVISA could make a judgement.
+
+16 It is possible that the provision was used to remove pure use claims from patents whose other claims were granted. In Argentina, as in the other countries, we cannot example claim narrowing with the data currently available, but hope to look at this in future research.
+
+---
+
+B.N. Sampat, K.C. Shodlen / Research Policy 46 (2017) 693-707
+
+701
+
+![Figure](figures/figure_002.png)
+
+Fig. 4. Detailed outcomes for secondary applications.
+
+Note: Grant rates conditional on filing.
+
+only a quarter of the secondary applications in our sample were examined after the new guidelines were introduced, they cannot be the main explanation for Argentina's low grant rate either.
+
+A salient feature in each country is withdrawal of applications before substantive examination. In a final set of analyses, focusing on secondary applications that were withdrawn in India, Brazil, and Argentina, we determined U.S. status of the matched twin application (for the applications that had national phase filings in the U.S.). As Fig. 5 shows, in each country withdrawn applications are disproportionately those that did not get granted in the U.S. or were granted and not renewed: 78% in India, 84% in Brazil, and 76% in Argentina. Given the preponderance of withdrawn applications among the non-grants in these countries this suggests that an important feature of the grant rate for secondary patents in developing countries is learning that applications are not successful or worth pursuing even in the U.S. The lag between global filing and examination in the developing countries may serve as a filter, a point to which we return below. $^17$
+
+## 5. Discussion
+
+The data reveal considerable heterogeneity in cross-national grant rates on secondary patents. The U.S. and European patent offices have the highest grant rates. Brazil and Argentina have the lowest grant rates, Japan and India are in the middle. This clustering is consistent throughout all of the analyses, looking at the full set of secondary applications, at segments stratified according to various
+
+measures of importance, and the twin applications filed in all six countries.
+
+The primary aim of the analyses was to assess the effectiveness of developing countries' specific instruments to restrict secondary patents. Evidence from both the comparative grant rates and detailed analyses of patent prosecution points to the limited direct effects of these measures in India and Brazil.
+
+However, the explanations for why Section 3(d) and Prior Consent are having only small direct roles may be different in the two countries. In India, both the lack of difference in primary vs. secondary grant rates, and the low utilization of 3(d) are consistent with a standard account of institutional weakness in developing countries: formal rules are often not consistently enforced ( Levitsky and Murillo, 2009 ) . The patent office operates under severe resource constraints, and is faced with the challenge of working through substantial quantities of applications on new, previously unpatentable subject matter. Moreover, India's patent examiners are tied into global networks, through training and via reliance on foreign databases and prosecution materials ( Kapczynski, 2009 ; Drahos, 2010 ; Sampat and Amin, 2013 ) . Such conditions, may limit the extent to which examiners employ Section 3(d) in the course of evaluating pharmaceutical patent applications.
+
+Another explanation for the limited role played by Section 3(d) might be that the scope for independent action of this provision may be constrained. Nearly all rejections in our dataset that cited Section 3(d) also gave other grounds for denying the patents. Although Section 3(d) was involved in these rejections, this specific provision may not have been needed to reject these applications, which were rejected on other grounds too. To the extent that 3(d) is similar, in terms of the sorts of applications it is used to reject, to conventional patentability criteria, such as lack of novelty or inven-
+
+17 Since Japan has a formal deferred examination system applications there may also benefit from post-application learning (Yamauchi and Nagaoka, 2015). Consistent with this, of those granted and renewed in the U.S. 70% are granted in Japan, compared to just 28% of those granted but not renewed in the U.S. This suggests that the high non-grant rate in Japan too may reflect withdrawal of applications deemed no longer worth pursuing, though absent detailed outcome data in Japan it is difficult to test this directly.
+
+---
+
+702
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+![Figure](figures/figure_002.png)
+
+Fig. 5. US Status Distribution for applications withdrawn or abandoned in developing countries. Note: For applications with national stage filings in U.S.
+
+tive step, then the limited role played by 3(d) per se becomes less surprising. 18
+
+Brazil's measures to restrict secondary patenting also appear to have had a limited direct effect. Though Brazil has a low grant rate, the differences between primary and secondary grant rates are indistinguishable, and our detailed analyses revealed negligible involvement by ANVISA. Difficulties of inter-agency coordination, often cited as a factor complicating enforcement (e.g. Dimitrov, 2009 ) could provide one explanation for this surprising finding. The Prior Consent system relies on the participation of two separate state agencies, and disagreements between INPI and ANVISA over Brazil's approach to pharmaceutical patents have been intense ( Shadlen, 2011 ; Kunisawa, 2009 ) . However, these disagreements and conflicts cannot account for the finding the finding that the effects of ANVISA on patenting outcomes are so small. Examination in Brazil is sequential, not simultaneous; minimal coordination is required. For most of applications in our dataset, prosecuted under the old workflow, all the work was done by INPI itself. Few of these applications ever cleared the first hurdle at INPI to be passed to ANVISA, but rather they were rejected or withdrawn while still at the patent office. For reasons explained above, the backlog at the patent office seems like a better explanation of limited role of ANVISA than problems with inter-agency coordination. $^19$
+
+In Argentina, unlike in either of the other developing countries, we do see differences in primary-secondary grant rates. The analyses of prosecution details also suggest that both the prohibitions on second use patents and the more restrictive examination guide-
+
+lines that came into place in 2012 may be having their intended impact. One explanation for this is that Argentina’s approach is comparatively easier to employ. It does not rely on application of a different sort of patentability criteria (as in India) or the participation of multiple agencies (as in Brazil), but simply constitutes instructions to patent examiners for how to go about rejecting secondary patents using traditional patentability criteria (e.g. lack of novelty or inventive step).
+
+However effective Argentina's policies toward secondary patents may have been, since only a minority of the applications in our dataset was subject to these measures they cannot explain the whole story. As in Brazil, one explanation for the low overall grant rate seems to be slow prosecution and applicants' decisions to give up on applicants they deem to be less important. In Argentina this effect may be bolstered by a policy aimed specifically at clearing the backlog. The patent office often issues requirements that applicants report on the status of applications filed abroad. If firms fail to respond, the pending applications are converted to abandoned or withdrawn.
+
+Another part of the story in Argentina is that, even before the new guidelines formally came into place they may have informed examiner practice. That is, the official establishment of new guidelines in 2012 may have reflected a new approach to examination that had already been adopted. Whereas in India and Brazil laws on the books have had limited direct impacts on patent prosecution in practice, in Argentina changes in patent prosecution in practice may have later been codified in laws on the books.
+
+## 6. Conclusion
+
+In this paper we used novel data on a large sample of filings emanating from PCT applications in three developed countries and three developing countries with restrictions on secondary patenting. We coded the applications as secondary (or primary) and identified a large set of “twin” filings across multiple jurisdictions
+
+18 This explanation would raise the broader question about why 3(d) is so controversial if its scope is small.
+
+15 Now that ANVISA sees applications first, under the new workflow, it will become more involved. It is important to keep in mind the important difference in the pool of applications ANVISA would have seen under the old workflow (applications approved by INPI) and under the new workflow (all applications for pharmaceutical patents).
+
+---
+
+B.N. Sampat, K.C. Shodlen / Research Policy 46 (2017) 693-707
+
+703
+
+to compare grant rates and evaluate restrictions toward secondary patents. We also examined the details of patent prosecution in the three developing countries to better understand the role of specific arrangements toward secondary patents, alone and in comparison to other influences on patenting outcomes.
+
+We found considerable differences in cross-national grant rates for secondary patents, despite the harmonization aimed for by TRIPS. Conditional upon filing, grant rates for secondary patents in the six countries ranged from 5% (Brazil) to 52% (U.S.). In all countries grant rates are higher for applications filed more broadly and (in all countries but the U.S.) for applications more vigorously pursued in the U.S., indicating grant rates reflect the extent of applicant effort in pursuing the applications. But even for these more important applications, differences in grant rates of secondary patents persist. These findings suggest that more attention be paid to differences in countries' pharmaceutical patent systems, to look “inside the black box” of patent examination to try to understand how countries' patent systems function in practice.
+
+We did not find much evidence that developing countries' policies to restrict secondary patents are driving outcomes. In the case of two countries, India and Brazil, the policies seem to be having little effect at all: grant rates for secondary patents are not lower than grant rates for primary patents in these countries, which they should be if policies designed specifically to achieve that outcome were working, and the detailed analyses of patent prosecution also reveal Section 3(d) and Prior Consent to be playing only limited direct roles. Argentina's measures to restrict secondary patents appear to be more effective: grant rates for secondary patents are considerably lower than those for primary patents, patents on second medical use are not granted, and the secondary patent grant rate following the introduction of the new guidelines has further declined. But the overall impact on secondary patenting outcomes of Argentina's measures is small: the bulk of applications in our sample (pre-2012, non-use applications) were not subject to either the prohibition on second use patents or the new examination guidelines.
+
+The detailed analyses also illustrated the potentially important role of other features of countries' patent systems. In all three countries large backlogs of applications and timing of TRIPS implementation contribute to delays in initiating examination, so that it often does not begin until information from developed country offices has become available. In this context, if resource-contained patent offices were inclined to rubber stamp developed country actions, as some previous analysts have suggested ( Drahos, 2010 ) , this would lead to high grant rates for secondary applications. But our analyses suggests another effect of the backlog: the delays they introduce may serve as a filter, leading firms to abandon less important applications based on information from other countries. That is, backlogs may unintentionally mimic features of deferred examination systems, leading to lower grant rates.
+
+One implication of our analyses is thus that administrative and procedural features of patent systems may play important roles in patent prosecution, and may actually exert a greater influence on grant rates than laws on the books per se. Although this is consistent with recent work on the USPTO ( Lemley and Sampat , 2012 ; Frakes and Wasserman forthcoming ), since this was not the main focus of our research, more work is needed to examine exactly how these features influence outcomes in developing countries.
+
+While we found minimal direct effects of countries' provisions toward secondary patents, there may be more indirect effects that our data do not capture. In India, it is possible that 3(d) influences prosecution by affecting how examiners apply other aspects of the patent law. That is, Section 3(d) might alter Indian patent examiners' overall approach to assessing applications and using conventional patentability criteria, such that applications rejected on grounds above and beyond 3(d) might not be rejected at all if not
+
+for this distinct provision in Indian patent law. And in Brazil, prior to the introduction of the new workflow in 2012, it is possible that ANVISA's participation in examination may have created incentives for rejection at the patent office. INPI examiners knew that favorable decisions would subsequently be scrutinized by counterparts at ANVISA while rejections and other applications without a decision would not be.
+
+Another potential indirect effect is on selective filing. We believe this is likely limited given the timing of when the applications were filed and when the policies were introduced. A similar possibility is selective withdrawal of secondary applications in response to countries' policies toward these types of applications, though we showed above that most withdrawal in India, Brazil, and Argentina is for applications that were revealed to be not worth renewing even in the U.S. While we do not view this particular indirect effect as being important, we cannot rule it out completely.
+
+There are other limitations as well. While our analysis focuses on grant rates, it is possible that the main impact of the provisions is on narrowing claims in granted patents, or that the provisions matter most for specific kinds of secondary patents. It is also likely that there are changes over time. Our research focuses on applications filed from 2000 to 2002, for reasons discussed. But as with all policies it is possible that there are implementation lags, and that countries' measures to restrict secondary patents matter more for more recently filed applications. Another limitation, reflecting the difficulty of obtaining reliable data on patent prosecution in developing countries, is that we focus on a relatively small set of developing countries. Extending to a broader set of comparable countries, including those without restrictions is also a promising avenue for future research.
+
+The universalization of pharmaceutical patenting through TRIPS generated significant controversy. The spread of secondary patenting has been the subject of particular concern among academics, activists, and policymakers, based on fears that such patents may restrict competition and thus inhibit access to medicines. Despite the attention that secondary patenting has received, we know little about what is happening on the ground. Our analyses provide the first cross-national comparative data on secondary patent grant rates, and on the actual impact of policies aiming to restrict them in developing countries. They reveal surprising but robust patterns of cross-national differences in grant rates. Subject to the limitations listed above, the analyses also suggest that the restrictions in developing countries are having limited direct effect. From a methodological perspective, they also point to the benefits of combining application-level analyses with prosecution details to understand the functioning of patent systems and the impact of patent policies.
+
+## Acknowledgements
+
+This research was supported by a grant from the Economic and Social Research Council (ES/K010999/1). Kim Grauer, Maria Saldarriaga, Elia Trippel and Jennifer Reid provided excellent research assistance. We thank Scott Hemphill, Rochelle Dreyfuss, Michael Frakes and Melissa Wasserman for useful comments, and are grateful to Carsten Fink and Julio Raffo at WIPO for helping us access the WIPO Statistics Database.
+
+## Appendix A.
+
+### • Coding guide
+
+A coding guide was provided to the coder to categorize the 5193 PCT applications. It is adapted from a guide designed by Scott
+
+---
+
+704
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+Hemphill that was used to code US patent grants (Hemphill and Sampat, 2011). Below, we excerpt the first page of the coding guide.
+
+## General
+
+We want to code the information in the published application (the WO document). To do so, click through the link provided for each application, which will take you to the Google transcription of the application. (This is useful since Google typically does translation for us, and the layout is pretty clean.) If you need the actual PDF file, you can access it through the PatentScope and/or Espacenet links provided in the Google patent file. We anticipate you will use information in the independent and dependent claims for the coding, supplemented by information in the title, abstract, description as needed. If you use any information beyond this please indicate this in the notes field.
+
+## • Coding
+
+Our main goal is to code applications by type. There are five broad categories of claims. A patent can, and often does, include more than one category of claims:
+
+- • A: active ingredient (see specific descriptions of A1–A4 below)
+• B: formulation or composition C: method of use
+• D: other, but related to the drug
+• E: biologic
+For each patent, indicate all categories that apply to a patent. For active ingredient claims, we want to distinguish the four subcategories:
+
+- • A1: active ingredient.
+• A2: is for polymorphs or other crystal forms.
+• A3: is for enantiomers or other isomers.
+• A4: salt, metabolite, or intermediate. Also pre-metabolites and
+derivatives
+## Identifying national phase application numbers and outcomes
+
+### EPO, JPO, and US
+
+We obtained EPO, JPO, and US national phase numbers from the WIPO Statistical Database, the same source we used to construct the basic dataset.
+
+We also obtained outcomes data from PATSTAT. For a random sample of 100 applications, we verified these sources provided essentially identical grant rate information as was determinable from the EP Register, the JPO Website, and USPTO Public PAIR. The U.S. grant rate calculated from PATSTAT is based on all grants from a given priority, so includes grants to all “child” applications (continuations) which we also verified against PAIR.
+
+### India
+
+We obtained national phase applications in India from PatentScope, and Indian outcomes from the IPO Website. Any granted application was classified as Granted. We considered an Indian application to be Abandoned/Withdrawn if the status on the IPO website is withdrawn without stated reason, or withdrawn under 11(B)4. Section 11(B)4 withdrawals are those where no request for examination was made. Given the time elapsed since filing, we assume that applications “Not Yet Published” were withdrawn before examination. We also grouped Section 9(1) Abandonments as withdrawals: these are cases where a complete specification was not filed.
+
+We classify applications as Rejected if they were abandoned under Section 21(1). Section 21(1) abandonments are typically those where there was a failure to respond to objections in a First Examination Report (FER) within the time limits prescribed. Our
+
+logic here is that these applications were abandoned because of the examiners' objections. It is also possible that the lack of response was for other reasons (e.g. the firm went out of business, the technology no longer interesting to the firm, or problems with the application were discovered at another patent office). Accordingly our analysis overstates rejection rates. Refusals through Controller decisions (including those indicated as Section 15 and 16 rejections) were also classified as rejections. Refusals through Controller Decisions result when a controller is unsatisfied with an applicant's response to the FER and/or the Controller refuses an application where there is a pre-grant opposition. As discussed more below, we focus on these Rejected cases when we examine how 3(d) is affecting rejection rates.
+
+We classified any application that was Awaiting Examination or Under Examination as Pending. The majority of these Pending applications are Awaiting Examination. Given that a request for examination (RFE) must be filed by now we could have also grouped these with withdrawn applications. Doing so would not affect calculation of grant rate or our assessment of the role 3(d).
+
+How might 3(d) affect whether or not an application is granted? In the process described above, 3(d) could directly lead to rejections in three main ways: (1) The examiner raises 3(d) in an FER, resulting in abandonment of the application, or (2) The controller raises 3(d) on reviewing arguments from response to FER, generating a rejection, or (3) A pre-grant opposition raises 3(d) objections, which are upheld in a Controller Report rejecting the application. Importantly, withdrawals of applications before RFEs are filed cannot be directly due to 3(d), since there are no examination documents prior to RFEs.
+
+To examine the direct role of 3(d) in rejections, we collected information from FER and Controller Reports for applications that have rejections on the merits on the role of 3(d). This set includes all non-granted applications, except those withdrawn before a request for examination was made. For each of these “Rejected” applications we determined if 3(d) was listed as a reason for rejection, and, if so, if this was the only grounds for rejection: We thus further categorize the Rejected applications as: Rejected, No 3(d); Rejected 3(d) Only; Rejected 3(d) Plus.
+
+### Brazil
+
+We obtained Brazilian national phase application numbers from the Derwent World Patents Index (and verified against information from PATSTAT). We obtained Brazilian outcomes by searching the INPI website. We dropped a small number of applications where PCT information on the national website did not match the original PCT number.
+
+Classifying outcomes in Brazil is complicated because of the nature of Brazil's pharmaceutical patent system. As discussed in the text, Brazil has a shared examination system, with pharmaceutical patent applications examined by both the National Institute for Industrial Property (INPI) and the Ministry of Health's health surveillance agency (ANVISA). Pharmaceutical patents can only be granted if both INPI and ANVISA approve. From 2001–2012 INPI would examine applications first, and only if it approved the application (i.e. issued a technical report indicating that the application fulfilled the criteria for patentability in the patent law) would it be sent to ANVISA for subsequent review. As of May 2012 the workflow was inverted, such that INPI immediately forwards all pharmaceutical patent applications to ANVISA, where they are reviewed and then returned to INPI for subsequent examination.
+
+To track outcomes, and to see ANVISA's role in outcomes, we searched all applications at both INPI and ANVISA. The INPI website provides data on each transaction that occurs at the patent office during the course of examination, allowing us to determine if applications were granted, rejected, withdrawn or abandoned, or still pending. While it is possible to know if an application was sent to ANVISA in the course of examination, from the INPI website
+
+---
+
+B.N. Sampat, K.C. Shodlen / Research Policy 46 (2017) 693-707
+
+705
+
+alone it is not possible to determine what actions were taken by the examiners at the health agency. To overcome this constraint we also consulted two ANVISA documents that indicate the actions that the health agency has taken on each application it has received under the old workflow (through May 2012) and the new workflow (since May 2012). Using data from these two sources we determined whether Brazilian patent applications were granted, rejected, abandoned/withdrawn, or pending, and ANVISA's role.
+
+Granted patents were approved by INPI and ANVISA. Applications with non-grant final determinations may be rejected or abandoned/withdrawn. For applications rejected by INPI, we determined whether the application was rejected by INPI alone or whether ANVISA was involved. Applications rejected solely by INPI are recorded as Rejected, INPI. The category Abandoned/Withdrawn refers to applications that were classified as “arquivado” by INPI, ordinarily on account of applicants not paying fees, never requesting examination in the first place, or not responding to INPI communication regarding non-substantive aspects of the application (“exigência”). When applicants fail to respond to substantive objections (“ciência”) these are classified as Rejected, INPI.
+
+Neither Rejected, INPI nor Abandoned/Withdrawn categories involve ANVISA directly. Rejected, PC refers to applications with final determinations of either reject or abandoned/withdrawn that, at some point in the process, were received by ANVISA. Rejected, PC includes applications initially approved by the INPI and sent to ANVISA, but where ANVISA did not consent to a grant and the INPI subsequently rejected the application. This category also includes applications: (1) initially approved by the INPI and sent to ANVISA, but where in the course of ANVISA examining the application became “arquivado” at INPI, (2) where ANVISA finished its examination and denied consent, but rather than being rejected by INPI it ended abandoned or withdrawn; and (3) applications under the new workflow that ANVISA approved but were either rejected at INPI or abandoned/withdrawn before INPI made a final decision. Thus if an application was received by ANVISA at any point in the process and ended with a non-grant final determination it is classified as Rejected, PC.
+
+Pending applications lack final determination. For applications filed under this old workflow, this includes a handful of socalled “frozen” applications ( Shadlen , 2011 ) where ANVISA denied consent but INPI did not issue a verdict and the application remained at the patent office, neither granted, rejected, nor abandoned/withdrawn.
+
+The interpretation of pending applications is complicated by the introduction of the new workflow in 2012, whereby pharmaceutical patent applications go to ANVISA before INPI examines them. Pending applications may be: at ANVISA awaiting initial examination having been approved by INPI (old workflow), at INPI awaiting final determination after having been returned with a negative ruling by ANVISA (old workflow), at ANVISA awaiting initial examination (new workflow), at INPI awaiting examination having been returned with ANVISA's consent (new workflow).
+
+## Argentina
+
+For Argentina, we used information from PATSTAT on all national filings, linked via priority numbers. Recall that Argentina is not a PCT country, so there are no “national phase” filings.
+
+We obtained from the Argentinian patent office a dataset of all patent applications filed in Argentina from 2000 to 2005, with bibliographic and priority details, as well as information on final status. We then matched these against the Argentinian application numbers in PATSTAT to determine which of the applications in our sample were filed in Argentina. For Argentina we have four outcomes: Granted, Rejected, Abandoned/Withdrawn, and Pending. The Argentine patent office reports three different non-grant outcomes: abandoned, voluntary withdrawal, and forced withdrawal.
+
+Despite the seemingly self-explanatory labels, these do not map systematically on to different situations, and their use tends to be inconsistent as well, so to avoid misinterpretation we combine these three into the single category of Abandoned/Withdrawn. (Of these secondary applications filed in Argentina that we classify as abandoned/withdrawn, just under two thirds of were “forced withdrawal” and just under one third “abandoned,” with less than 5% “voluntary wiithdrawal.”)
+
+For the subset of analyses where we consider the effects of Argentina's new examination guidelines, we used the date of the patent office decision. After the new guidelines were introduced in May 2012, all examination decisions were to be made in conformity with these new rules, including applications where examination had already commenced. As our data from the patent office include the dates of the patent office's decisions (not just the dates decisions were communicated to applicants, but the decisions themselves), we identify applications with decision dates after 8 May 2012 and distinguish these as being treated under the new guidelines. All Pending applications are, by definition, part of the new guidelines subset too.
+
+## Regression models
+
+As a complement to the graphical presentation in the paper, we also examined the main questions in the paper in a regression framework. Doing so allows us to control for application specific characteristics, and facilitates inference testing.
+
+We estimated linear probability models relating the probability that an application i filed in country j is granted, with heteroskedasticity-consistent standard errors, clustered on the PCT application.
+
+Model 1 examines differences in secondary patent grant rates by country. It is the regression analog of Fig. 1 (a) in the paper, but also controls for PCT filing year, the number of claims in the PCT application, family size, and whether their are any domestic inventors. The estimation sample includes 9794 observations for the 2964 secondary applications, one for each national filing in the US, EPO, Japan, India, Brazil, and Argentina. It shows that relative to the reference category, US applications, secondary grant rates are significantly lower in Japan, Brazil, India, and Argentina, with the largest differences in Brazil and Argentina. But there is no statistically significant difference between the U.S. and EPO. For each of the developing countries, we can reject the hypotheses that the grant rate is the same as in the US and EPO at the 1 % level. We cannot reject the null that the Indian rate is equivalent to Japan (p = 0.59), but we can reject this for Brazil and Argentina (grant rate equivalent to Japan) each at the 1 % level. Consistent with what we saw in the figures, the grant rate increases with family size. And we also see evidence here that having domestic inventors increases the grant rate, consistent with the idea of "home country bias" in international patenting ( Jensen et al. , 2006 ) . But even after controlling for these factors, the patterns and magnitudes of differences in secondary grant rates between countries are similar to what Fig. 1 (a) showed. One thing to note is that the differences between India and Japan are less pronounced after controlling for family size, which we also saw in Fig. 2 in the paper.
+
+Model 2 is the main difference-in-difference model, the regression analog of comparing 1(a) and 1(b) in the paper. Here, the estimation sample includes all applications, primary and secondary. The model allows for differences in primary and secondary grant rates to vary by country. Only three countries have a significant difference in grant rates between primary and secondary patents: the US (where secondary patents have a 22 percentage point lower grant rate), Argentina (where secondary patents have a 8 percentage point lower grant rate), and India (where secondary patents have a 6 percentage point higher grant rate). As in the
+
+---
+
+706
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+figures in the text, there is no evidence that the gap between secondary and primary grant rates is differentially larger in India or Brazil than in the developed countries. For Argentina, the gap is significantly lower than that in the EPO (p < 0.01) and Japan (p < 0.05).
+
+Model 3 is estimated over secondary applications only, and includes application specific fixed effects. These fixed effects absorb any unobserved differences across applications, i.e. the coefficients on the country specific grant rates are based on within-application variation. (Of the variables listed above, the only one with within application variation, other than the country dummies, is the indicator for whether the application is domestic.) Note that unlike the secondary twins analyses from Fig. 1 (c) and (e) of the paper, identification here comes not from applications filed in all countries, but from grant rates for any applications filed in multiple countries. The overall patterns across countries are similar to those observed from Model 1. Secondary grant rates are significantly lower than the reference category (the US) in Japan, India, Brazil, and Argentina, each at the 1% level. (However there is no significant difference between the US and EPO.) The secondary patent grant rate is lowest in Brazil, followed by Argentina, then India. There is no statistically significant difference between the Argentinian and Brazilian grant rates (p = 0.07), but each have significantly lower grant rates than India at the 1% level.
+
+<table><tr><td></td><td>(1)Granted</td><td>(2)Granted</td><td>(3)Granted</td></tr><tr><td>EP</td><td>0.00594 (0.624)</td><td>−0.205*** (0.000)</td><td>0.0157 (0.256)</td></tr><tr><td>JP</td><td> $-0.235^{\cdots}(0.000)$ </td><td> $-0.427^{\cdots}(0.000)$ </td><td> $-0.208^{\cdots}(0.000)$ </td></tr><tr><td>BR</td><td> $-0.584^{\cdots}(0.000)$ </td><td> $-0.815^{\cdots}(0.000)$ </td><td> $-0.505^{\cdots}(0.000)$ </td></tr><tr><td>IN</td><td> $-0.224^{\cdots}(0.000)$ </td><td> $-0.501^{\cdots}(0.000)$ </td><td> $-0.146^{\cdots}(0.000)$ </td></tr><tr><td>AR</td><td> $-0.553^{\cdots}(0.000)$ </td><td> $-0.696^{\cdots}(0.000)$ </td><td> $-0.468^{\cdots}(0.000)$ </td></tr><tr><td>PCT Year 2001</td><td> $-0.0295^{\circ}(0.042)$ </td><td>−0.00462 (0.678)</td><td></td></tr><tr><td>PCT Year 2002</td><td>−0.0161 (0.239)</td><td>0.0269** (0.009)</td><td></td></tr><tr><td>Number of Claims</td><td>−0.000332 (0.129)</td><td> $-0.000338^{\circ}(0.033)$ </td><td></td></tr><tr><td>Family size</td><td>0.0265*** (0.000)</td><td>0.0276*** (0.000)</td><td></td></tr><tr><td>Any Domestic Inventor</td><td>0.0922*** (0.000)</td><td>0.0985*** (0.000)</td><td></td></tr><tr><td>Secondary, US</td><td></td><td> $-0.216^{\cdots}(0.000)$ </td><td></td></tr><tr><td>Secondary, EP</td><td></td><td>−0.00327 (0.813)</td><td></td></tr><tr><td>Secondary, JP</td><td></td><td>−0.0239 (0.073)</td><td></td></tr><tr><td>Secondary, BR</td><td></td><td>0.0123 (0.294)</td><td></td></tr><tr><td>Secondary, IN</td><td></td><td>0.0552* (0.044)</td><td></td></tr><tr><td>Secondary, AR</td><td></td><td> $-0.0758^{\cdots}(0.000)$ </td><td></td></tr><tr><td>Constant</td><td>0.289*** (0.000)</td><td>0.470*** (0.000)</td><td>0.509*** (0.000)</td></tr><tr><td>Observations</td><td>9794</td><td>16174</td><td>9794</td></tr><tr><td>PCT fixed effects</td><td>No</td><td>No</td><td>Yes</td></tr></table>
+
+
+p-values in parentheses.
+
+'p < 0.05. ''p < 0.01. '''p < 0.001.
+
+## References
+
+Abud, M.J., Hall, B., Helmers, C., 2015. An empirical analysis of primary and secondary pharmaceutical patents in Chile. PLoS One 10, e0124257, http://dx. doi.org/10.1371/journal.pone.0124257.
+
+Amin, T., Kesselheim, A.S., 2012. Secondary patenting of branded pharmaceuticals: a case study of how patents on two HIV drugs could be extended for decades. Health Aff. (Millwood) 31, 2286–2294, http://dx.doi.org/10.1377/hlthaff.2012. 0107.
+
+Barbosa, D.B., 2013. Patents and the Emerging Markets of Latin America: Brazil. In Emerging Markets and the World Patent Order. In: Abbott, Frederick M., Correa, Carlos M., Drahos, Peter (Eds.). Edward Elgar Publishing, pp. 135–151.
+
+Basheer, S., Reddy, P., 2008. The Efficacy of Indian Patent Law: Ironing Out the Creases in Section 3(d) (SSRN Scholarly Paper No. ID 1086254). Social Science Research Network, Rochester, NY.
+
+Basso, M., 2006. Intervention of health authorities in patent examination: the Brazilian practice of the prior consent. Int. J. Intellect. Prop. Manage. 1, 54–74. http://dx.doi.org/10.1504/IJIPM.2006.011022.
+
+Berndt, E.R., Cockburn, I.M., 2014. The hidden cost of low prices: limited access to new drugs in India. Health Aff. 33, 1567–1575, http://dx.doi.org/10.1377/ hlthaff.2013.130.
+
+Burdon, M., Sloper, K., 2003. The art of using secondary patents to improve protection. J. Med. Mark. 3, 226–238, http://dx.doi.org/10.1057/palgrave.jmm. 5040125.
+
+Carley, M., Hedge, D., Marco, A., 2015. What is the probability of receiving a U.S. Patent. Yale J. Law Technol. 17, 203.
+
+Christie, A.F., Dent, C., Liddicoat, J., 2016. The examination effect: a comparison of the outcome of patent examination in the US, Europe and Australia. John Marshall Rev. Intell. Prop. Law 16, 21–43.
+
+Correa, C.M., Balleri, C., Giulietti, M., Lavopa, F., Musetti, C., Palopoli, G., Pippo, T., De la Puente, C., Lowenstein, V., 2011. Patentes, suministro de medicamentos y protección de la Salud Pública; Patents, supply of medicines and protection of public health. Rev. Argent. salud publica 2, 19–27.
+
+Correa, C.M., 2007. Guidelines for the Examination of Pharmaceutical Patents: Developing a Public Health Perspective. ICTSD.
+
+Correa, C.M., 2014. Tackling the Proliferation of Patents: How to Avoid Undue Limitations to Competition and the Public Domain (No Research Paper 52). South Centre.
+
+de Rassenfosse, G., Jaffe, A., Webster, E., 2016. Low-quality Patents in the Eye of the Beholder: Evidence from Multiple Examiners (No w22244). National Bureau of Economic Research, Cambridge, MA.
+
+Dimitrov, M., 2009. Piracy and the State: The Politics of Intellectual Property Rights in China. Cambridge University Press.
+
+Drahos, P., 2008. Trust me: patent offices in developing countries. Am. J. Law Med. 34, 151.
+
+Drahos, P., 2010. The Global Governance of Knowledge: Patent Offices and Their Clients. Cambridge University Press.
+
+Eisenberg, R.S., 2008. Pharma's Nonobvious Problem. Lewis & Clark L. Rev, Articles 12, 375-430.
+
+Ellery, T., Hansen, N., 2012. Pharmaceutical Lifecycle Management: Making the Most of Each and Every Brand. Wiley.
+
+European Commission, 2009. Pharmaceutical Sector Inquiry Report.
+
+Frakes, M., Wasserman, M., forthcoming. Is the Time Allocated to Review Patent Applications Inducing Examiners to Grant Invalid Patents? Evidence from Micro-Level Application Data. Review of Economics and Statistics.
+
+GAO, 2016. Patent Office Should Define Quality, Reassess Incentives, and Improve Clarity (No. GAO-16-490). Government Accountability Office, Washington, D.C.
+
+Hemphill, C.S., Sampat, B.N., 2011. When do generics challenge drug patents? J. Empir. Legal Stud. 8, 613–649, http://dx.doi.org/10.1111/j.1740-1461.2011. 01235.x.
+
+Hopkins, M.M., Mahdi, S., Patel, P., Thomas, S.M., 2007. DNA patenting: the end of an era? Nat. Biotechnol. 25, 185-187, http://dx.doi.org/10.1038/nbt0207-185.
+
+Howard, L., 2007. Use of patents in drug lifecycle management. J. Generic Med. 4, 230–236, http://dx.doi.org/10.1057/palgrave.jgm.4950065.
+
+Jaffe, A.B., Lerner, J., 2004. Innovation and Its Discontents: How Our Broken Patent System Is Endangering Innovation and Progress, and What to Do About It. Princeton University Press.
+
+Jensen, P.H., Palangkaraya, A., Webster, E., 2006. Disharmony in international patent office decisions. Fed. Circuit Bar. J. 15, 679–704.
+
+Kapczynski, A., Park, C., Sampat, B., 2012. Polymorphs and prodrugs and salts (Oh my!): An empirical analysis of secondary pharmaceutical patents. PLoS One 7, e49470, http://dx.doi.org/10.1371/journal.pone.0049470.
+
+Kapczynski, A., 2009. Harmonization and its discontents: a case study of TRIPS implementation in India's pharmaceutical sector. Calif. Law Rev. 97, 1571–1649.
+
+Kapczynski, A., 2013. Engineered in India: patent law 2.0. New Engl. J. Med. 369, 2163, http://dx.doi.org/10.1056/NEJMc1311602.
+
+Kesselheim, A.S., 2007. Intellectual property policy in the pharmaceutical sciences: the effect of inappropriate patents and market exclusivity extensions on the health care system. AAPS J. 9, E306–E311, http://dx.doi.org/10.1208/ aapsj0903033.
+
+Kunisawa, V.Y.M., 2009. Patenting pharmaceutical inventions on second medical uses in Brazil. J. World Intell. Prop. 12, 297–316, http://dx.doi.org/10.1111/j. 1747-1796.2009.00362.x.
+
+Löfgren, H., Williams, O.D., 2013. The New Political Economy of Pharmaceuticals: Production Innovation and TRIPS in the Global South. Palgrave Macmillan.
+
+Lanjouw, J.O., Pakes, A., Putnam, J., 1998. How to count patents and value intellectual property: the uses of patent renewal and application data. J. Ind. Econ. 46, 405–432, http://dx.doi.org/10.1111/1467-6451.00081.
+
+Lemley, M.A., Moore, K.A., 2004. Ending abuse of patent continuations. Boston Univ. Law Rev. 84, 63-118, http://dx.doi.org/10.2139/ssrn.462404.
+
+Lemley, M.A., Sampat, B., 2012. Examiner characteristics and patent office outcomes. Rev. Econ. Stat. 94, 817–827, http://dx.doi.org/10.1162/REST.a. 00194.
+
+Levitsky, S., Murillo, M.V., 2009. Variation in institutional strength. Annu. Rev. Polit. Sci. 12, 115–133, http://dx.doi.org/10.1146/annurev.polisci.11.091106.121756.
+
+Moir, H.V.J., 2013. Empirical evidence on the inventive step. Eur. Intell. Prop. Rev. 35, 246-252.
+
+PhRMA, 2016. Special 301 Submission 2016. http://phrma-docs.phrma.org/files/dmfile/special-301-review.pdf.
+
+Rathod, S.K., 2010. Ever-greening: a status check in selected countries. J. Generic Med. 7, 227–242.
+
+Reichman, J.H., 2009. Intellectual property in the twenty-first century: will the developing countries lead or follow? Houston Law Rev. 46, 1115-1185.
+
+Sampat, B.N., Amin, T., 2013. How do public health safeguards in indian patent law affect pharmaceutical patenting in practice? J. Health Politics Policy Law 38, 735–755.
+
+---
+
+B.N. Sampat, K.C. Shadlen / Research Policy 46 (2017) 693-707
+
+707
+
+Sampat, B.N., Shadlen, K.C., 2015a. TRIPS implementation and secondary pharmaceutical patenting in Brazil and India. St. Comp. Int. Dev. 50, 228–257. http://dx.doi.org/10.1007/s12116-015-9181-7.
+
+Sampat, B.N., Shadlen, K.C., 2015b. Drug patenting in India: looking back and looking forward. Nat. Rev. Drug Discov. 14, 519-520.
+
+Schultz, M., Madigan, K., 2016. The Long Wait for Innovation: The Global Patent Pendency Problem Center for the Protection of Intellectual Property. http:// cpip.gmu.edu/wp-content/uploads/2016/10/Schultz-Madigan-The-Long-Waitfor-Innovation-The-Global-Patent-Pendency-Problem.pdf .
+
+Shadlen, K.C., 2011. The political contradictions of incremental innovation: lessons from pharmaceutical patent examination in Brazil. Polit. Soc. 39, 143–174. http://dx.doi.org/10.1177/0032329211402601.
+
+Silva, H.M., 2008. Avaliação da análise dos pedidos de patentes farmacêuticas feita pela Anvisa no cumprimento do mandato legal da anuência prévia. ENSP, Rio de Janeiro.
+
+South Centre, 2011. The Doha Declaration on TRIPS and Public Health Ten Years Later: The State of Implementation Policy Brief 7, November 2011.
+
+Sternitzke, C., 2010. Knowledge sources, patent protection, and commercialization of pharmaceutical innovations. Res. Policy 39, 810–821, http://dx.doi.org/10. 1016/j.respol.2010.03.001.
+
+UNAIDS, 2011. Doha+ 10 TRIPS flexibilities and access to antiretroviral therapy: Lessons from the past, opportunities for the future.
+
+USPTO, 2015. Maintenance Fee Register. https://www.google.com/googlebooks/ uspto-patents-maintenance-fees.html#1981-present .
+
+Webster, E., Jensen, P.H., Palangkaraya, A., 2014. Patent examination outcomes and the national treatment principle. RAND J. Econ. 45, 449–469, http://dx.doi.org/ 10.1111/1756-2171.12053.
+
+Yamauchi, I., Nagaoka, S., 2015. An economic analysis of deferred examination system: evidence from a policy reform in Japan. Int. J. Ind Organ. 39, 19–28, http://dx.doi.org/10.1016/j.ijindorg.2015.01.003.
+

processed/00130_W2796464785/document.md

@@ -0,0 +1,570 @@
+![Figure](figures/figure_000.png)
+
+![Figure](figures/figure_001.png)
+
+6 OPEN ACCESS
+
+Citation: Sampat BN, Shadlen KC (2018) Indian pharmaceutical patent prosecution: The changing role of Section 3(d). PLoS ONE 13(4): e0194714. https://doi.org/10.1371/journal.pone.0194714
+
+Editor: Joshua L. Rosenbloom, Iowa State University, UNITED STATES
+
+Received: October 28, 2017
+
+Accepted: March 8, 2018
+
+Published: April 2, 2018
+
+Copyright: © 2018 Sampat, Shadlen. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
+
+Data Availability Statement: All relevant data and code are available from: https://dataverse.harvard. edu/dataverse/boffindata. Detailed instructions for how to access all of the data files and instructions for replication have been included as a Supporting Information file.
+
+Funding: The research was funded by the Economic and Social Research Council, Grant ES/ K010999/1 (http://www.researchcatalogue.esrc.ac. uk/grants/ES.K010999.1/read) (KCS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
+
+RESEARCH ARTICLE
+
+# Indian pharmaceutical patent prosecution: The changing role of Section 3(d)
+
+Bhaven N. Sampat$^{1,2c}$, Kenneth C. Shadlen$^{3c*}$
+
+1 Department of Health Policy and Management, Mailman School of Public Health, Columbia University, New York, NY, United States of America, 2 NBER, Cambridge, MA, United States of America, 3 Department of International Development, London School of Economics and Political Science, London, United States of America
+
+☺ These authors contributed equally to this work.
+
+* k.shadlen@lse.ac.uk
+
+## Abstract
+
+India, like many developing countries, only recently began to grant pharmaceutical product patents. Indian patent law includes a provision, Section 3(d), which tries to limit grant of “secondary” pharmaceutical patents, i.e. patents on new forms of existing molecules and drugs. Previous research suggests the provision was rarely used against secondary applications in the years immediately following its enactment, and where it was, was redundant to other aspects of the patent law, raising concerns that 3(d) was being under-utilized by the Indian Patent Office. This paper uses a novel data source, the patent office’s first examination reports, to examine changes in the use of the provision. We find a sharp increase over time in the use of Section 3(d), including on the main claims of patent applications, though it continues to be used in conjunction with other types of objections to patentability. More surprisingly, see a sharp increase in the use of the provision against primary patent applications, contrary to its intent, raising concerns about potential over-utilization.
+
+## Introduction
+
+India began to allow pharmaceutical products to become patented in 2005, in compliance with the country's obligations under the World Trade Organization's Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS). In doing so, the Indian government inserted a controversial provision into the patent law, Section 3(d), which tries to limit the grant of “secondary” pharmaceutical patents, i.e. patents on new forms of existing molecules and drugs.
+
+Section 3(d) has been the source of considerable conflict. One prominent case that brought the world's attention to 3(d) was the Indian Patent Office's (IPO) decision to reject a secondary patent on Novartis' cancer drug “Gleevec” (imatinib mesylate), a decision that cited Section 3 (d) as one of the grounds for rejection. Novartis challenged the constitutionality of Section 3 (d) and appealed the IPO's decision, actions that in turn inspired health activists to embark on a campaign against Novartis and in support of the provision. The legality of 3(d) was upheld, and the decision to reject the Gleevec patent was confirmed by the Intellectual Property Appellate Board in 2009 and then, ultimately, the Indian Supreme Court in 2013 [ 1 , 2 ] .
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+1/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+Competing interests: The authors have declared that no competing interests exist.
+
+Debates and controversies over 3(d) have not been limited to this one drug. The provision has triggered heated and polarized views on pharmaceutical patents in India, and more broadly in developing countries adopting pharmaceutical patents in compliance with TRIPS. On the one hand, many legal scholars, civil society groups, and international organizations have lauded India's policy choice, citing 3(d) as a prominent example of a country complying with its international obligations but doing so in a way that can preserve generic competition. In that spirit India's Section 3(d) is commonly held out as model to follow, and other countries where pharmaceutical patenting is also new are encouraged to act similarly [ 3 – 5 ] . On the other hand, many foreign governments and the transnational pharmaceutical industry regard 3(d) with disdain. The US Government routinely cites 3(d) as among the reasons for including India on the “Priority Watch List” in the United States Trade Representative’s annual Special 301 Report [ 6 ] , for example, and the provision has drawn repeated criticism from international drug firms and their representatives [ 7 , 8 ] . The concern that 3(d) makes it difficult to get a patent in India is widespread in the scholarly literature as well [ 9 , 10 ] . However, these analyses did not look specifically at the role of 3(d) itself, but measures of patent protection on molecules which could be influenced by other factors, including the timing of TRIPS implementation in India.
+
+Previous empirical analyses that did look directly at 3(d) found little independent role of 3 (d) in shaping Indian pharmaceutical patent outcomes [ 11 , 12 ] . Specifically, these analyses found that the provision was involved in a relatively small number of cases, and, where it was, it was almost always used together with other more conventional reasons for rejecting patent applications, such as lack of novelty or inventive step. However, the previous analyses were based on pharmaceutical application filings and examination decisions in the early years after the introduction of pharmaceutical patenting in India. It is possible that the role of 3(d) has changed over time, given implementation lags and updated guidance to IPO examiners. Moreover, where 3(d) and other grounds for rejecting patents were employed, previous analyses were unable to untangle which were the main reasons for rejection.
+
+This paper uses new micro-level prosecution data to examine changes over time in 3(d) and to assess the independent role of this provision. While analyses of patent prosecution process are now common for applications filed at the United States Patent and Trademark Office. [ 13 , 14 ] , there are few empirical analyses of developing country patent prosecution. This is particularly crucial for analyzing patent prosecution in the context of TRIPS, given concerns that developing countries' practices may differ substantially from their laws on the books [ 15 ] .
+
+As we seek to understand how the patent office functions and, specifically, the role of Section 3(d) in patent prosecution processes, we focus not just on the IPO's final decisions, but also examiners' initial reports, as well as the exchanges that occur between applicants and the patent office following issuance of the initial reports. Focusing on the first examination reports (FERs) provides a fuller picture of the role of 3(d) in patent prosecution, allowing us to understand how 3(d) is used by examiners and how applicants respond to 3(d) objections that are raised in the course of examination. Another novel aspect of our approach is that we examine the role of 3(d) and other substantive grounds for rejection in targeting the first claim of patent applications. This allows us to assess whether 3(d) struck the core of the application, and whether it did so on its own or in conjunction with other aspects of patent law. Analyses of FERs, which we see relatively early in the prosecution process, also avoids the problem of censoring which complicates assessment of grant rates. This is particularly important for examining changes over time. There is a trade-off, however, as we do not see final decisions in most cases, as we discuss more below.
+
+We find a sharp increase in the prevalence of 3(d) in FERs over time, including on applications’ most important claims. However, 3(d) rarely works alone: it continues to be invoked
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+2/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+along with other more conventional objections, even when it is used on an application's main claim. While the provision does appear to make obtaining a patent more difficult and the prosecution process longer, it is hard to know whether this is due to the independent effects of 3 (d), the types of applications that draw 3(d) objections, or the types of examiners that invoke 3 (d). Surprisingly, we also find evidence that 3(d) is more commonly used for primary patents than secondary patents, suggesting that it is functioning differently than intended.
+
+The paper has 5 sections. Section 2 provides brief background and context on the introduction of pharmaceutical patents in India and Section 3(d), along with an overview of the patent prosecution process. Section 3 describes the data and empirical approach. Section 4 presents results, examining the changing utilization of 3(d) over time in FERs, the relationship between 3(d) and novelty and inventive step, the association between 3(d) in FERs and final outcomes, and the use of 3(d) on primary vs. secondary patent applications. Section 5 presents discussion of the main findings, indicates directions for future research, and links research on the role of Section 3(d) to broader issues regarding the implications of pharmaceutical patents in India for access to medicines in poor countries in the context of TRIPS.
+
+## TRIPS, pharmaceutical patents, and Section 3(d)
+
+The World Trade Organization’s (WTO) Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) requires all countries to grant pharmaceutical patents. With the exception of “Least Developed Countries,” all WTO members that did not already allow pharmaceutical patents as of 1995, when TRIPS went into effect, had until 2005 to begin doing so. During the transition period, from 1995 until the date that a country made pharmaceuticals patentable, TRIPS required members to receive and hold applications in a “mailbox.” Thus, if in a given country pharmaceutical patents were to become available as of 1999, from 1995 to 1999 the country would accept applications in the mailbox, and these would be examined as of 1999, along with other applications received from that date onwards.
+
+India was one of the countries that most resisted TRIPS during the Uruguay Round trade negotiations of the late 1980s and early 1990s. India opposed the inclusion of rules on countries’ intellectual property policies and practices in the international trade regime, and once the “trade-IP” linkage was established and TRIPS negotiations began, India adamantly resisted the subsequent obligation that all countries allow pharmaceuticals to be patented [ 16 – 20 ] . Although process patents were available in India, product patents had been prohibited since 1970. The absence of patent protection in India coincided with substantial development of the local pharmaceutical sector, and TRIPS was thus perceived as a serious threat [ 19 , 21 , 22 ] . Perhaps not surprisingly, when forced to allow drug patents but allowed a transition period before doing so, India waited until 2005 to make pharmaceutical products patentable, the maximum period allowed. Indeed, India is one of the only countries to use the full transition period and delay pharmaceutical patenting until 2005. And, also in grudging compliance with the country’s new international obligations, as of 1999 India also began receiving applications in a mailbox, to be examined as of 2005 when the product patent regime was in operation.
+
+In 2005, at the point of introducing the final amendments to the Patents Act to allow for pharmaceutical patents, the Indian government included Section 3(d), a provision that establishes a high barrier for secondary patents. Specifically, 3(d) stipulates that many secondary patents are not considered as inventions, and thus not eligible for patents, unless the applicants demonstrate that these have greater efficacy:
+
+The following are not inventions within the meaning of this Act. . . The mere discovery of a new form of a known substance which does not result in the enhancement of the known
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+3/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+efficacy of that substance or the mere discovery of any new property or new use for a known substance or the mere use of a known process, machine or apparatus unless such known process results in a new product or employs at least one new reactant. For the purposes of this clause, esters, ethers, polymorphs, metabolites, pure form, particle size, isomers, mixtures of isomers, complexes, combinations, and other derivatives of known substance shall be considered to be the same substance, unless they differ significantly in properties with regard to efficacy.
+
+Section 3(d) was implemented explicitly to address concerns that additional patents on existing substances could be used to extend market exclusivity and delay generic competition. Basheer and Reddy [ 23 ] report that the Minister of Commerce at the time the patent law was being finalized introduced 3(d) to prevent “ever-greening” [ 24 – 26 ] . While some actors sought a more restrictive approach, for example prohibiting all secondary patents, the designers of 3 (d) sought a middle ground that would allow patents on modified forms of existing compounds so long as they demonstrated improvements (“efficacy”) over the earlier, known substance. This intermediate position was subsequently supported by a government-established committee that was asked to report on whether India should prohibit patents on all “incremental innovations” [ 27 ] .
+
+Thus to obtain a pharmaceutical patent in India, not only do applicants have to satisfy traditional criteria that are common across all countries, e.g. novelty and inventive step, but also meet Section 3(d) requirements. As indicated in the introduction, Section 3(d) has received considerable attention, but its effects have tended to be exaggerated by both supporters and critics. We use micro-level data to shed new light on India's new pharmaceutical patent system and the role of 3(d).
+
+Before proceeding to the data and analyses, a quick review of the Indian pharmaceutical patent prosecution process may be useful. Applicants must request examination by the IPO within 4 years after their application's international priority date; failure to do so leads to applications being classified as "withdrawn." When the IPO examines applications, a first examination report is typically issued within six months. FERs range from a few lines to long and detailed documents with extensive discussions of claims. FERs are like "first office actions" in the U.S., which list objections such as novelty and inventive step, as well as other less substantive grounds such as lack of clarity and mistakes in the application. If an applicant does not respond to the FER the application is "abandoned." When the applicant does respond, amending or eliminating claims, or rebutting the objections raised by examiners, the IPO then issues a second report and, typically, invites the applicant to a hearing. If the applicant overcomes these objections the patent is granted. If, however, the applicant stops pursuing the application after initially having replied to the FER, for example the applicant does not respond to the IPO's second report or does not attend the hearing, or does take these steps but is unable to convince the patent office of the merits of the case, the application is refused.
+
+## Data and empirical approach
+
+We started with a set of pharmaceutical applications that were filed globally via the Patent Cooperation Treaty (PCT), both to focus on relatively important applications and to allow for comparability of Indian outcomes to those in other jurisdictions. Accordingly, we began with the September 2015 version of the OECD Triadic Patent Families database, which covers all applications filed in the European Patent Office, US Patent and Trademark Office, and Japanese Patent Office. Using this database, we focused on all “pharmaceutical” applications with priority years (first global filing years) 2000–2012. We then collected information from the
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+4/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+WIPO statistics database on all Indian national stage applications; since at the time we collected the data the Indian data were truncated in 2012, we focus the subset with Indian applications filed through 2011. For tractability, we focus on applications with priority PCT month July. This resulted in 1,964 PCT applications, mapping to 1,993 Indian national stage applications. (Since India took full advantage of the transitional period to introduce pharmaceutical patents that was allowed by TRIPS, as explained above, the applications in our dataset that were filed in India from 2000–2005 were held in a “mailbox” until examination commenced in 2005.)
+
+We collected Indian outcomes on all 2000 – 2011 applications from the Indian patent database as of May 2017. We record five mutually exclusive categories: applications can be granted, pending (still waiting final determination), withdrawn before examination, abandoned after a first examination report issued, or refused. As explained above, if an applicant pursues the application after receiving the FER but is unsuccessful in overcoming the objections raised, the application is considered formally refused. We also collect data on duration of prosecution for granted patents.
+
+As explained, a novel contribution of our work is that we analyze the first examination reports issued by the patent office after applications have undergone their first substantive review. For all applications with FERs we determined if the reports included any 3(d) objections, and also whether they included any novelty or inventive step objections. We also determined whether there were 3(d) objections on the first claim, and, for a subset of applications, whether there were novelty or inventive step objections on Claim 1 as well.
+
+While most of our analyses of 3(d) focus on FERs, we also use the full prosecution record of some applications to gain a stronger sense of the role of 3(d). For all applications where there was a 3(d) objection on claim 1 of the FER and a final outcome of refusal, and for a random selection of applications with 3(d) objections on claim 1 that ultimately were granted by the patent office, we read through the correspondence between applicants and the patent office (e.g. replies to FERs, subsequent examination reports, controller's reports) to understand how applicants respond to 3(d) objections and the role of 3(d) throughout the prosecution process.
+
+To examine the different roles of 3(d) for different types of applications, we code each of the applications in our sample as to whether they claim a new compound ("primary" patent applications) or, alternatively, a modified form, composition, or use of an existing compound ("secondary" patent applications) using the coding scheme from previous research [ 11 , 12 ] . The claims coding also revealed a handful of pure process applications. After dropping these we were left with 1853 applications. S1 File provides details on data construction, as well as links to the datasets, our coding of patent application type, outcomes, and prosecution histories, and computer code to fully reproduce the results below.
+
+## Results
+
+We use these data to address the following questions:
+
+- 1. How has the use of 3(d) by examiners in FERs changed over time?
+
+2. How much overlap is there between 3(d) and novelty/inventive step objections in FERs?
+
+3. How does the inclusion of 3(d) objections in FERs, alone or in conjunction with novelty or
+inventive step, correlate with different outcomes?
+
+4. What kinds of patent applications draw 3(d) objections in FERs?
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+5/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+## The changing role of 3(d) over time
+
+To examine the role of 3(d) over time, we focused on applications that have FERs. The share of applications with an FER drops over time (for example, from about 78 percent in the 2001– 2004 period to 52 percent in the 2008–2011 period). This is not surprising, as examination has not yet begun on a larger share of more recent applications. We were able to locate FERs for nearly all abandoned, granted, and refused applications (as well as a third of the pending applications, where examination has begun but not yet concluded), yielding 1,283 FERs. Overall, 37 percent of the applications with FERs are granted, 45 percent abandoned, 5 percent refused, and 13 percent pending.
+
+The solid line in Fig 1 shows the share of applications with an FER with any 3(d) objection, by application year. The sharp increase over time, from less than 40 percent of the early applications to more than 80 percent of the most recent applications, demonstrates an increased utilization of 3(d) by Indian patent examiners. While previous research, based on even earlier sets of applications, revealed a low incidence of 3(d) [ 11 , 12 , 28 , 29 ] , this is clearly no longer the case.
+
+We also examined the extent to which 3(d) targeted the first claim of the application. Patent applications include “independent” and “dependent” claims, with the latter following and based on the former. Claim 1 is, by definition, an independent claim, and typically the broadest and most important claim [ 30 – 32 ] . One challenge to studying the incidence of 3(d) objections on claim 1 is that in many instances the FERs do not indicate specific claims, e.g. the objection is simply that “claims fail to satisfy” Section 3(d). The share of ambiguous FERs of this sort, where Section 3(d) is raised but in a way that does not indicate specific claims, declines
+
+![Figure](figures/figure_006.png)
+
+Fig 1. Share of applications with 3(d) objections in first examination report (FER). Based on 1283 applications with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g001
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+6/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+sharply, from about 39 percent in 2001–2004 to 14 percent in 2008–2011. This trend reveals greater sharpness and focus in the patent office's use of 3(d) and examination practices more generally.
+
+The bottom line of Fig 1 shows that the share of applications where there are specific 3(d) objections on claim 1 is also rising sharply over time. By the end of the period under study, in excess of 60 percent of the FERs include 3(d) objections on the first claim. Not only is 3(d) increasing over time, but it is typically, and indeed increasingly, striking at the core of the patent application.
+
+Taken together, we find (1) a sharp growth in the use of 3(d) over time, (2) that it tends to target first claim of the application, and increasingly so. Unlike previous work suggesting underutilization of 3(d), here we find evidence that IPO is using 3(d) to try to reject almost everything, in the first examination report at least.
+
+## Is 3(d) redundant?
+
+The data presented so far suggest that 3(d) is a major way in which the Indian Patent Office tries to limit patent grants, and increasingly used over time. This is consistent with concerns that 3(d) makes it harder to obtain patents in India than other jurisdictions (as it was meant to do). However, one wrinkle is that we do not know what work is being done by 3(d) itself. Examiners may also be objecting to patents on other, more traditional grounds, such as lack of novelty or inventive step. Indeed, previous research has suggested just that, that Section 3(d) was rarely used alone, but rather in conjunction with other ways of rejecting applications [ 11 , 12 ] . We explore this here too, both overall and for the main claim. Specifically, we also identified novelty and inventive step objections on the 427 FERs for applications filed between 2006 and 2007. We focused on applications for which there were electronic FERs, dropping 9, leaving 427. An advantage of looking at this time period is that the applications are more likely to have FERs (86 percent do) and the FERs are more likely to have clearer delineation of specific objections on specific claims.
+
+First, we looked for whether applications with any 3(d) objections had other sorts of objections. Fig 2 shows the distribution:
+
+The majority of FERs (67 percent) cite both 3(d) and novelty/inventive step, and only 4 percent cite neither. At the application level, Section 3(d) rarely is used on its own: of the FERs with 3(d) objections the vast majority of these applications (96 percent) also draw novelty or inventive step objections.
+
+Although this overlap suggests redundancy between 3(d) and more traditional patentability criteria, it is possible that, within a given application, these different criteria are being invoked against different claims. Might 3(d) be used for important claims independently of novelty and inventive step? To investigate this, we examined overlap between 3(d) and novelty/inventive step at the level of claim 1. We use the same four categories as above to classify each FER (no objection, objection not directed at claim 1, objection including claim 1, objection presented without specific claims), now for 3(d) as well as novelty/inventive step. Fig 3 shows the distribution:
+
+Where 3(d) is raised against claim 1, the same claim is also subject to a novelty or inventive step objection more than 88 percent of the time (and for another 7 percent of these novelty or inventive step objections are raised, but the language of the FERs does not allow us to know for certainty if examiners had claim 1 in mind). Only 4 percent of the time when 3(d) was raised against claim 1 were there no novelty or inventive step objections in the FERs at all. The relationships are less sharp in the other direction. Where novelty or inventive step are raised against claim 1, the same claim is also subject to a 3(d) objection 56 percent of the time (and
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+7/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_002.png)
+
+Fig 2. Distribution of objections in first examination report for 2006–7 applications. Based on 427 applications filed in 2006 and 2007 with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g002
+
+another 8 percent where 3(d) was raised without specifying the claims), while for 27 percent of the applications where conventional patentability criteria were raised against the first claim, 3 (d) was not mentioned.
+
+That the IPO tends to pair 3(d) objections with those based on conventional patenting criteria makes it difficult to know whether 3(d) per se is making examination stricter (e.g. by getting patent examiners to think more carefully about novelty and inventive step), or whether more rigorous examination is leading to more 3(d) and also other types of rejections. This is difficult to know without some source of random variation in 3(d), which we have not been able to identify. However, below we present evidence on associations between 3(d) and final outcomes that provide a sense of whether applications with and without 3(d) objections in first examination reports fare differently.
+
+## The association between 3(d) objections in FERs and final outcomes
+
+The finding that 3(d) objections are present in the vast majority of FERs does not mean that these applications are all rejected (or, if rejected, on 3(d) grounds). As in the US, where the first office action is almost always a “non-final rejection,” in India applicants can and do rebut 3(d) and other objections, or amend claims to address them. If they do so successfully they obtain patent protection.
+
+How often do applications with 3(d) objections get granted? Fig 4 shows that applications with both 3(d) and novelty/inventive step objections have a grant rate of 24 percent, compared to 63 percent for the small number of applications that have neither 3(d) nor novelty/inventive step objections (p $<.01$ ). The grant rate for applications with novelty/inventive step objections
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+8/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_002.png)
+
+Fig 3. Distribution of objections in first examination reports for 2006–2007 applications. Based on 427 applications filed in 2006 and 2007 with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g003
+
+alone (34 percent) is about 10 percentage points higher than those with novelty/inventive step and 3(d) objections (p = .051).
+
+Fig 5 shows the same information for the detailed claim level objection grounds, sorted in descending order of frequency. Compared to applications with only novelty/inventive step specifically on claim 1, where the grant rate is 36 percent, those with both novelty/inventive step and 3(d) objections on claim 1 have a much lower grant rate of 19 percent (p $<.01$ ).
+
+We also considered, for granted patents, whether 3(d) objections (anywhere in the FER) add time to the process. To do so, for granted patents we collected data on the timing from FER to certificate of issue. Fig 6 shows the mean grant lag by type of objections in the FER:
+
+Overall, applications with 3(d) objections that are eventually granted take about 200 more days to be granted than applications where FERs did not mention 3(d) (p = .01). And applications with just novelty/inventive step objections in initial review that end up granted proceed to this outcome on average 178 days more quickly than when they also have 3(d) objections (p = .51).
+
+These results suggest that adding 3(d) to the mix is associated with a more difficult prosecution process (more likely to get rejected, or more likely to abandon the application), and, conditional on getting a patent, dealing with 3(d) is associated with longer lags. But, as indicated above, it is hard to know whether this is because of 3(d) per se. It is also possible that the types of applications that attract 3(d) are less likely to be granted or would be subject to more rigorous examination anyhow, for reasons that we are unable to observe, or that examiners who use 3(d) are more rigorous.
+
+Still another possibility is that early use of 3(d) creates opportunities for the patent office to probe deeper about whether applications meet even standard patentability grounds. There is
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+9/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_002.png)
+
+Fig 4. Grant rates for 2006–2007 applications by type of objection, broad. Based on 427 applications filed in 2006 and 2007 with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g004
+
+some evidence of this from the prosecution record of the applications that were ultimately refused. Recall that applications are classified as “refused” only after the applicant has responded to the FER (otherwise they are classified as “abandoned”). These applications also then ordinarily feature a further examination report by the IPO that includes an invitation for a hearing and, ultimately, a Controller’s Decision with a final outcome. We reviewed the full prosecution histories of 20 of the 22 applications in our dataset that had 3(d) objections on claim 1 in their FERs and a final patent office decision of refusal. (For two of these applications the relevant prosecution materials beyond FERs were not available.) In six of these the postFER prosecution process appeared to feature rigorous application of standard patentability criteria: in three instances examiners’ 3(d) objections were successfully overcome, but the application was still rejected on grounds of lacking novelty or inventive step, and in another three instances novelty and inventive step objections were not raised in the FERs but were cited as grounds for final refusal (in one case solely, without 3(d)). By contrast, in three other instances the FERs cited novelty or inventive step but these were no longer mentioned as grounds for refusal in the final patent office decisions. In the remaining 11 applications, both 3(d) and standard patentability criteria were used the same ways in FERs and final decisions.
+
+## What kinds of applications get 3(d) objections?
+
+Previous analyses of 3(d) have focused mainly on its effects on secondary patent applications, which is natural since these are the applications it was meant to target. Together with the results (above) on the growth of 3(d) objections, prominent cases of 3(d) being used against primary patents (including, in a preliminary ruling, sobusfovir)—those covering drugs ' original molecules–raise the question of whether it is being used more expansively.
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+10/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_002.png)
+
+Fig 5. Grant rates for 2006–2007 applications by type of objection, detailed. Based on 427 applications filed in 2006 and 2007 with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g005
+
+Here we return to the full sample of applications with FERs (not just 2006 – 07), and we use our coding of whether the applications are primary or secondary. As noted above, and discussed in more detail elsewhere [ 11 , 12 ] we categorize as “ primary ” applications those that include at least one claim on a new compound. Secondary applications include those on polymorphs and crystal forms, enantiomers and isomers, salts, metabolites and derivatives, and other modified forms, compositions, or uses of an existing compound that do not also have a new compound claim. (See discussion in S1 File for more detail.)
+
+Overall about 56 percent of the applications in our sample are primary, and 44 percent secondary. Yet as Fig 7 shows applications for primary patents account for a disproportionate share, 69 percent, of the applications drawing 3(d) objections on the first claim. Indeed, while 30 percent of secondary applications drew 3(d) objections on the first claim, 54 percent of the primary applications did (p $<.01$ ).
+
+Moreover, Fig 8 shows that the share of primary applications with 3(d) objections and 3(d) objections on claim 1, is increasing over time, with a less pronounced trend for secondary applications.
+
+The use of 3(d) on primary applications is surprising. To explore this further we reviewed a random sample of 20 granted patents where 3(d) objections on claim 1 had been raised in the FERs, again reviewing the entire prosecution history. One way that applicants overcome 3(d) objections is by presenting comparative data to satisfy 3(d)'s requirement that new forms of known substances demonstrate increased efficacy. Or, in the absence of such data, applicants amend or remove claims such that 3(d) is no longer relevant. In many instances, however,
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+11/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_003.png)
+
+Fig 6. Application-grant lags. Based on grants from 2006-7 filings.
+
+https://doi.org/10.1371/journal.pone.0194714.g006
+
+applicants reject the relevance of this provision altogether. In 16 of the 20 cases, applicants insisted (successfully, as the patents were granted) that their compounds were new and therefore not subject to 3(d).
+
+In doing so, applicants commonly reprimand the IPO for inappropriately applying 3(d), reminding the patent office that this provision was intended to prevent patents on new forms of known drugs, and not as a block on new compounds altogether. As an illustration, consider the reply by Eli Lilly on the FER for application 2252/MUMNP/2009, insisting that its claimed invention is novel and accusing the IPO of abusing 3(d):
+
+“Section 3(d) was designed to make a higher bar of innovation for patentability of new salts, esters, etc. of know substances (e.g. pharmaceuticals), to avoid trivial, inconsequential alterations being made to the FORM of such substances and thus extending market exclusivity of known substances. It is NOT meant to create a higher bar for new substances by deeming all new compounds to be merely derivatives of known compounds. . . . It is submitted that Section 3(d) is intended to block patenting by individuals who would make trivial changes to the form of a known compound, such as making a different salt, ester or other prodrug, which would be expected to retain the original activity of the parent compound. This provision was NOT intended to be a secondary, higher bar to patentability of novel compounds over and above inventive step. . . . The present improper use of 3(d) attempts to define every new compound as merely a derivative of some know structural chemical core thereby barring patentability, despite the compound's novelty and inventiveness in the unpredictable chemical arts” (pp. 2–3, emphasis in original).
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+12/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_002.png)
+
+Fig 7. 3(d) objections by application type. Based on 1283 applications with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g007
+
+## Discussion and conclusion
+
+The data reveal substantial increases in the use of 3(d) over time in FERs, overall and with specific regard to the main claim. Clearly, the IPO is relying extensively on 3(d) to raise a higher barrier for obtaining pharmaceutical patents.
+
+While the increased reliance on 3(d) may reflect characteristics of the applications filed in India, this may also reflect explicit policy. In the initial years of India's new pharmaceutical patent regime, many observers asserted that, notwithstanding the high-profile Gleevec case, 3 (d) tended to be under-utilized [ 11 , 28 , 29 ] . The association representing India's leading pharmaceutical firms published a report, authored by the former director of intellectual property in the Ministry of Commerce calling for more aggressive application of Section 3(d), for example, and subsequently worked with the IPO to revise the examination guidelines to that effect [ 28 ] . And the defense of 3(d) provided by the Appellate Board and then the Supreme Court [ 1 , 2 ] may have contributed to this too, by giving examiners greater confidence to use this provision. It is difficult to ascertain the effects of constituent pressures, revised guidelines, and legal support, though it is reasonable to believe they have contributed to the increased use of 3(d).
+
+But Section 3(d) is rarely used alone. Even when 3(d) is invoked as a reason why a patent should not be granted, it is rarely invoked as the only reason. Examiners also use other, traditional, grounds to deny patents, such as lack of novelty or inventive step. Previous work, at the application level, suggested that this was common [ 11 , 12 ] , and the current findings, based on FERs, are consistent with that research: looking at applications filed in 2006 – 07 for which we could obtain FERs, we find that when 3(d) objections are raised, in nearly all (94 percent) instances so too are objections based on lack of novelty or inventive step.
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+13/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+![Figure](figures/figure_002.png)
+
+Fig 8. 3(d) objections by application type, over time. Panel (a) Primary applications; Panel (b) Secondary applications. Based on 1283 applications with a first examination report.
+
+https://doi.org/10.1371/journal.pone.0194714.g008
+
+Overlap between 3(d) and other patentability criteria at the application level does not necessarily imply redundancy in use, as different provisions of the patent law may be applied to different claims within a single application. Researching the use of 3(d) and other provisions at the claims level is difficult, on account of the quality of FERs. In the initial years of patent examination FERs tended to be too vague, simply indicating that “claims” do not satisfy the tests of 3(d) or other aspects of the patent law, without indicating which claims a given objection was referring to. Looking at a set of applications during the time period when FERs tended to be more specific (but early enough so that FERs have been produced), our findings at the claims level are consistent with what we observed at the application level: in nearly all cases of 3(d) being used against the first claim in an application, so too were novelty and inventive step. Most of the time, whether looking at applications as a whole or the first claim, 3(d) objections are accompanied by novelty or inventive step objections as well. Although there is increasing use of 3(d), it is not independent use.
+
+We also examined association between 3(d) and final outcomes. More than half of the applications that draw 3(d) objections at first examination are ultimately granted. However, grants are less likely if a 3(d) objection is raised in the FER than otherwise, and the grant lag is longer when 3(d) is in the FER.
+
+For abandoned applications, where applicants do not reply to the FER, it is not possible to know how 3(d) affected the final outcome relative to other factors leading applicants to decide not to pursue the patent. For refused applications, however, we can start to understand the role of 3(d) by reviewing the post-FER prosecution record. In contrast to abandoned applications, refused applications (and granted patents, as discussed below) ordinarily feature a reply
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+14/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+to the examination report by the applicant and, in turn, a further report by the IPO that includes an invitation for a hearing and, ultimately, a Controller's Decision with a final outcome. Our review of the full prosecution processes for the refused applications in our dataset where the FERs included 3(d) objections on the first claim revealed rigor throughout the examination process in the use of standard patentability criteria. In the case of applications where 3(d) was initially invoked in the FER but the application was refused on other grounds, for example, the initial use of 3(d) early in the examination process may have created an opportunity for examiners to discover or consider in depth other problems with the application that were not fully appreciated in the initial examination. In sum, patent applications drawing 3(d) objections in their FERs are less likely to be granted and, even when granted, experience a more complicated path to grant on account of 3(d)'s involvement.
+
+Our finding that 3(d) is invoked against the main claims of primary applications is surprising. Section 3(d) demands that “new forms of known substances” demonstrate increased “efficacy” in order to be patented. Applying Section 3(d) thus entails two steps: first, a decision has to be made as to whether the claimed invention is subject to this rule (i.e. if it is derived from a known substance), and, if so, a second decision has to be made as to the efficacy of the claimed invention relative to the known substance. Much of the analysis and debate over this provision has regarded the first step, of determining what efficacy means, how much increased efficacy is enough, relative to what reference point, and how to demonstrate this [ 23 , 33 – 38 ] . But prior to decisions about efficacy is the first order question about whether the claimed invention is a “new form of a known substance” and subject to 3(d) at all. One would expect primary patents to clear the hurdle set by this first order question. Given the rationale behind introducing Section 3(d) in the first place, to prevent the accumulation of secondary patents that could extend periods of exclusivity, and the associated debates over whether India should bar all secondary patents or only those that fail to demonstrate increased “efficacy” [ 27 , 39 ] , it is reasonable to expect primary patents to fall largely out of the purview of this provision. For that reason, we have previously argued that if 3(d) is implemented as intended, new drugs are likely to receive patents in India – but only one patent [ 40 ] .
+
+But 3(d) may be functioning differently than expected. Section 3(d) is being invoked commonly by examiners against applications for primary patents. Even if many of these patents end up granted, the incidence of 3(d) in FERs against primary applications seems surprising. This result suggests 3(d) may be used in a way that is different from what it was designed to accomplish.
+
+Understanding how and why 3(d) is used against what we are calling primary patents is important, and warrants more research. All applications are subject to 3(d) assessment: if examiners determine that the claimed invention is “new” then 3(d) is not involved, but if they conclude that the claimed invention is derived from a compound that is already known, then the application is subject to the provision (and will be evaluated against its efficacy standard). In the first step of this process, an examiner’s assessments of what constitutes a new compound vs. a new form of a known compound is what matters. This distinction is not always clear cut. By one perspective, once a new class of compounds is created, most compounds that follow and have similar basic chemical structure are “derivatives” and therefore subject to 3(d). These would be primary patents in typical categorization (and in our coding). If these are subject to 3 (d) then this provision could have broader effects beyond secondary patents as traditionally defined. To be sure, there must (at least hypothetically) be genuinely “new” compounds that are not be vulnerable to 3(d), but it is uncertain where exactly the line between non-vulnerable and vulnerable primary patents might be [ 23 ] .
+
+The possibility of 3(d) being used to block primary patents received considerable attention in January 2015, when the IPO used 3(d) to reject a compound patent on sofosbuvir
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+15/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+("Solvadi"), Gilead's important Hepatitis C treatment. In this instance, the IPO regarded the compound that Gilead sought to patent as being a close derivative of an existing molecule, and consequently rejected the application due to lack of efficacy. Although Gilead eventually prevailed on appeal and received a patent, the events are revealing. During the course of the prosecution, Gilead denied the relevance of 3(d), maintaining that the claimed was new and therefore not subject to questions of efficacy. During the course of the prosecution, Gilead denied the relevance of 3(d) on precisely the same grounds as what we observe in our analyses of 3(d) and primary patents, and also criticized the IPO's "improper use" of this provision in precisely the same terms cited above [ 41 ] .
+
+While 3(d) objections against primary patents may not necessarily persist through the prosecution process and these patents may be granted, its use in this way may still create challenges for the drug industry. Convincing the IPO that the claimed invention is not a derivative and therefore that 3(d) is not relevant (or demonstrating therapeutic efficacy where 3(d) is relevant), is not a formality; doing so takes time and resources. It is no surprise, then, that the transnational pharmaceutical industry intensely dislikes Section 3(d). As discussed above, some responses to FERs explicitly assert that 3(d) is being applied too aggressively in patent examination.
+
+While this article has focused on the use of Section 3(d) by examiners in FERs, our research is also suggestive for how the IPO's use of Section 3(d) may affect outcomes. As explained, it may be that applications that draw 3(d) objections have characteristics that make them less likely to be granted anyway. Or perhaps variation is at the level of examiners themselves, that some are more inclined to invoke 3(d) and that those who use 3(d) tend to be more difficult to satisfy overall. We are unable to know the precise channels with the data and types of variation available.
+
+Theoretically, there are several indirect mechanisms through which 3(d) could be influencing outcomes, in addition to its direct effects. First, invocations of Section 3(d) in FERs may function, in effect, similarly to how “ deferred examination ” rules are known to function [ 42 ] , and thus lead to applications that may otherwise be granted to be abandoned or refused. Applicants have to reply to 3(d) objections to obtain a granted patent, and in the course of preparing their replies, or deciding whether to do so, they may determine that it is not worth their time and effort to pursue the patent (often for commercial reasons that may have little to do with the quality of the application and likelihood of grant). Applications of this sort, had they been pursued, may have been granted, but they were not, and therefore end up as non-grants.
+
+The use of Section 3(d) early in the examination process may also generate opportunities for raised scrutiny at subsequent stages of the prosecution process. After an FER cites 3(d), subsequent exchanges between applicants and the IPO may reveal other deficiencies and weaknesses in the claims, deficiencies that were not noticed in the initial examination. The presence of a 3(d) objection in the FER may be associated with more rigorous subsequent examination by the IPO.
+
+More broadly, it is possible that this single provision makes the entire examination process more rigorous. Given the attention that 3(d) has received, as a mechanism to blunt the impact of pharmaceutical patenting, it may signal to examiners to be more rigorous and search a bit harder to find reasons to object to patent applications. After all, the same actors–pharmaceutical patent examiners and patent office controllers–are responsible for applying 3(d) as well as conventional elements of patent law. In contrast to a previous legal literature which emphasizes the emergence of a shared “interpretative culture” among patent offices such that nonmeritorious claims come to be seen as patentable [ 15 , 43 ] , the presence of Section 3(d) in Indian patent law may lead to more rigorous application of novelty and inventive step. Indeed, some of the novelty and inventive step objections raised by the IPO may not have been noticed
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+16/19
+
+---
+
+:PLOS ONE
+
+Indian pharmaceutical patent prosecution
+
+or cited, in the FERs or later in the prosecution process, if not for 3(d) making the examination process more rigorous in the first place. Identification of the direct and indirect effects of 3(d) on outcomes is a subject for future research.
+
+Beyond its local significance, analysis of how the Indian pharmaceutical patent system functions has important global implications. The extent to which drug companies can expect patent protection in developing countries will hinge on how 3(d) works, in general, and on its application toward different types of patents. If 3(d) is interpreted to cover traditionally “primary” patent applications as well as secondary ones, it will be difficult to obtain patent protection in India even for new drugs. (By new drugs, we mean drugs approved by regulators as new molecular entities, which typically have at least one patent application we would classify as primary.) If the provision is focused mainly on secondary applications, its impact will be mainly on the number of patents per drug and thus duration of patent protection, rather than on whether new drugs receive any patent protection at all.
+
+These different scenarios also have implications for access to medicines beyond India. Indian pharmaceutical firms have played a central role as provider of affordable, high-quality, generic medicines throughout the developing world [ 44 , 45 ] . While many of the drugs supplied by Indian firms are older medicines where any patents have long since expired, Indian firms also supply their own versions of more recent drugs that, while patented in some countries, lack protection in India because they pre-date the product patent regime. But now India grants patents: all drugs with post-1995 priority dates are in principle (under TRIPS) eligible for patents in India [ 40 ] . India's ability to continue providing generic drugs to developing countries—and its role as “Pharmacy of the Developing World”—will depend crucially on how broadly Section 3(d) is interpreted and implemented. Specifically, if 3(d) serves as a block on primary patents this will make Indian generic production and distribution of new drugs easier. If, instead, 3(d) is focused on secondary patents, going forward Indian generic production and distribution will not be possible until later in the lifecycle of most new drugs, after their primary patents expire.
+
+## Supporting information
+
+$1 File. "Data Appendix.pdf" This file provides information on data construction and code to reproduce the all results in the text.
+
+(PDF)
+
+## Acknowledgments
+
+The research was funded by the Economic and Social Research Council, Grant ES/K010999/1. We thank Jazmin Macias-Arteaga, Michael Novello, Antonio Postigo-Angon, Chetali Rao, Jennifer Reid, and Bhavish Vora for excellent research assistance.
+
+## Author Contributions
+
+Conceptualization: Bhaven N. Sampat, Kenneth C. Shadlen.
+
+Data curation: Bhaven N. Sampat, Kenneth C. Shadlen.
+
+Formal analysis: Bhaven N. Sampat, Kenneth C. Shadlen.
+
+Funding acquisition: Bhaven N. Sampat, Kenneth C. Shadlen.
+
+Investigation: Bhaven N. Sampat, Kenneth C. Shadlen.
+
+Methodology: Bhaven N. Sampat, Kenneth C. Shadlen.
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+17/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+Project administration: Bhaven N. Sampat, Kenneth C. Shadlen. Resources: Bhaven N. Sampat, Kenneth C. Shadlen. Supervision: Kenneth C. Shadlen. Validation: Bhaven N. Sampat, Kenneth C. Shadlen. Visualization: Bhaven N. Sampat, Kenneth C. Shadlen. Writing – original draft: Bhaven N. Sampat, Kenneth C. Shadlen. Writing – review & editing: Bhaven N. Sampat, Kenneth C. Shad.
+
+## References
+
+1. Intellectual Property Appellate Board. IPAB Order Dated 26-Jun-2009 in Novartis v. Union of India [Internet]. Scribd. 2009 [cited 2017 Oct 19]. Available from: https://www.scribd.com/document/ 133340456/IPAB-Order-Dated-26-Jun-2009-in-Novartis-v-Union-of-India
+
+2. Supreme Court of India. Novartis Patent Judgement [Internet]. Scribd. 2013 [cited 2017 Oct 19]. Available from: https://www.scribd.com/document/133343411/Novartis-patent-Judgement
+
+3. Reichman JH. Intellectual Property in the Twenty-First Century: Will the Developing Countries Lead or Follow? Houst Law Rev. 2009; 46:1115-85. PMID: 21423816
+
+4. UNAIDS. Doha+10 TRIPS flexibilities and access to antiretroviral therapy: Lessons from the past, opportunities for the future. [Internet]. 2011 [cited 2013 Aug 22]. Available from: http://www.unaids.org/ en/media/unaids/contentassets/documents/unaidspublication/2011/JC2260_DOHA+10TRIPS_en.pdf
+
+5. Amin TM. Re-visiting the Patents and Access to Medicines Dichotomy: An Evaluation of TRIPS Implementation and Public Health Safeguards in Developing Countries. In: Aginam O, Harrington J, Yu PK, editors. The Global Governance of HIV/AIDS: Intellectual Property and Access to Essential Medicines. Cheltenham ; Northampton, MA: Edward Elgar Publishing Ltd; 2013.
+
+6. USTR. Special 301 Report [Internet]. Washington, D.C.: United States Trade Representative; various years [cited 2016 Jul 28]. Available from: https://ustr.gov/issue-areas/intellectual-property/Special-301
+
+7. Das S. Big Pharma groupings urges US to ask India to change its IPR policies. The Economic Times [Internet]. 2015 Feb 12 [cited 2017 Oct 19]; Available from: https://economictimes.indiatimes.com/ industry/healthcare/biotech/pharmaceuticals/big-pharma-groupings-urges-us-to-ask-india-to-changeits-ipr-policies/articleshow/46206065.cms
+
+8. Shankar R. PhRMA concerned over Sect 3(d) of India's Patents Act as it fears other emerging markets may emulate it as a model. Pharmabiz.com [Internet]. 2016 Feb 12; Available from: http://www. pharmabiz.com/PrintArticle.aspx?aid=93430&sid=1
+
+9. Berndt ER, Cockburn IM. The Hidden Cost Of Low Prices: Limited Access To New Drugs In India. Health Aff (Millwood). 2014 Jan 9; 33(9):1567-75.
+
+10. Duggan M, Garthwaite C, Goyal A. The Market Impacts of Pharmaceutical Product Patents in Developing Countries: Evidence from India †. Am Econ Rev. 2016 Jan; 106(01):99-135.
+
+11. Sampat BN, Shadlen KC. TRIPS Implementation and Secondary Pharmaceutical Patenting in Brazil and India. Stud Comp Int Dev. 2015 May 15; 50(2):228-57.
+
+12. Sampat BN, Shadlen KC. Secondary pharmaceutical patenting: A global perspective. Res Policy. 2017 Apr; 46(3):693-707.
+
+13. Lemley MA, Sampat B. Examiner Characteristics and Patent Office Outcomes. Rev Econ Stat. 2012 Aug; 94(3):817-27.
+
+14. Frakes MD, Wasserman MF. Is the Time Allocated to Review Patent Applications Inducing Examiners to Grant Invalid Patents?: Evidence from Micro-Level Application Data. Rev Econ Stat. 2017; 99 (3):550-63.
+
+15. Drahos P. The Global Governance of Knowledge: Patent Offices and Their Clients. Cambridge University Press; 2010. 369 p.
+
+16. Matthews D. Globalising Intellectual Property Rights: The TRIPS Agreement. Routledge; 2002. 217 p.
+
+17. Pugatch MP. The International Political Economy of Intellectual Property Rights. Edward Elgar Publishing; 2004. 298 p.
+
+18. Watal J. From Punta Del Este to Doha and Beyond: Lessons from the TRIPS Negotiating Processes. WIPO J. 2011; 3(1):24-35.
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+18/19
+
+---
+
+![Figure](figures/figure_000.png)
+
+Indian pharmaceutical patent prosecution
+
+19. Sinha A. Globalizing India: How Global Rules and Markets are Shaping India's Rise to Power. Cambridge University Press; 2016. 355 p.
+
+20. Reddy P, Chandrashekaran S. Create, copy, disrupt: India's intellectual property dilemmas. New Delhi: Oxford University Press; 2017.
+
+21. Chaudhuri S. The WTO and India's Pharmaceuticals Industry: Patent Protection, TRIPS, and Developing Countries. New Delhi ; New York: Oxford University Press; 2005. 376 p.
+
+22. Horner R. Strategic decoupling, recoupling and global production networks: India's pharmaceutical industry. J Econ Geogr. 2013 Jul 13; 14(6):1117-40.
+
+23. Basheer S, Reddy P. The 'Efficacy' of Indian Patent Law: Ironing out the Creases in Section 3(d) [Internet]. Rochester, NY: Social Science Research Network; 2008 Jan [cited 2015 Feb 9]. Report No.: ID 1086254. Available from: http://papers.ssrn.com/abstract=1086254
+
+24. Kapczynski A. Harmonization and its Discontents: A Case Study of TRIPS Implementation in India's Pharmaceutical Sector. Calif Law Rev. 2009; 97:1571-649.
+
+25. Gopakumar KM. The Need to Curb Patents on Known Substances. Econ Polit Wkly. 2013 Aug 1; 48 (32):55-7.
+
+26. Sengupta A. Two Decades of Struggle. Econ Polit Wkly. 2013 Aug 1; 48(32):43-5.
+
+27. Mashelkar R, Mehta G, Datta A, Madhava Menon N, Sharma M. Revised Report of the Technical Expert Group on Patent Law Issues. 2009.
+
+28. James T. Patent Protection and Innovation: Section 3(d) of the Patents Act and Indian Pharmaceutical Industry. Mumbai: Indian Pharmaceutical Alliance; 2009.
+
+29. Chaudhuri S, Park C, Gopakumar KM. Five Years into the Product Patent Regime: India's Response. New York: United Nations Development Program; 2010.
+
+30. Tong X, Frame JD. Measuring national technological performance with patent claims data. Res Policy. 1994 Mar 1; 23(2):133-41.
+
+31. Quinn G. Understanding Patent Claims| Patents & Patent Law [Internet]. IPWatchdog.com. 2014 [cited 2017 Oct 17]. Available from: http://www.ipwatchdog.com/2014/07/12/understanding-patent-claims/id=50349/
+
+32. EPO. Guidelines for Examination [Internet]. European Patent Office. [cited 2017 Oct 17]. Available from: https://www.epo.org/law-practice/legal-texts/html/guidelines/e/f_iv_3_4.htm
+
+33. Turrill ZL. Finding the Patent Balance: The Novatis Glivec Case and the Trips Compliance of India's Section 3(D) Efficacy Standard Note. Georget J Int Law. 2012 2013; 44:1555-88.
+
+34. Banerji S, Choudhry RK. Supreme Court offers limited definition of 'efficacy' under Patents Act [Internet]. International Law Office. 2013 [cited 2017 Jul 19]. Available from: http://www. internationallawoffice.com/Newsletters/Intellectual-Property/India/Saikrishna-Associates/SupremeCourt-offers-limited-definition-of-efficacy-under-Patents-Act
+
+35. Correa C. Is Section 3(d) Consistent with TRIPS? Econ Polit Wkly Econ Polit Wkly Econ Polit Wkly Econ Polit Wkly Econ Polit Wkly. 2013 Aug 10; 48(32):49-52.
+
+36. Du D. Novartis AG v. Union of India: Evergreening, Trips, and Enhanced Efficacy under Section 3(d). J Intellect Prop Law. 2013 2014; 21:223-64.
+
+37. Arora S, Chaturvedi R. Section 3(d): Implications and Key Concerns For Pharmaceutical Sector. J Intellect Prop Rights. 2016; 21(1):16-26.
+
+38. Grover A. Analysing the Supreme Court Judgment. Econ Polit Wkly. 2013 Aug 10; 48(32):46-9.
+
+39. Basheer S. Limiting the Patentability of Pharmaceutical Inventions and Micro-Organisms: A Trips Compatibility Review [Internet]. Rochester, NY: Social Science Research Network; 2005 Nov [cited 2017 Jul 17]. Report No.: ID 1391562. Available from: https://papers.ssrn.com/abstract=1391562
+
+40. Sampat BN, Shadlen KC. Drug patenting in India: looking back and looking forward. Nat Rev Drug Discov. 2015; 14:519-20. https://doi.org/10.1038/nrd4681 PMID: 26184492
+
+41. IPO. In the matter of Application no. 6087/DELNP/2005 filed in India on 27/12/2005 for Grant of Patent: Decision [Internet]. 2015 [cited 2017 Oct 18]. Available from: http://164.100.176.38/decision/6087DELNP-2005-8837/6087DNP2005.pdf
+
+42. Yamauchi I, Nagaoka S. An economic analysis of deferred examination system: Evidence from a policy reform in Japan. Int J Ind Organ. 2015 Mar; 39:19-28.
+
+43. Drahos P. 'Trust Me': Patent Offices in Developing Countries. Am J Law Med. 2008; 34(2/3):151.
+
+44. Waning B, Diedrichsen E, Moon S. A lifeline to treatment: the role of Indian generic manufacturers in supplying antiretroviral medicines to developing countries. J Int AIDS Soc. 2010; 13:35. https://doi.org/ 10.1186/1758-2652-13-35 PMID: 20840741
+
+45. MSF. Hands Off Our Medicines [Internet]. Hands Off Our Medicines. [cited 2017 Oct 18]. Available from: https://handsoff.msf.org/
+
+PLOS ONE | https://doi.org/10.1371/journal.pone.0194714 April 2, 2018
+
+19/19
+

processed/00231_W1970180215/document.md

@@ -0,0 +1,418 @@
+# EXAMINER CHARACTERISTICS AND PATENT OFFICE OUTCOMES
+
+Mark A. Lemley and Bhaven Sampat*
+
+Abstract — In this paper, we show that there are important differences across patent examiners at the U.S. Patent and Trademark Office. We show that more experienced examiners cite less prior art, are more likely to grant patents, and are more likely to grant patents without any rejections. These results suggest that the most important decisions made by the patent office are significantly affected by the happenstance of which examiner gets an application. They also point to human resource policies as potentially important levers, hitherto neglected, in patent system reform.
+
+## I. Introduction
+
+U NDER the standard economic rationale, patents provide incentives to conduct research and disclose inventions in return for a limited right to exclude others from making, selling, or using these inventions (Nordhaus, 1969) . Under optimally designed patent systems, the static social costs from the market power conferred by patents are offset by the dynamic welfare gains from increased invention and disclosure. From this perspective, if patents were granted on inventions that were already known, they would create harms without concomitant social gains (Lemley & Shapiro, 2005) . In the United States, examiners at the Patent and Trademark Office (PTO) are charged with serving as gatekeepers to ensure that only novel and nonobvious inventions are granted patents. $^1$
+
+In this paper, using information from several novel data sets, we show that there are important differences across patent examiners and that these relate to the most important decisions made by the PTO. In particular, we find that more experienced examiners are less active in searching for prior art and have a much higher grant rate.
+
+The results could be problematic as a public policy matter because they suggest that the decision to allow or reject a patent application may not be driven by the merits of that application but rather by the luck of the draw. At the same time, they suggest that human resource policies and incentive structures at the PTO could affect patent grant rates, an important finding amid growing concern that it grants too many low-quality patents and is in need of significant
+
+reform (Jaffe & Lerner, 2004; Cohen & Merrill, 2003; FTC, 2003).
+
+Several scholars have previously studied examiner heterogeneity and its effects on patenting. Cockburn, Kortum, and Stern (2003) studied the effect of particular examiners on the characteristics of issued patents and their survival in litigation. They find that examiner fixed effects explain a significant percentage of the variation in the characteristics of issued patents and that examiner differences affected litigation outcomes: patents issued by certain examiners were more likely to be upheld in court than those issued by others. They conclude that “ there may be as many patent offices as there are patent examiners. ” In another study, Doug Lichtman (2004) studied the role of the PTO in compelling amendments during the prosecution process, an issue directly relevant to prosecution history estoppel and the application of the doctrine of equivalents. Lichtman identified the extent to which issued claims differed from the claims as originally drafted. This study too found examiner effects to be important: whether an applicant amended its claims depended in significant measure on which examiner reviewed the application. Lichtman concluded that different examiners had different styles, with some examiners systematically more likely than others to compel applicants to narrow their claims.
+
+In this paper, rather than just looking at issued patents, we identify a group of patent applications and follow them through the process to examine the impact of examiner characteristics on patent application outcomes. Specifically, we assess how examiner characteristics, and examiner experience in particular, affect the outcomes at the patent office for each of the nearly 10,000 applications filed in January 2001. 2 We also investigate how examiner characteristics affect the number of patent citations and publication citations they reference in their granted patents. These measures are proxies for how much scrutiny examiners are giving to applications and, in particular, how thoroughly they are searching the prior art.
+
+In the next section, we describe the patent prosecution process to provide context for our analysis and motivate construction of the variables we examine. In section III , we describe how we collected the examiner data and provide descriptive statistics. Section IV presents results from linear probability models relating examiner experience and application characteristics to prior art citations. Section V relates examiner experience to the grant rate and whether applications were amended during prosecution.
+
+One concern about our analysis is selection: if specific types of applications are assigned to specific examiners, this could confound our results. Section VI examines this possi-
+
+Received for publication January 8, 2009. Revision accepted for publication December 20, 2010.
+
+* Lemley: Stanford Law School and Durie Tangri LLP; Sampat: Columbia University.
+
+We thank Wes Cohen, Ernie Beffel, Chris Cotropia, Al Engleberg, John Golden, Stuart Graham, Arti Rai, Scott Stern, and participants at the NBER Productivity Lunch, the 2008 NBER Summer Institute, the Georgia Tech Roundtable on Entrepreneurship Research, and UNC Law School Conference on Empirical Patent Scholarship for useful comments. Sarah Craven, Reuven Garrett, and Matthew Wasserman provided excellent research assistance. Patrick Crosby at xblabs.com helped obtain some the data used in this analysis. B.S. thanks the Ford Foundation for financial support.
+
+"As Thomas Jefferson, the first commissioner of the Patent Office, suggested in considering the originality of an 1813 invention on "Elevators, Conveyors, and Hopper-Boys," patent examiners have the difficult job of "drawing a line between the things which are worth to the public the embarrassment of an exclusive patent, and those which are not."
+
+2 As a result of changes in the law, January 2001 is the first month for which the relevant data were available.
+
+The Review of Economics and Statistics, August 2012, 94(3): 817–827 © 2012 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology
+
+Electronic copy available at: http://ssrn.com/abstract-1329091
+
+---
+
+818
+
+THE REVIEW OF ECONOMICS AND STATISTICS
+
+bility and suggests that our results are not driven by selection bias. Section VII explores some possible causal mechanisms driving our results, and section VIII concludes with a discussion of potential implications for policy and caveats.
+
+## II. The Patent Prosecution Process
+
+Roughly 450,000 new patent applications are filed every year. Each of these applications contains, among other things, written descriptions of the invention and specific claims defining the boundaries of the property right that the applicant hopes to obtain. In the United States, applicants have a duty of candor to disclose any previous patents and publications, or “prior art,” that are material to the patentability of the claimed invention. Accordingly, most (but not all) patent applicants also disclose patent and nonpatent prior art as part of their patent application. Though there is a duty of disclosure in the United States, there is no affirmative requirement that applicants conduct prior art searches; Sampat (2010) and Alcacer et al. (2009) argue that incentives for applicants to do so vary across inventions and industries.
+
+Once applications arrive at the PTO, they are divided into technology classes, or art units. Supervisory patent examiners (SPEs) within each of the art units assign particular applications to particular examiners based on a rather loose set of rules (Manual of Patent Examining Procedure sec. 903.08(b)). One issue relevant to our empirical analyses — or any other attempt to relate examiner characteristics to PTO outcomes — is whether there is sorting, that is, whether particular types of applications are assigned to particular types of examiners. Merges (2001) suggests that while sorting could be good from a policy perspective, there is a strong “ all patents are created equal ” tradition at the PTO militating against this. Our interviews with SPEs suggest that there is some sorting, but that familiarity with particular technologies and docket flow management, rather than judgments of an application's quality or patent-worthiness, are the dominant considerations.
+
+The patent examiner assigned the application reviews it and conducts his or her own search for prior art that might make the application unpatentable. This involves searching databases of previous U.S. patents, either manually or through algorithms available to examiners. Examiners may also search foreign patents and the nonpatent literature (such as scientific and technical journals) to find prior art that might compromise patentability. Searching the nonpatent prior art in particular may be more difficult: Thomas (2001) argues that “ in comparison to much of the secondary literature [nonpatent prior art], patents are readily accessible, conveniently identified, and printed in a common format. Identification of a promising secondary reference, and full comprehension of its contents, often prove to be more difficult tasks ” (p. 318).
+
+Examiners then assess the novelty and nonobviousness of the claims in the application relative to what is disclosed in the complete list of prior art — the prior art references from the applicant, plus any discovered through the examiner’s own search. Examiners generally issue an initial rejec-
+
+tion of the application, setting out the problems they find in one or more of the claims (Lemley & Sampat, 2009) . The applicant responds, generally by amending the patent claims or by disputing the rejection. After the response, the examiner may then allow the application or issue a final rejection. Even a final rejection is not really final, however; applicants can respond by amending their application or requesting an interview with the examiner to press their case. They may also file one of a variety of continuation applications to continue to argue for patentability.
+
+If the patent issues, the patent lists a range of bibliographic data, including the final claims and information on the applicant and examiner involved in the examination. Issued patents also list each of the prior art references considered during the prosecution process and, since 2001, indications of whether these references came from applicant disclosures or examiner searches (Alcacer & Gittleman, 2006; Sampat, 2010; Alcacer et al., 2009) .
+
+The large number of applications facing the PTO means that examiners are subject to sharp time constraints; the process of reading and evaluating an application, searching for prior art, writing a rejection, responding to an amendment with a second office action, having an interview, and fulfilling various formal requirements can take three to four years on average (Allison & Lemley, 2000) , but the examiner spends an average of only eighteen hours over those years working on any given application (Lemley, 2001) .
+
+The incentives facing examiners are complicated and the object of considerable policy debate (Jaffe & Lerner, 2004; GAO, 2005) . Examiners' promotion and bonus decisions are tied to the number of “ counts ” they accumulate. These counts can be earned by first office actions, or by disposal of cases, which occur when an application is granted or abandoned. Other examiner activities, including searching for prior art, compelling further amendments after a first office action, or issuing final rejections, are not rewarded by the count system. Some have argued that these rules create a bias toward granting rather than rejecting applications (Jaffe & Lerner, 2004) .
+
+These incentives operate differently over an examiner's career. Examiners are hired at different grades (GS-5, GS-7, GS-9, or GS-11) on the government employee pay schedule, depending on their technical experience and educational background. The counts per hour they are required to attain to achieve satisfactory performance reviews (and thus further promotion) and annual bonuses increase sharply with pay grade. $^3$
+
+Another difference between more junior and senior examiners is the scrutiny they receive. A junior examiner's
+
+3 A recent GAO report provides this illustration: “ A GS-12 patent examiner working on data processing applications is expected to achieve two counts in 31.6 hours, whereas a GS-12 patent examiner working on plastic molding applications is expected to do so in 20.1 hours. GS-7 patent examiners working on those types of applications, however, are expected to achieve two counts in 45.1 and 28.7 hours, respectively. ” http://www.gao.gov/htext/d08527t.html.
+
+Electronic copy available at: http://ssrn.com/abstract-1329091
+
+---
+
+EXAMINER CHARACTERISTICS AND PATENT OFFICE OUTCOMES
+
+819
+
+work is subject to review from more senior examiners; indeed, though they do the bulk of the examination, they are listed as secondary examiners on applications they examine. Once promoted to the GS-14 level (generally between four and six years after hire, depending on what level they were hired at), they receive signatory authority, or the right to sign off on an application independently.
+
+The main empirical question we explore in this paper is whether more experienced examiners behave differently in how they process applications and in their propensity to grant applications. This provides a window into whether the human resource dimensions of the PTO matter. Examiner experience is also interesting since there is a widespread belief that despite the incentives discussed above, more experienced examiners are better, and that retaining experienced examiners, or intensifying their mentorship of junior examiners, is the key to solving the patent quality crisis (Jaffe & Lerner, 2004) . $^4$
+
+## III. Data and Descriptive Statistics
+
+We used data from Delphion, a patent data vendor, to identify every original utility patent application filed in January 2001 and published before April 2006. Evaluating applications rather than just issued patents allows us to obtain richer data on the behavior of examiners, data that were not available before 2001. After eliminating plant, design, and reissue patents, applications directed at foreign filing, and follow-on applications (continuations and divisionals), we were left with 9,960 original patent applications. We collected information on the status of these applications as of June 2008 and other information about the prosecution process as of April 2006 from the Patent Office's Patent Application Information Retrieval (PAIR) database. 5
+
+As reported in Lemley and Sampat (2009) , 72 % of the applications had resulted in patents by June 2008. About a quarter of the applications, 27 % , were not patented because they were abandoned by the applicant. These abandonments could be because the applicant could not overcome an examiner’s rejection, but the applications could also be abandoned for other reasons (perhaps the applicant's firm went out of business). $^6$ The remaining 1.2 % of the applications were pending as of June 2008.
+
+In this paper, we focus on the set of these applications that received a final disposition — the 99 % of the applications that are patented or abandoned. In our main analyses, we relate whether these 9,846 applications were granted by June 2008 to examiner characteristics.
+
+Since the PTO employs its examiners, and examiner names are listed on issued patents as well as in application status databases, it might seem straightforward to identify the examiners associated with each application. However, official PTO examiner data are reported in a haphazard format and are rife with errors, as Cockburn et al. (2003) point out. To take just one example, we identified one examiner whose name was spelled no fewer than twenty different ways on the front page of issued patents. $^7$
+
+To solve this problem, we acquired the PTO Employee Directories from 1992 to the present. These directories list the examiner name in standardized format. We then used a mix of programming and hand-matching to link these clean examiner names to examiner names in the applications filed in January 2001. In our empirical analyses, we treat the examiner for each application or patent as the person who did the most direct work on that application: the secondary examiner if there was one, or the primary examiner if there was no secondary assigned.
+
+Access to the employee directories also allowed us to determine exactly when the examiner began working at the PTO, which we used to construct experience measures. These measures are right-censored, since we lack access to employee directories before 1992. Though the applications were filed in 2001, there can be lags before they are docketed to an examiner (90 % of the applications in our sample were docketed by the end of 2001 and 98 % by the end of 2002). Accordingly, in our analyses, we calculate experience as the difference between the first year an examiner appears in the employee directory and the year the application is docketed to the examiner. $^8$
+
+In view of the right-censoring, in our empirical analyses, we examine the effects of examiner experience nonparametrically, including dummy variables for zero or one year experience, two to four years, five to seven years, or eight or more years, though the main results are similar if we use different cohort groupings or treat “ experience ” as a continuous variable.
+
+4 For example, a recent New York Times opinion piece, “Inventing a Better Patent System,” notes “the quality of American patents has been deteriorating for years” and that “one root of the problem is that patent examiners, many of whom are young or lack practical experience, are not qualified to evaluate whether complex claims in biotech or physics meet the most critical tests: whether the claim is novel relative to prior art, and whether this would be obvious to a person skilled in the art.” http:// www.nytimes.com/2009/11/17/opinion/17pozen.html.
+
+3 Unfortunately, changes to the PAIR interface make it difficult to update the prosecution data beyond April 2006.
+
+" There is no actual way for the PTO to finally reject a patent (Lemley & Moore, 2004). We consider a patent application to have been finally abandoned if the applicant has filed a notice of abandonment or has not responded to a PTO rejection or filed an appeal within six months, the time limit for doing so.
+
+7 Ponnathapura Achutamurthy's name was listed as Achutamurthy Donnathapu, Achutamurthy P., Achutamurthy Ponnathapau, Achutamurthy Ponnathapu, Achutamurthy Ponnathapua, Achutamurthy Ponnathapuea, Achutamurthy Ponnathapur, Achutamurthy Ponnathapura, Achutamurthy Ponnathaput, Achutamurthy Ponnathapura, Achutamurthy Punnathapu, Achutamurtry Ponnathapu, Achutamurty Ponnathapu, Achuthamurthy Ponnathapu, Achutmurthy Ponnathapu, Achutyamurthy Ponnathapu, Murthy Ponnathapu, Murthy Ponnathapu Achuta, Murthy Ponnathapuachut, and Ponnathapuachuta.
+
+6 For a nontrivial share of the applications, 16.95 % , this experience measure was negative, in most of these cases taking on a value of $-1$ . This is because the employee directories are by fiscal rather than calendar years and also that they are occasionally updated with lags. In these cases, we recoded experience to zero. Note that all of our results are unchanged if we instead use 2001 minus year first in directory as our experience measure.
+
+Electronic copy available at: http://ssrn.com/abstract-1329091
+
+---
+
+820
+
+THE REVIEW OF ECONOMICS AND STATISTICS
+
+TABLE 1.-Descriiriven STATESTICS
+
+<table><tr><td>Variable</td><td>N</td><td>Mean</td><td>SD</td><td>Min</td><td>Max</td></tr><tr><td>Examiner has zero or 1 year experience</td><td>9,847</td><td>0.318</td><td>0.466</td><td>0</td><td>1</td></tr><tr><td>2–4 years experience</td><td>9,847</td><td>0.283</td><td>0.451</td><td>0</td><td>1</td></tr><tr><td>5–7 years experience</td><td>9,847</td><td>0.099</td><td>0.299</td><td>0</td><td>1</td></tr><tr><td>8 or more years experience</td><td>9,847</td><td>0.299</td><td>0.458</td><td>0</td><td>1</td></tr><tr><td>Application patented?</td><td>9,847</td><td>0.723</td><td>0.447</td><td>0</td><td>1</td></tr><tr><td>Patented without rejections?</td><td>7,121</td><td>0.178</td><td>0.382</td><td>0</td><td>1</td></tr><tr><td>Number of examiner patent cites</td><td>6,528</td><td>5.75</td><td>4.65</td><td>0</td><td>59</td></tr><tr><td>Number of examiner publication cites</td><td>6,528</td><td>0.158</td><td>1.17</td><td>0</td><td>46</td></tr><tr><td>Number of applicant patent cites</td><td>6,528</td><td>4.86</td><td>13.9</td><td>0</td><td>337</td></tr><tr><td>Number of applicant publication cites</td><td>6,528</td><td>1.37</td><td>5.9</td><td>0</td><td>188</td></tr></table>
+
+
+One outcome measure we examine is the extent to which examiners search, as measured by the number of references in issued patents coming from examiners (rather than applicants). Since 2001, the front pages of issued patents identify whether references emanate from applicant disclosures or examiner searches. Previous research (Sampat, 2010; Alcacer et al., 2009) shows that examiners account for a surprisingly large share of citations to patents, though the examiner share varies across fields. Cotropia, Lemley, and Sampat (2010) show that the examiner-provided citations are the ones that are most likely to be used in deciding whether to grant a patent.
+
+Some of this research uses the applicant share of references in a patent as a proxy for how intensely applicants are searching for prior art (Sampat, 2010; Alcacer et al., 2009) . In our analyses, for the 6,528 patents resulting from the applications in our sample, we take an analogous approach to assessing examiner search intensity: relating the number of examiner citations to patents and the number of examiner citations to publications to examiner characteristics. 9 In some of the models, we also control for the number of applicant references to patented and nonpatent prior art. 10
+
+The other main outcome measures we use relates to what happens to the application, the most important decision made by the PTO. In particular, for each of the 9,846 applications, we relate examiner characteristics to whether the application was granted (“Granted”) and whether it was granted on first office action (“Granted with no rejections”).
+
+Table 1 shows descriptive statistics on each of the main independent variables, calculated at the patent application or patent level.
+
+The modal experience cohort is the first: most of our January 2001 patents have relatively new examiners. Also
+
+notable is that there are few examiners in the five-to-sevenyear cohort, consistent with concerns that numerous examiners leave the PTO relatively early in their careers (GAO, 2005) . But over a quarter of the examining corps are veterans, having stayed with the PTO for over eight years. Similar trends are evident if we calculate these figures across examiners: of the 2,797 unique examiners on these applications, 35 % have no experience, 28 % two to four years, 9 % five to seven years, and 28 % eight or more years. $^11$
+
+Of the applications, 72 % were granted. Of those, about 18 % were granted on first office action, without any rejections.
+
+For granted patents, we also have applicant and examiner citation data. Examiners are much more active in citing patented relative to nonpatented prior art, both absolutely and as compared to applicants. These figures are consistent with arguments that examiners are better at searching patented prior art than nonpatent prior art (Thomas, 2001) . $^12$
+
+## IV. Examiner Experience and Citation Patterns
+
+We begin by relating an examiner's experience to how deeply she searches, as measured by how many citations she adds during patent prosecution. Since citations are available only for issued patents, this measure is conditional on patent issuance and calculated over all 6,528 patents resulting from the applications in our sample.
+
+Before even turning to measured examiner characteristics, we can see the importance of examiner heterogeneity in simpler models. Table 2 summarizes information from least-squares regressions relating the number of examiner citations to patents and the number of examiner citations to
+
+9 While 7117 of the applications were granted—as measured by a notice of allowance being mailed—there is a lag between allowance and when the patent is published: only 6,528 of the resulting patents were published at the time we collected the data. The variables based on information from front pages of issued patents are collected only for these 6,528 patents.
+
+$^{10}$ In a previous version, we looked at the share of applications supplied by an examiner. On the suggestion of a referee that the share conflates applicant and examiner side characteristics, we now use the number of examiner citations. Results using the share measures are qualitatively similar and available on request.
+
+11 There is considerable within- and across-field variation in experience. Unreported analyses reveal that in the computer art units, examiners are overwhelmingly newly hired. There is also a large proportion of newly hired examiners in communications. This may reflect either greater turnover in those art units (a function in part of other job opportunities in those industries) or the growth in the number of applications (and thus hiring) in those fields. By contrast, examiners in the chemical and the mechanical art units have substantially more tenure, and indeed the largest number of them have been at the PTO for more than eight years. The biotechnology and organic chemistry art unit also includes a large number of examiners with a long tenure at the PTO, though not as many as in the mechanical and chemical industries.
+
+12 Examiners account for 43.6% of citations to patented prior art in their issued patents, but only 9.5% of all citations to nonpatent prior art.
+
+---
+
+EXAMINER CHARACTERISTICS AND PATENT OFFICE OUTCOMES
+
+821
+
+Table 2.-Summary Information from OLS Regressions Relating the Number of Examiner Citations in a Patent to Art Unit Effects and Examiner Effects
+
+<table><tr><td></td><td colspan="2">Number of Examiner Citations to Patents</td><td colspan="2">Number of Examiner Citations to Publications</td></tr><tr><td></td><td> $R^{2}$ </td><td>Adjusted  $R^{2}$ </td><td> $R^{2}$ </td><td>Adjusted  $R^{2}$ </td></tr><tr><td>Model with art-unit fixed effects only</td><td>0.219</td><td>0.182</td><td>0.221</td><td>0.185</td></tr><tr><td>Model with art-unit and examiner fixed effects</td><td>0.694</td><td>0.527</td><td>0.625</td><td>0.421</td></tr><tr><td>Joint significance of examiner fixed effects</td><td colspan="2">F (2210,4227) = 3.7</td><td colspan="2">F (2210,4227) = 3.1</td></tr><tr><td></td><td colspan="2">p &lt; 0.01</td><td colspan="2">p &lt; 0.01</td></tr><tr><td>N</td><td>6,528</td><td>6,528</td><td>6,528</td><td>6,528</td></tr></table>
+
+
+TABLE 3.-OLS REGRESSIONS RELATING THE NUMBER OF EXAMINER CITATIONS TO EXAMINER EXPERIENCE AND APPLICANT CITATIONS
+
+<table><tr><td></td><td>Model 3.1: Number of Examiner Cites to Patents</td><td>Model 3.2: Number of Examiner Cites to Publications</td><td>Model 3.3: Number of Examiner Cites to Patents</td><td>Model 3.4: Number of Examiner Cites to Publications</td></tr><tr><td>2–4 years Experience</td><td>−0.342 (0.246)</td><td>−0.064 (0.039)</td><td>−0.349 (0.246)</td><td>−0.064 (0.039)</td></tr><tr><td>5–7 years Experience</td><td>−0.974*** (0.293)</td><td>−0.011 (0.059)</td><td>−0.989*** (0.292)</td><td>−0.011 (0.057)</td></tr><tr><td>8+ years Experience</td><td>−1.709*** (0.241)</td><td>−0.078** (0.032)</td><td>−1.712*** (0.241)</td><td>−0.081** (0.031)</td></tr><tr><td>Number of applicant cities to patents</td><td></td><td></td><td>−0.012*** (0.004)</td><td>−0.002 (0.002)</td></tr><tr><td>Number of applicant cities to publications</td><td></td><td></td><td>−0.003 (0.013)</td><td>0.022 (0.015)</td></tr><tr><td>Constant</td><td>6.55*** (0.195)</td><td>0.205*** (0.028)</td><td>6.613*** (0.195)</td><td>0.186*** (0.029)</td></tr><tr><td>Art unit effects</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 6,528</td></tr></table>
+
+
+Robust standard errors (clustered on examiners) reported in parentheses. Asterisks denote statistical significance: *10 % , **5 % , ***1 % .
+
+publications (nonpatent prior art) to art unit indicators (for over 300 art units, to control for technology class) and examiner indicators. The large increments to $R^2$ and adjusted $R^2$ in the models that include examiner effects provide strong initial evidence of examiner heterogeneity in prior art citation patterns. Note also that for both patent and publication citations, the examiner effects are jointly significant at the 99 % level.
+
+Next, we estimated models relating the number of examiner citations in a patent to examiner experience. Here again, these models are estimated over each of the granted patents resulting from the January 2001 applications and include fixed effects for each art unit. Table 3 shows results.
+
+Model 3.1 shows that the number of examiner patent citations decreases monotonically with examiner experience, with the most experienced examiners (eight or more years) citing nearly two fewer patents than the least experienced (the left-out category). Model 3.2 shows that the magnitude of the experience effect is much smaller for publication citations, though the most experienced examiners have a significantly lower volume of such cites.
+
+Since examiner citations in a patent could themselves be affected by the extent of prior art an applicant provides (Sampat, 2010) , for each dependent variable we also estimated models where we included the number of applicant citations in the patent (to patents and publications separately) as independent variables. Interestingly, there is little
+
+evidence of substitutability here: more applicant cites do not seem to be associated with fewer examiner cites. (Model 3.3 shows a statistically significant but small negative relationship between the number of applicant cites to patents and the number of examiner cites.) Moreover, including these variables does little to change the magnitude of the experience effects documented in models 3.1 and 3.2.
+
+Overall, the data suggest that within art units, more experienced examiners tend to cite less prior art, with the magnitude of this effect more pronounced for citations to patented prior art than for publication citations.
+
+## V. Examiner Experience and the Patent Grant Rate
+
+Next, using the full sample of applications (those granted and not granted), we relate examiner experience to the most important decision made by the PTO: whether to grant a patent. Table 4 shows results. Model 4.1 shows results from linear probability models relating whether an application was granted to examiner experience and art unit effects. The grant rate increases monotonically with experience, with the two most experienced groups having an 11 percentage point higher grant rate.
+
+We cannot conclude from these results whether the more or less experienced examiners, or even the more or less thorough examiners, have it right, absent priors on whether
+
+---
+
+822
+
+THE REVIEW OF ECONOMICS AND STATISTICS
+
+Table 4.— Linear Probability Models Relating Whether an Application Is Granted and Whether Granted with No Reactions, to Examiner Experience
+
+<table><tr><td></td><td>Model 4.1: Was the Application Granted?</td><td>Model 4.2: Was the Application Granted with No Rejections?</td></tr><tr><td>2–4 years experience</td><td>0.057*** (0.013)</td><td>0.060*** (0.013)</td></tr><tr><td>5–7 years experience</td><td>0.110*** (0.018)</td><td>0.111*** (0.020)</td></tr><tr><td>8+ years experience</td><td>0.112*** (0.013)</td><td>0.133*** (0.015)</td></tr><tr><td>Constant</td><td>0.663*** (0.009)</td><td>0.105*** (0.009)</td></tr><tr><td>Art unit effects N</td><td>Yes; jointly significant 9,846</td><td>Yes; jointly significant 7,117</td></tr></table>
+
+
+Robust standard errors (clustered on examiners) reported in parentheses. Asterisks denote statistical significance: *10 % , **5 % , ***1 % .
+
+the grant rate is currently too high or too low. But it is suggestive that examiners are doing more work, and rejecting applications with more rigor, at early stages in their career, and both doing less work and allowing more patents as their tenure increases.
+
+But it may also be that whether a patent is granted is too blunt a measure of the quality of examination. As discussed in section II, claims can change over the course of patent prosecution. Amendments to the application compelled by examiners (in the face of prior art) can narrow the scope of the patent. Another measure of the rigor or quality of patent prosecution is the extent to which this narrowing occurs.
+
+To examine this, for the 7,121 applications that were allowed, we collected data from the Patent Application Information Retrieval (PAIR) database on whether there were any rejections during patent prosecution and investigated how the likelihood of grant with at least one rejection relates to experience. We were unable to locate prosecution histories for four of these, reducing our sample to 7,117. For those where we could, 82 % of these granted applications received a nonfinal rejection, and 26 % a final rejection, during patent prosecution. But 18 % did not receive any rejections before they were issued; they were issued on the first office action, as is.
+
+Model 4.2 in table 4 shows results from linear probability models relating whether a patent was issued with no rejections to experience (conditional on a patent having issued). The likelihood of granting without rejections increases sharply with experience, with the most experienced cohort 13 percentage points more likely to do so.
+
+The finding that examiner experience increases the probability of having had no rejections, conditional on issuing a patent, also undermines one possible explanation for the relationship between examination and grant rates reported above: that experienced examiners are better at separating the wheat from the chaff, or identifying what is patentable in an application, getting quickly to the patentable piece of an application by negotiating with the applicant to amend
+
+his claims appropriately. First office action allowances involve little such parsing or deliberation.
+
+## VI. Selection
+
+Taken together, the data from the previous two sections show that not only do more experienced examiners cite less prior art, they also have a significantly higher grant rate and are significantly more likely to grant on the first office action (that is, grant without rejections).
+
+One potential threat to the validity of these results is selection. If different examiners are systematically assigned to “ easier ” or “ harder ” applications, or those more or less patentworthy, our estimates would be biased. Our interviews with supervisory examiners suggest that an examiner’s familiarity with a specific technology appears to play a role in assignment decisions. But for selection bias to explain our results, a different type of selection effect would be necessary. It would require, first, that the SPEs be in a position at the outset to assess whether an application was more or less likely to be patentable and, second, that they would be motivated to give those more likely to be patentable to the most experienced examiners.
+
+The PTO does not publish rules regarding the assignment of applications within art units to particular individuals. To investigate the possibility of selection effects, we conducted written interviews with roughly two dozen current and former examiners and supervisory patent examiners. Those interviews revealed no evidence of deliberate selection or assignment of applications on the basis of the characteristics of the application. Applications are assigned to art units in the first instance on the basis of their technology classifications; the head of the art unit then decides which subclass the application belongs to and assigns it accordingly. Within subclasses, SPEs for the most part assigned applications randomly, assigning applications to particular examiners on the basis of the last digit of the application serial number. Because application serial numbers are assigned sequentially in the central PTO, this assignment system, while not truly random, is not subject to manipulation by applicants or supervisory examiners. A minority of supervisory examiners assigned applications on the basis of docket management, giving the oldest unassigned application to an examiner when that examiner finished examining a prior application. This minority approach would have the effect of disproportionately assigning applications to examiners who are quickest to resolve cases but does not involve assignment on the basis of observable characteristics. Only one interviewee mentioned the possibility of selection of applications, and that person mentioned it to dismiss it — referring to supervisory examiners having to “ resist cherrypicking what appeared to be the easiest cases. ”
+
+Thus, our interviews reveal no evidence that supervisory examiners do any kind of substantive evaluation of the applications before assigning them to particular examiners, and the press of work makes it implausible that they could
+
+---
+
+EXAMINER CHARACTERISTICS AND PATENT OFFICE OUTCOMES
+
+823
+
+Table 5.-Application-Level Characteristics versus Examiner Experience
+
+<table><tr><td></td><td>Model 5.1: Pages</td><td>Model 5.2: Family Size</td><td>Model 5.3: Applicant Patent Cites</td><td>Model 5.4: Applicant Publication Cites</td><td>Model 5.5: Applicant Patent Volume in 2000</td></tr><tr><td>2–4 years</td><td>0.450 (0.447)</td><td>0.113 (0.071)</td><td>−0.642 (0.502)</td><td>−0.070 (0.227)</td><td>0.284 (31.428)</td></tr><tr><td>5–7 years</td><td>0.176 (0.535)</td><td>0.259** (0.105)</td><td>−1.194** (0.605)</td><td>−0.107 (0.258)</td><td>−13.676 (44.379)</td></tr><tr><td>8+ years</td><td>0.241 (0.524)</td><td>0.088 (0.080)</td><td>−0.260 (0.531)</td><td>0.092 (0.186)</td><td>5.714 (31.972)</td></tr><tr><td>Constant</td><td>14.574*** (0.298)</td><td>2.972*** (0.051)</td><td>5.278*** (0.387)</td><td>1.379*** (0.158)</td><td>525.395*** (23.983)</td></tr><tr><td>Art unit effects N</td><td>Yes; jointly significant 9,846</td><td>Yes; jointly significant 9,846</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 4,503</td></tr></table>
+
+
+Robust standard errors (clustered on examiners) reported in parentheses. Asterisks denote statistical significance: *10 % , **5 % , ***1 % .
+
+do enough of a review to make a judgment as to whether an incoming application was likely to make it through the office. Nor would such a selection bias be logical. For example, recall our finding that more experienced examiners are much more likely to grant applications without amendment: if supervisory examiners were in fact engaged in some sort of sorting, we would expect it to work in the opposite direction than our data suggest, with the toughest applications rather than the easiest being assigned to the more seasoned examiners.
+
+Nonetheless, in this section, we explore selection more systematically. First, using both the application data and granted patent data, we look for evidence of selection on observables such as the length of the patent, family size, or number of claims or prior art citations. Second, we directly control for how patentworthy an application is by looking at whether the same applications were granted in Europe (at the European Patent Office, EPO).
+
+## A. Selection on Observables
+
+There are two difficulties in examining potential selection on observable variables. The first is that for patent applications, much of the front-page data available for issued patents (including citations and assignee information) are not available. Accordingly, in addition to examining selection on observables for our applications, we also do supplementary analyses on the subset that eventually issued as patents.
+
+More generally, for both applications and patents, it is difficult to identify variables that would measure the patentworthiness of an invention. Patent-worthiness is ultimately based on a reading of the claims and a judgment (ideally, by a person having ordinary skill in the art) about whether they are novel and nonobvious in light of the prior art. 13
+
+Lacking the ability to make that determination for each of the patents in our sample, we instead collected data on things we could measure to test for selection on observable
+
+characteristics. We started with examining the two variables on which we do have data at the time of application: the number of pages in the application and the patent family size, that is, the number of countries in which an application was filed. The latter has been used as a measure of patent value in the applied literature. Table 5 shows results.
+
+Models 5.1 and 5.2 relate each of these variables to the nonparametric experience measures, and art unit effects. In model 5.1 there is no clear pattern in any of the coefficients, and none are statistically significant. There is thus little evidence that more experienced examiners get systematically longer or shorter applications. For the “ family size ” measure, the coefficients also vary haphazardly with experience, and only one of the estimates (for five to seven years) is statistically significant at the 5 % level. Thus, here too there is no evidence of any systematic selection.
+
+As another check for selection, we looked at applicationlevel characteristics for the subset of granted patents. (Granted patents include more information, including the variables below, than applications.) This analysis requires characteristics of the application that could plausibly influence patent-worthiness but would not be associated with the effects of examiner characteristics on patent prosecution or features of the granted patents. Accordingly, we do not examine measures that could reflect the impact of the patent prosecution process itself (such as forward citation counts to an issued patent); we instead focus on “time zero” measures that reflect characteristics of the application as filed, rather than the patent as granted.
+
+Specifically, we examine the number of references to patented prior art provided by the applicant as part of the application, the number of references to nonpatented prior art, and the number of patents the applicant was issued in the previous year (2000), a measure of the applicant's experience. $^14$ Models 5.3 to 5.5 show results from regressions of these variables on examiner experience. These
+
+13 If there were a ready variable or set of variables that proxied for this, the PTO's task would be much easier.
+
+14 Data on the application patent volume variable are available for only the 4,503 patents that were assigned at issue and thus have assignee information listed.
+
+---
+
+824
+
+THE REVIEW OF ECONOMICS AND STATISTICS
+
+TABLE 6.-U.S. VERSUS EPO OUTCOMES
+
+<table><tr><td></td><td>Not Patented by PTO</td><td>Patented by PTO</td><td>Total</td></tr><tr><td>Rejected by EPO</td><td>344</td><td>525</td><td>869</td></tr><tr><td>Patented by EPO</td><td>128</td><td>1143</td><td>1271</td></tr><tr><td>Pending at EPO</td><td>79</td><td>512</td><td>591</td></tr><tr><td>Total</td><td>551</td><td>2,180</td><td>2,731</td></tr></table>
+
+
+models too show no evidence that more experienced examiners get different types of applications.
+
+The analyses thus show little evidence of selection on observables, using either the limited observables we have for applications overall or the more comprehensive set of application-level measures we have for applications that eventually mature into patents. But none of these variables is a perfect measure of the patent-worthiness of the application. For reasons discussed above, it is likely impossible to measure patent-worthiness directly. 15 In the next section, we examine a proxy variable, whether the same application was granted by the EPO.
+
+### B. Selection: Evidence from the EPO
+
+We also collected information from Delphion on whether the applications in our sample were also filed at the EPO, which examines and grants European patents for the 32 states in the European Patent Convention. For applications that were not granted by the EPO, we supplemented the Delphion data with information from the European Register on whether they were rejected or withdrawn or were currently pending. $^16$ Following Webster et al. (2007) , we limited our sample to the 8,905 applications filed in January 2001 that had a single priority application, to be certain that the corresponding EPO applications were on the same invention.
+
+Of these U.S. applications, 2,761 were filed at the EPO, which had granted patents on 46 % of them as of June 2008. Table 6 shows a cross-tabulation of EPO and PTO outcomes as of June 2008. Of the applications that were granted in the United States, slightly more than half (52.1 % ) have been granted by the EPO. 17 By contrast, of those granted by the EPO, the vast majority (88 % ) are also granted by the PTO.
+
+Table 7 shows the effects of the experience and search intensity measures on probability of U.S. grant, conditional on EPO status. Model 7.1 shows that relative to applications patented at the EPO (the left-out category), applications
+
+rejected at the EPO have a 27 percentage point lower probability of being patented in the United States. But controlling for EPO status, more experienced examiners have a statistically and qualitatively higher likelihood of grant than less experienced examiners.
+
+We also assessed the effects of experience on likelihood of U.S. patent grant separately for applications that were patented at the EPO (model 7.2) and rejected at the EPO (model 7.3). Model 7.2 shows that examiner experience does not affect the likelihood that applications patented at the EPO are patented in the United States: there is no statistically significant difference between older and newer examiners' probability of approving applications that were also approved by the EPO. By contrast, model 7.3 shows that more experienced examiners are more likely to grant applications that the EPO rejected.
+
+We are reluctant to treat EPO decisions as a gold standard to which U.S. decisions should be compared, given that the EPO too faces concerns about lax patent quality. But it is interesting that the experience effect does not reflect younger examiners being more likely to reject applications patented at the EPO; instead, older examiners are more likely to patent applications rejected by the EPO. That the experience effect persists when we control for EPO decisions is further reason to believe there is not a selection effect at work.
+
+## VIII. Other Explanations for the Experience Effect
+
+The analyses in the previous section suggest that the experience effect does not reflect sorting of applications across examiners. What, then, is its source?
+
+One possible explanation is selective retention. As we pointed out, the PTO faces significant employee attrition, particularly among examiners who have been with the agency less than five years (GAO, 2007) . If examiners who were more diligent, more thorough, more technically sophisticated, or more highly educated were more likely to leave the PTO earlier in their careers, perhaps because they have better job opportunities, this could provide one explanation for our results. Under this interpretation, the experience effect would reflect different features of examiners who stay versus leave, not anything about changing incentives or capabilities facing a given examiner over her career at the PTO. Since we lack any additional demographic information about our examiners, this is difficult to assess directly. Moreover, since we have only a cross-section of applications at a given point in time, we cannot examine changes in an examiner's grant behavior over her career.
+
+However, the PTO employee directories do give information on when and if the examiners in our January 2001 sample left the agency, assuming this happened before 2007. Using these data, we determined that 20.1 % of the examiners in our sample had left by the end of 2006. Attrition is more pronounced for the newest examiners (consistent with
+
+15 For the same reason, it is unlikely that supervisory patent examiners (SPEs) are judging patent-worthiness in their initial assignment decisions either.
+
+16 http://www.epoline.org/.
+
+17 This number differs somewhat from that in Jensen et al. (2006) , who show that 63 % of granted applications in the United States are also granted by the EPO in the most recent cohort for which they have data (priority year 1995). But this figure is trending downward over time (from 1990 to 1995) in their data, and, based on our data from 2001, appears to have continued to do so.
+
+---
+
+EXAMINER CHARACTERISTICS AND PATENT OFFICE OUTCOMES
+
+825
+
+TABLE 7.-LINEAR PROBABILITY MODELS RELATING WHETHER AN APPLICATION IS GRANTED TO EPO OUTCOMES, AND EXPERBENCE
+
+<table><tr><td></td><td>Model 7.1:Patented in United States?</td><td>Model 7.2: Patented in United States and Patented by EPO</td><td>Model 7.3: Patented in United States and Rejected by EPO</td></tr><tr><td>2–4 years experience</td><td>0.0369*(0.021)</td><td>–0.004(0.026)</td><td>0.051(0.053)</td></tr><tr><td>5–7 years experience</td><td>0.080***(0.028)</td><td>0.011(0.032)</td><td>0.174**(0.071)</td></tr><tr><td>8+ years experience</td><td>0.073***(0.021)</td><td>0.028(0.026)</td><td>0.118**(0.054)</td></tr><tr><td>Rejected by EPO</td><td>–0.269***(0.019)</td><td></td><td></td></tr><tr><td>Pending by EPO</td><td>–0.009(0.018)</td><td></td><td></td></tr><tr><td>Constant</td><td>0.844***(0.016)</td><td>0.890***(0.018)</td><td>0.536***(0.036)</td></tr><tr><td>Art unit effects N</td><td>Yes; jointly significant2,731</td><td>Yes; jointly significant1,271</td><td>Yes; jointly significant869</td></tr></table>
+
+
+Robust standard errors (clustered on examiners) reported in parentheses. Asterisks denote statistical significance: $^{*} 10 \%, ** 5 \%, *** 1 \%$.
+
+TABLE 8.-OLS Models RELATING Prosecution CHARACTERISTICS AND Outcomes UTCOMES TO Whether EXAMINER Will Leave IN Five Years EARS
+
+<table><tr><td></td><td>Model 8.1: Examiner Patent Cites</td><td>Model 8.2: Examiner Publication Cites</td><td>Model 8.3: Was the Application Patented?</td><td>Model 8.4: Was the Application Patented with No Rejections?</td></tr><tr><td>Examiner will leave in 5 years</td><td>−0.174 (0.212)</td><td>0.069 (0.048)</td><td>−0.006 (0.013)</td><td>.051*** (.014)</td></tr><tr><td>2–4 years experience</td><td>−0.366 (0.249)</td><td>−0.058 (0.040)</td><td>0.056*** (0.013)</td><td>.064*** (.013)</td></tr><tr><td>5–7 years experience</td><td>−1.014*** (0.296)</td><td>−0.001 (0.059)</td><td>0.109*** (0.018)</td><td>.117*** (.020)</td></tr><tr><td>8+ years experience</td><td>−1.725*** (0.243)</td><td>−0.076** (0.032)</td><td>0.112*** (0.013)</td><td>.136*** (.015)</td></tr><tr><td>Applicant patent cites</td><td>−0.011*** (0.004)</td><td>−0.002 (0.002)</td><td></td><td></td></tr><tr><td>Applicant publication cites</td><td>−0.003 (0.013)</td><td>−0.023 (0.015)</td><td></td><td></td></tr><tr><td>Constant</td><td>6.657*** (0.208)</td><td>0.169*** (0.033)</td><td>0.664*** (0.009)</td><td>.094*** (.009)</td></tr><tr><td>Art unit effects N</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 6,528</td><td>Yes; jointly significant 9,846</td><td>Yes; jointly significant 7,117</td></tr></table>
+
+
+Robust standard errors (clustered on examiners) reported in parentheses. Asterisks denote statistical significance: * 10 % , **5 % , ***1 % .
+
+the figures in GAO, 2007) with 27.1% of the examiners hired in 2001 having left by the end of 2006.
+
+To assess whether examiners who would exit are different from others, we estimated models relating the two measures of search intensity and grant rates to a dummy variable indicating whether the examiner would leave the PTO by 2006. Table 8 shows the results.
+
+Models 8.1 and 8.2 respectively relate the number of patent and publication references in a patent to whether the examiner would exit by 2006, conditional on experience. Conditional on experience, examiners who would leave within five years have a lower number of patent citations and a higher number of publication citations, though neither effect is statistically significant. For patented applications, we also examined if whether an application was granted (model 8.3) and the probability of granting with no rejections (model 8.4) related to whether an examiner would leave. The relationship between whether an examiner will
+
+leave and the grant rate is essentially zero and statistically insignificant. And examiners who would exit have a higher likelihood of granting without rejections, the opposite of what we would expect were selective retention the explanation for why more experienced examiners tend to be more likely to grant without rejections. Collectively, these results suggest that selective retention is unlikely to be the source of the experience effect we documented above.
+
+Another possibility is examiner tenure. After promotion, examiners are not subject to the same level of scrutiny. Among other things, with full signatory authority, they can sign off on their own applications without review. This could plausibly cause them to be more lax.
+
+It is difficult to test for a tenure effect directly. A first problem is data: we lack direct information about whether examiners are tenured or untenured. But using data from issued patents, we can determine which examiners were definitely tenured by 2001. Specifically, we can infer an
+
+---
+
+826
+
+THE REVIEW OF ECONOMICS AND STATISTICS
+
+examiner's probable tenure year as the first year after which she was listed as primary examiner on all the patents she issued that were applied for in that year. In other words, we determined the first application year after which an examiner consistently had signatory authority.
+
+Not surprisingly, the likelihood an examiner is tenured increases sharply with experience, with 9.6 % of examiners with two to four years of experience tenured by 2001, 41.6 % of examiners with five to seven years of experience, and 90.7 % of examiners with eight or more years experience. The strong relationship between tenure and experience also makes it difficult to separately identify a tenure effect separate from an experience effect. More important, tenure is nonrandom: for a given level of experience, those who are tenured are likely to be systematically different from those who are not.
+
+While we cannot rule out a tenure effect, the result that there are strong differences in patent application outcomes even between the two- to four-year cohort (the bulk of whom are not tenured) and brand-new examiners suggests that it alone is unlikely to drive the experience effect.
+
+Another potential explanation is the quota system, which imposes increasing demands on examiners and decreased time per count as they accumulate experience at the PTO. It may be that more experienced examiners are doing less prior art searching and granting more patents because they are required to process more applications in the same time frame. Still another possible explanation is that more senior examiners are less familiar with current trends in their field and therefore more likely to find applications patentable. Whether the experience effect documented in the main models reflects the quota system, technological obsolescence, or tenure, it suggests that human resource policies at the PTO could affect outcomes.
+
+## IX. Implications
+
+Consistent with the qualitative and quantitative results of Cockburn et al. (2003) , our data emphasize that examiners matter. In particular, more experienced examiners cite less art, issue fewer initial rejections, and are more likely to grant patents.
+
+Taken alone, the result that more senior examiners are more likely to grant could suggest that they can more quickly figure out what is patentable in an application. But our data on prior art citation patterns do not support that conclusion. The finding that more senior examiners systematically cite less prior art reinforces the inference that senior examiners are doing less work, rather than that they are merely getting it right more often than junior examiners. $^18$ The fact that seniority is correlated with more first-action allowances — allowances where there was no rejection — is
+
+also inconsistent with the idea that more experienced examiners are simply negotiating the applicant to a narrower, patentable outcome; in the first-action allowance cases, there is no negotiation at all.
+
+We show that these results are unlikely to reflect either sorting of applications across examiners or selective retention of examiners. But our data cannot separately distinguish between a technological obsolescence effect, an experience effect driven by the count/quota requirements over an examiner's career, or a tenure effect, each of which would have different policy implications. But under any of these explanations, our results suggest that human resource policies have important effects on PTO outcomes. From this perspective, the tenure system, the count system, and examiner recruitment and retention policies should be a more prominent part of current patent reform deliberations.
+
+Our findings may also have significant implications for the industry-specific results we discussed in our prior paper (Lemley & Sampat, 2009) . While there is no question that there are industry-specific differences throughout patent prosecution, some of the differences we identified in the prior article may turn out to owe their origin to differences in examiners. In preparing this study, we determined that the computer industry had by far the highest percentage of new examiners: more than 60 % of examiners in that art unit had less than a year of experience compared with less than 20 % in mechanics and chemistry. In our prior paper, we found that the computer industry had a surprisingly low grant rate: lower than any other industry. At least some of that result may be explained by the prevalence of new examiners in that industry. If it is generalizable, this result may have another effect: booms in patenting in new industries may be self-limiting. The more applications are filed in an art unit, the more new examiners the art unit will have to hire. And because new examiners are more likely to reject patents, this will drive down the grant rate in that art unit, limiting the number of patents that result.
+
+Finally, our data suggest that whether the patent office grants or rejects a patent is significantly related to the happenstance of which examiner is assigned the application. That may not be an encouraging result if our goal is a system that rewards deserving patent applicants while denying patents to the rest. $^19$ Nor is it clear that in this environment, all issued patents should enjoy the same presumption of validity.
+
+18 Similarly, while one might have sought to explain the lower citation patterns by more experienced examiners as greater parsimony learned from experience, the fact that those reduced citations are accompanied by a greater propensity to grant patents undermines that explanation.
+
+19 As a referee usefully pointed out to us, to the extent that the PTO's objectives are broader than quality—instead including speed and quantity of applications processed, for example—heterogeneity across experienced and less experienced examiners would be less of a problem. Thus, even if some examiners search less and are more likely to grant quality patents than others, they may compensate by handling more applications per unit of time. There is some evidence that more experienced examiners at least receive more applications: examiners with zero to one year of experience received an average of 3.2 new applications in January 2001, compared to 3.6 for examiners with two to four years of experience, 3.6 for examiners with five to seven years of experience, and 3.9 for examiners with more than eight years of experience.
+
+---
+
+EXAMINER CHARACTERISTICS AND PATENT OFFICE OUTCOMES
+
+827
+
+## REFERENCES
+
+Alcacer, J., and M. Gittleman, "Patent Citations as a Measure of Knowledge Flows: The Influence of Examiner Citations," this REVIEW 88:4 (2006), 774-779.
+
+Alcacer, J., M. Gittleman, and B. Sampat, "Applicant and Examiner Citations to Patents: An Exploration and Overview," Research Policy 38 (2009), 415-427.
+
+Allison, J. R., and M. A. Lemley, "Who's Patenting What? An Empirical Exploration of Patent Prosecution," Vanderbilt Law Review 56 (2000), 2099-2174.
+
+Cockburn, I., S. Kortum, and S. Stern, "Are All Patent Examiners Equal? Examiners, Patent Characteristics, and Litigation Outcomes," In W. M. Cohen and S. A. Merrill (Eds.), Patents in the Knowledge-Based Economy (Washington, DC: National Academies Press, 2003).
+
+Cohen, W. M., and S. Merrill, Patents in the Knowledge-Based Economy (Washington, DC: National Academies Press, 2003).
+
+Cotropia, C., M. Lemley, and B. Sampat, "Do Applicant Citations Matter?" working paper (2012).
+
+Federal Trade Commission (FTC), “To Promote Innovation: The Proper Balance of Competition and Patent Law and Policy,” Federal Trade Commission technical report (2003).
+
+Government Accountability Office (GAO), "USPTO Has Made Progress in Hiring Examiners, but Challenges to Retention Remain" (2005).
+
+——— "Hiring Efforts Are Not Sufficient to Reduce the Patent Application Backlog" (2007).
+
+Jaffe, A. B., and J. Lerner, Innovation and Its Discontents: How Our Broken Patent System Is Endangering Innovation and Progress, and
+
+What to Do about It (Princeton, NJ: Princeton University Press, 2004).
+
+Jensen, P. H., A. Palangkaraya, and E. Webster, "Disharmony in International Patent Office Decisions," Federal Circuit Bar Journal 15 (2006), 679-704.
+
+Lemley, M. A., "Rational Ignorance at the Patent Office," Northwestern University Law Review 95 (2001), 1495-1532.
+
+Lemley, Mark A., and C. Shapiro, "Probabilistic Patents," Journal of Economic Perspectives 19 (2005), 75-98.
+
+Lemley, M. A., and K. A. Moore, "Ending Abuse of Patent Continuations," Boston University Law Review 82 (2004), 63-118.
+
+Lemley, M. A., and B. Sampat, "Is the Patent Office a Rubber Stamp?" Emory Law Journal 58 (2009).
+
+Lichtman, D., "Rethinking Prosecution History Estoppel," University of Chicago Law Review 71 (2004), 181-208.
+
+Merges, R., "As Many as Six Impossible Patents before Breakfast: Property Rights for Business Concepts and Patent System Reform," Berkeley Technology Law Journal 14 (2001), 577615.
+
+Norhaus, W., Invention, Growth, and Welfare (Cambridge, MA: MIT Press, 1969).
+
+Sampat, B., "When Do Applicants Search for Prior Art?" Journal of Law and Economics 53 (2010), 399-416.
+
+Thomas, J., "Collusion and Collective Action in the Patent System: A Proposal for Patent Bounties," University of Illinois Law Review 2001(1), 305-353.
+
+Webster, E., A. Palangkaraya, and P. H. Jenson, "Characteristics of International Patent Application Outcomes," Economic Letters 95 (2007), 362-368.
+

processed/00235_W2059930869/document.md

@@ -0,0 +1,419 @@
+Volume 5 • Number 1 • 2002 VALUE IN HEALTH
+
+# The Economics of Gene Therapy and of Pharmacogenetics
+
+Patricia Danzon, PhD,1 and Adrian Towse, MA, MPhil2
+
+1Wharton School, University of Pennsylvania, Philadelphia, PA, USA; 2Office of Health Economics, London, England, UK
+
+## ABSTRACT
+
+This paper focuses on the economic issues arising from two uses of genomics: 1) the development of gene therapy; 2) and use of pharmacogenetics to identify a patient’s genotype before treatment to exclude those who will not benefit or who may be harmed. We conclude that private-sector investment aimed at developing gene therapy for monogenic diseases is likely to be socially suboptimal. Short-term administration regimens yielding long-term therapeutic benefits are likely to meet payer resistance to large “one-off” costs because of budget constraints or, in competitive systems, concerns that the savings would accrue to future insurers or would attract high-cost patients. For some monogenic diseases, patient numbers may be too small to support commercial development without changes to orphan drug legislation or payer willingness to accept higher cost-effectiveness
+
+thresholds. In the case of pharmacogenetics, we conclude that it can often be socially optimal to test before treatment, particularly if the proportion of nonresponders is high, if there is a potential for serious adverse reactions, or if the test is inexpensive. Genetic testing that fragments the patient population could reduce incentives for R&D unless prices are adjusted to reflect the higher expected benefits of targeted treatment per patient. Even in situations where prices are adjusted, patient populations may be too small to make commercial development viable. This problem with small numbers is analogous to that associated with gene therapy for monogenic diseases and may require similar remedies if society values developing treatments for these diseases.
+
+Keywords: gene therapy, monogenic disease, orphan drugs, pharmaceutical pricing, pharmacogenetics.
+
+## Introduction
+
+Genomics offers great promise for providing new and more effective therapies and diagnostic tests for patients. Many new medicines will arise from the use of pharmacogenomics in conventional drug discovery as knowledge of the human genome increases understanding of disease. This paper focuses on the economic issues arising from two relatively novel uses of genomics:
+
+- 1. the development of gene therapy, where the
+aim is to insert genes that will produce or regu-
+late the expression of proteins that are related
+to the patient's disease. This approach is most
+immediately relevant to monogenic diseases,
+many of which are currently incurable. The
+hope is that gene therapy will provide long-
+term therapeutic benefits;
+2. the use of pharmacogenetics to identify a pa-
+tient’s genotype before treatment, to identify
+- those who will not benefit or who may be
+harmed. Here genomics is used to identify genetic
+traits that may lead either to nonresponse or to
+adverse reactions to specific medicines for any in-
+dication, including the great majority of diseases
+that are polygenic but in which genetic makeup
+can affect response to specific treatments.
+We examine whether existing reimbursement and regulatory regimes will encourage the socially optimal development of these two uses of genomics and suggest possible policy changes. The high costs and uncertainties associated with gene therapy and its very novel modes of action, small patient numbers, and long-lived therapeutic effects may lead to suboptimal levels of commercial research. In the case of pharmacogenetics, testing could be socially beneficial but nevertheless reduce innovative firm's incentives to develop new drugs. This is in part due to the low numbers of patients treated—because drugs would be targeted at a subset of the patient population—and to the costs of testing. This may be offset by reduced R&D costs per drug developed or by unit price increases that reflect the greater specificity of use and consequent greater expected health gain per patient treated.
+
+Address correspondence to: Adrian Towse, Director, Office of Health Economics, 12 Whitehall London SW1A 2DY. E-mail: atowse@ohe.org
+
+The Office of Health Economics is supported by the Association of the British Pharmaceutical Industry.
+
+c ISPOR 1098-3015/02/$15.00/5 5-13
+
+5
+
+---
+
+6
+
+Danzon and Towse
+
+## Formal Statement of the Problem
+
+Assume that a new therapy (g) is considered costeffective by payers for their patients relative to an existing treatment (0) if
+
+$$\Delta C_{g, 0} / \Delta E_{g, 0}<k_{p}\quad (1)$$
+
+where
+
+$\Delta C_{g,0}=P_{g}-P_{0}+\Delta C_{g,0}\frac{d}{4}\Delta C_{g,0}^{I}$
+
+$P_{g}$, $P_{0}=$ respectively, the prices of the gene therapy and existing treatment; $\Delta C_{g,0}^{d}=$ change in other direct treatment costs; $\Delta C_{g,0}^{I}=$ change in indirect costs; $\Delta E_{g,0}=$ change in quality-adjusted life years (QALYs) or other outcome measures; and $k_p=$ threshold cost per QALY at which an intervention is considered cost-effective for the treatment of patient group p.
+
+We assume $\Delta C_{g,0}$ and $\Delta E_{g,0}$ are discounted at a socially optimal rate. If $k_p$ is optimally chosen (i.e., reflects the willingness to pay for additional investment in treatment of patient group p), then equation 1 defines the condition for socially optimal reimbursement. We can use it to define the maximum price ( $P_g^{max}$ ) at which the new therapy is cost effective:
+
+$$P_{z}^{\max }=k_{p} \Delta E_{z, 0}+\left(P_{0}+\Delta C_{z, 0}^{d}+\Delta C_{z, 0}^{1}\right)\quad (2)$$
+
+$$ or  P_{g}^{\max }=b \quad (2)^{\prime}$$
+
+where b = k p $\Delta$ E $_{g,0}$ + (P $_{0}$ + $\Delta$ C $_{g,0}$ d + $\Delta$ C $_{g,0}$ l ) and represents the social benefits (positive or negative) from treatment g compared with treatment 0.
+
+Assume that payers typically set the actual price at some fraction $\alpha$ of the maximum price, where $\alpha$ reflects the share of the social gain that accrues to the innovator firm and $(1-\alpha)$ is the share captured by the payer, so that
+
+$$P_{g}=\alpha b=\alpha P_{g}^{\max }\quad (3)$$
+
+Consider now the perspective of the firm planning to invest in developing a new therapy (g). We abstract from the fact that in practice $\alpha$ and $P_g^{max}$ may not be known with certainty because of a lack of transparency in the price-setting process. Ignoring this uncertainty and using a net present value investment-valuation approach, the producer's breakeven profit constraint can be written:
+
+$$\Pi^{T}=\Sigma^{T}[(P_{2}-M)Q^{t}N^{t}(1+r)^{-t}]-F(r,L,p)\quad (4)$$
+
+where $\langle  \Pi \rangle =$ discounted present value of net revenue over the T years of the product's market life (“profit”);
+
+M = variable cost per treatment to the producer, assumed to be invariant over time; N = number of patients treated per year; Q = number of treatments per patient per year, such that NQ is the number of treatments sold per year; F = discounted present value of the firm's R&D cost, with $F_{r}$, $F_{L}>0$, $F_{p}<0$; r = cost of capital; L = expected years from discovery to launch; p = probability of success in showing safety and efficacy in clinical trials; other fixed costs are zero.
+
+To assess the implications of the payer's costeffectiveness requirement on the producer's breakeven constraint, we substitute the price from equation 3 into the firm's break-even constraint in equation 4 . Therefore,
+
+$$\Pi^{T}=\Sigma^{T}[(\alpha b-M)Q^{l}N^{l}(I+r)^{-t}]-F(r,L,p).\quad (4)^{\prime}$$
+
+Equation 4' provides a framework in which to consider how the characteristics of gene therapy and pharmacogenetics may affect their commercial viability compared with other pharmaceutical R&D.
+
+## Gene Therapy
+
+Let us consider effects on revenue and R&D costs.
+
+The total operating net revenue depends on:
+
+- 1. the social value of treatment b (the cost savings
+and gain in quality of life) and the producer's
+share, $\alpha$ ;
+2. the number of patients treated per year N,
+which may be low because most monogenic
+diseases (i.e., those involving only a single gene)
+affect a relatively small number of patients;
+3. Q, which may be small because of the long-
+term benefits of gene therapies, implying that
+each patient may require treatment to be ad-
+ministered only once or twice a year rather
+than once or twice a day, the standard regimen
+for many pharmacotherapeutic agents adminis-
+tered to chronically ill patients; and
+4. the variable cost of producing and delivering
+the treatment, M, which is likely to be signifi-
+cantly higher for gene therapies than for other
+drugs owing to both the novel delivery systems
+required and the small patient base over which
+to realize scale economies.
+The extended duration of benefits and consequent reduction in frequency of administration
+
+---
+
+Gene Therapy and Pharmacogenetics
+
+7
+
+(low Q per patient for long-lasting therapies) would be irrelevant if payers were willing to assume that the cost of new therapies would be fixed relative to the full social benefits: P = α Pmax. In this case, the price for gene therapy treatment would increase proportionately with the duration of its effects. A low Q would be associated with a relatively high benefit, b, per administered treatment compared with a once-a-day alternative therapy, thus preserving neutral incentives for efficient investment in R&D with respect to duration of benefits. This ideal may, however, be undermined in practice for several reasons.
+
+First, payers tend to scrutinize and bargain aggressively over products that are priced relatively high per dose or per patient treated. If so, the producer’s share is a decreasing proportion of the maximum price: $\mathrm{d}\alpha$ / $\mathrm{dP}^{\max} < 0$ . Second, turnover of patient populations could make competing health insurers reluctant to pay for long-lasting therapies because the insurer that pays for the initial treatment does not capture the full savings in future treatment costs if patients subsequently switch to other insurers. Thus the risk of adverse selection could reinforce the incentive of insurers to avoid offering long-lived therapies that target high-cost patients, such as gene therapy. This should be less of a problem in countries where patients have a limited choice of health plans, such as Canada and the United Kingdom. However, in these systems managers and doctors face annual budget constraints that limit their ability to invest in treatments that have higher immediate costs but longer-term benefits.
+
+The private cost of R&D F(r, L, p), may be atypically high for gene therapy compared with conventional therapy despite the relatively smaller requirements in terms of trial sizes. The extremely novel mode of action means that the probability of success, p, is very low and the expected duration of the R&D process, L, is relatively long. Several hundred clinical trials of gene therapy have been initiated but none has been successfully completed so far. The deaths in 1999 of two trial participants [1] have made expectations even more pessimistic.
+
+Thus, values for each element of the expression for $\Pi^\mathrm{T}$ are likely to be lower in the case of gene therapies for monogenic diseases than corresponding values for conventional therapies for major diseases. The high risk and costs of R & D to companies are a reflection of real social costs arising from basic research into new technologies, the benefits of which cannot be fully expropriated by
+
+the company. If reimbursement systems are biased against therapies providing long-term benefits, under-investment compared to the social optimum may result.
+
+An examination of clinical trials in gene therapy tends to support our concerns that low patient numbers and investor perception of payer resistance to long-lived therapies may be influencing the allocation of R & D efforts. The majority of gene therapy trials are in cancer indications (oncology), followed closely by trials directed at cures for AIDS and HIV infection [2] . Many trials are focused on variants of treatment that would require repeat administration rather than providing a one-off cure. Only one monogenic disorder, cystic fibrosis, is the subject of significant clinical development, due in part to the early discovery of the gene for this disease. Unlike all other areas of drug development, most clinical trials in gene therapy are undertaken with at least partial public funding [3] . This suggests that the initial promise of gene therapy — that of delivering cures for monogenic diseases — is unlikely to be realized without changes in incentives or significant public investment.
+
+The problem of insufficient commercial incentives for investment in drugs for small populations is not unique to gene therapy. Orphan drug legislation enacted in the United States in 1983 provides special incentives, including 5-year market exclusivity for the orphan indication and special tax credits, for drugs to treat diseases that affect fewer than 200,000 US patients. The European Union has now established an orphan drug regime. However, long-lived gene therapies are disadvantaged relative to conventional therapies by the current legislation if payers resist price increases in proportion to the duration of the effects. If the 200,000 patient threshold is the number of patients expected to use an orphan drug per day per year, then, arguably, this number should be adjusted for long-lasting gene therapies. For example, if the gene therapy lasts 5 years, then in steady state only one-fifth of the population with the disease would be treated per year. Thus, to provide neutral incentives the orphan drug threshold for long-lasting therapies should be $n\times 200$ ,000, where $n$ is the average duration of benefits for one treatment with gene therapy relative to that of conventional therapies. This will increase the likelihood that a gene therapy receives orphan status and the associated benefit of market exclusivity, which should increase the price the company receives for the gene therapy, thereby in-
+
+---
+
+8
+
+Danzon and Towse
+
+creasing $\alpha$, the share of the social benefit that the company obtains.
+
+Thus, the characteristics of gene therapy—long and uncertain R&D, a small patient base, and infrequent treatment—may lead to suboptimal commercial investment in these therapies. Reimbursement systems introduce a bias against gene therapy if payers respond to budgetary or commercial pressures by focusing on short-term drug-budget costs without due weight to long-term health benefits and societal savings. Although society has signaled a willingness to pay additional subsidies to encourage treatments for orphan diseases, current legislation is not neutral between treatments administered on a daily basis and those administered sporadically or once or twice in a lifetime. We consider the public policy implications in the final section of the paper.
+
+## Pharmacogenetics
+
+Pharmacogenetic testing is designed to identify patients’ genotypes so that drugs can be targeted to the subgroup whose genetic makeup makes them most likely to benefit. This raises the expected effectiveness per patient treated and hence the costeffectiveness of the drug by eliminating the cost of treating patients whose genetic makeup makes them unlikely to benefit (nonresponders) or likely to suffer harm (adverse responders). For example:
+
+- 1. New tests based on the ApoE gene may identify
+patients who are more likely to benefit from
+drugs designed to slow the symptomatic degen-
+eration associated with Alzheimer's disease [4] .
+2. Testing for the human CysLT 1 receptor for cys-
+teinyl leukotrienes may predict the effective-
+ness in individual patients of the three new cys-
+teinyl leukotriene antagonists for asthma [5] .
+3. The presence of the B 1 variant of the CETP
+gene appears to predict the response of patients
+with coronary atherosclerosis to statin treat-
+ment. In a clinical trial statins slowed disease
+progression in B 1 B 1 carriers but not in B 2 B 2 car-
+riers [6] .
+Payers would rationally adopt pharmacogenetic testing before treatment if the savings from treating fewer patients and avoiding complications exceeded the costs of testing. For drug companies pharmacogenetic testing means lower patient volume and hence lower revenues per drug, other things being equal. This reduction in gross sales may be exacerbated if payers subtract the costs of the genetic screening from the price that they are
+
+willing to pay or reimburse for the drug. A key issue, therefore, is the extent to which we might expect drug prices to increase as a consequence of testing, as might be the case if the innovator were able to get a share of any increased societal benefit arising from testing.
+
+To examine the potential impact of testing from societal, payer, and company perspectives we adapt the formal model set out earlier to introduce responders and nonresponders, adverse reactions, and testing costs.
+
+Let $N_1$ be the number of patients who benefit from the drug. $N_2$ is the number who do not benefit but who cannot be identified without testing, so
+
+$$N=\left(N_{1}+N_{2}\right)$$
+
+Let $b_1 = k_p \Delta E_{1g,0} + (P_0 + \Delta C_{1g,0}^d + \Delta C_{1g,0}^l)$ i.e., the health gain plus cost savings per patient relative to current treatment. Only the responders, $N_1$ , obtain these benefits.
+
+Let $a_2 = k_p \Delta E_{2g,0} + (P_0 + \Delta C_{2g,0}^d + \Delta C_{2g,0}^l)$ denote the adverse health effect ( $\Delta E_2 < 0$ ) plus the consequential costs of the adverse reaction for each patient in group $N_2$ .
+
+We let $P_{g1}^{max}$ be the maximum price of the drug without testing, $P_{g2}^{max}$ be the maximum price of the drug with testing, and $P_t$ be the price of the test. Thus, adapting equation 2', the maximum price of the drug without testing is
+
+$$P_{g 1}^{\max }=\left(b_{1} ~N_{1}-a_{2} ~N_{2}\right) / N . \quad (5)$$
+
+With testing, the payer's maximum price for a drug targeted solely at the responders is
+
+$$P_{g2}^{\max}=b_{1}-P_{1}N/N_{1}.\quad (6)$$
+
+The change in the maximum price, which is the social value of testing, is therefore
+
+$$\begin{aligned}
+\DeltaP_{g2}^{\max}&=P_{g2}^{\max}-P_{g1}^{\max}\\
+&=b_{1}(1-N_{1}/N)-P_{t}N/N_{1}+a_{2}N_{2}/N\\
+&=(b_{1}+a_{2})(N_{2}/N)-P_{t}N/N_{1}.&(7)
+\end{aligned}$$
+
+This social value of testing reflects three factors:
+
+- 1. the expected health benefit per patient treated,
+which is equal to b 1 (the benefit per responder)
+times the proportion of nonresponders N 2 /N;
+2. the averted costs of treating the adverse effects
+of the drug on nonresponders, a 2 , times the
+proportion of nonresponders N 2 /N;
+3. a cost of testing the whole patient population,
+P, N, amortized over the N 1 responders.
+The first two components will be non-negative and testing will be beneficial if these effects exceed
+
+---
+
+Gene Therapy and Pharmacogenetics
+
+9
+
+the third element, the cost of testing. Formally, testing offers benefit to society if $\Delta P_{5}^{max}>0$, or
+
+$$(b_{1}+a_{2})(N_{2}/N)>P_{t}N/N_{1}.\quad (8)$$
+
+Thus testing is of benefit if the savings from avoiding treatment and side effects for the N2 nonresponders exceeds the cost of testing all patients. In the simplest case, if a2 is zero (no side effects) and we substitute for P $_{g1}$ max from equation 5 this can be rewritten:
+
+$$N_{2} /\left(N_{1}+N_{2}\right)>P_{1} / P_{g 1}^{max } . \quad (9)$$
+
+Testing is worthwhile from a societal perspective if the ratio of nonresponders to the total population exceeds the ratio of the cost of the test to the social value of the drug in the absence of testing.
+
+Returning to the general case where there may be adverse events ( $a_2 \geq 0$ ), and assuming that the price obtained by the company reflects the same share $\alpha$ of the social benefit of the drug with and without testing, then equation 7 shows that testing increases the price of the drug obtained by the company:
+
+$$\alpha \Delta P_{g}^{\max }=\alpha(b+a)\left(N_{2} / N\right)-\alpha P_{1} ~N / N_{1} . \quad (10)$$
+
+The company gets a share $\alpha$ of the social gain of avoiding the drug costs and adverse events associated with treating nonresponders but also bears the share of $\alpha$ of the costs of testing. It is rewarded and motivated to produce more specific drugs. In this case only the share $\alpha$ of the cost of the test $P_t$ is borne by the company, with the payer accepting to pay the remainder $(1-\alpha)$ . This follows from the assumption that the cost of the test is deducted to arrive at social benefit before it is divided between the producer and the payer. Note that in the simplest case of zero adverse reactions and test costs (i.e., if $a_2=0$ , $P_t=0$ ) the price would rise in proportion to the increase in proportion of patients expected to benefit:
+
+$$\alpha \Delta P_{g}^{\max }=\alpha b_{1}\left(~N_{2} / N\right)  or  P_{g 2} / P_{g 1}=N / N_{1} .$$
+
+In practice, payers may be unlikely to permit the price of the drug, adjusted for the cost of testing and of side effects averted, to increase in proportion to its expected benefit per patient. We suggested above that payers scrutinize most stringently those products that are priced relatively high; actual price is then a decreasing proportion of the maximum price: $d\alpha$ / $dP^{max} < 0$ , in which case $P_{5}$ would be lower than that suggested by equation 10 . The cost of the test, $P_{5}$ , the ratio of nonresponders to responders, $N_{2}/N_{1}$ , the severity of the adverse reactions, and the value of $\alpha$ are all
+
+crucial to the ability of a company to obtain a price premium for a more targeted product and to face, ex ante, neutral incentives for developing targeted products versus more indiscriminate products with lower expected benefits per patient treated. We consider in turn the payer and company perspectives.
+
+## The Payer Perspective
+
+We can note the general conditions in which testing is of benefit to the payer. Let B1 denote the potential payer benefit per period with no test, B2 denote the potential payer benefit with testing, Pg1 be the price the payer is paying for the drug in the absence of a test, and Pg1 + ΔP be the price with a test. Note that we assume that payers focus on health effects and all costs. Adapting equations 5 and 6 above yields
+
+$$B_{1}=N_{1}b_{1}-N_{2}a_{2}-N P_{g1}\quad (11)$$
+
+$$B_{2}=N_{1}b_{1}-N_{1}(P_{\xi1}+\DeltaP)-N P_{t}.\quad (12)$$
+
+In assessing the cost-effectiveness of testing to the payer, potential savings from avoiding adverse events and not paying for drugs to treat nonresponders have to be offset by the costs of testing and any higher price that is charged by the innovator. The protocol with testing offers greater benefit to the payer than does indiscriminate treatment of all patients if
+
+$$B_{2}-B_{1}>0,  or$$
+
+$$N_{2}(a_{2}+P_{gl})>N_{1}\Delta P_{z}+NP_{1}.\quad (13)$$
+
+If there are no adverse effects, i.e., $a_2 = 0$ , and if the payer does not give the company a price increase following the introduction of testing, so that $\Delta P_g = 0$ , then equation 13 reduces to
+
+$$N_{2}/N>P_{t}/P_{g}.\quad (14)$$
+
+Testing is worthwhile for the payer if the ratio of nonresponders to the total population exceeds the ratio of the price of the test to the price of the drug. More generally, the extent to which payers are prepared to award higher prices, i.e., $\Delta P_g > 0$ is crucial to creating consistent incentives for companies to develop new products using pharmacogenetics given society's cost-effectiveness threshold $k_p$ . Of course if companies are able to develop other products that treat nonresponders at prices that result in a positive incremental cost-effectiveness below the cost-effectiveness threshold $k_p$ , overall payer expenditures will rise. This may give rise to rationing issues if payer budgets are constrained (so $k_p$ will rise) or to more aggressive bar-
+
+---
+
+10
+
+Danzon and Towse
+
+gaining by payers seeking to reduce $\alpha$ . We do not consider this issue further. We now consider the impact on the company.
+
+## The Company Perspective
+
+Assume that the innovative firm faces two choices. It could ignore the possibility of pharmacogenetic testing and develop a traditional drug. This drug would be targeted indiscriminately to all patients with the disease in question, of which a proportion receives no benefit and may be harmed. The innovative firm's alternative choice is to develop and sell a genetic test that would identify the N 1 patients who will benefit and produce a drug targeted to them. Assume that the test can be sold at a price P 1 and produced at constant marginal cost C 1 . Adapting equation 4, assuming that Q = 1 for simplicity, let $\Pi_1$ , P 5 be the producer's profit and price, respectively, with no testing, and $\Pi_2$ , P 5 + $\Delta$ P 5 , the profit and price with testing, and F 2 , the R&D cost with testing. Therefore,
+
+$$\Pi_{1}^{\top}=\Sigma^{\top}[(P_{g1}-M)N^{t}(1+r)^{-4}]-F_{1}\quad (15)$$
+
+$$\begin{split}
+\Pi_{2}^{\top}=&\Sigma^{T}[(P_{gl}+\DeltaP-M)N_{1}^{t}(1+r)^{-t}]-\\
+&F_{2}+N^{t}(P_{t}-C_{t})(1+r)^{-t}.&(16)
+\end{split}$$
+
+The producer's profit is greater with the test than without only if $\Pi_2^\top - \Pi_1^\top > 0$ , or
+
+$$\begin{aligned}
+&\Sigma^{T}(P_{g1}+\DeltaP_{g}-M)N_{1}^{1}(1+r)^{-1}+\\
+&(F_{1}-F_{2})+N^{1}(P_{1}-C_{1})(1+r)^{-1}>\\
+&\Sigma^{T}[(P_{g1}-M)N^{1}](1+r)^{-1}.&(17)
+\end{aligned}$$
+
+Equation 17 shows that if the final drug price is unchanged, i.e., $\Delta P_g = 0$ , the innovative firm has no incentive to invest in pharmacogenetic testing in development that will result in a narrower indication unless there are savings in R & D costs ( $F_1$ – $F_2$ ) or profits to be made on the provision of the tests. Savings in R & D costs may be possible if, for example, genetic testing permits phase III trials to be targeted to fewer patients who are more likely to benefit. Thus, efficacy may be demonstrated with much smaller trials. However, there may be additional costs if either the link between the gene and the response or the reliability of the test has to be validated. It is also possible that with genetic testing the drug could be designed such that it is effective for a larger fraction of the patient population. In that case, the tendency for pharmacogenetics to reduce the average size of the target population per drug would be mitigated. Moreover, if the proportion of patients N 2 /N who fail to benefit is expected to be relatively large, an untargeted
+
+drug might fail to qualify for reimbursement because of poor cost-effectiveness.
+
+More realistically, with free entry to the business of developing genetic tests, pretreatment tests are likely to be developed irrespective of the actions of the innovator when they can yield a net saving to the payer. This occurs when the cost of testing the entire patient population N P t is less than the savings from avoiding treatment of nonresponders P g N 2 plus any savings from averting harm, a 2 N 2 . It is likely, therefore, that drug producers will have incentives to do this testing themselves as part of drug development rather than wait for others to do it after drugs reach the market. In the latter case, the producer suffers the loss of sales but gets none of the potential benefits of smaller trials or an improved drug design. Nevertheless, to the extent that pharmacogenetic testing tends to reduce the patient population per drug, some drugs may not be worth developing once testing becomes an option if the reduction in expected revenues due to population fragmentation exceeds the reduction in R & D costs.
+
+The key issue, then, is what happens to drug prices. If pre- and post-testing prices reflect expected social benefits, the price of the drug will increase in proportion to the expected benefits (net of testing costs) as specificity increases and the risk of zero benefit or positive harm declines as a consequence of genetic testing.
+
+The willingness of payers to award higher prices for targeted benefits (i.e., maintaining a constant value of $\alpha$ ) will be essential to retaining neutrality in investment incentives.
+
+## Examples
+
+Two drug launches illustrate the potential impact of testing on manufacturers and payers. In one case, nebacumab (CentoxinTM, Centocor, Malvern, PA) no test was available, whereas in the case of trastuzumab (Herceptin, Genentech, San Francisco, CA), tests are available.
+
+Nebacumab. Nebacumab was launched in 1991 in most European countries [7] as a treatment for sepsis. However, it only worked in those cases where the sepsis was due to Gram-negative bacteremia, or approximately one-third of all sepsis cases. With a cost of $4000 per patient, doctors found themselves under pressure to use the drug on all cases of sepsis, despite knowing that for every 1000 patients treated, $2.67 million was being spent on
+
+---
+
+Gene Therapy and Pharmacogenetics
+
+11
+
+drugs for patients who could not benefit. It eventually became clear that nebacumab was harmful to patients without Gram-negative bacteremia. A trial in which nebacumab was given to 538 sepsis patients showed that, although the 28-day mortality of the 200 patients with Gram-negative bacteremia was reduced to 30%, overall mortality was not significantly different from the 49% mortality in the placebo group. This implies that mortality in the other 338 patients was 60%, an increase of 11%. Centocor withdrew the product from the European market and withdrew its FDA application. In the absence of a bedside diagnostic test to promptly identify patients with Gram-negative bacteremia, the product was of no value to sepsis patients as a group and of no value to payers.
+
+Recall that the necessary condition for testing to be beneficial for the payer as compared with not testing is B $_2-$ B $_1>0$ . Taking equation 13 and letting $\Delta P_g = 0$ , then
+
+$$N_{2}/N>P_{1}/(a_{2}+P_{g1}).\quad (18)$$
+
+In these circumstances testing is worthwhile from a payer perspective if the ratio of nonresponders to the total population exceeds the ratio of the price of the test to the price of the drug plus the cost of the adverse reactions experienced by the nonresponders. The maximum price at which a test is worthwhile to a payer is therefore
+
+$$P_{1}<[N_{2}/N](a_{2}+P_{g1}).\quad (19)$$
+
+In the nebacumab case, using the data above we obtain the following rough values: N 2 / N = 0.67, a 2 = [0.11 (20 × $10,000)] = $22,000 if we assume that each death costs 20 QALYs, $\Delta$ E = 20, and k $_p$ is $10,000. We ignore extra treatment costs, and P $_\mathrm{pl}$ = $4000. Thus the maximum value of a test is P $_\mathrm{t}$ $<$ 0.67 $\times$ $26,000, or P $_\mathrm{t}$ $<$ $17,420.
+
+However, this assumes that at a drug price of $4000, nebacumab was cost-effective and that the benefit to the payer, B 2 , as set out in equation 12, was positive. If we assume that the health gains per patient and the marginal cost per QALY threshold are as in the case of the adverse reactions, then b 1 = [0.19 (20 × $10,000)] = $38,000.
+
+From equation 11, the payer benefit per patient treated with no testing is
+
+$$\begin{aligned}
+B_{1} & =0.33 \times \$ 38,000-0.67 \times \$ 22,000-\$ 4000 \\~& =\$ 12,540-\$ 14,740-\$ 4000=-\$ 6200
+\end{aligned}$$
+
+Thus, at the price of $4000, the drug is not costeffective without testing. For the drug to be costeffective with testing and with treatment confined
+
+to patients with Gram-negative bacteremia, from equation 12,
+
+$$38,000-\$ 4000-N / N_{1} P_{t}>0$$
+$$P_{t}<0.33 \times \$ 34,000$$
+$$P_{t}<\$ 11,333$$
+
+Thus, using these assumptions, with a bedside diagnostic test costing up to $ 11,333 per test, nebacumab would have been cost-effective to payers at a price of $ 4000. Without the test the product was not cost-effective at any price.
+
+Trastuzumab. Trastuzumab is a new product for the treatment of breast cancer. It benefits only those patients with lesions that express increased quantities of the HER-2 protein, or approximately 25 % of patients [8] . Three diagnostic tests have been approved by the FDA [9] . Each costs less than $ 100 per test. There are no adverse reactions in the patient group that does not respond. Thus, equation 14 shows that, conditional on the decision to use the product, using the test is of benefit to payers if $P_E > P_r/[N_2/N]$ . This means that $P_E > 100/0.75$ , $P_E > $ 133.
+
+The price of trastuzumab in the United States is $1382 for a 440-mg injection, and patients need to take therapy throughout the period in which the disease is being treated. Thus, testing clearly makes economic sense compared to not testing. However, this analysis does not consider whether there is an overall payer benefit from using the product with testing, i.e., if B 2 > 0. We do not have the information on which to make that assessment, nor do we know if the product will provide positive returns to the company.
+
+We can also use equation 14 to ask what proportion of nonresponders would represent breakeven for the payer. If we assume that the total treatment cost per patient is $ 7000 (five sets of treatment) then P t /P F l = 0.015; i.e., it is worthwhile from the payer's perspective to test if more than 1.5 % of patients are nonresponders. Thus testing is highly worthwhile because the test is very inexpensive relative to the price of the drug.
+
+The nebacumab example clearly illustrates the potential gains to payers and manufacturers afforded by diagnostic tests that can distinguish potential responders from nonresponders so that treatment is targeted solely to the responders. In this case, an appropriate test was not available and the product had to be abandoned, at least for the purpose for which it was originally developed. If a
+
+---
+
+12
+
+Danzon and Towse
+
+test had been available to distinguish patients who would benefit from those who would be harmed, the price of the drug could have been higher, depending on the cost of the test, and would still have been cost-effective for payers. Of course we do not know if nebacumab would have earned a commercial return for its developers even in these circumstances. The trastuzumab example again indicates the value of a test to payers. It shows that with a low test cost, testing makes sense even when nonresponders are a small fraction of potential patients, suggesting that manufacturers could benefit from having such tests at the time of drug development. Such knowledge can be used either to design the drug to fit a broader spectrum of patients or to abandon products early if they can only benefit a small fraction of patients and hence will never cover their development costs. Of course the implication is that with testing becoming feasible and, in all probability, supplied competitively by third parties, drug producers will face smaller target populations. In some cases the resulting target population may be too small for the drug to be commercially viable unless payers increase prices to reflect the increase in expected benefits per patient treated. In the absence of such price adjustments, patients who would have benefited may forgo treatment unless R&D costs for targeted drugs can be significantly reduced. Even with such adjustments, the patient population may be too small to enable R&D costs to be recovered.
+
+## Implications for Public Policy
+
+In the case of gene therapy, we concluded that private sector investment in developing cures for monogenic diseases is likely to be socially suboptimal for several reasons:
+
+- 1. Long-lived therapies were likely to meet payer
+resistance to large one-off costs because of
+budget constraints or, in competitive systems,
+concerns that the savings would accrue to
+other insurers or that such therapies would at-
+tract high-cost patients.
+2. Current orphan drug legislation to encourage
+development of treatments for diseases with
+low patient numbers is therefore not neutral as
+between once-a-day and long-lived therapies.
+3. The novel nature of the treatment implies an
+atypically high risk of failure and a long delay
+to success, making these therapies unattractive
+to private investors.
+Evidence on trial activity appears to confirm our assessment. Public investment is already playing a major role in the development of gene therapies for monogenic diseases, and this may be the best policy to address the development risk. However, adjusting reimbursement norms and orphan drug laws so that they are neutral between long-lived and oncea-day therapies might be a better way of achieving the appropriate mix of private and public funds once public funding has established proof of concept. In addition to the benefits offered by orphan drug status and public funding of trials to establish proof of concept, one must also consider the issue of whether payer cost-effectiveness thresholds for monogenic diseases should be higher than for other diseases. It is therefore important that the full social benefit of such an adjustment is obtained by signaling to companies that this is the case.
+
+In the case of pharmacogenetics, testing will often be socially optimal, particularly if the proportion of nonresponders is high, if serious adverse reactions can arise, or if the test is inexpensive. The problem of patient fragmentation that results from genetic testing is most appropriately addressed by adjusting prices to reflect higher benefits of targeted treatment. However, two potential problems remain:
+
+- 1. Payers may be reluctant to adjust prices up-
+ward for targeted treatments to reflect the in-
+crease in expected benefits per patient treated
+that results from treating only those patients
+who are genetically appropriate candidates.
+Doing so requires companies and payers to use
+economic evaluation to identify the higher
+value associated with such targeting.
+2. If genetic testing reduces populations eligible
+for treatment but does not significantly reduce
+the costs of R&D through smaller trials re-
+quired to show efficacy, and if prices are not
+adjusted, then an increasing number of poten-
+tial treatments may be shelved for lack of com-
+mercial viability at normal payer thresholds.
+Even where prices are adjusted, patient populations may be too small to make commercial development viable. The problem with small numbers is analogous to that associated with gene therapy for monogenic diseases and may require similar solutions if society wishes to find cures for these diseases.
+
+## References
+
+1 Langreth R. Gene therapy is dealt a set back by the FDA. Wall Street Journal. October 11, 1999.
+
+2 Lehman Brothers. Gene Therapy: the Ultimate Ap-
+
+---
+
+Gene Therapy and Pharmacogenetics
+
+13
+
+plication of Genomics. New York: Lehman Brothers Inc., 1997.
+
+3 Martin PA, Thomas SM. The commercial development of gene therapy in Europe and the USA. Hum Gene Ther 1998;9:87-114.
+
+4 Bell J. The human genome. In: Marinker M, Peckham M. eds. Clinical Futures. London: BMJ Books, 1998.
+
+5 Lynch KRO, Neill GP, Liu Q, et al. Characterisation of the human cysteinyl leukotriene Cys LT1 receptor. Nature 1999;399:789-93.
+
+6 Kuivenhoven JA, Jukema JW, Zwinderman AH, et al. The role of a common variant of the cholesteryl ester transfer protein gene in the progression of coronary atherosclerosis. N Eng J Med 1998; 338:86-93.
+
+7 Centoxin: a lesson on patient selection? Pharm Pricing Rev 1999;4:34.
+
+8 Wisecarver JL. HER-2/neu Testing Comes of Age. Am J Clin Path 1999;111:229-301.
+
+9 McNeil C. How should HER-2 status be determined? J Nat Cancer Inst 1999;91:10.
+

processed/00235_W2059930869/metadata.json

@@ -0,0 +1,1735 @@
+{
+  "source_file": "/tmp/tmp58muunju/00235_W2059930869_The_Economics_of_Gene_Therapy_and_of_Pharmacogenetics.pdf",
+  "total_pages": 9,
+  "pages": [
+    {
+      "page_number": 1,
+      "elements": [
+        {
+          "bbox": [
+            494,
+            34,
+            606,
+            57
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Volume 5 • Number 1 • 2002\nVALUE IN HEALTH"
+        },
+        {
+          "bbox": [
+            73,
+            75,
+            580,
+            100
+          ],
+          "label": "sec_0",
+          "reading_order": 1,
+          "text": "The Economics of Gene Therapy and of Pharmacogenetics"
+        },
+        {
+          "bbox": [
+            71,
+            125,
+            547,
+            162
+          ],
+          "label": "para",
+          "reading_order": 2,
+          "text": "Patricia Danzon, PhD,1 and Adrian Towse, MA, MPhil2\n\n1Wharton School, University of Pennsylvania, Philadelphia, PA, USA; 2Office of Health Economics, London, England, UK"
+        },
+        {
+          "bbox": [
+            73,
+            182,
+            144,
+            196
+          ],
+          "label": "sec_1",
+          "reading_order": 3,
+          "text": "ABSTRACT"
+        },
+        {
+          "bbox": [
+            73,
+            210,
+            333,
+            432
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "This paper focuses on the economic issues arising from\ntwo uses of genomics: 1) the development of gene ther-\napy; 2) and use of pharmacogenetics to identify a pa-\ntient’s genotype before treatment to exclude those who\nwill not benefit or who may be harmed. We conclude\nthat private-sector investment aimed at developing gene\ntherapy for monogenic diseases is likely to be socially\nsuboptimal. Short-term administration regimens yielding\nlong-term therapeutic benefits are likely to meet payer re-\nsistance to large “one-off” costs because of budget con-\nstraints or, in competitive systems, concerns that the sav-\nings would accrue to future insurers or would attract\nhigh-cost patients. For some monogenic diseases, patient\nnumbers may be too small to support commercial de-\nvelopment without changes to orphan drug legislation\nor payer willingness to accept higher cost-effectiveness"
+        },
+        {
+          "bbox": [
+            347,
+            210,
+            607,
+            403
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "thresholds. In the case of pharmacogenetics, we conclude\nthat it can often be socially optimal to test before treat-\nment, particularly if the proportion of nonresponders is\nhigh, if there is a potential for serious adverse reactions,\nor if the test is inexpensive. Genetic testing that frag-\nments the patient population could reduce incentives for\nR&D unless prices are adjusted to reflect the higher ex-\npected benefits of targeted treatment per patient. Even in\nsituations where prices are adjusted, patient populations\nmay be too small to make commercial development via-\nble. This problem with small numbers is analogous to\nthat associated with gene therapy for monogenic diseases\nand may require similar remedies if society values devel-\noping treatments for these diseases."
+        },
+        {
+          "bbox": [
+            347,
+            403,
+            606,
+            432
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Keywords: gene therapy, monogenic disease, orphan drugs,\npharmaceutical pricing, pharmacogenetics."
+        },
+        {
+          "bbox": [
+            73,
+            455,
+            149,
+            471
+          ],
+          "label": "sec_1",
+          "reading_order": 7,
+          "text": "Introduction"
+        },
+        {
+          "bbox": [
+            73,
+            475,
+            333,
+            588
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "Genomics offers great promise for providing new\nand more effective therapies and diagnostic tests\nfor patients. Many new medicines will arise from\nthe use of pharmacogenomics in conventional drug\ndiscovery as knowledge of the human genome in-\ncreases understanding of disease. This paper fo-\ncuses on the economic issues arising from two rel-\natively novel uses of genomics:"
+        },
+        {
+          "bbox": [
+            73,
+            592,
+            333,
+            733
+          ],
+          "label": "list",
+          "reading_order": 9,
+          "text": "1. the development of gene therapy, where the\naim is to insert genes that will produce or regu-\nlate the expression of proteins that are related\nto the patient's disease. This approach is most\nimmediately relevant to monogenic diseases,\nmany of which are currently incurable. The\nhope is that gene therapy will provide long-\nterm therapeutic benefits;\n2. the use of pharmacogenetics to identify a pa-\ntient’s genotype before treatment, to identify"
+        },
+        {
+          "bbox": [
+            363,
+            453,
+            607,
+            554
+          ],
+          "label": "list",
+          "reading_order": 10,
+          "text": "those who will not benefit or who may be\nharmed. Here genomics is used to identify genetic\ntraits that may lead either to nonresponse or to\nadverse reactions to specific medicines for any in-\ndication, including the great majority of diseases\nthat are polygenic but in which genetic makeup\ncan affect response to specific treatments."
+        },
+        {
+          "bbox": [
+            347,
+            557,
+            607,
+            820
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "We examine whether existing reimbursement\nand regulatory regimes will encourage the socially\noptimal development of these two uses of genom-\nics and suggest possible policy changes. The high\ncosts and uncertainties associated with gene ther-\napy and its very novel modes of action, small pa-\ntient numbers, and long-lived therapeutic effects\nmay lead to suboptimal levels of commercial re-\nsearch. In the case of pharmacogenetics, testing\ncould be socially beneficial but nevertheless reduce\ninnovative firm's incentives to develop new drugs.\nThis is in part due to the low numbers of patients\ntreated—because drugs would be targeted at a\nsubset of the patient population—and to the costs\nof testing. This may be offset by reduced R&D\ncosts per drug developed or by unit price increases\nthat reflect the greater specificity of use and conse-\nquent greater expected health gain per patient\ntreated."
+        },
+        {
+          "bbox": [
+            73,
+            755,
+            332,
+            794
+          ],
+          "label": "foot",
+          "reading_order": 12,
+          "text": "Address correspondence to: Adrian Towse, Director, Of-\nfice of Health Economics, 12 Whitehall London SW1A\n2DY. E-mail: atowse@ohe.org"
+        },
+        {
+          "bbox": [
+            73,
+            796,
+            332,
+            822
+          ],
+          "label": "foot",
+          "reading_order": 13,
+          "text": "The Office of Health Economics is supported by the Asso-\nciation of the British Pharmaceutical Industry."
+        },
+        {
+          "bbox": [
+            71,
+            832,
+            230,
+            847
+          ],
+          "label": "foot",
+          "reading_order": 14,
+          "text": "c ISPOR 1098-3015/02/$15.00/5 5-13"
+        },
+        {
+          "bbox": [
+            343,
+            832,
+            354,
+            846
+          ],
+          "label": "foot",
+          "reading_order": 15,
+          "text": "5"
+        }
+      ]
+    },
+    {
+      "page_number": 2,
+      "elements": [
+        {
+          "bbox": [
+            73,
+            50,
+            84,
+            66
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "6"
+        },
+        {
+          "bbox": [
+            500,
+            49,
+            606,
+            66
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "Danzon and Towse"
+        },
+        {
+          "bbox": [
+            74,
+            78,
+            269,
+            94
+          ],
+          "label": "sec_1",
+          "reading_order": 2,
+          "text": "Formal Statement of the Problem"
+        },
+        {
+          "bbox": [
+            74,
+            99,
+            332,
+            142
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "Assume that a new therapy (g) is considered cost-\neffective by payers for their patients relative to an\nexisting treatment (0) if"
+        },
+        {
+          "bbox": [
+            155,
+            148,
+            330,
+            164
+          ],
+          "label": "equ",
+          "reading_order": 4,
+          "text": "$$\\Delta C_{g, 0} / \\Delta E_{g, 0}<k_{p}\\quad(1)$$"
+        },
+        {
+          "bbox": [
+            74,
+            168,
+            333,
+            335
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "where\n\n$\\Delta C_{g,0}=P_{g}-P_{0}+\\Delta C_{g,0}\\frac{d}{4}\\Delta C_{g,0}^{I}$\n\n$P_{g}$, $P_{0}=$ respectively, the prices of the gene therapy\nand existing treatment;\n$\\Delta C_{g,0}^{d}=$ change in other direct treatment costs;\n$\\Delta C_{g,0}^{I}=$ change in indirect costs;\n$\\Delta E_{g,0}=$ change in quality-adjusted life years\n(QALYs) or other outcome measures; and\n$k_p=$ threshold cost per QALY at which an inter-\nvention is considered cost-effective for the treat-\nment of patient group p."
+        },
+        {
+          "bbox": [
+            74,
+            345,
+            333,
+            457
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "We assume $\\Delta C_{g,0}$ and $\\Delta E_{g,0}$ are discounted at a\nsocially optimal rate. If $k_p$ is optimally chosen (i.e.,\nreflects the willingness to pay for additional in-\nvestment in treatment of patient group p), then\nequation 1 defines the condition for socially opti-\nmal reimbursement. We can use it to define the\nmaximum price ( $P_g^{max}$ ) at which the new therapy is\ncost effective:"
+        },
+        {
+          "bbox": [
+            114,
+            461,
+            330,
+            480
+          ],
+          "label": "equ",
+          "reading_order": 7,
+          "text": "$$P_{z}^{\\max }=k_{p} \\Delta E_{z, 0}+\\left(P_{0}+\\Delta C_{z, 0}^{d}+\\Delta C_{z, 0}^{1}\\right)\\quad(2)$$"
+        },
+        {
+          "bbox": [
+            98,
+            484,
+            330,
+            500
+          ],
+          "label": "equ",
+          "reading_order": 8,
+          "text": "$$ or  P_{g}^{\\max }=b \\quad(2)^{\\prime}$$"
+        },
+        {
+          "bbox": [
+            73,
+            503,
+            333,
+            547
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "where b = k p $\\Delta$ E $_{g,0}$ + (P $_{0}$ + $\\Delta$ C $_{g,0}$ d + $\\Delta$ C $_{g,0}$ l ) and represents\nthe social benefits (positive or negative) from treatment\ng compared with treatment 0."
+        },
+        {
+          "bbox": [
+            74,
+            547,
+            333,
+            616
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Assume that payers typically set the actual price\nat some fraction $\\alpha$ of the maximum price, where $\\alpha$\nreflects the share of the social gain that accrues to\nthe innovator firm and $(1-\\alpha)$ is the share captured\nby the payer, so that"
+        },
+        {
+          "bbox": [
+            148,
+            620,
+            330,
+            637
+          ],
+          "label": "equ",
+          "reading_order": 11,
+          "text": "$$P_{g}=\\alpha b=\\alpha P_{g}^{\\max }\\quad(3)$$"
+        },
+        {
+          "bbox": [
+            74,
+            641,
+            333,
+            753
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "Consider now the perspective of the firm plan-\nning to invest in developing a new therapy (g). We\nabstract from the fact that in practice $\\alpha$ and $P_g^{max}$\nmay not be known with certainty because of a lack\nof transparency in the price-setting process. Ignor-\ning this uncertainty and using a net present value in-\nvestment-valuation approach, the producer's break-\neven profit constraint can be written:"
+        },
+        {
+          "bbox": [
+            98,
+            756,
+            330,
+            775
+          ],
+          "label": "equ",
+          "reading_order": 13,
+          "text": "$$\\Pi^{T}=\\Sigma^{T}[(P_{2}-M)Q^{t}N^{t}(1+r)^{-t}]-F(r,L,p)\\quad(4)$$"
+        },
+        {
+          "bbox": [
+            74,
+            779,
+            333,
+            823
+          ],
+          "label": "para",
+          "reading_order": 14,
+          "text": "where\n$\\langle \\Pi \\rangle =$ discounted present value of net revenue over\nthe T years of the product's market life (“profit”);"
+        },
+        {
+          "bbox": [
+            346,
+            76,
+            607,
+            258
+          ],
+          "label": "para",
+          "reading_order": 15,
+          "text": "M = variable cost per treatment to the producer,\nassumed to be invariant over time;\nN = number of patients treated per year;\nQ = number of treatments per patient per year,\nsuch that NQ is the number of treatments sold per\nyear;\nF = discounted present value of the firm's R&D\ncost, with $F_{r}$, $F_{L}>0$, $F_{p}<0$;\nr = cost of capital;\nL = expected years from discovery to launch;\np = probability of success in showing safety and\nefficacy in clinical trials;\nother fixed costs are zero."
+        },
+        {
+          "bbox": [
+            347,
+            270,
+            606,
+            341
+          ],
+          "label": "para",
+          "reading_order": 16,
+          "text": "To assess the implications of the payer's cost-\neffectiveness requirement on the producer's break-\neven constraint, we substitute the price from equa-\ntion 3 into the firm's break-even constraint in\nequation 4 . Therefore,"
+        },
+        {
+          "bbox": [
+            367,
+            344,
+            605,
+            362
+          ],
+          "label": "equ",
+          "reading_order": 17,
+          "text": "$$\\Pi^{T}=\\Sigma^{T}[(\\alpha b-M)Q^{l}N^{l}(I+r)^{-t}]-F(r,L,p).\\quad(4)^{\\prime}$$"
+        },
+        {
+          "bbox": [
+            347,
+            366,
+            606,
+            437
+          ],
+          "label": "para",
+          "reading_order": 18,
+          "text": "Equation 4' provides a framework in which to\nconsider how the characteristics of gene therapy\nand pharmacogenetics may affect their commer-\ncial viability compared with other pharmaceutical\nR&D."
+        },
+        {
+          "bbox": [
+            347,
+            455,
+            432,
+            472
+          ],
+          "label": "sec_1",
+          "reading_order": 19,
+          "text": "Gene Therapy"
+        },
+        {
+          "bbox": [
+            347,
+            477,
+            605,
+            505
+          ],
+          "label": "para",
+          "reading_order": 20,
+          "text": "Let us consider effects on revenue and R&D costs.\n\nThe total operating net revenue depends on:"
+        },
+        {
+          "bbox": [
+            347,
+            511,
+            608,
+            787
+          ],
+          "label": "list",
+          "reading_order": 21,
+          "text": "1. the social value of treatment b (the cost savings\nand gain in quality of life) and the producer's\nshare, $\\alpha$ ;\n2. the number of patients treated per year N,\nwhich may be low because most monogenic\ndiseases (i.e., those involving only a single gene)\naffect a relatively small number of patients;\n3. Q, which may be small because of the long-\nterm benefits of gene therapies, implying that\neach patient may require treatment to be ad-\nministered only once or twice a year rather\nthan once or twice a day, the standard regimen\nfor many pharmacotherapeutic agents adminis-\ntered to chronically ill patients; and\n4. the variable cost of producing and delivering\nthe treatment, M, which is likely to be signifi-\ncantly higher for gene therapies than for other\ndrugs owing to both the novel delivery systems\nrequired and the small patient base over which\nto realize scale economies."
+        },
+        {
+          "bbox": [
+            347,
+            791,
+            606,
+            822
+          ],
+          "label": "para",
+          "reading_order": 22,
+          "text": "The extended duration of benefits and conse-\nquent reduction in frequency of administration"
+        }
+      ]
+    },
+    {
+      "page_number": 3,
+      "elements": [
+        {
+          "bbox": [
+            73,
+            49,
+            253,
+            67
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Gene Therapy and Pharmacogenetics"
+        },
+        {
+          "bbox": [
+            594,
+            49,
+            606,
+            65
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "7"
+        },
+        {
+          "bbox": [
+            73,
+            76,
+            333,
+            258
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "(low Q per patient for long-lasting therapies)\nwould be irrelevant if payers were willing to as-\nsume that the cost of new therapies would be fixed\nrelative to the full social benefits: P = α Pmax. In\nthis case, the price for gene therapy treatment\nwould increase proportionately with the duration\nof its effects. A low Q would be associated with a\nrelatively high benefit, b, per administered treat-\nment compared with a once-a-day alternative ther-\napy, thus preserving neutral incentives for efficient\ninvestment in R&D with respect to duration of\nbenefits. This ideal may, however, be undermined\nin practice for several reasons."
+        },
+        {
+          "bbox": [
+            73,
+            258,
+            333,
+            547
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "First, payers tend to scrutinize and bargain ag-\ngressively over products that are priced relatively\nhigh per dose or per patient treated. If so, the pro-\nducer’s share is a decreasing proportion of the\nmaximum price: $\\mathrm{d}\\alpha$ / $\\mathrm{dP}^{\\max} < 0$ . Second, turnover\nof patient populations could make competing\nhealth insurers reluctant to pay for long-lasting\ntherapies because the insurer that pays for the ini-\ntial treatment does not capture the full savings in\nfuture treatment costs if patients subsequently\nswitch to other insurers. Thus the risk of adverse\nselection could reinforce the incentive of insurers\nto avoid offering long-lived therapies that target\nhigh-cost patients, such as gene therapy. This\nshould be less of a problem in countries where pa-\ntients have a limited choice of health plans, such\nas Canada and the United Kingdom. However, in\nthese systems managers and doctors face annual\nbudget constraints that limit their ability to invest\nin treatments that have higher immediate costs but\nlonger-term benefits."
+        },
+        {
+          "bbox": [
+            73,
+            547,
+            333,
+            711
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "The private cost of R&D F(r, L, p), may be\natypically high for gene therapy compared with\nconventional therapy despite the relatively smaller\nrequirements in terms of trial sizes. The extremely\nnovel mode of action means that the probability\nof success, p, is very low and the expected dura-\ntion of the R&D process, L, is relatively long. Sev-\neral hundred clinical trials of gene therapy have\nbeen initiated but none has been successfully com-\npleted so far. The deaths in 1999 of two trial par-\nticipants [1] have made expectations even more\npessimistic."
+        },
+        {
+          "bbox": [
+            73,
+            711,
+            333,
+            822
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "Thus, values for each element of the expression\nfor $\\Pi^\\mathrm{T}$ are likely to be lower in the case of gene\ntherapies for monogenic diseases than correspond-\ning values for conventional therapies for major\ndiseases. The high risk and costs of R & D to com-\npanies are a reflection of real social costs arising\nfrom basic research into new technologies, the\nbenefits of which cannot be fully expropriated by"
+        },
+        {
+          "bbox": [
+            347,
+            77,
+            606,
+            135
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "the company. If reimbursement systems are biased\nagainst therapies providing long-term benefits, un-\nder-investment compared to the social optimum\nmay result."
+        },
+        {
+          "bbox": [
+            347,
+            135,
+            607,
+            422
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "An examination of clinical trials in gene ther-\napy tends to support our concerns that low patient\nnumbers and investor perception of payer resis-\ntance to long-lived therapies may be influencing\nthe allocation of R & D efforts. The majority of\ngene therapy trials are in cancer indications (on-\ncology), followed closely by trials directed at cures\nfor AIDS and HIV infection [2] . Many trials are\nfocused on variants of treatment that would re-\nquire repeat administration rather than providing\na one-off cure. Only one monogenic disorder, cys-\ntic fibrosis, is the subject of significant clinical de-\nvelopment, due in part to the early discovery of\nthe gene for this disease. Unlike all other areas of\ndrug development, most clinical trials in gene ther-\napy are undertaken with at least partial public\nfunding [3] . This suggests that the initial promise\nof gene therapy — that of delivering cures for mo-\nnogenic diseases — is unlikely to be realized without\nchanges in incentives or significant public invest-\nment."
+        },
+        {
+          "bbox": [
+            347,
+            422,
+            607,
+            822
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "The problem of insufficient commercial incen-\ntives for investment in drugs for small populations\nis not unique to gene therapy. Orphan drug legis-\nlation enacted in the United States in 1983 pro-\nvides special incentives, including 5-year market\nexclusivity for the orphan indication and special\ntax credits, for drugs to treat diseases that affect\nfewer than 200,000 US patients. The European\nUnion has now established an orphan drug re-\ngime. However, long-lived gene therapies are dis-\nadvantaged relative to conventional therapies by\nthe current legislation if payers resist price in-\ncreases in proportion to the duration of the ef-\nfects. If the 200,000 patient threshold is the num-\nber of patients expected to use an orphan drug per\nday per year, then, arguably, this number should\nbe adjusted for long-lasting gene therapies. For ex-\nample, if the gene therapy lasts 5 years, then in\nsteady state only one-fifth of the population with\nthe disease would be treated per year. Thus, to\nprovide neutral incentives the orphan drug thresh-\nold for long-lasting therapies should be $n\\times 200$ ,000, where $n$ is the average duration of bene-\nfits for one treatment with gene therapy relative to\nthat of conventional therapies. This will increase\nthe likelihood that a gene therapy receives orphan\nstatus and the associated benefit of market exclu-\nsivity, which should increase the price the com-\npany receives for the gene therapy, thereby in-"
+        }
+      ]
+    },
+    {
+      "page_number": 4,
+      "elements": [
+        {
+          "bbox": [
+            73,
+            50,
+            84,
+            66
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "8"
+        },
+        {
+          "bbox": [
+            499,
+            49,
+            606,
+            66
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "Danzon and Towse"
+        },
+        {
+          "bbox": [
+            73,
+            77,
+            332,
+            108
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "creasing $\\alpha$, the share of the social benefit that the\ncompany obtains."
+        },
+        {
+          "bbox": [
+            73,
+            108,
+            333,
+            328
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "Thus, the characteristics of gene therapy—long\nand uncertain R&D, a small patient base, and in-\nfrequent treatment—may lead to suboptimal com-\nmercial investment in these therapies. Reimburse-\nment systems introduce a bias against gene therapy\nif payers respond to budgetary or commercial pres-\nsures by focusing on short-term drug-budget costs\nwithout due weight to long-term health benefits\nand societal savings. Although society has signaled\na willingness to pay additional subsidies to en-\ncourage treatments for orphan diseases, current\nlegislation is not neutral between treatments ad-\nministered on a daily basis and those administered\nsporadically or once or twice in a lifetime. We\nconsider the public policy implications in the final\nsection of the paper."
+        },
+        {
+          "bbox": [
+            74,
+            345,
+            181,
+            362
+          ],
+          "label": "sec_1",
+          "reading_order": 4,
+          "text": "Pharmacogenetics"
+        },
+        {
+          "bbox": [
+            73,
+            366,
+            333,
+            493
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "Pharmacogenetic testing is designed to identify pa-\ntients’ genotypes so that drugs can be targeted to\nthe subgroup whose genetic makeup makes them\nmost likely to benefit. This raises the expected ef-\nfectiveness per patient treated and hence the cost-\neffectiveness of the drug by eliminating the cost of\ntreating patients whose genetic makeup makes\nthem unlikely to benefit (nonresponders) or likely\nto suffer harm (adverse responders). For example:"
+        },
+        {
+          "bbox": [
+            73,
+            495,
+            333,
+            692
+          ],
+          "label": "list",
+          "reading_order": 6,
+          "text": "1. New tests based on the ApoE gene may identify\npatients who are more likely to benefit from\ndrugs designed to slow the symptomatic degen-\neration associated with Alzheimer's disease [4] .\n2. Testing for the human CysLT 1 receptor for cys-\nteinyl leukotrienes may predict the effective-\nness in individual patients of the three new cys-\nteinyl leukotriene antagonists for asthma [5] .\n3. The presence of the B 1 variant of the CETP\ngene appears to predict the response of patients\nwith coronary atherosclerosis to statin treat-\nment. In a clinical trial statins slowed disease\nprogression in B 1 B 1 carriers but not in B 2 B 2 car-\nriers [6] ."
+        },
+        {
+          "bbox": [
+            73,
+            695,
+            333,
+            822
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "Payers would rationally adopt pharmacogenetic\ntesting before treatment if the savings from treat-\ning fewer patients and avoiding complications ex-\nceeded the costs of testing. For drug companies\npharmacogenetic testing means lower patient vol-\nume and hence lower revenues per drug, other\nthings being equal. This reduction in gross sales\nmay be exacerbated if payers subtract the costs of\nthe genetic screening from the price that they are"
+        },
+        {
+          "bbox": [
+            347,
+            77,
+            607,
+            162
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "willing to pay or reimburse for the drug. A key is-\nsue, therefore, is the extent to which we might ex-\npect drug prices to increase as a consequence of\ntesting, as might be the case if the innovator were\nable to get a share of any increased societal benefit\narising from testing."
+        },
+        {
+          "bbox": [
+            347,
+            162,
+            607,
+            230
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "To examine the potential impact of testing from\nsocietal, payer, and company perspectives we\nadapt the formal model set out earlier to introduce\nresponders and nonresponders, adverse reactions,\nand testing costs."
+        },
+        {
+          "bbox": [
+            347,
+            230,
+            607,
+            272
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Let $N_1$ be the number of patients who benefit\nfrom the drug. $N_2$ is the number who do not bene-\nfit but who cannot be identified without testing, so"
+        },
+        {
+          "bbox": [
+            441,
+            279,
+            510,
+            294
+          ],
+          "label": "equ",
+          "reading_order": 11,
+          "text": "$$N=\\left(N_{1}+N_{2}\\right)$$"
+        },
+        {
+          "bbox": [
+            347,
+            297,
+            606,
+            353
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "Let $b_1 = k_p \\Delta E_{1g,0} + (P_0 + \\Delta C_{1g,0}^d + \\Delta C_{1g,0}^l)$ i.e.,\nthe health gain plus cost savings per patient rela-\ntive to current treatment. Only the responders, $N_1$ ,\nobtain these benefits."
+        },
+        {
+          "bbox": [
+            347,
+            353,
+            606,
+            410
+          ],
+          "label": "para",
+          "reading_order": 13,
+          "text": "Let $a_2 = k_p \\Delta E_{2g,0} + (P_0 + \\Delta C_{2g,0}^d + \\Delta C_{2g,0}^l)$ de-\nnote the adverse health effect ( $\\Delta E_2 < 0$ ) plus the\nconsequential costs of the adverse reaction for\neach patient in group $N_2$ ."
+        },
+        {
+          "bbox": [
+            347,
+            410,
+            607,
+            479
+          ],
+          "label": "para",
+          "reading_order": 14,
+          "text": "We let $P_{g1}^{max}$ be the maximum price of the drug\nwithout testing, $P_{g2}^{max}$ be the maximum price of\nthe drug with testing, and $P_t$ be the price of the\ntest. Thus, adapting equation 2', the maximum\nprice of the drug without testing is"
+        },
+        {
+          "bbox": [
+            404,
+            483,
+            605,
+            501
+          ],
+          "label": "equ",
+          "reading_order": 15,
+          "text": "$$P_{g 1}^{\\max }=\\left(b_{1} ~N_{1}-a_{2} ~N_{2}\\right) / N . \\quad(5)$$"
+        },
+        {
+          "bbox": [
+            347,
+            504,
+            606,
+            534
+          ],
+          "label": "para",
+          "reading_order": 16,
+          "text": "With testing, the payer's maximum price for a\ndrug targeted solely at the responders is"
+        },
+        {
+          "bbox": [
+            417,
+            538,
+            605,
+            556
+          ],
+          "label": "equ",
+          "reading_order": 17,
+          "text": "$$P_{g2}^{\\max}=b_{1}-P_{1}N/N_{1}.\\quad(6)$$"
+        },
+        {
+          "bbox": [
+            347,
+            559,
+            606,
+            589
+          ],
+          "label": "para",
+          "reading_order": 18,
+          "text": "The change in the maximum price, which is the\nsocial value of testing, is therefore"
+        },
+        {
+          "bbox": [
+            373,
+            592,
+            605,
+            639
+          ],
+          "label": "equ",
+          "reading_order": 19,
+          "text": "$$\\begin{aligned}\n\\DeltaP_{g2}^{\\max}&=P_{g2}^{\\max}-P_{g1}^{\\max}\\\\\n&=b_{1}(1-N_{1}/N)-P_{t}N/N_{1}+a_{2}N_{2}/N\\\\\n&=(b_{1}+a_{2})(N_{2}/N)-P_{t}N/N_{1}.&(7)\n\\end{aligned}$$"
+        },
+        {
+          "bbox": [
+            361,
+            643,
+            602,
+            660
+          ],
+          "label": "para",
+          "reading_order": 20,
+          "text": "This social value of testing reflects three factors:"
+        },
+        {
+          "bbox": [
+            347,
+            669,
+            607,
+            782
+          ],
+          "label": "list",
+          "reading_order": 21,
+          "text": "1. the expected health benefit per patient treated,\nwhich is equal to b 1 (the benefit per responder)\ntimes the proportion of nonresponders N 2 /N;\n2. the averted costs of treating the adverse effects\nof the drug on nonresponders, a 2 , times the\nproportion of nonresponders N 2 /N;\n3. a cost of testing the whole patient population,\nP, N, amortized over the N 1 responders."
+        },
+        {
+          "bbox": [
+            347,
+            792,
+            606,
+            823
+          ],
+          "label": "para",
+          "reading_order": 22,
+          "text": "The first two components will be non-negative\nand testing will be beneficial if these effects exceed"
+        }
+      ]
+    },
+    {
+      "page_number": 5,
+      "elements": [
+        {
+          "bbox": [
+            74,
+            49,
+            253,
+            67
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Gene Therapy and Pharmacogenetics"
+        },
+        {
+          "bbox": [
+            594,
+            50,
+            606,
+            65
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "9"
+        },
+        {
+          "bbox": [
+            74,
+            76,
+            332,
+            109
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "the third element, the cost of testing. Formally,\ntesting offers benefit to society if $\\Delta P_{5}^{max}>0$, or"
+        },
+        {
+          "bbox": [
+            132,
+            113,
+            330,
+            128
+          ],
+          "label": "equ",
+          "reading_order": 3,
+          "text": "$$(b_{1}+a_{2})(N_{2}/N)>P_{t}N/N_{1}.\\quad(8)$$"
+        },
+        {
+          "bbox": [
+            74,
+            133,
+            332,
+            217
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "Thus testing is of benefit if the savings from\navoiding treatment and side effects for the N2 non-\nresponders exceeds the cost of testing all patients.\nIn the simplest case, if a2 is zero (no side effects)\nand we substitute for P $_{g1}$ max from equation 5 this\ncan be rewritten:"
+        },
+        {
+          "bbox": [
+            138,
+            223,
+            330,
+            239
+          ],
+          "label": "equ",
+          "reading_order": 5,
+          "text": "$$N_{2} /\\left(N_{1}+N_{2}\\right)>P_{1} / P_{g 1}^{max } . \\quad(9)$$"
+        },
+        {
+          "bbox": [
+            74,
+            243,
+            332,
+            299
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Testing is worthwhile from a societal perspective\nif the ratio of nonresponders to the total population\nexceeds the ratio of the cost of the test to the social\nvalue of the drug in the absence of testing."
+        },
+        {
+          "bbox": [
+            74,
+            299,
+            332,
+            397
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "Returning to the general case where there may\nbe adverse events ( $a_2 \\geq 0$ ), and assuming that the\nprice obtained by the company reflects the same\nshare $\\alpha$ of the social benefit of the drug with and\nwithout testing, then equation 7 shows that test-\ning increases the price of the drug obtained by the\ncompany:"
+        },
+        {
+          "bbox": [
+            99,
+            400,
+            330,
+            418
+          ],
+          "label": "equ",
+          "reading_order": 8,
+          "text": "$$\\alpha \\Delta P_{g}^{\\max }=\\alpha(b+a)\\left(N_{2} / N\\right)-\\alpha P_{1} ~N / N_{1} . \\quad(10)$$"
+        },
+        {
+          "bbox": [
+            74,
+            421,
+            333,
+            630
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "The company gets a share $\\alpha$ of the social gain\nof avoiding the drug costs and adverse events as-\nsociated with treating nonresponders but also\nbears the share of $\\alpha$ of the costs of testing. It is re-\nwarded and motivated to produce more specific\ndrugs. In this case only the share $\\alpha$ of the cost of\nthe test $P_t$ is borne by the company, with the payer\naccepting to pay the remainder $(1-\\alpha)$ . This fol-\nlows from the assumption that the cost of the test\nis deducted to arrive at social benefit before it is\ndivided between the producer and the payer. Note\nthat in the simplest case of zero adverse reactions\nand test costs (i.e., if $a_2=0$ , $P_t=0$ ) the price\nwould rise in proportion to the increase in propor-\ntion of patients expected to benefit:"
+        },
+        {
+          "bbox": [
+            103,
+            634,
+            299,
+            651
+          ],
+          "label": "equ",
+          "reading_order": 10,
+          "text": "$$\\alpha \\Delta P_{g}^{\\max }=\\alpha b_{1}\\left(~N_{2} / N\\right)  or  P_{g 2} / P_{g 1}=N / N_{1} .$$"
+        },
+        {
+          "bbox": [
+            74,
+            655,
+            333,
+            822
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "In practice, payers may be unlikely to permit\nthe price of the drug, adjusted for the cost of test-\ning and of side effects averted, to increase in pro-\nportion to its expected benefit per patient. We\nsuggested above that payers scrutinize most strin-\ngently those products that are priced relatively\nhigh; actual price is then a decreasing proportion\nof the maximum price: $d\\alpha$ / $dP^{max} < 0$ , in which\ncase $P_{5}$ would be lower than that suggested by\nequation 10 . The cost of the test, $P_{5}$ , the ratio of\nnonresponders to responders, $N_{2}/N_{1}$ , the severity\nof the adverse reactions, and the value of $\\alpha$ are all"
+        },
+        {
+          "bbox": [
+            347,
+            77,
+            607,
+            177
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "crucial to the ability of a company to obtain a\nprice premium for a more targeted product and to\nface, ex ante, neutral incentives for developing tar-\ngeted products versus more indiscriminate prod-\nucts with lower expected benefits per patient\ntreated. We consider in turn the payer and com-\npany perspectives."
+        },
+        {
+          "bbox": [
+            348,
+            186,
+            454,
+            203
+          ],
+          "label": "sec_1",
+          "reading_order": 13,
+          "text": "The Payer Perspective"
+        },
+        {
+          "bbox": [
+            347,
+            203,
+            607,
+            328
+          ],
+          "label": "para",
+          "reading_order": 14,
+          "text": "We can note the general conditions in which test-\ning is of benefit to the payer. Let B1 denote the po-\ntential payer benefit per period with no test, B2 de-\nnote the potential payer benefit with testing, Pg1 be\nthe price the payer is paying for the drug in the ab-\nsence of a test, and Pg1 + ΔP be the price with a\ntest. Note that we assume that payers focus on\nhealth effects and all costs. Adapting equations 5\nand 6 above yields"
+        },
+        {
+          "bbox": [
+            391,
+            333,
+            605,
+            349
+          ],
+          "label": "equ",
+          "reading_order": 15,
+          "text": "$$B_{1}=N_{1}b_{1}-N_{2}a_{2}-N P_{g1}\\quad(11)$$"
+        },
+        {
+          "bbox": [
+            388,
+            353,
+            605,
+            370
+          ],
+          "label": "equ",
+          "reading_order": 16,
+          "text": "$$B_{2}=N_{1}b_{1}-N_{1}(P_{\\xi1}+\\DeltaP)-N P_{t}.\\quad(12)$$"
+        },
+        {
+          "bbox": [
+            347,
+            374,
+            607,
+            486
+          ],
+          "label": "para",
+          "reading_order": 17,
+          "text": "In assessing the cost-effectiveness of testing to\nthe payer, potential savings from avoiding adverse\nevents and not paying for drugs to treat nonre-\nsponders have to be offset by the costs of testing\nand any higher price that is charged by the inno-\nvator. The protocol with testing offers greater\nbenefit to the payer than does indiscriminate treat-\nment of all patients if"
+        },
+        {
+          "bbox": [
+            426,
+            491,
+            492,
+            505
+          ],
+          "label": "equ",
+          "reading_order": 18,
+          "text": "$$B_{2}-B_{1}>0,  or $$"
+        },
+        {
+          "bbox": [
+            402,
+            512,
+            605,
+            528
+          ],
+          "label": "equ",
+          "reading_order": 19,
+          "text": "$$N_{2}(a_{2}+P_{gl})>N_{1}\\Delta P_{z}+NP_{1}.\\quad(13)$$"
+        },
+        {
+          "bbox": [
+            347,
+            531,
+            607,
+            590
+          ],
+          "label": "para",
+          "reading_order": 20,
+          "text": "If there are no adverse effects, i.e., $a_2 = 0$ , and if\nthe payer does not give the company a price in-\ncrease following the introduction of testing, so\nthat $\\Delta P_g = 0$ , then equation 13 reduces to"
+        },
+        {
+          "bbox": [
+            431,
+            594,
+            605,
+            610
+          ],
+          "label": "equ",
+          "reading_order": 21,
+          "text": "$$N_{2}/N>P_{t}/P_{g}.\\quad(14)$$"
+        },
+        {
+          "bbox": [
+            347,
+            614,
+            607,
+            823
+          ],
+          "label": "para",
+          "reading_order": 22,
+          "text": "Testing is worthwhile for the payer if the ratio\nof nonresponders to the total population exceeds\nthe ratio of the price of the test to the price of the\ndrug. More generally, the extent to which payers\nare prepared to award higher prices, i.e., $\\Delta P_g > 0$\nis crucial to creating consistent incentives for com-\npanies to develop new products using pharmaco-\ngenetics given society's cost-effectiveness thresh-\nold $k_p$ . Of course if companies are able to develop\nother products that treat nonresponders at prices\nthat result in a positive incremental cost-effective-\nness below the cost-effectiveness threshold $k_p$ ,\noverall payer expenditures will rise. This may give\nrise to rationing issues if payer budgets are con-\nstrained (so $k_p$ will rise) or to more aggressive bar-"
+        }
+      ]
+    },
+    {
+      "page_number": 6,
+      "elements": [
+        {
+          "bbox": [
+            74,
+            50,
+            91,
+            66
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "10"
+        },
+        {
+          "bbox": [
+            500,
+            49,
+            605,
+            66
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "Danzon and Towse"
+        },
+        {
+          "bbox": [
+            74,
+            77,
+            333,
+            122
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "gaining by payers seeking to reduce $\\alpha$ . We do not\nconsider this issue further. We now consider the\nimpact on the company."
+        },
+        {
+          "bbox": [
+            74,
+            144,
+            199,
+            160
+          ],
+          "label": "sec_1",
+          "reading_order": 3,
+          "text": "The Company Perspective"
+        },
+        {
+          "bbox": [
+            73,
+            160,
+            333,
+            383
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "Assume that the innovative firm faces two choices.\nIt could ignore the possibility of pharmacogenetic\ntesting and develop a traditional drug. This drug\nwould be targeted indiscriminately to all patients\nwith the disease in question, of which a propor-\ntion receives no benefit and may be harmed. The\ninnovative firm's alternative choice is to develop\nand sell a genetic test that would identify the N 1\npatients who will benefit and produce a drug tar-\ngeted to them. Assume that the test can be sold at\na price P 1 and produced at constant marginal cost\nC 1 . Adapting equation 4, assuming that Q = 1 for\nsimplicity, let $\\Pi_1$ , P 5 be the producer's profit and\nprice, respectively, with no testing, and $\\Pi_2$ , P 5 +\n$\\Delta$ P 5 , the profit and price with testing, and F 2 , the\nR&D cost with testing. Therefore,"
+        },
+        {
+          "bbox": [
+            103,
+            387,
+            330,
+            406
+          ],
+          "label": "equ",
+          "reading_order": 5,
+          "text": "$$\\Pi_{1}^{\\top}=\\Sigma^{\\top}[(P_{g1}-M)N^{t}(1+r)^{-4}]-F_{1}\\quad(15)$$"
+        },
+        {
+          "bbox": [
+            103,
+            409,
+            330,
+            441
+          ],
+          "label": "equ",
+          "reading_order": 6,
+          "text": "$$\\begin{split}\n\\Pi_{2}^{\\top}=&\\Sigma^{T}[(P_{gl}+\\DeltaP-M)N_{1}^{t}(1+r)^{-t}]-\\\\\n&F_{2}+N^{t}(P_{t}-C_{t})(1+r)^{-t}.&(16)\n\\end{split}$$"
+        },
+        {
+          "bbox": [
+            73,
+            445,
+            332,
+            476
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "The producer's profit is greater with the test than\nwithout only if $\\Pi_2^\\top - \\Pi_1^\\top > 0$ , or"
+        },
+        {
+          "bbox": [
+            129,
+            480,
+            330,
+            527
+          ],
+          "label": "equ",
+          "reading_order": 8,
+          "text": "$$\\begin{aligned}\n&\\Sigma^{T}(P_{g1}+\\DeltaP_{g}-M)N_{1}^{1}(1+r)^{-1}+\\\\\n&(F_{1}-F_{2})+N^{1}(P_{1}-C_{1})(1+r)^{-1}>\\\\\n&\\Sigma^{T}[(P_{g1}-M)N^{1}](1+r)^{-1}.&(17)\n\\end{aligned}$$"
+        },
+        {
+          "bbox": [
+            73,
+            530,
+            333,
+            822
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "Equation 17 shows that if the final drug price is\nunchanged, i.e., $\\Delta P_g = 0$ , the innovative firm has\nno incentive to invest in pharmacogenetic testing\nin development that will result in a narrower indi-\ncation unless there are savings in R & D costs ( $F_1$ –\n$F_2$ ) or profits to be made on the provision of the\ntests. Savings in R & D costs may be possible if, for\nexample, genetic testing permits phase III trials to\nbe targeted to fewer patients who are more likely\nto benefit. Thus, efficacy may be demonstrated\nwith much smaller trials. However, there may be\nadditional costs if either the link between the gene\nand the response or the reliability of the test has to\nbe validated. It is also possible that with genetic\ntesting the drug could be designed such that it is\neffective for a larger fraction of the patient popu-\nlation. In that case, the tendency for pharmacoge-\nnetics to reduce the average size of the target pop-\nulation per drug would be mitigated. Moreover, if\nthe proportion of patients N 2 /N who fail to bene-\nfit is expected to be relatively large, an untargeted"
+        },
+        {
+          "bbox": [
+            347,
+            77,
+            605,
+            108
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "drug might fail to qualify for reimbursement be-\ncause of poor cost-effectiveness."
+        },
+        {
+          "bbox": [
+            347,
+            108,
+            607,
+            395
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "More realistically, with free entry to the busi-\nness of developing genetic tests, pretreatment tests\nare likely to be developed irrespective of the ac-\ntions of the innovator when they can yield a net\nsaving to the payer. This occurs when the cost of\ntesting the entire patient population N P t is less\nthan the savings from avoiding treatment of non-\nresponders P g N 2 plus any savings from averting\nharm, a 2 N 2 . It is likely, therefore, that drug pro-\nducers will have incentives to do this testing them-\nselves as part of drug development rather than\nwait for others to do it after drugs reach the mar-\nket. In the latter case, the producer suffers the loss\nof sales but gets none of the potential benefits of\nsmaller trials or an improved drug design. Never-\ntheless, to the extent that pharmacogenetic testing\ntends to reduce the patient population per drug,\nsome drugs may not be worth developing once\ntesting becomes an option if the reduction in ex-\npected revenues due to population fragmentation\nexceeds the reduction in R & D costs."
+        },
+        {
+          "bbox": [
+            347,
+            395,
+            607,
+            492
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "The key issue, then, is what happens to drug\nprices. If pre- and post-testing prices reflect ex-\npected social benefits, the price of the drug will in-\ncrease in proportion to the expected benefits (net\nof testing costs) as specificity increases and the risk\nof zero benefit or positive harm declines as a con-\nsequence of genetic testing."
+        },
+        {
+          "bbox": [
+            347,
+            492,
+            607,
+            548
+          ],
+          "label": "para",
+          "reading_order": 13,
+          "text": "The willingness of payers to award higher\nprices for targeted benefits (i.e., maintaining a\nconstant value of $\\alpha$ ) will be essential to retaining\nneutrality in investment incentives."
+        },
+        {
+          "bbox": [
+            348,
+            573,
+            397,
+            590
+          ],
+          "label": "sec_1",
+          "reading_order": 14,
+          "text": "Examples"
+        },
+        {
+          "bbox": [
+            347,
+            590,
+            607,
+            673
+          ],
+          "label": "para",
+          "reading_order": 15,
+          "text": "Two drug launches illustrate the potential impact of\ntesting on manufacturers and payers. In one case, ne-\nbacumab (CentoxinTM, Centocor, Malvern, PA) no\ntest was available, whereas in the case of trastu-\nzumab (Herceptin, Genentech, San Francisco, CA),\ntests are available."
+        },
+        {
+          "bbox": [
+            347,
+            694,
+            607,
+            822
+          ],
+          "label": "para",
+          "reading_order": 16,
+          "text": "Nebacumab. Nebacumab was launched in 1991\nin most European countries [7] as a treatment for\nsepsis. However, it only worked in those cases where\nthe sepsis was due to Gram-negative bacteremia,\nor approximately one-third of all sepsis cases. With\na cost of $4000 per patient, doctors found them-\nselves under pressure to use the drug on all cases\nof sepsis, despite knowing that for every 1000 pa-\ntients treated, $2.67 million was being spent on"
+        }
+      ]
+    },
+    {
+      "page_number": 7,
+      "elements": [
+        {
+          "bbox": [
+            73,
+            49,
+            253,
+            67
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Gene Therapy and Pharmacogenetics"
+        },
+        {
+          "bbox": [
+            588,
+            50,
+            605,
+            66
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "11"
+        },
+        {
+          "bbox": [
+            73,
+            77,
+            333,
+            299
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "drugs for patients who could not benefit. It even-\ntually became clear that nebacumab was harmful\nto patients without Gram-negative bacteremia. A\ntrial in which nebacumab was given to 538 sepsis\npatients showed that, although the 28-day mortal-\nity of the 200 patients with Gram-negative bacter-\nemia was reduced to 30%, overall mortality was\nnot significantly different from the 49% mortality\nin the placebo group. This implies that mortality\nin the other 338 patients was 60%, an increase of\n11%. Centocor withdrew the product from the\nEuropean market and withdrew its FDA applica-\ntion. In the absence of a bedside diagnostic test to\npromptly identify patients with Gram-negative\nbacteremia, the product was of no value to sepsis\npatients as a group and of no value to payers."
+        },
+        {
+          "bbox": [
+            73,
+            299,
+            333,
+            356
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "Recall that the necessary condition for testing\nto be beneficial for the payer as compared with\nnot testing is B $_2-$ B $_1>0$ . Taking equation 13 and\nletting $\\Delta P_g = 0$ , then"
+        },
+        {
+          "bbox": [
+            141,
+            360,
+            330,
+            376
+          ],
+          "label": "equ",
+          "reading_order": 4,
+          "text": "$$N_{2}/N>P_{1}/(a_{2}+P_{g1}).\\quad(18)$$"
+        },
+        {
+          "bbox": [
+            73,
+            380,
+            333,
+            479
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "In these circumstances testing is worthwhile\nfrom a payer perspective if the ratio of nonre-\nsponders to the total population exceeds the ratio\nof the price of the test to the price of the drug plus\nthe cost of the adverse reactions experienced by\nthe nonresponders. The maximum price at which\na test is worthwhile to a payer is therefore"
+        },
+        {
+          "bbox": [
+            140,
+            484,
+            330,
+            500
+          ],
+          "label": "equ",
+          "reading_order": 6,
+          "text": "$$P_{1}<[N_{2}/N](a_{2}+P_{g1}).\\quad(19)$$"
+        },
+        {
+          "bbox": [
+            73,
+            504,
+            333,
+            601
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "In the nebacumab case, using the data above we\nobtain the following rough values: N 2 / N = 0.67,\na 2 = [0.11 (20 × $10,000)] = $22,000 if we as-\nsume that each death costs 20 QALYs, $\\Delta$ E = 20,\nand k $_p$ is $10,000. We ignore extra treatment\ncosts, and P $_\\mathrm{pl}$ = $4000. Thus the maximum value\nof a test is P $_\\mathrm{t}$ $<$ 0.67 $\\times$ $26,000, or P $_\\mathrm{t}$ $<$ $17,420."
+        },
+        {
+          "bbox": [
+            73,
+            601,
+            333,
+            711
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "However, this assumes that at a drug price of\n$4000, nebacumab was cost-effective and that the\nbenefit to the payer, B 2 , as set out in equation 12,\nwas positive. If we assume that the health gains\nper patient and the marginal cost per QALY\nthreshold are as in the case of the adverse reac-\ntions, then b 1 = [0.19 (20 × $10,000)] =\n$38,000."
+        },
+        {
+          "bbox": [
+            74,
+            711,
+            332,
+            740
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "From equation 11, the payer benefit per patient\ntreated with no testing is"
+        },
+        {
+          "bbox": [
+            101,
+            745,
+            303,
+            771
+          ],
+          "label": "equ",
+          "reading_order": 10,
+          "text": "$$\\begin{aligned}\nB_{1} & =0.33 \\times \\$ 38,000-0.67 \\times \\$ 22,000-\\$ 4000 \\\\~& =\\$ 12,540-\\$ 14,740-\\$ 4000=-\\$ 6200\n\\end{aligned}$$"
+        },
+        {
+          "bbox": [
+            73,
+            779,
+            333,
+            822
+          ],
+          "label": "para",
+          "reading_order": 11,
+          "text": "Thus, at the price of $4000, the drug is not cost-\neffective without testing. For the drug to be cost-\neffective with testing and with treatment confined"
+        },
+        {
+          "bbox": [
+            347,
+            77,
+            606,
+            109
+          ],
+          "label": "para",
+          "reading_order": 12,
+          "text": "to patients with Gram-negative bacteremia, from\nequation 12,"
+        },
+        {
+          "bbox": [
+            399,
+            114,
+            555,
+            188
+          ],
+          "label": "equ",
+          "reading_order": 13,
+          "text": "$$38,000-\\$ 4000-N / N_{1} P_{t}>0$$\n$$P_{t}<0.33 \\times \\$ 34,000$$\n$$P_{t}<\\$ 11,333$$"
+        },
+        {
+          "bbox": [
+            347,
+            194,
+            606,
+            265
+          ],
+          "label": "para",
+          "reading_order": 14,
+          "text": "Thus, using these assumptions, with a bedside\ndiagnostic test costing up to $ 11,333 per test, ne-\nbacumab would have been cost-effective to payers\nat a price of $ 4000. Without the test the product\nwas not cost-effective at any price."
+        },
+        {
+          "bbox": [
+            347,
+            270,
+            607,
+            437
+          ],
+          "label": "para",
+          "reading_order": 15,
+          "text": "Trastuzumab. Trastuzumab is a new product for\nthe treatment of breast cancer. It benefits only\nthose patients with lesions that express increased\nquantities of the HER-2 protein, or approximately\n25 % of patients [8] . Three diagnostic tests have\nbeen approved by the FDA [9] . Each costs less than\n$ 100 per test. There are no adverse reactions in\nthe patient group that does not respond. Thus, equa-\ntion 14 shows that, conditional on the decision to\nuse the product, using the test is of benefit to pay-\ners if $P_E > P_r/[N_2/N]$ . This means that $P_E > 100/0.75$ ,\n$P_E > $ 133."
+        },
+        {
+          "bbox": [
+            347,
+            437,
+            607,
+            588
+          ],
+          "label": "para",
+          "reading_order": 16,
+          "text": "The price of trastuzumab in the United States is\n$1382 for a 440-mg injection, and patients need\nto take therapy throughout the period in which\nthe disease is being treated. Thus, testing clearly\nmakes economic sense compared to not testing.\nHowever, this analysis does not consider whether\nthere is an overall payer benefit from using the\nproduct with testing, i.e., if B 2 > 0. We do not\nhave the information on which to make that as-\nsessment, nor do we know if the product will pro-\nvide positive returns to the company."
+        },
+        {
+          "bbox": [
+            347,
+            588,
+            607,
+            711
+          ],
+          "label": "para",
+          "reading_order": 17,
+          "text": "We can also use equation 14 to ask what pro-\nportion of nonresponders would represent breakeven\nfor the payer. If we assume that the total treat-\nment cost per patient is $ 7000 (five sets of treat-\nment) then P t /P F l = 0.015; i.e., it is worthwhile\nfrom the payer's perspective to test if more than\n1.5 % of patients are nonresponders. Thus testing\nis highly worthwhile because the test is very inex-\npensive relative to the price of the drug."
+        },
+        {
+          "bbox": [
+            347,
+            711,
+            607,
+            822
+          ],
+          "label": "para",
+          "reading_order": 18,
+          "text": "The nebacumab example clearly illustrates the\npotential gains to payers and manufacturers af-\nforded by diagnostic tests that can distinguish po-\ntential responders from nonresponders so that treat-\nment is targeted solely to the responders. In this\ncase, an appropriate test was not available and the\nproduct had to be abandoned, at least for the pur-\npose for which it was originally developed. If a"
+        }
+      ]
+    },
+    {
+      "page_number": 8,
+      "elements": [
+        {
+          "bbox": [
+            74,
+            50,
+            91,
+            66
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "12"
+        },
+        {
+          "bbox": [
+            499,
+            49,
+            606,
+            66
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "Danzon and Towse"
+        },
+        {
+          "bbox": [
+            73,
+            76,
+            333,
+            505
+          ],
+          "label": "half_para",
+          "reading_order": 2,
+          "text": "test had been available to distinguish patients who\nwould benefit from those who would be harmed,\nthe price of the drug could have been higher, de-\npending on the cost of the test, and would still\nhave been cost-effective for payers. Of course we\ndo not know if nebacumab would have earned a\ncommercial return for its developers even in these\ncircumstances. The trastuzumab example again in-\ndicates the value of a test to payers. It shows that\nwith a low test cost, testing makes sense even\nwhen nonresponders are a small fraction of poten-\ntial patients, suggesting that manufacturers could\nbenefit from having such tests at the time of drug\ndevelopment. Such knowledge can be used either\nto design the drug to fit a broader spectrum of pa-\ntients or to abandon products early if they can\nonly benefit a small fraction of patients and hence\nwill never cover their development costs. Of course\nthe implication is that with testing becoming feasi-\nble and, in all probability, supplied competitively\nby third parties, drug producers will face smaller\ntarget populations. In some cases the resulting tar-\nget population may be too small for the drug to be\ncommercially viable unless payers increase prices\nto reflect the increase in expected benefits per pa-\ntient treated. In the absence of such price adjust-\nments, patients who would have benefited may\nforgo treatment unless R&D costs for targeted drugs\ncan be significantly reduced. Even with such ad-\njustments, the patient population may be too small\nto enable R&D costs to be recovered."
+        },
+        {
+          "bbox": [
+            74,
+            539,
+            240,
+            557
+          ],
+          "label": "sec_1",
+          "reading_order": 3,
+          "text": "Implications for Public Policy"
+        },
+        {
+          "bbox": [
+            73,
+            560,
+            333,
+            617
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "In the case of gene therapy, we concluded that pri-\nvate sector investment in developing cures for mo-\nnogenic diseases is likely to be socially suboptimal\nfor several reasons:"
+        },
+        {
+          "bbox": [
+            73,
+            622,
+            333,
+            817
+          ],
+          "label": "list",
+          "reading_order": 5,
+          "text": "1. Long-lived therapies were likely to meet payer\nresistance to large one-off costs because of\nbudget constraints or, in competitive systems,\nconcerns that the savings would accrue to\nother insurers or that such therapies would at-\ntract high-cost patients.\n2. Current orphan drug legislation to encourage\ndevelopment of treatments for diseases with\nlow patient numbers is therefore not neutral as\nbetween once-a-day and long-lived therapies.\n3. The novel nature of the treatment implies an\natypically high risk of failure and a long delay\nto success, making these therapies unattractive\nto private investors."
+        },
+        {
+          "bbox": [
+            347,
+            77,
+            607,
+            327
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Evidence on trial activity appears to confirm our\nassessment. Public investment is already playing a\nmajor role in the development of gene therapies for\nmonogenic diseases, and this may be the best policy\nto address the development risk. However, adjust-\ning reimbursement norms and orphan drug laws so\nthat they are neutral between long-lived and once-\na-day therapies might be a better way of achieving\nthe appropriate mix of private and public funds\nonce public funding has established proof of con-\ncept. In addition to the benefits offered by orphan\ndrug status and public funding of trials to establish\nproof of concept, one must also consider the issue\nof whether payer cost-effectiveness thresholds for\nmonogenic diseases should be higher than for other\ndiseases. It is therefore important that the full social\nbenefit of such an adjustment is obtained by signal-\ning to companies that this is the case."
+        },
+        {
+          "bbox": [
+            347,
+            327,
+            607,
+            437
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "In the case of pharmacogenetics, testing will often\nbe socially optimal, particularly if the proportion of\nnonresponders is high, if serious adverse reactions\ncan arise, or if the test is inexpensive. The problem\nof patient fragmentation that results from genetic\ntesting is most appropriately addressed by adjusting\nprices to reflect higher benefits of targeted treat-\nment. However, two potential problems remain:"
+        },
+        {
+          "bbox": [
+            348,
+            441,
+            607,
+            651
+          ],
+          "label": "list",
+          "reading_order": 8,
+          "text": "1. Payers may be reluctant to adjust prices up-\nward for targeted treatments to reflect the in-\ncrease in expected benefits per patient treated\nthat results from treating only those patients\nwho are genetically appropriate candidates.\nDoing so requires companies and payers to use\neconomic evaluation to identify the higher\nvalue associated with such targeting.\n2. If genetic testing reduces populations eligible\nfor treatment but does not significantly reduce\nthe costs of R&D through smaller trials re-\nquired to show efficacy, and if prices are not\nadjusted, then an increasing number of poten-\ntial treatments may be shelved for lack of com-\nmercial viability at normal payer thresholds."
+        },
+        {
+          "bbox": [
+            347,
+            654,
+            607,
+            740
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "Even where prices are adjusted, patient popula-\ntions may be too small to make commercial develop-\nment viable. The problem with small numbers is\nanalogous to that associated with gene therapy for\nmonogenic diseases and may require similar solu-\ntions if society wishes to find cures for these diseases."
+        },
+        {
+          "bbox": [
+            348,
+            759,
+            408,
+            773
+          ],
+          "label": "sec_1",
+          "reading_order": 10,
+          "text": "References"
+        },
+        {
+          "bbox": [
+            354,
+            776,
+            606,
+            802
+          ],
+          "label": "reference",
+          "reading_order": 11,
+          "text": "1 Langreth R. Gene therapy is dealt a set back by the\nFDA. Wall Street Journal. October 11, 1999."
+        },
+        {
+          "bbox": [
+            354,
+            802,
+            606,
+            818
+          ],
+          "label": "reference",
+          "reading_order": 12,
+          "text": "2 Lehman Brothers. Gene Therapy: the Ultimate Ap-"
+        }
+      ]
+    },
+    {
+      "page_number": 9,
+      "elements": [
+        {
+          "bbox": [
+            74,
+            49,
+            253,
+            67
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "Gene Therapy and Pharmacogenetics"
+        },
+        {
+          "bbox": [
+            588,
+            50,
+            606,
+            66
+          ],
+          "label": "header",
+          "reading_order": 1,
+          "text": "13"
+        },
+        {
+          "bbox": [
+            90,
+            78,
+            330,
+            104
+          ],
+          "label": "reference",
+          "reading_order": 2,
+          "text": "plication of Genomics. New York: Lehman Broth-\ners Inc., 1997."
+        },
+        {
+          "bbox": [
+            80,
+            104,
+            330,
+            143
+          ],
+          "label": "reference",
+          "reading_order": 3,
+          "text": "3 Martin PA, Thomas SM. The commercial develop-\nment of gene therapy in Europe and the USA.\nHum Gene Ther 1998;9:87-114."
+        },
+        {
+          "bbox": [
+            80,
+            143,
+            330,
+            178
+          ],
+          "label": "reference",
+          "reading_order": 4,
+          "text": "4 Bell J. The human genome. In: Marinker M, Peck-\nham M. eds. Clinical Futures. London: BMJ Books,\n1998."
+        },
+        {
+          "bbox": [
+            80,
+            179,
+            330,
+            219
+          ],
+          "label": "reference",
+          "reading_order": 5,
+          "text": "5 Lynch KRO, Neill GP, Liu Q, et al. Characterisa-\ntion of the human cysteinyl leukotriene Cys LT1\nreceptor. Nature 1999;399:789-93."
+        },
+        {
+          "bbox": [
+            355,
+            78,
+            606,
+            142
+          ],
+          "label": "reference",
+          "reading_order": 6,
+          "text": "6 Kuivenhoven JA, Jukema JW, Zwinderman AH,\net al. The role of a common variant of the choles-\nteryl ester transfer protein gene in the progression\nof coronary atherosclerosis. N Eng J Med 1998;\n338:86-93."
+        },
+        {
+          "bbox": [
+            355,
+            142,
+            606,
+            168
+          ],
+          "label": "reference",
+          "reading_order": 7,
+          "text": "7 Centoxin: a lesson on patient selection? Pharm\nPricing Rev 1999;4:34."
+        },
+        {
+          "bbox": [
+            355,
+            168,
+            605,
+            193
+          ],
+          "label": "reference",
+          "reading_order": 8,
+          "text": "8 Wisecarver JL. HER-2/neu Testing Comes of Age.\nAm J Clin Path 1999;111:229-301."
+        },
+        {
+          "bbox": [
+            355,
+            193,
+            606,
+            219
+          ],
+          "label": "reference",
+          "reading_order": 9,
+          "text": "9 McNeil C. How should HER-2 status be deter-\nmined? J Nat Cancer Inst 1999;91:10."
+        }
+      ]
+    }
+  ],
+  "metadata": {}
+}

processed/00238_W2025122215/document.md

@@ -0,0 +1,525 @@
+![Figure](figures/figure_000.png)
+
+Technological Forecasting & Social Change 94 (2015) 187-201
+
+Contents lists available at ScienceDirect
+
+Technological Forecasting & Social Change
+
+![Figure](figures/figure_004.png)
+
+# Knowledge combination modeling: The measurement of knowledge similarity between different technological domains
+
+![Figure](figures/figure_006.png)
+
+Hiroko Nakamura$^{a,*}$,$^{1}$, Shinji Suzuki$^{a,}$, Ichiro Sakata$^{b}$, Yuya Kajikawa$^{c}$
+
+a Center for Aviation Innovation Research (CAIR), The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8654, Japan
+
+b Innovation Policy Research Center, The University of Tokyo, Tokyo, Japan
+
+c Graduate School of Innovation Management, Tokyo Institute of Technology, Tokyo, Japan
+
+## A RTICLE INFO
+
+### Article history:
+
+Received 17 January 2014 Received in revised form 7 April 2014 Accepted 14 September 2014 Available online 8 October 2014
+
+### Keywords:
+
+Patent analysis Citation analysis Bibliometrics Breadth search Knowledge recombination
+
+## A B S T R A C T
+
+This paper proposes the DB-Combination model that considers three different knowledge combinations in depth (D) and breadth (B) based on similarities of two technological knowledge domains. We also investigate three methodologies A1, A2 and A3 to highlight the three knowledge combinations. To identify technological knowledge domains, citation analysis on patent information was used for A1 and A2 and pre-existing patent classification analysis was used for A3. And to measure the similarity between identified technological knowledge domains, text similarity measurements, existing intra-industrial citation tracing and IPC share similarity comparison were used for A1, A2 and A3 respectively. The usability of the model and methodologies were demonstrated through a case study on technological knowledge of the automobile industry and the aircraft industry. While these methodologies still need to be improved, it was demonstrated that the three measurements can highlight candidates of the three knowledge combination proposed in DB-Combination model. This research contributes to accelerate breadth knowledge recombination in a complex technology industry.
+
+© 2014 Elsevier Inc. All rights reserved.
+
+## 1. Introduction
+
+It is said that innovation comes from a recombination of knowledge ( Dosi , 1982 ; Nelson and Winter , 1977 ; Schumpeter , 1934 ) and that combining one's own knowledge with that of different industry and different technology domains has the possibility of bringing new knowledge creation ( Schoenmakers and Duijsters , 2010 ; Gassmann and Zeschky , 2008 ; Dosi , 1982 ) . On the other hand, the technological domain of industries with a complex system has a wide range of technological subdomains ( Fleming and Sorenson , 2001 ; Eriksson , 2000 ) . It is not easy for engineers of an industry to search for a new candidate of knowledge combination in knowledge of another industry with a complex system; firstly they should need to identify the technological sub-domains of the other, and then,
+
+to select sub-domains to look further. The information of subdomains also must be updated frequently ( Herrero et al. , 2010 ) . The cost of the collection and the integration of different knowledge ( Nakamura et al. , 2011a ; Kajikawa et al. , 2006 ; Tijssen , 1992 ) and the uncertainty of success ( Schilling and Green , 2011 ; Moorthy and Polley , 2010 ) are problems that limit practitioners to explore new opportunities.
+
+To support practitioners to have such exploration, namely breadth activities, and to bring innovation, the authors focus on one side of breadth activities; namely, searching the technological knowledge of other industries and integrating it to their own knowledge. We propose a knowledge combination model and discuss methodologies that effectively identify a technological sub-domain that can be combined between the two industries.
+
+We focus on patents as the information source of technological knowledge of focused industries because firstly patent is considered to be the best available indicator for R&D invention related to technology and outcomes of innovation activities (OECD, 1994). And secondly we believed that there is a
+
+* Corresponding author. E-mail address: techhn@mail.ecc.u-tokyo.ac.jp (H. Nakamura).
+
+1 Tel.: +81 3 5841 0744
+
+http://dx.doi.org/10.1016/j.techfore.2014.09.009 0040-1625/© 2014 Elsevier Inc. All rights reserved.
+
+---
+
+188
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+potential need in practitioners for scientometrics that can support breadth search with patents. Practitioners contacted in this study explained that, although a patent is essential to protect practitioners' intellectual products and investigate competitors' strategy, difficulty in searching information from patent data hinders frequent use of patents as a source of knowledge. And computer-based bibliometrics approaches are taken because it can process vast amount of data and it is expected to ease breadth search ( Alavi and Leidner , 2001 ; Smallheiser and Swanson , 1998 ; Smalheiser , 2012 ; Herrero et al. , 2010 ; Cantu and Ceballos , 2010 ; Fleming and Sorenson , 2001 ; Kostoff , 2008 ) .
+
+This paper is organized as follows: The next section reviews previous literature. The third section proposes a knowledge combination model and three measurements. The fourth section conducts a case study on technological knowledge of automobile and aeronautic industry and shows identified technological domains and highlighted pairs of technological domains. The section also discusses the results with automobile and aeronautic experts. The Discussion section compares the three methodologies. The final section concludes this paper with the findings.
+
+## 2. Literature
+
+Patent is often used in the innovation literatures. For example, patent is used as the indicators of technological knowledge of focused industries in the literatures mapping technological portfolios of a company or an industry. Leydesdorff et al. (2012) discussed methodologies to map the technological portfolios and the relation between the identified technology using International Patent Classification (IPC) and patent citation analysis approach. Schoen et al. (2012) discussed methodologies to map technological domains of major R&D companies and the dynamics of knowledge over firms and regions. Patent is also used as the indicators of technological knowledge of the focused company in the literatures investigating the impact of technological knowledge breadth and depth to company performance (ex. Schoenmakers and Duijsters, 2010; Leiponen and Helfat, 2010; Moorthy and Polley, 2010; Fleming and Sorenson, 2001; Nesta and Saviotti, 2005) . Patent is also used as the indicator of the efficiency of innovation policies of the focused countries, domains, or sectors in the literatures analyzing economic development, technological change, speed and change of industrial structures (ex. Criscuolo, 2006; Soete and Wyatt, 1983) . Patent is used in these literatures because it is objective and their standard changes slowly ( Griliches, 1990) .
+
+Our approach seems to be similar to the first group of literatures, such as those of Leydesdorff et al. (2012) and Schoen et al. (2012) . However we use patent not as the indicator to obtain the overview of knowledge existing in a domain but as a source of technological knowledge to produce new knowledge combination. The aim of this paper is neither to describe technological trend and portfolio by utilizing patent as indicator nor to evaluate performance of firms and industries by utilizing patent as evidence. The aim of this paper is to explore opportunities for business development by utilizing patents as information resource.
+
+Citation analysis that we use in this paper has been developed because of the need for scientific information retrieval and has established itself as one of the most effective
+
+approaches in identifying technological domains in academic documents and in creating the overview because a citation can reflect the self-organizing dynamics of scholars' communication ( Leydesdorff , 2008 ; Kajikawa et al. , 2008 , 2006 ; Nicolaisen , 2007 ; Cronin , 2001 ) . It is widely known that dissemination and exchange of communication are important for the development of science ( Everett and Pecotich , 1993 ) . Using citation information to evaluate scientific activities, nonetheless, brings various concerns such as risk of skewed analysis results due to the existence of citations for critics, self-citations, an Englishbias, and availability of literatures ( Martin , 2012 ; Everett and Pecotich , 1993 ; Garfield , 1979 ) . Despite these concerns, citation analysis is utilized among researchers and decision-makers related to science and innovation management as quantifiable and objective approaches that can compensate and validate the experts' judgments ( Nerur et al. , 2008 ) , and can be used in administration fields (see Garfield , 1979 for example), and its further development and use are expected by practitioners.
+
+Because patent citations are made with the consideration of legal and economic matters, there is still criticism about applying citation analysis to patents because citation behavior is different between academic journals and patents and also citation analysis inclines too much toward documents with links ( Leydesdorff , 2008 ; Michel and Bettels , 2001 ; Meyer , 2000 ; Kostoff , 1998 ) . However, there are many attempts to apply such methods to patent analysis because the pre-existing classification system of patents is based on technological and functional characteristics and is often difficult to understand the overview of the system the patents are related from the classification ( Griliches , 1990 ) . Furthermore, when technology rapidly progresses and changes, it is not easy for existing classification schema to capture such a change. Narin ( 1994 ) discussed the high similarity between the analysis of scientific papers and analysis of patents because both are suited especially to the analysis of national productivity, inventor productivity, referencing cycles, and citation impact.
+
+Considering the criticism and support regarding patent citation analysis, we conducted two different approaches to identify technological domains from patent data; namely, the citation analysis approach and another approach with an international classification standard of patents as we can see in the next section.
+
+## 3. Methodology
+
+### 3.1. Depth and breadth knowledge combination model
+
+The methodology proposed in this paper aims to support the following knowledge recombination process of practitioners, that is, identifying the technological sub-domains of other industries, selecting sub-domains to combine and researching for bringing new knowledge. And to do that, firstly we propose a knowledge combination model between two technological domains, named the DB-Combination model ( Fig. 1 ). We assume that, limited to the technological knowledge and the combination between different industries, knowledge recombination in depth and breadth discussed in the previous innovation literatures such as those ofAlavi and Leidner (2001), Gassmann and Zeschky (2008), Schoenmakers and Duijsters (2010), Dosi (1982), can be modeled depending on the "similarity" between the domains as shown in Fig. 1 . Fig. 1
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+189
+
+![Figure](figures/figure_002.png)
+
+Fig. 1. DB-Combination Model for knowledge integration between two technological domains in depth and breadth.
+
+considers the similarity of any pairs of sub-domains of each technological domain; industry A and of industry B, in the horizon.
+
+Knowledge recombination in deep (D) search is categorized to DB-D in Fig. 1 that shows combining knowledge between a very similar pair of sub-domains. A deeper understanding and improvement of the technological domains will be brought by such combining. It is likely that such knowledge combining has been already challenged so that only incremental innovation is expected to occur ( Katila and Ahuja , 2002 ) . On the contrary, knowledge recombination in breadth are categorized to DB-T and DB-C because such recombination can be considered to be brought by either knowledge transfer (T) of unique technologies from one technology domain to another (pair DB-T) or knowledge combination (C) between different technologies sub-domains with weak similarity (DB-C). Unlike knowledge combining in depth, knowledge combining in breadth is expected to broaden the scope of research and to provide a chance of new knowledge combination or transfer/replacement ( Katila and Ahuja , 2002 ) .
+
+Our proposal is similar to literature-based discovery (LBD) process studied in bibliometrics mining academic literature. According to Gordon et al. (2002) , LBD can be either an open or closed discovery process. An open discovery process is characterized by the generation of a hypothesis, starting with a research question or scientific problem, and explores to extract plausible candidates relating with the starting concept (like DB-C in our model). On the other hand, a closed discovery process is the testing of a hypothesis, starting from two entities at both ends where common intermediate terms are extracted and evaluated to verify important connections between them (like DB-D in our model).
+
+The similarity of sub-domains should have various factors of technological knowledge including problems that the technological knowledge aims to solve, the processes that the technological knowledge needs to solve a problem, the environment where the technological knowledge is used, and the components and materials of the technology itself. In the following sub-sections, we propose methodologies to measure the similarity of sub-domains of different industries, using patents. Finding sub-domains in different industries that are weakly similar to a sub-domain of an industry is more difficult for practitioners than finding very similar sub-domains or very unique sub-domains so that a similarity measurement of a DB-C
+
+combination, for example, is expected to be especially effective in helping practitioners bring forth advancement in technology.
+
+In order to investigate the effectiveness of the methodologies, we conduct a case study to analyze technological knowledge combination between the automobile industry and the aircraft industry, and interview experts on the results in order to evaluate the effectiveness of the model and methodologies. Automobile industry companies are facing severe global competition. The knowledge on how to reduce the cost of production is critical and the relocation and offshoring of parts production and assembly are always investigated through the departments of business management ( Herrero et al. , 2010 ) . Research, engineering, and product planning departments are also under pressure to increase productivity and creativity. One research center of a Japanese automobile company showed their interest in our research concept and asked for a system that can detect technological areas or trends that they themselves cannot recognize the importance as yet. On the other hand, the aircraft industry is a high-tech industry and regarded as a source of technology renewal ( Eriksson , 2000 ) . There are many examples of the spillover of technology from the aircraft industry to the automobile industry, such as composite material production and head-up display. Recently, the aircraft industry has faced the need of radical changes to achieve environmental and business sustainability and to seek technological solutions for safe and economical carbon-neutral growth ( Nakamura et al. , 2012 , 2011b ) . “ More and All Electric ” system concept in the aviation industry, that is, replacing more or all the heavy hydraulic systems of aircrafts by efficient electric systems, is one of the technological directions investigated in order to reduce fuel consumption ( Nakamura et al. , 2011b ) . The automobile industry can be considered as advanced in this area because the industry has already commercialized all-electric vehicles. Knowledge transfer or combination is now expected not only in the direction of the aircraft industry to the automobile industry, but also vice versa, from automobile to aircraft.
+
+### 3.2. Data
+
+Patent data were retrieved from Thomson Reuters' Thomson Innovation with Derwent World Patents Index (DWPI). Thomson Innovation is a comprehensive worldwide patent database, which covers patents recorded at more than 80 patent authorities and includes the United States Patent and
+
+---
+
+190
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+Trademark Office (USPTO) (DWPI data is available from 1963), the World Intellectual Property Organization (WIPO) (DWPI from 1978), the European Patent Office (EPO) (DWPI from 1978) and the Japanese Patent Office (JPO) (DWPI from 1963). The DWPI bundles patents recorded at 47 worldwide patent authorities as a protection for the same invention as a sort of family so that duplicate data retrieval can be avoided during search of patents crossing over several patent offices' databases. DWPI also provides manually added English abstracts from patent documents issued in more than 30 foreign languages so that it can allow text analysis of inventions from non-English language sources.
+
+We retrieved automobile and aviation patents which include one of following texts in applicant name, for the automobile industry: "honda motor", "honda giken", "toyota jidosha", "toyota motor", "mazda kk", "mazda motor", "nissan motor", "nissan jidosha", "mitsubishi jidosha" or "mitsubishi motor" and for aircraft industry: boeing, airbus, "rolls royce", "pratt and whitney", "rockwell collins", "hamilton sundstrand", "parker hannifin", "messier and dowty" and mtu. 242,305 and 27,989 patents were retrieved in July 2012 for automobile and aircraft by using the queries above. About the queries for the automobile industry, we listed more than one name for one company. Standardization of applicant name is being carried in many authorities with the spread of online application but there are different representations of applicant names in the past data.
+
+Compared to the aircraft industry, the automobile industry tends to patent more so that we limited the dataset for the automobile industry to patents of Japanese automobile companies. Queries for the aircraft industry cover major aircraft prime manufactures, major engine manufactures and major tier 1 suppliers of flight and data management, electric power systems, and mechanical components. To make the query process simple, we excluded General Electric or other major component suppliers that have other large business branches. We will discuss the limitation of our research as affected by queries in the Discussion section, but experts of each industry that we contacted in the survey validated that these queries are adequate for investigation of methodologies used to find knowledge integration candidates among the automobile and aircraft technology domains.
+
+### 3.3. Methodologies
+
+In order to identify a knowledge combination with the DB-Combination model from patent data, we take three computer-based bibliometrics approaches, A1, A2 and A3. Our DB-Combination model requires methodologies to identify technological sub-domains of industries from patent data and similarity measurements between identified sub-domains of two different industries. A1 and A2 take a citation analysis approach, and A3 takes IPC analysis approach for data structuring. For similarity measurement, A1 takes a text similarity measurement approach, A2 also takes an existing intra-industrial citation tracing approach, and A3 takes an IPC share similarity comparison approach ( Table 1 ).
+
+### 3.3.1. Technological sub-domain identification
+
+We conduct two different approaches to identify technological domains from the obtained patent data, using the citation analysis approach and an IPC analysis approach.
+
+Table 1 Three approaches.
+
+<table><tr><td>Method</td><td>Technological sub-domains identification</td><td>Similarity measurement</td></tr><tr><td>A1</td><td>Citation analysis</td><td>Text similarity measurement</td></tr><tr><td>A2</td><td>Citation analysis</td><td>Intra-industrial citation tracing</td></tr><tr><td>A3</td><td>IPC analysis</td><td>IPC share similarity comparison</td></tr></table>
+
+
+3.3.1.1. Citation analysis approach. The first approach is a citation network analysis. The analysis procedure is schematically illustrated in Fig. 2 . The patent and citation data are converted into a non-weighted, non-directed network in which a patent is represented as a node and backward citations to patents as links. The maximum connected component (MC) of the network is extracted. To minimize noise data and the quantity of data, we regarded patents not citing or cited by other patents in the component as digressional from the mainstream of those technological domains and eliminated them. Finally, the network was divided into clusters depending on the density of links using a topological clustering method ( Newman , 2004 ; Newman and Girvan , 2004 ) . After clustering the network, we characterized each cluster by the expert-based approach. A1 and A2 ( Table 1 ) take this approach.
+
+3.3.1.2. IPC analysis approach. Classification is important in the patent system to facilitate a search for "prior art" ( Leydesdorff , 2008 ) . The Strasbourg Agreement concerning the International Patent Classification established the IPC in 1971. The IPC is a hierarchical system of language independent symbols which divides technology into eight sections with 70,000 subdivisions of twelve-digit codes. The appropriate symbols are allotted by the national or regional industrial property office of 57 states (WIPO homepage).
+
+A3 takes the IPC sub-class of four-digit codes. $X_i$ is the share of sub-class $i$ in industry $X$ . $num(X$ , IPC $(i))$ is the number of patents of sub-class $i$ of industry $X$ , and $num(X)$ is the total number of patents of industry $X$ . A patent can have several IPC codes, where the total of the share for industry $X$ exceeds 100 % . This paper assumes the share as the rate of importance of the sub-class in an industry and therefore, we filter minor subclasses with the share threshold.
+
+$$X_{i}=\frac{num(X,IPC(i))}{num(X)}.\quad (1)$$
+
+### 3.3.2. Similarity measurement
+
+3.3.2.1. Cosine text similarity measurement. A1 research takes the text similarity approach, assuming that the similarity in characteristics such as background problems, processes to solve the problem, operational conditions, or compounds, can be measured by similarity in the text.
+
+First, the DWPI title and abstract of each text were analyzed and the frequency of word i in Cluster s ( $FreW_{si}$ ) was evaluated by the following formula:
+
+$$F r e W_{s i}=\frac{n_{s}}{n_{s}} \times \log \left(\frac{N}{N_{i}}\right) . \quad (2)$$
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+191
+
+![Figure](figures/figure_002.png)
+
+Fig. 2. Schematic diagram of the citation network analysis.
+
+In Eq. (2) , $n_{si}$ represents the number of word $i$ that appeared in the DWPI title and the abstract of the patents of Cluster $s$ obtained in citation analysis. $n_s$ represents the number of words that appeared in the title and the abstract of patents of Cluster $s$ . $N$ is the number of clusters in total. $N_i$ is the number of clusters in which a patent contains the word $i$ in the title and the abstract. The similarity of the text is evaluated by a cosine similarity that is often used in text mining and regards each text as a vector with the length of $FreW_{si}$ . Cosine similarity $Cos(a,m)$ between two clusters $a$ and $m$ is defined as Eq. (3) . A large $Cos(a,m)$ represents a relatively high similarity.
+
+$$similarity=\cos (a, m)=\frac{\sum FreW_{a i} \times FreW_{x a i}}{\sqrt{\left(FreW_{a i}\right)^{2}} \sqrt{\left(FreW_{m i}\right)^{2}}} . \quad (3)$$
+
+3.3.2.2. Existing intra-industry citation tracing. In A2, we assume that the similarity of different technology sub-domains can be measured in citations between automobile patents and aircraft patents. We consider that, if there is a similarity, and a potential of knowledge combination, some practitioners must have already recognized the similarity and challenged it. In A2, unlike in A1 which identifies technological sub-domains from automobile and aircraft patents separately in the citation analysis approach, these patents are combined and classified into clusters. Therefore, if there are patents with intra-industry citations, we must have clusters containing patents of both industries. We highlight such intra-industry clusters as similar technology sub-domains between the industries.
+
+3.3.2.3. IPC share similarity comparison. It is obvious that patents with the same IPC are similar in some characteristics of technological knowledge. We highlight IPC domains that have a similar IPC share in two datasets as formula (4). It is similar to Jaffe (1986) that characterized the technological position of a firm by the distribution of the firms' patents over patent classes. While Jaffe (1986) used a vector to see
+
+the overall characteristics, we focused on a factor or a patent class that have similar positions. If the share of IPC i over the firm's patents is similar between compared two industries, $r_{i}$ is close to 0.5.
+
+$$r_{i}=\frac{\sqrt{A v i_{i} M o t o_{i}}}{A v i_{i}+M o t o_{i}} . \quad (4)$$
+
+### 3.3.3. Interviews
+
+In order to research the effectiveness of three proposed approaches for practitioners to conduct breadth activities and to find possible knowledge combination domains, we discussed the results of the three approaches with experts. We interviewed four various levels of aeronautic researchers at Japan Aerospace Exploration Agency (JAXA) and two senior engineers at Toyota Central R&D Labs, INC in May 2012 in Japan. We visited their offices and had two-hour face-to-face interview for each center. We started the interview with questions about the importance and activities of breadth searches, and asked whether the highlighted areas in A1, A2, and A3 identified possible knowledge combination pairs and provided useful information that support practitioners in creating new knowledge.
+
+## 4. Results
+
+### 4.1. Technological sub-domain identification
+
+Among the three approaches ( Table 1 ), the A1 and A2 approaches used citation analysis to identify technological domains from the data. The patent dataset of the automobile and aircraft industries was analyzed separately in the A1 research, and was combined in the A2 research.
+
+In the A1 research, the MC of automobile and aircraft consists of 60,458 patents and 8281 patents, respectively, and were divided into 303 and 104 clusters. The number of patents in a cluster, that is, the size of clusters, varies from 6 to 5187 for
+
+---
+
+192
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+Table 2
+
+Overview of automobile technology domains identified in approach A1.
+
+<table><tr><td></td><td>CL Size</td><td>Yr</td><td>Characteristic of cluster</td></tr><tr><td></td><td>Freq. terms</td><td>Core patent</td><td>Core patent DWPI title example (for top 5 clusters)</td></tr><tr><td rowspan="4">1</td><td>15187</td><td>2003.8</td><td>Hybrid system (drive control)</td></tr><tr><td>Hybrid, battery, hybrid vehicle, torque, engine, electric, clutch, braking, generator, drive</td><td>JP20002225578A</td><td>Hybrid vehicle powered both by internal combustion engine and electric motor, has power transmission unit that maintains continuity of input power level during power source switch over</td></tr><tr><td></td><td>JP2000346187A</td><td>Power transmission for hybrid vehicle, includes drive shaft and output shaft whose revolution numbers are maintained more or less same, by controlling gear shift ratio</td></tr><tr><td></td><td>JP200002327A</td><td>Transmission speed controller of hybrid vehicles performs regenerative control of electric motor or generator, such that revolution number of transmission is decreased or increased, respectively when shift-up or shift-down is detected by speed change detector</td></tr><tr><td rowspan="4">2</td><td>4778</td><td>2003.3</td><td>Drive control</td></tr><tr><td>Steering, lane, image, object, driver, road, preceding vehicle, assistance, information, preceding</td><td>JP10211886A</td><td>Power steering system for e.g. electrically driven vehicle, hydraulic vehicle has steering suppression unit that prevents steering operation after decision unit has determined degree of danger based on detected position, distance, and relative velocity of obstruction</td></tr><tr><td></td><td>US20050125137A1</td><td>Vehicle deceleration control apparatus, has controller executing deceleration control by brake system that applies braking force to vehicle and shift operation that shifts automatic transmission of vehicle into low speed ratio</td></tr><tr><td></td><td>JP2002267470A</td><td>Information presentation system calculates position of view point of driver and window of vehicle with respect to specific position, based on the information related to specific position</td></tr><tr><td rowspan="4">3</td><td>4350</td><td>2002.8</td><td>Exhaust emission control</td></tr><tr><td>Catalyst, exhaust, gas, exhaust gas, oxide, purification, nitrogen oxide, nitrogen, nox, fuel ratio</td><td>JP8338228A</td><td>Exhaust emission control device of diesel engine stores nitric oxide when exhaust gas temperature is low and converts into nitrogen dioxide when high using oxidation catalyst along with oil and heater</td></tr><tr><td></td><td>JP7217474A</td><td>Exhaust gas purifying apps, for internal combustion engines comprises an air–fuel ratio controller to temporarily make rich exhaust gas flowing to nitrogen oxide absorber</td></tr><tr><td></td><td>JP2004174490A</td><td>Manufacture of catalyst material, e.g. for exhaust gas purification, involves mixing acidic solution comprising ions of cerium, zirconium and catalyst metal with aqueous ammonia, for co-precipitating and baking obtained mixture</td></tr><tr><td rowspan="4">4</td><td>3072</td><td>2002.6</td><td>IC engine valve control</td></tr><tr><td>Valve, engine, combustion, intake, internal combustion, internal combustion engine, valve timing, combustion, compression, internal</td><td>JP2003206771A</td><td>Internal combustion engine has camshafts comprising cams and movable bearings at axial portion, arranged on both sides of cylinder, to accommodate cam portions and bearings in holes formed in cylinder block and lower case</td></tr><tr><td></td><td>JP7293216A</td><td>Value gear of internal combustion engine for car has driving cam which is shut from suction valve supported by eccentric part in cylindrical cover</td></tr><tr><td></td><td>JP2004218522A</td><td>Driving control apparatus of internal combustion engine, controls supply of air into cylinder corresponding machine compression ratio, by delaying/advancing valve closing time, based on driving load of engine</td></tr><tr><td rowspan="4">5</td><td>2881</td><td>2005.2</td><td>Fuel cell system</td></tr><tr><td>Fuel cell, cell, fuel cell system, cell system, fuel, hydrogen, gas, cell stack, fuel cell stack, stack</td><td>JP7235324A</td><td>Fuel battery drive device increases dynamic pressure of oxygen gas supplied to electrode temporarily, when electrode gates get due to adhesion of solidified water drops</td></tr><tr><td></td><td>JP2000243417A</td><td>Fuel cell apparatus comprises removal unit to remove impurities which reduce specific value of power generated</td></tr><tr><td></td><td>JP2003331893A</td><td>Fuel cell system has control valve which supplies air to fuel cell stack, so that air purges through fuel electrode, during starting and electric power generation completion states</td></tr><tr><td>6</td><td>2557</td><td>1992.6</td><td>Automatic transmission control</td></tr><tr><td>Transmission, automatic transmission, clutch, automatic, shift, hydraulic, gear, valve, engine, pressure</td><td></td><td></td></tr><tr><td rowspan="4">7</td><td>2416</td><td>2001.8</td><td>Body structure</td></tr><tr><td>Frame, panel, bumper, front, vehicle body, collision, member, body, pillar, floor</td><td></td><td></td></tr><tr><td>8</td><td>2397</td><td>Combustion control</td></tr><tr><td>Injection, fuel, combustion, fuel injection, ignition, engine, internal combustion, internal combustion engine, combustion, valve</td><td></td><td>internal combustion engine, combustion, valve</td></tr><tr><td rowspan="2">9</td><td>2007</td><td>1991.3</td><td>Air–fuel ratio control</td></tr><tr><td>Air fuel, fuel ratio, air fuel ratio, fuel engine, air, ratio, intake, fuel injection, injection</td><td></td><td></td></tr><tr><td rowspan="2">10</td><td>2004</td><td>2004</td><td>Battery and cooling system</td></tr><tr><td>Battery, cell, fuel cell, cooling, floor, battery pack, pack, seat, frame, fuel</td><td></td><td></td></tr></table>
+
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+193
+
+Table 3 Overview of aircraft technology domains identified in approach A1.
+
+<table><tr><td rowspan="2">CL</td><td rowspan="2">Size Freq. terms</td><td rowspan="2">Yr Core patent</td><td rowspan="2">Characteristic of cluster Core patent DWPI title example (for top 5 clusters)</td></tr><tr><td rowspan="3">1</td><td>903</td><td>2002.5</td><td>Cabin equipment</td></tr><tr><td>Seat, cabin, door, passenger, compartment</td><td>US5083727A</td><td>Aircraft cabin system for selectively locating interior units with adjustable line connectable between aircraft cabin interior and adjacent service distribution coupling</td></tr><tr><td>Fuselage, gear, floor, luggage, overhead</td><td>US6464169B1</td><td>Overhead galley and crew rest facility for aircraft has cart transferring mechanism provided in overhead section and which supports galley cart between planar cart plane and cart lift</td></tr><tr><td rowspan="3">2</td><td>665</td><td>US4055317A</td><td>Aft main deck split level galley is used in aircraft and has raised deck giving increased floor space in concave rear pressure bulkhead</td></tr><tr><td>Welding, stir welding, friction stir welding, stir, composite</td><td>2001.1</td><td>Structure assembly/fabrication</td></tr><tr><td>Friction stir, ceramic, friction, sol, workpiece</td><td>US4622445A</td><td>Honeycomb panel inductive brazing placing alloy between face sheet and core and heating to alloy melting point</td></tr><tr><td rowspan="3">3</td><td>Friction stir, ceramic, friction, sol, workpiece</td><td>US5041321A</td><td>Mfg. fiber formed ceramic insulation useful in aerospace systems by multiple impregnation of soft felt that formed from slurry of fiber with sol–gel glass binder, and adding fiber–reinforced glass layer</td></tr><tr><td></td><td>US5645744A</td><td>Apparatus for induction processing of workpiece comprises forming dies non-susceptible to induction heating combined with susceptor sheets susceptible to induction heating</td></tr><tr><td></td><td></td><td>Flight Information System</td></tr><tr><td rowspan="3"></td><td>577</td><td>Weather, display, radar, weather radar, radar system</td><td>US7675461B1</td><td>Terrain data circuit for use in aircraft, has display control circuit generating composite terrain image based on data from radar system and data from database, where display signal representative of terrain is provided based on image</td></tr><tr><td></td><td>Terrain, antenna, weather radar, signal, data</td><td>US8049644B1</td><td>Aircraft warning system for depicting terrain awareness and warning system alert information in terrain advisory display, has processing circuit providing indication of first or second warming signal, or caution signal on display Method of adjusting position of antenna used in radar mounted in e.g. aircraft, involves reducing position error by adjusting antenna position using terrain angle within beam</td></tr><tr><td></td><td></td><td>Composite structure</td></tr><tr><td rowspan="3">4</td><td>568</td><td>Composite, tape, stringer, fuselage, skin</td><td>US20060060705A1</td><td>Shell structure for composite fuselage of aircraft, has fitting whose ends are attached to stiffeners and skins of primary and secondary panels, for joining primary panel to secondary panel</td></tr><tr><td></td><td>Mandrel, structure, material, composite, composite material</td><td>US20060108058A1</td><td>Composite shell structure manufacturing system for aircraft, has composite material applicator that moves with respect to support structure to apply composite material on interior mold surface of lay-up mandrel</td></tr><tr><td></td><td></td><td>US20090139641A1</td><td>Shell structure manufacturing method for fuselage of smaller aircraft e.g. fighter aircraft, involves positioning stiffeners on inner surface of skin lay-up, and concurring skin</td></tr><tr><td rowspan="3">5</td><td>558</td><td>1998.7</td><td>lay-up and stiffeners to bend stiffeners to skin lay-up</td></tr><tr><td>Engine, nozzle, jet engine, jet, exhaust</td><td>US20100126139A1</td><td>Notzle system for gas turbine engine having longitudinal axis for aircraft, has fan nozzle which can be pivoted about axis oriented transversely relative to longitudinal axis to vary fan duct nozzle throat area</td></tr><tr><td></td><td>Turbine, flow, gas, gas turbine, nacelle</td><td>US4044973A</td><td>Nacelle mounting for turbofan jet engine has rear nacelle structure formed in two D-shape ducts hinged to open as clam shells</td></tr><tr><td></td><td></td><td>US4458863A</td><td>Strut supported inlet for turbofan has space frame thrust linkage transferring loads to wing strut</td></tr><tr><td>6</td><td>431</td><td>2003.4</td><td>Refueling system</td></tr><tr><td></td><td>Refueling, antenna, tanker aircraft, tanker, boom, phased array, phased, flight refueling, maintenance, array antenna</td><td>1999.5</td><td>Wing system</td></tr><tr><td>7</td><td>428</td><td></td><td>Wing system</td></tr><tr><td></td><td>Flap, wing, lift, edge flap, aircraft wing, high lift, slat, trailing edge,edge, trailing</td><td></td></tr><tr><td>8</td><td>341</td><td>2004.4</td><td>Environment Control System, APU</td></tr><tr><td></td><td>Air, fuel cell, cooling, fuel, air conditioning, heat, conditioning, exchanger, heat exchanger, decompression</td><td></td></tr><tr><td>9</td><td>259</td><td>1998.6</td><td>Assembly tools</td></tr><tr><td></td><td>Drilling, workpiece, fastener, tool, hole, rivet, riveting, carriage, electromagnetic, machine</td><td></td></tr><tr><td rowspan="2">10</td><td>202</td><td>1999.8</td><td>Landing gear system</td></tr><tr><td></td><td>Gear, brake, braking, wheel, landing gear, landing, control, braking system, command</td><td></td></tr></table>
+
+
+the automobile clusters and from 4 to 903 for the aircraft clusters. As the size of the top 4 clusters of the automobile was relatively large, we analyzed the sub-clusters of the four clusters, applying the same clustering approach. Each cluster was divided into 42, 29, 39, and 35 sub-clusters. Tables 2 and 3
+
+show the overview of the top 10 clusters in automobile and aircraft patents respectively. To identify the characterization of clusters, we analyzed the core patents that have frequent links (non-direction) with the other patents in the cluster and the frequent words that were calculated by formula (3). The
+
+---
+
+194
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+example of core patents is listed for the top 5 clusters in the Tables. The size and the average patented years can also provide characteristic trends of each cluster.
+
+In the A2 approach, 69,281 patents constructed the MC from the combined dataset of automobile and aircraft industry patents and were divided into 420 clusters. The size of clusters varied from 4 to 6254. The overview of the top 10 clusters is listed in Table 4 .
+
+In the A3 approach, 8 sections, 112 classes and 676 subclasses and 8 sections, 119 classes and 554 sub-classes were found in the automobile and aircraft dataset. To reduce the data volume, thresholds were set at two stages. First, IPC sub-classes with more than 2% of share for either the automobile or aircraft were taken. Secondly, from the filtered IPC sub-classes at the first stage, IPC sub-classes with more than 0.5% of share for both the automobile and aircraft were taken. In total, 62 sub-classes
+
+Table 4 Overview of the automobile and aircraft combined technology domains identified in approach A2.
+
+<table><tr><td>CL</td><td>Size</td><td>Yr</td><td>Characteristic of cluster</td></tr><tr><td></td><td>Freq. terms</td><td>Core patent</td><td>Core patent DWPI title example (for top 5 clusters)</td></tr><tr><td rowspan="4">1</td><td>6254</td><td>2003.9</td><td>Hybrid system (drive control)</td></tr><tr><td>Hybrid, hybrid vehicle, battery, engine, torque</td><td>JP2002225578A</td><td>Hybrid vehicle powered both by internal combustion engine and electric motor, has power transmission unit that maintains continuity of input power level during power source switch over</td></tr><tr><td>Electric, motor, drive, power, braking</td><td>JP2000346186A</td><td>Power transmission for hybrid vehicle, includes drive shaft and output shaft whose revolution numbers are maintained more or less same, by controlling gear shift ratio</td></tr><tr><td></td><td>JP200002327A</td><td>Transmission speed controller of hybrid vehicles performs regenerative control of electric motor or generator, such that revolution number of transmission is decreased or increased, respectively when shift-up or shift-down is detected by speed change detector</td></tr><tr><td rowspan="4">2</td><td>6138</td><td>1998.1</td><td>IC engine system</td></tr><tr><td>Fuel, catalyst, exhaust, engine, injection</td><td>US4949695A</td><td>Fuel evaporative purge system malfunction detecting device has value in purge passage which opens when engine operates and sensor to detect negative vacuum and compare it with intake vacuum</td></tr><tr><td>Injection, air fuel, fuel ratio, combustion, exhaust gas</td><td>US4571683A</td><td>Learning air–fuel ratio control system for electronic engine has second learning term corrected during idling period and first when predetermined load is exceeded</td></tr><tr><td></td><td>US4561400A</td><td>Air–fuel ratio control method controlling fuel injection rate by learning correction coefficient for obstruction of air flow motor to compensate for aging of meter Driving support system</td></tr><tr><td rowspan="4">3</td><td>5282</td><td>2003.3</td><td>Power steering system for e.g. electrically driven vehicle, hydraulic vehicle has steering suppression unit that prevents steering operation after decision unit has determined degree of danger based on detected position, distance, and relative velocity of obstruction</td></tr><tr><td>Driver, road, object, assistance, driving, preceding vehicle</td><td>JP2002267470A</td><td>Information presentation system calculates position of view point of driver and window of vehicle with respect to specific position, based on the information related to specific position</td></tr><tr><td></td><td>JP2001310719A</td><td>Lane deviation prevention apparatus for vehicles, regulates yawing moment along specified direction, based on braking force difference between right and left wheels</td></tr><tr><td rowspan="4">4</td><td>4375</td><td>2002.1</td><td>Aircraft system</td></tr><tr><td>Aircraft, wing, composite, fuselage, skin</td><td>US7675461B1</td><td>Terrain data circuit for use in aircraft, has display control circuit generating composite terrain image based on data from radar system and data from database, where display signal representative of terrain is provided based on image</td></tr><tr><td>Panel, flap, flight, cabin, antenna</td><td>US8049644B1</td><td>Aircraft warning system for depicting terrain awareness and warning system alert information in terrain advisory display, has processing circuit providing indication of first or second warning signal, or caution signal on display</td></tr><tr><td></td><td>US8077078B1</td><td>Measurement method for determining altitude of aircraft using altitude of runway and radar sweep by radar system, and based on vertical angle and range between runway and aircraft, and runway altitude</td></tr><tr><td rowspan="4">5</td><td>3467</td><td>2003.8</td><td>Combustion control system</td></tr><tr><td>Fuel, combustion, injection, ignition, engine</td><td>JP10266878A</td><td>Controller for four-stroke engine has exhaust valve whose closing time is delayed when demand load of system becomes low</td></tr><tr><td>Fuel injection, internal comb., int. comb. Engine, comb. Engine, valve timing</td><td>JP2000179368A</td><td>Fuel supply procedure in gasoline internal combustion engine, involves increasing proportion of supply of fuel having high and low octane numbers during high and low load running respectively</td></tr><tr><td></td><td>JP11210539A</td><td>Internal combustion engine has controller which regulates gas in combustion chamber to compression stroke end stage within target temperature range based on output of decision circuit that judges whether self-adhering fire of mixed air will increase</td></tr><tr><td rowspan="2">6</td><td>3444</td><td>2005.1</td><td>Fuel cell system</td></tr><tr><td>Fuel cell, cell, fuel, fuel cell system, cell system, hydrogen, gas, cell stack, stack</td><td></td></tr><tr><td rowspan="2">7</td><td>2473</td><td>2001.9</td><td>Body structure</td></tr><tr><td>Front, frame, vehicle body, bumper, collision, member, body, rear, pillar, structure</td><td></td></tr><tr><td rowspan="2">8</td><td>2297</td><td>2004.1</td><td>Battery and cooling system</td></tr><tr><td>Battery, cell, fuel cell, fuel, cooling, floor, battery pack, stack, seat, cell stack</td><td></td></tr><tr><td rowspan="2">9</td><td>2082</td><td>1993.5</td><td>Automatic transmission control</td></tr><tr><td>Transmission, automatic transmission, shift, gear, automatic clutch, hydraulic, variable transmission, speed, torque</td><td></td></tr><tr><td rowspan="2">10</td><td>2008</td><td>2001.5</td><td>IC engine valve control</td></tr><tr><td>Valve, compression ratio, engine, cam, combustion, internal comb., int. comb. Engine, comb. Engine, valve timing, timing</td><td></td></tr></table>
+
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+195
+
+were taken as the major sub-classes and cover 207,950 automobile and 22,231 aircraft patents. Table 5 lists the 10 biggest IPC sub-classes found in automobile and aircraft patents. The titles appearing in this paper for each IPC code were adopted from the WIPO homepage.
+
+### 4.2. Similarity measurement
+
+### 4.2.1. A1 approach
+
+In the A1 approach, the similarity measurement was conducted among the top 35 clusters and the top 15 subclusters of the top 4 clusters of the automobile and the top 25 clusters of aircraft. Table 6 shows part of the results. The average of cosine similarity except for the sub-clusters was 0.28 and the standard deviation was 0.11. The average of cosine similarity between the sub-clusters of the automobile and clusters of aircraft was 0.39, and the standard deviation was 0.1034. We highlighted cluster pairs with more than 0.5 of cosine similarity and automobile sub-clusters and aircraft cluster pairs with more than 0.6 of cosine similarity. As a result, 48 pairs of clusters were highlighted. We discussed the potential for the combination of knowledge.
+
+As a result, automobile clusters of driving control and hybrid car driving control such as in cluster 6 (Automatic Transmission Control) and the sub-clusters of cluster 1 (the Hybrid System) and 2 (Drive Control) were highlighted with the aircraft clusters of flight control such as in cluster 12 (Throttle Control), and 13 (Active Flow Control). The use of terms related to system control was similar among highlighted clusters.
+
+However, according to the interviews with aeronautical researchers, basically, 'driving control for automobiles' represents actuating defined motion according to input while 'flight control for aircrafts' represents actuating motion according to operating conditions to achieve the inputted flight attitude. Based on the experts' judgments, the authors found that, in this case, the terms in patents are similar, but the engineering philosophy is very different so that there is less opportunity for knowledge recombination among highlighted automobile and aircraft control domains.
+
+On the other hand, there was another type of pairs identified in A1, whose relations were not clear at a glance for authors and interviewers. For example, high similarity was detected between the sub-clusters of automobile cluster 3 (Exhaust Emission Control) and cluster 4 (IC Engine Valve Control) and aircraft cluster 15 (Fiber-reinforced Materials) and cluster 24 (Heat Blanket). We investigated the factors for this high similarity appeared in these very different pairs at a glance, and some similar uses of terms were detected such as temperatures, heating, and pressure. The performance of an exhaust emission purifying system becomes lower at between 300 °C to 400 °C and over 500 °C (Patent JP2002-126453). Improvement of exhaust emission purifier efficiency in a wide range of temperatures is expected in the automobile. On the other hand, application of fiber-reinforced materials such as carbon fiber-reinforced polymers and fiber-reinforced ceramics is important for aircraft development because light, strong, and heat-resistant characteristics of these materials are very suited to the aircraft operation condition. Investigating the relation between exhaust gas purification and composite materials further, some patents were found that focus on the heat resistance nature of fiber-reinforced ceramics and apply fiberreinforced ceramics as exhaust purifying catalyst support and increase the performance of the purifier (ex. Patent JP2001179110) under a wide range of temperatures.
+
+At first, the relation between those clusters was unclear, but the similarity between problems to be solved in the exhaust emission purifier and in the properties of fiber-reinforced materials showed the possibility of knowledge combination and a breakthrough in the exhaust gas purifier. When we showed the results of our investigation on such pairs, automobile engineers found that highlighting such similarity of problems is very useful for them to transfer their technology to other industries and vice versa.
+
+### 4.2.2. A2 approach
+
+A2 approach resulted into 689 patents citing other industries and 612 patents cited by other industries in the automobile and aircraft dataset. In the MC of the combined
+
+Table 5 Major automobile and aircraft IPC sub-classes identified in approach A3.
+
+<table><tr><td>IPC</td><td>Class (ex. B23)</td><td>Sub-class (ex. B23P)</td><td>S-Avi</td><td>S–M</td><td></td></tr><tr><td>B23P</td><td>Machine tools</td><td>Combined operation</td><td>3.9%</td><td>1.0%</td><td>A</td></tr><tr><td>B29C</td><td>Working of plastic</td><td>Shaping of plastic</td><td>4.6%</td><td>2.1%</td><td>A</td></tr><tr><td>B32B</td><td>Layard product</td><td>Layard product</td><td>5.0%</td><td>0.8%</td><td>A</td></tr><tr><td>B60K</td><td>Vehicles in general</td><td>Arrangement of propulsion units</td><td>0.4%</td><td>10.2%</td><td>M</td></tr><tr><td>B60L</td><td></td><td>Propulsion of electrically-propelled vehicles</td><td>0.3%</td><td>5.2%</td><td>M</td></tr><tr><td>B60R</td><td></td><td>Vehicles</td><td>0.6%</td><td>9.6%</td><td>M</td></tr><tr><td>B62D</td><td>Land vehicles for traveling</td><td>Motor vehicles</td><td>0.2%</td><td>7.8%</td><td>M</td></tr><tr><td>B64C</td><td>Aircraft</td><td>Aeroplanes</td><td>15.4%</td><td>0.1%</td><td>A</td></tr><tr><td>B64D</td><td></td><td>Equipment</td><td>13.6%</td><td>0.0%</td><td>A</td></tr><tr><td>F01D</td><td>Machines or engine</td><td>Non-positive-displacement</td><td>11.4%</td><td>0.2%</td><td>A</td></tr><tr><td>F01N</td><td></td><td>Exhaust apparatus for machines</td><td>0.5%</td><td>5.3%</td><td>M</td></tr><tr><td>F02B</td><td>Combustion engines</td><td>Internal-combustion piston engines</td><td>0.9%</td><td>5.0%</td><td>M</td></tr><tr><td>F02C</td><td></td><td>Gas-turbine plants</td><td>9.2%</td><td>0.2%</td><td>A</td></tr><tr><td>F02D</td><td></td><td>Controlling combustion engines</td><td>0.6%</td><td>13.6%</td><td>M</td></tr><tr><td>F02K</td><td></td><td>Jet-propulsion plants</td><td>3.9%</td><td>0.0%</td><td>A</td></tr><tr><td>F02M</td><td></td><td>Supplying combustion engines</td><td>1.1%</td><td>6.2%</td><td>M</td></tr><tr><td>F04D</td><td>Pumps for liquids</td><td>Non-positive-displacement pumps</td><td>3.5%</td><td>0.2%</td><td>A</td></tr><tr><td>F16H</td><td>Engineering elements</td><td>Gearing</td><td>0.9%</td><td>6.9%</td><td>M</td></tr><tr><td>G06F</td><td>Computing</td><td>Electric digital data processing</td><td>7.4%</td><td>2.9%</td><td>A</td></tr><tr><td>H01M</td><td>Basic electric elements</td><td>Batteries</td><td>1.1%</td><td>8.5%</td><td>M</td></tr></table>
+
+
+---
+
+196
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+Table 6
+
+Cosine similarity results in approach A1.
+
+<table><tr><td></td><td>Motor:1</td><td>Motor:2</td><td>Motor:3</td><td>Motor:4</td><td>Motor:5</td><td>Motor:6</td><td>Motor:7</td><td>Motor:8</td><td>Motor:9</td><td>Motor:10</td></tr><tr><td>Avi:1</td><td>0.22</td><td>0.19</td><td>0.19</td><td>0.23</td><td>0.10</td><td>0.24</td><td>0.41</td><td>0.17</td><td>0.23</td><td>0.50</td></tr><tr><td>Avi:2</td><td>0.19</td><td>0.13</td><td>0.19</td><td>0.19</td><td>0.14</td><td>0.19</td><td>0.28</td><td>0.17</td><td>0.21</td><td>0.31</td></tr><tr><td>Avi:3</td><td>0.20</td><td>0.36</td><td>0.15</td><td>0.17</td><td>0.11</td><td>0.20</td><td>0.12</td><td>0.20</td><td>0.21</td><td>0.20</td></tr><tr><td>Avi:4</td><td>0.18</td><td>0.21</td><td>0.14</td><td>0.19</td><td>0.07</td><td>0.18</td><td>0.36</td><td>0.12</td><td>0.16</td><td>0.31</td></tr><tr><td>Avi:5</td><td>0.35</td><td>0.18</td><td>0.41</td><td>0.40</td><td>0.11</td><td>0.34</td><td>0.32</td><td>0.37</td><td>0.47</td><td>0.36</td></tr><tr><td>Avi:6</td><td>0.27</td><td>0.25</td><td>0.17</td><td>0.17</td><td>0.10</td><td>0.30</td><td>0.14</td><td>0.19</td><td>0.25</td><td>0.17</td></tr><tr><td>Avi:7</td><td>0.30</td><td>0.23</td><td>0.19</td><td>0.30</td><td>0.20</td><td>0.27</td><td>0.21</td><td>0.25</td><td>0.24</td><td>0.27</td></tr><tr><td>Avi:8</td><td>0.33</td><td>0.20</td><td>0.32</td><td>0.38</td><td>0.41</td><td>0.36</td><td>0.18</td><td>0.37</td><td>0.53</td><td>0.46</td></tr><tr><td>Avi:9</td><td>0.28</td><td>0.23</td><td>0.17</td><td>0.19</td><td>0.08</td><td>0.25</td><td>0.22</td><td>0.26</td><td>0.28</td><td>0.27</td></tr><tr><td>Avi:10</td><td>0.46</td><td>0.44</td><td>0.21</td><td>0.22</td><td>0.07</td><td>0.45</td><td>0.18</td><td>0.22</td><td>0.27</td><td>0.15</td></tr><tr><td>Avi:11</td><td>0.23</td><td>0.31</td><td>0.19</td><td>0.20</td><td>0.08</td><td>0.23</td><td>0.39</td><td>0.17</td><td>0.16</td><td>0.39</td></tr><tr><td>Avi:12</td><td>0.43</td><td>0.33</td><td>0.23</td><td>0.44</td><td>0.16</td><td>0.55</td><td>0.17</td><td>0.28</td><td>0.45</td><td>0.22</td></tr><tr><td>Avi:13</td><td>0.32</td><td>0.34</td><td>0.28</td><td>0.37</td><td>0.23</td><td>0.37</td><td>0.24</td><td>0.33</td><td>0.41</td><td>0.38</td></tr><tr><td>Avi:14</td><td>0.25</td><td>0.19</td><td>0.31</td><td>0.30</td><td>0.08</td><td>0.22</td><td>0.36</td><td>0.28</td><td>0.24</td><td>0.42</td></tr><tr><td>Avi:15</td><td>0.25</td><td>0.13</td><td>0.32</td><td>0.29</td><td>0.26</td><td>0.31</td><td>0.21</td><td>0.29</td><td>0.27</td><td>0.26</td></tr></table>
+
+
+"Automobile technology fields
+
+<table><tr><td>Driving system</td><td>2, 33, 14, 34, 6 (AT), 18 (Motor), 19 (Suspension)</td></tr><tr><td>Engine</td><td>3 (Exhaust Purification), 4 (Valve Cntl), 8 (Combustion), 9, 11 (Fuel Injection), 31 (Cooling Sys.), 32 (Fuel Tank)</td></tr><tr><td>Structure</td><td>7, 21 (Collision Absorption), 15 (Airbag), 23 (Door Lock), 26 (Hood), 28 (Seat Storage), 35 (Assembry)</td></tr><tr><td>Hybrid/fuel system</td><td>1 (Drive Cntl.), 5, 16, 20 (Fuel Cell), 10, 25 (Battery)</td></tr><tr><td>Others</td><td>13 (Motor Cycle), 28 (Walking Robot), 30 (Nap Detector), 33 (Noise Control)</td></tr></table>
+
+
+*Aircraft technology fields
+
+<table><tr><td>Flight control</td><td>3 (Flight Info. Sys.), 6 (Refueling Sys.), 7 (Wing Sys.), 10 (Landing Gear), 12 (Throttle Cntl.), 13 (Active Flow Cntl.)</td></tr><tr><td>Propulsion</td><td>5 (Nacell), 16 (Blade Containment), 19 (Gas Turbine Safely Devices), 20 (Gas Turbine Component), 23 (Combustion), 25 (Fuel Injection)</td></tr><tr><td>Structure</td><td>2 (Assembly Method), 4 (Composite Material), 9 (Assembly Tools), 11 (Composite Material, Fastening), 15 (Fiber Reinforced Material), 17 (Structure Inspection), 24 (Heat Blancket)</td></tr><tr><td>Others</td><td>1 (Cabine, Storage), 8 (APU Sys.), 14 (Harness), 18 (Starter Generator), 21 (Fiber-Opt Magnetic Field), 22 (Toilet Sys.)</td></tr></table>
+
+
+dataset, 421 patents cite other industries and 345 patents are cited by other industries. We analyzed the top 35 clusters and highlighted clusters that have both automobile and aircraft patents. Table 7 shows part of the results and the percentage of each industry patent. 6 clusters were highlighted in total.
+
+As to the results of A2, cluster 4 was the Aircraft System cluster, where many aircraft patents belong. According to the sub-cluster analysis, the cluster consists of various sub-clusters such as Composite Structures, Composite Material Welding, Engine Mounting, Cabin Compartments, and Aircraft Wing System and Avionics. Among 4375 patents of cluster 4, 110 patents are automobile patents. Many automobile patents are also found in Composite Structures and Composite Material Welding sub-clusters and the topics of the automobile patents found in these sub-clusters are the production of body structure with a fiber-reinforced composite, surface defect inspection, and stir welding. There are also automobile patents in Avionics and in that case the main topics of these automobile patents are navigation display processing. On the other hand, clusters 6 and 9 are the actual automobile clusters, such as the Fuel Cell System cluster and the Automatic Transmission Control cluster. Among 3444 patents and 2082 patents of clusters 6 and 9, 42 patens and 69 patents are aircraft patents.
+
+Composite materials and the navigation system are well known examples of technology transfer from the aircraft
+
+industry to the automobile industry. There is also technology transfer recognized in braking systems such as anti-lock systems and disk brakes from the aircraft industry to the automobile industry. In addition, the fuel cell and electric generation system is often considered to be more advanced in the automobile industry than in the aircraft industry so that aeronautic researchers are carefully watching technology development in the automobile industry. The A2 approach highlighted both past and emerging knowledge transfer from one industry to another and also highlighted knowledge exchanges in very similar processes such as in the Assembly.
+
+Fig. 3 shows the sub-clusters of cluster 6 in A2 and visualizes the use of a large graph layout (LGL) ( Adai et al. , 2004 ) . $^2$ Researchers of JAXA showed interest in the results. When they needed to analyze patents related to Fuel Cell technology, they searched the data with keywords such as “fuel cell” and obtained many patents containing the keywords. However, the researchers had problems because it was difficult to categorize the obtained data in detail. The researchers said that results in
+
+2 LGL is based on a spring layout algorithm where links play the role of spring connecting nodes. As a result of this layout, the group of patents citing each other is located in closer positions and only the intra-cluster links for each cluster are shown with the same color, in the order in which the clusters are intuitively understood to clarify the position of each cluster.
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+197
+
+Table 7 Percentages of automobile and aviation patents in identified domains in approach A2.
+
+<table><tr><td></td><td>Motor</td><td>Aviation</td></tr><tr><td>1</td><td>99.90%</td><td>0.10%</td></tr><tr><td>2</td><td>99.90%</td><td>0.10%</td></tr><tr><td>3</td><td>100.00%</td><td>0.00%</td></tr><tr><td>4</td><td>2.50%</td><td>97.50%</td></tr><tr><td>5</td><td>99.70%</td><td>0.30%</td></tr><tr><td>6</td><td>98.80%</td><td>1.20%</td></tr><tr><td>7</td><td>99.80%</td><td>0.20%</td></tr><tr><td>8</td><td>99.60%</td><td>0.40%</td></tr><tr><td>9</td><td>96.70%</td><td>3.30%</td></tr><tr><td>10</td><td>99.90%</td><td>0.10%</td></tr></table>
+
+
+Fig. 3 provided them with an interesting overview of patents related to automobile Fuel Cell technology that they would not have been able to analyze within their own limited resources.
+
+### 4.2.3. A3 approach
+
+In the A3 approach, IPC sub-classes with more than 0.48 of $r_{i}$ were highlighted (Table 8 ). 19 sub-classes were highlighted.
+
+As tothe results of A3, it was easy to find common topics between the automobile and aviation industries in the highlighted IPC sub-class because it is the very objective of IPC classification. For example, G01L is a Measuring Force class and how to exclude the impact of torsions of shafts in measuring the force and conciliation frequency of a hybrid
+
+engine shaft and turbine engine shaft is one of its common topics. H01L is semiconductor device class and a common topic in improving the efficiency of how to cool semiconductors. H04N became a Pictorial Communication and edge problem when several pictures were combined.
+
+![Figure](figures/figure_009.png)
+
+Fig. 3. Structure overview of fuel cell sub-clusters, cluster 6 in A2.
+
+At the interview, automobile engineers commented that, even though this approach was simple, some of the highlighted areas were interesting to look at. Furthermore, they commented that, to increase the value of information in this approach, for example, the analysis of the co-occurrence of IPC categories in the patents of each industry, will be useful. As several IPC categories can be defined in one patent, highlighting differences in the co-occurrence of IPC between two industries can give an industry new idea of technological application already challenged in other industries.
+
+## 5. Discussion
+
+Table 9 compares A1, A2 and A3 results. For approaches of technological sub-domain identification, the citation analysis approach limited the identification to the MC so that the coverage of data was much inferior to the IPC approach. If a comprehensive overview of technological sub-domains in other industries is needed, the IPC approach can satisfy the needs as a list of technological knowledge. Moreover, the overview of the IPC approach can be obtained easily with spread sheet software. On the other hand, our citation analysis enables us to identify technology sub-domains as associated parts of the systems so that the analysis can support engineers in obtaining breadth knowledge related to the technology subdomains being transferred from or to and can accelerate a broadening scope of projects and adoption of technology.
+
+However, wider data coverage of IPC approach does not promise effectiveness in knowledge combination. It can be interpreted from Table 4 that the Fuel Cell technology of cluster 6 is a very new domain compared to other listed clusters. In addition, analysis of sub-clusters of cluster 6 can dismantle different technological knowledge consisting of the Fuel Cell system as was appreciated by JAXA engineers ( Fig. 3 ). The Fuel Cell Structure shown in the citation analysis was descriptive and different from the IPC structure. Many fuel cell patents in cluster 6 in A2 were categorized to sub-classes: F02D (Controlling Combustion Engines), F02B (Internal-Combustion Piston Engines), F02M (Supplying Combustion Engines), or group F02D41/F02D45 (Electrical control of supply of combustible mixture or its constituents), F02D13 (Controlling the engine output power by varying inlet or exhaust valve operating characteristics). IPC categories are periodically revised but not very flexible in adapting to the dynamism of technological knowledge. The principal aim of IPC is to facilitate a search of patents in function. Therefore, using IPC categories to see an overview of a system is difficult.
+
+For measuring similarities between domains, compared to the other two approaches, the A1 approach had a higher likelihood of highlighting DB-C combinations. The A1 approach highlighted similarity either in functions (ex. control system) or properties (ex. heat resistance). While it seemed difficult to combine knowledge of the same functions from different philosophies, bringing knowledge with different functions but the same property or the same type of problems together can have the possibility of success in knowledge recombination in
+
+---
+
+198
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+Table 8 IPC-sub classes with high $\tau_{0}$
+
+<table><tr><td>IPC</td><td>Class (ex. B23)</td><td>Sub-class (ex. B23P)</td><td>S-Avi</td><td>S-M</td><td>R</td></tr><tr><td>B05D</td><td>Spraying</td><td>Processes for applying liquids</td><td>1.3%</td><td>0.9%</td><td>0.50</td></tr><tr><td>B21D</td><td>Mechanical metal-working</td><td>Processing of sheet metal</td><td>1.8%</td><td>1.1%</td><td>0.49</td></tr><tr><td>B22F</td><td>Casting</td><td>Working metallic powder</td><td>0.8%</td><td>0.6%</td><td>0.49</td></tr><tr><td>B23K</td><td>Machie tools</td><td>Welding</td><td>2.9%</td><td>2.0%</td><td>0.49</td></tr><tr><td>C04B</td><td>Cements</td><td>Lime, magnesia</td><td>0.6%</td><td>0.6%</td><td>0.50</td></tr><tr><td>C23C</td><td>Coating metallic material</td><td>Coating metallic material</td><td>1.8%</td><td>1.1%</td><td>0.48</td></tr><tr><td>F16C</td><td>Engineering elements</td><td>Shafts</td><td>1.5%</td><td>1.1%</td><td>0.49</td></tr><tr><td>G01C</td><td>Measuring</td><td>Measuring distance</td><td>2.5%</td><td>1.8%</td><td>0.49</td></tr><tr><td>G01L</td><td></td><td>Measuring force</td><td>0.8%</td><td>0.7%</td><td>0.50</td></tr><tr><td>G01M</td><td></td><td>Testing static</td><td>1.8%</td><td>1.1%</td><td>0.49</td></tr><tr><td>G01P</td><td></td><td>Measuring linear</td><td>0.6%</td><td>0.6%</td><td>0.50</td></tr><tr><td>G01R</td><td></td><td>Measuring electric variables</td><td>1.4%</td><td>0.8%</td><td>0.48</td></tr><tr><td>G05B</td><td>Controlling</td><td>Control system in general</td><td>1.6%</td><td>1.1%</td><td>0.49</td></tr><tr><td>G06T</td><td>Computing</td><td>Image data processing</td><td>0.6%</td><td>1.0%</td><td>0.48</td></tr><tr><td>H01L</td><td>Basic electric elements</td><td>Semiconductor devices</td><td>1.5%</td><td>2.1%</td><td>0.49</td></tr><tr><td>H01R</td><td></td><td>Electrically conductive connections</td><td>1.0%</td><td>0.6%</td><td>0.48</td></tr><tr><td>H02K</td><td>Generation of electric power</td><td>Dynamo-electric machines</td><td>1.4%</td><td>2.0%</td><td>0.49</td></tr><tr><td>H02P</td><td></td><td>Control of electric motors</td><td>1.0%</td><td>1.5%</td><td>0.49</td></tr><tr><td>H04N</td><td>Electric communication technique</td><td>Pictorial communication</td><td>0.7%</td><td>0.8%</td><td>0.50</td></tr></table>
+
+
+breadth. To measure similarity in operating conditions or properties such as temperature, pressure or affinity, selecting terms to measure the text similarity is favorable and is expected to appear in future research.
+
+On the other hand, as we have already discussed, the A2 approach highlighted past and emerging knowledge transfer from one industry to another (DB-T combination), such as in Avionics, Composite Materials, and Fuel Cell, and knowledge exchanges in very similar processes (DB-D combination) such as in Assembly. We can expect the A2 approach to provide useful technological information related to on-going technology transfer from one industry to another. We also expect that analysis of the development of links in past knowledge transfer will help to manage future knowledge transfer and such analysis is expected to be studied in the near future.
+
+The A3 approach measured similarities, while it reduced the amount of data that needed to be looked at, and it is not surprising to find common topics between two industries' patents because they are categorized into the same IPC that have been established to facilitate finding DB-D pairs.
+
+While the methodologies still need to be improved, the three measurements can highlight DB-C, DB-T and DB-D pairs in the proposed DB-Combination model ( Fig. 4 ). In our case, i.e., knowledge combination between automobile and aircraft industry, A1 approach is effective to explore DB-C, because it can capture implicit technological relationships by textsimilarity measurement. On the other hand, A3 approach based on IPC similarity can extract explicit relationships and can be used to deepen existing common technologies, which we call DB-D. A2 approach also utilizes explicit technological
+
+relationships by citation-similarity measurement, but its effect is DB-D and DB-T. Therefore, we can conclude that practitioners should utilize each approach to supplement the others or select relevant approach to fit their purpose.
+
+However, it must be noted that the results of three approaches can largely depend on the dataset. In this paper, patents from a limited number of companies of the automobile industry and the aircraft industry were investigated. Expansion of the dataset is needed to explore patents in these industries more completely. And analysis on different sectors might give different results. In the above analysis, we used automobile patents only by Japanese automobile contractors. This is because according to our interview with experts, in the automobile industry, patents tend to cite other patents issued by their competitor in the same country, while in the aviation industry manufacturing is globalized and such a tendency is relatively not expected. And in the above analysis, we do not include major supplier of the automobile industry like Aisin, Denso, and Bosch. This is because the difference of the supply chain system between automobile and aircraft industries. Our aim is to capture the mainstream of technology on an industry. In the development of an aircraft, the aircraft prime manufacturers often take risk and revenue sharing partnerships with suppliers and large parts of technology development are done in suppliers. To avoid such a citation bias, we limited the scope of automobile patents and included patents by the suppliers in the aviation industry.
+
+However, there is a possibility that such a dataset construction strategy can affect the results. To test it, we added the following applicant names and reexamined the results; Ford or
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+199
+
+Table 9
+
+A1, A2 and A3 data coverage.
+
+<table><tr><td></td><td>A1</td><td>A2</td><td>A3</td></tr><tr><td>Dataset</td><td colspan="3">Automobile: 243,305, aircraft: 27,989</td></tr><tr><td>Patent numbers that are used for technological sub-domain identification (% over the total)</td><td>Automobile MC: 60,458Aircraft MC: 8281(25.4%)</td><td>Combined MC: 69,281(25.6%)</td><td>270,294(100%)</td></tr><tr><td>Identified cluster</td><td>Automobile: 303Aircraft: 104</td><td>420</td><td>676</td></tr><tr><td>Similarity analyzed cluster numbers</td><td>Automobile: 35Aircraft: 25(20.6%)</td><td>35(20.5%)</td><td>62(85%)</td></tr><tr><td>Highlighted similarity</td><td>48 pairs of clusters</td><td>6 clusters</td><td>19 sub-classes</td></tr></table>
+
+
+“General Motors” or GM or Chrysler or Daimlerchrysler or “Daimler Chrysler” or Daimler or “Bayerische Motoren Werke” or Peugeot or Citroen or Volkswagen or Renault or “Hyundai Motor” or Aisin or Denso or Bosch. When we analyzed such an expanded dataset, the number of patents for the automobile increased from 243,305 to 573,788.
+
+In the approach A1, the patents in the maximum component increased from 60,458 to 176,604. Even though the number of clusters also increased from 303 to 836, as we explained above, many clusters are of the same technology domains but ofthe different countries. For example, there are two large clusters of IC Engine Valve Control; one is mainly of Japanese applicants and the other is of US applicants. We also observed new technology domains; air conditioning system and multifunction switch system. These technology domains are often applied to cars in upper segments and these patents are mostly of German companies. As a result, similarity measurements with the new dataset highlight not only the combination found in the original dataset but also new pairs such as air conditioning system. Therefore, automobile patent clusters are apparently affected by such a country citation bias.
+
+In the approach A2, we obtained similar results with the original dataset. The patents in the maximum component
+
+increased from 69,281 to 188,032. In the original paper, 6 domains were highlighted and 5 of the 6 were found in the new dataset. The rest, Automatic Transmission domain, can be also highlighted by extracting a sub-cluster of a large automobile cluster. Therefore, the expansion of the dataset does not affect the result. But we must note that the number of the aircraft patents in the component also increased from 8577 to 9966, even though the same query is used for the aviation industry. Therefore, when we examine the details of such additional aviation patents, we expect to extract other technological domains where technologies in the aviation industry have high relatedness with technologies in the automobile industry.
+
+In the approach A3 with the new dataset, 18 IPC sub-classes were highlighted in the same condition to the original paper and 13 sub-classes were common to the original results. The difference occurred because the new dataset includes patents not from the automobile companies. It reduces the share of the automobile industry in technological domains such as painting and coating. However, if we set the threshold $r_i$ of highlighting conditions from 0.48 to 0.42, the same 38 sub-clusters were highlighted in both datasets.
+
+As shown above, the addition of patents from the automobile companies of other countries and from other companies increases the opportunity of technology transfer but doesn't
+
+![Figure](figures/figure_010.png)
+
+Fig. 4. Three methodologies on the DB-Combination model.
+
+---
+
+200
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+affect the conclusion derived by the original dataset. The results with the expanded dataset has the same tendency with those with the original dataset in the comparison between A1, A2, and A3 approaches, while some differences appear because of country citation bias in A1 and similarity setting in A3. These must be tackled by future research to test reliability and demonstrate effectiveness of our methodology in different contexts and datasets.
+
+We must also note that there is a criticism on using patent as the indicator of technological knowledge, even though patent is the best available indicator for R&D invention related to technology and outcomes of innovation activities ( OECD, 1994 ) . Knowledge of a company can be classified into tacit and explicit knowledge and a patent can represent only part of explicit knowledge ( Jones and Miller, 2007 ; Nonaka, 1994 , 1991 ) . Not all the explicit knowledge appears in the form of patents because first, the benefit from patenting must exceed the cost of registration and maintenance of the patent and secondly, some industries tend to not publish but keep secret technological knowledge ( Pavitt, 1985 ) .
+
+## 6. Summary
+
+To support practitioners to collect breadth technological knowledge in other industries and to combine it to their own knowledge, this paper proposed the DB-Combination model that considered the similarities of two technological knowledge domains and three different knowledge combinations in depth and breadth. We also investigated three methodologies A1, A2 and A3 to highlight the three combination pairs from patents by identifying technological knowledge sub-domains and measuring their similarities. The usability of the model and methodologies were demonstrated through a case study on the technological knowledge of the automobile industry and the aircraft industry. This paper used patent data as a source of technological knowledge and identified technological subdomains from patent data in two approaches; citation analysis (A1, A2) and IPC analysis (A3), and then measured the similarity between different sub-domains in three approaches; text similarity measurement (A1), existing intra-industrial citation tracing (A2) and IPC share similarity comparison (A3).
+
+A1 showed the potential of measuring DB-C combinations that are pairs of technology sub-domains with weak similarity that may bring a technological breakthrough if the similarities are of operating conditions or properties of technology. Using A2 it was possible to identify both past and current DB-T combinations that are a transfer of a unique technology subdomain of one industry to another. A3 measured pairs of DB-D combinations.
+
+The DB-Combination model with the integration of A1, A2 and A3 methodologies can become an effective innovation designing methodology that allows engineers and product managers to find useful technological knowledge from different industries and explore opportunities of technological breakthrough in depth or breadth.
+
+## Acknowledgment
+
+A part of this research is financially supported by theResearch Institute of Science and Technology for Society (RISTEX) of Japan Science and Technology Agency (JST).
+
+## References
+
+Adai, A.T., Date, S.V., Wieland, S., Marcotte, E.M., 2004. LGL: creating a map of protein function with an algorithm for visualizing very large biological networks. J. Mol. Biol. 340 (1), 179-190.
+
+Alavi, M., Leidner, D., 2001. Review: knowledge management and knowledge management systems: conceptual foundations and research issues. MIS Q. 25 (1), 107–136.
+
+Cantu, F.J., Ceballos, H.G., 2010. A multiagent knowledge and information network approach for managing research assets. Expert Syst. Appl. 37, 5272-5284.
+
+Criscuolo, P., 2006. The 'home advantage' effect and patent families. A comparison of OECD triadic patents, the USPTO and the EPO. Scientometrics 66 (1), 23-41.
+
+Cronin, B., 2001. Bibliometrics and beyond: some thoughts on web-based citation analysis. J. Inf. Sci. 27 (1), 1-7.
+
+Dosi, G., 1982. Technological paradigms and technological trajectories — a suggested interpretation of the determinants and directions of technical change. Res. Policy 11, 147–162.
+
+Eriksson, S., 2000. Technology spill-over from the aircraft industry: the case of Volvo Aero. Technovation 20, 653–664.
+
+Everett, J.E., Pecotich, A., 1993. Citation analysis mapping of journals in applied and clinical-psychology. J. Appl. Soc. Psychol. 23 (9), 750-766.
+
+Fleming, L., Sorenson, O., 2001. Technology as a complex adaptive system: evidence from patent data. Res. Policy 30, 1019-1039.
+
+Garfield, L., 1979. Is citation analysis a legitimate evaluation tool? Scientometrics 1 (4), 359-375.
+
+Gassmann, O., Zeschky, M., 2008. Opening up the solution space: the role of analogical thinking for breakthrough product innovation. Creat. Innov. Manag. 17 (2), 97–106.
+
+Gordon, M.D., Lindsay, R.K., Fan, W., 2002. Literature based discovery on the world wide web. ACM Trans. Internet Technol. 2 (4), 261-275.
+
+Griliches, Z., 1990. Patent statistics as economic indicators: a survey. J. Econ. Lit. 28 (4), 1661–1707 (90).
+
+Herrero, A., Corchado, E., Sáiz, L., Abraham, A., 2010. DIPKIP: a connectionist knowledge management system to identify knowledge deficits in practical cases. Comput. Intell. 26 (1), 26–56.
+
+Jaffe, A.B., 1986. Technological opportunity and spillovers of R&D evidence from firm's patents, profits, and market value. Am. Econ. Rev. 76 (5), 984-1001.
+
+Jones, B., Miller, B., 2007. Innovation diffusion in the new economy: the tacit component. Routledge, New York.
+
+Kajikawa, Y., Abe, K., Noda, S., 2006. Filling the gap between researchers studying different materials and different methods: a proposal for structured keywords. J. Inf. Sci. 32, 511–524.
+
+Kajikawa, Y., Yoshikawa, J., Takeda, Y., Matsushima, K., 2008. Tracking emerging technologies in energy research: toward a roadmap for sustainable energy. Technol. Forecast. Soc. Chang. 75 (6), 771–782.
+
+Katila, R., Ahuja, G., 2002. Something old, something new: a longitudinal study of search behavior and new product introduction. Acad. Manag. J. 45, 1183–1194.
+
+Kostoff, R.N., 1998. The use and misuse of citation analysis in research evaluation. Scientometrics 43, 27-43.
+
+Kostoff, R.N., 2008. Literature-related discovery (LRD): introduction and background. Technol. Forecast. Soc. Chang. 75 (2), 165–185.
+
+Leiponen, A., Helfat, C.E., 2010. Innovation objectives, knowledge sources, and the benefits of breadth. Strateg. Manag. J. 31 (2), 224–236.
+
+Leydesdorff, L., 2008. Patent classifications as Indicators of intellectual organization. J. Am. Soc. Inf. Sci. Technol. 59 (10), 1582-1597.
+
+Leydesdorff, L., Kushnir, D., Rafols, I., 2012. Interactive overlay maps for US patent (USPTO) data based on international patent classifications (IPC). Scientometrics http://dx.doi.org/10.1007/s11192-012-0923-2.
+
+Martin, B.R., 2012. The evolution of science policy and innovation studies. Res. Policy 41, 1219–1239.
+
+Meyer, M., 2000. What is special about patent citations? Differences between scientific and patent citations. Scientometrics 49, 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.
+
+Moorthy, S., Polley, D.E., 2010. Technological knowledge breadth and depth: performance impacts. J. Knowl. Manag. 14 (3), 359-377.
+
+Nakamura, H., Suzuki, S., Tomobe, H., Kajikawa, Y., Sakata, I., 2011a. Citation lag analysis in supply chain research. Scientometrics 87 (2), 221–232.
+
+Nakamura, H., Kajikawa, Y., Suzuki, S., 2011b. Innovation for sustainability in aviation: world challenges and visions. In: Nobre, F., Walker, D., Harris, R. (Eds.), Technological, Managerial and Organizational Core Competencies: Dynamic Innovation and Sustainable Advantage. IGI Global.
+
+Nakamura, H., Kajikawa, Y., Suzuki, S., 2012. Multi-level perspectives with technology readiness measures for aviation innovation. Sustain. Sci. http:// dx.doi.org/10.1007/s11625-012-0187-z .
+
+Narin, F., 1994. Patent bibliometrics. Scientometrics 30 (1), 147-155.
+
+---
+
+H. Nakamura et al. / Technological Forecasting & Social Change 94 (2015) 187-201
+
+201
+
+Nelson, R.R., Winter, S.G., 1977. In search of useful theory of innovation. Res. Policy 6, 36-76.
+
+Nerur, S.P., Rasheed, A.A., Natarajan, V., 2008. The intellectual structure of the strategic management field: an author co-citation analysis. Strateg. Manag. J. 29, 319–336.
+
+Nesta, L., Saviotti, P.P., 2005. Coherence of the knowledge base and the firm's innovative performance: evidence from the U.S. Pharm. Ind. 8 (1), 123-142.
+
+Newman, M.E.J., 2004. Fast algorithm for detecting community structure in networks. Phys. Rev. 69, 066133.
+
+Newman, M.E.J., Girvan, M., 2004. Finding and evaluating community structure in networks. Phys. Rev. 69, 026133.
+
+Nicolaisen, J., 2007. Citation analysis. Annu. Rev. Inf. Sci. Technol. 14 (3), 609–641.
+
+Nonaka, I., 1994. A dynamic theory of organizational knowledge creation. Organ. Sci. 5 (1), 14-37.
+
+Nonaka, I., November-December 1991. The knowledge-creating company. Harv. Bus. Rev. 96-104.
+
+OECD, 1994. The measurement of scientific and technological activities: using patent data as science and technology indicators. Organisation for Economic Co-operation and Development, (108 pages).
+
+Pavitt, K., 1985. Patent statistics as indicators of innovative activities: possibilities and problems. Scientometrics 7 (1-2), 77-99.
+
+Schilling, M.A., Green, E., 2011. Recombinant search and breakthrough idea generation: an analysis of high impact papers in the social sciences. Res. Policy 40 (10), 1321–1331.
+
+Schoen, A., Villard, L., Laurens, P., Cointet, J.-P., Heimeriks, G., Alkemade, F., 2012.
+
+The network structure of technological developments; technological distance as a walk on the technology map, 17th International Conference on Science and Technology Indicators.
+
+Schoenmakers, W., Duijsters, G.M., 2010. The technological origins of radical inventions. Res. Policy 39 (8), 1051-1059.
+
+Schumpeter, J., 1934. The Theory of Economic Development. Harvard University Press, Cambridge.
+
+Smalheiser, N.R., 2012. Literature-based discovery: beyond the ABCs. J. Am. Soc. Inf. Sci. Technol. 63 (2), 218–224.
+
+Smallheiser, N.R., Swanson, D.R., 1998. Using ARROWSMITH: a computerassisted approach to formulating and assessing scientific hypothesis. Comput. Methods Prog. Biomed. 57 (3), 149–153.
+
+Soete, L.G., Wyatt, S.M.E., 1983. The use of foreign patenting as an internationally comparable science and technology output indicator. Scientometrics 5 (1), 31-54.
+
+Tijssen, R.J.W., 1992. A quantitative assessment of interdisciplinary structures in science and technology: co-classification analysis of energy research. Res. Policy 21 (1), 27–44.
+
+Dr. Hiroko Nakamura is a Project Researcher at the Centre for Aviation Innovation Research and also at the Global Leader Program for Social Design and Management at the University of Tokyo. She previously worked for Nissan Motor Co. Ltd., as a product planner. She received her bachelor's, and master's degree, and Ph.D. from the same university and a Special Master's degree from the Ecole Centrale Paris. Her research interests include development of methodology for technology and innovation management.
+
+Dr. Shinji Suzuki is a Professor at the Department of Aeronautics and Astronautics and the Director of the Center for Aviation Innovation Research at the University of Tokyo. He received his Ph.D. degree in Engineering from the University of Tokyo in 1986, after his research career at Toyota's Central R&D Labs, Inc., in Japan. He received his bachelor's, and master's degrees from the same university. His main research interests are in the design and control aspects of air safety and unmanned aerial vehicles. He is currently an Executive Committee Member of ICAS.
+
+Dr. Ichiro Sakata is a Professor at the Department of Technology Management for Innovation, the Director of Innovation Policy Research Center and the cohead of Presidential Endowed Chair for Electricity Network Innovation by Digital Grid at the University of Tokyo. He received his bachelor's degree and Ph.D. from the same university, and master's degree from Brandeis University. He has 20 years'of working experience at the Japanese Ministry of Economy, Trade and Industry (METI). His research interests include technology management, technology roadmap and innovation network. He has published more than 100 papers in peer reviewed journals and conference proceedings.
+
+Dr. Yuya Kajikawa is an Associate Professor at Graduate School of Innovation Management, Tokyo Institute of Technology. He is also a visiting researcher at Graduate School of Engineering, The University of Tokyo. He received his bachelor's, and master's degrees, and Ph.D. from the University of Tokyo. His research interests include development of methodology for technology and innovation management. He has a number of publications in peer-reviewed journals and conference proceedings, which cover a variety of disciplines including engineering, information science, environmental science, and technology and innovation management. He serves as an Associate Editor of Technological Forecasting and Social Change.
+

processed/00253_W2013293823/document.md

@@ -0,0 +1,134 @@
+correspondence
+
+# Unrestricted free access works and must continue
+
+Bioinformatics researchers shouldn't need coercion to act responsibly and collegially.
+
+Sir — The policy on release of unpublished data from large genome centres has generated considerable discussion and some confusion, as your Editorial “Sacrifice for the greater good?” makes plain ( Nature 421, 875; 2003). In our view, data sets from large, centralized, expensive genome data-collection projects should be freely available to the entire scientific community, immediately and with no restrictions or conditions.
+
+Our position is that pre-publication release of large genome data sets is a special case, and not a principle that should be applied “throughout the world of biology”, as was asserted in your editorial. Large genome sequencing has become an expensive, factory-style operation, in which economies of scale can only be realized by very large centres. Large data production centres, established and supported by the scientific community, represent a different model of science from traditional investigator-initiated projects. We argue that they need to operate under different rules.
+
+The broader scientific community supports the highly centralized model represented by the US large-scale genome centres (funded via the National Institutes of Health and the Department of Energy) on the condition that everyone in this community gets equal access to the data. If this is the case, everyone wins: large data sets are generated at the lowest cost and greatest speed, and scientific work progresses on multiple fronts without delay. In contrast, if genome centres restrict their data and get preferential access to it, then some members of the community will no longer support monopolistic funding models (in which large centres sequence one genome after another without peer review of each project). Instead, they will demand the right to compete with these empires, especially for the most scientifically desirable genomes. Other scientists, especially bioinformaticians, will seek to relocate to the centres to gain the advantage of early data access. Data restrictions will therefore promote factionalization where we should be seeking efficiencies of scale, and centralization where we should be promoting diversity.
+
+We agree with your editorial that the proposed new policy, recently released for comment by the US genome centres (see www.genome.gov/page.cfm?pageID= 10506537), is ambiguous. It states that genome sequence data “should be available
+
+for all to use without restriction”. This statement, which notably uses the word “should” rather than “must”, is qualified by a lengthy discussion of conditions, including a reference to the sequence producers’ interest in “the first peerreviewed published analysis of the results of the sequencing project”. This reflects the real concern of the genome centres that prepublication data access may put the scientists there at a competitive disadvantage. We understand these concerns, but we believe that the qualifying discussion cannot coexist with the principle of “without restriction”. We propose that these qualifications are simply dropped to avoid confusion among data consumers, journals and journal reviewers. The Human Genome Project has been a spectacularly successful demonstration that the “Bermuda rules” of free access without restriction do work, for everyone.
+
+As bioinformaticians, we have an important role in this process. We reaffirm our own commitment to respecting the
+
+goals of the scientists at the genome centres, who should be consulted as part of any large-scale analysis of unpublished genome data, and included as collaborators where appropriate. It is a serious problem that these invaluable centres feel compelled to coerce such simple scientific courtesy from our community. We encourage our peers in bioinformatics to act responsibly, cooperatively and collegially, to help to assure open, unrestricted, immediate release from large community-driven data-collection projects.
+
+Steven Salzberg* , Ewan Birney†, Sean Eddy‡, Owen White*
+
+*The Institute for Genomic Research, 9712 Medical Center Drive, Rockville, Maryland 20850, USA salzberg@tigr.org †European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SD, UK ‡Howard Hughes Medical Institute and Department of Genetics, Washington University School of Medicine, 4566 Scott Avenue, St Louis, Missouri 63110, USA
+
+## Watching an IT icon slide into history
+
+Sir — In the famous photographs of James Watson and Francis Crick with their DNA model (reproduced in Nature 421, 15 & 417; 2003) a small icon of information technology can be seen in Crick's right hand — a slide rule. (One wonders, incidentally, why this was shown in the photograph instead of the “measuring stick” which, together with a plumb line, they used “to obtain the relative positions of all atoms” , as described by Watson in The Double Helix — anecdote has it that Crick told the
+
+insistent photographer that the slide rule was irrelevant.)
+
+![Figure](figures/figure_015.png)
+
+Ruled out: unlike Francis Crick, it seems that the slide rule is slowly being forgotten.
+
+At Newcastle University we have been teaching biology students the use of computers for nearly three decades, beginning each year with a brief account of the staggering advances in computational power since the middle of the twentieth century. Each year, we show them an example of what Crick held. It seems fitting, on the fiftieth anniversary of the reconstruction of DNA, and at the advent of the rise of computational biology (C. Surridge Nature 420, 205; 2002), to report that this year, for the first time, none of our first-year students recognizes a slide rule.
+
+W. John Cram School of Biology, University of Newcastle upon Tyne, Newcastle upon Tyne NE1 7RU, UK
+
+## Secrecy is increasing in step with competition
+
+Sir — Various authors and reports have recently claimed (for example, refs 1–3) that the increasing commercialization of academic science has led to an increase in secrecy. However, our comparison of two surveys of experimental biologists, mathematicians and physicists, conducted about 30 years apart, suggests a more
+
+NATURE | VOL.422|24 APRIL 2003 | www.nature.com/nature
+
+2003 Nature Publishing Group
+
+801
+
+A. BARRINGTON BROWN/SPL
+
+---
+
+correspondence
+
+complicated and interesting picture.
+
+As feared, secrecy (measured as unwillingness to discuss ongoing research with those outside the research group) has increased.
+
+In 1966 (ref. 4), 50% of 1,042 respondents reported feeling safe in talking with all others about their current research, but by 1998, when we surveyed 202 scientists from the same three fields (details of methods and results available from J.W.), the equivalent number was 26%. Experimental biologists have become particularly secretive, with only 14% willing to talk openly about their current research in 1998. Secrecy is strongly predicted by scientific competition (measured as concern over having one's research results anticipated). The effects of commercial activity, on the other hand, are quite mixed. Patenting has no effect; industry funding is associated with greater secrecy; but having industry collaborators is associated with less secrecy.
+
+These university–industry collaborations can be viewed as part of a strategy to share findings and expertise with the wider scientific and technical community. For companies, timeliness and customization of information are often more important than exclusivity, so they are willing to tolerate, even encourage, their academic collaborators' participation in the shared conversation of a scientific field, thereby giving the company access to the whole community's expertise. In contrast to these collaborations, industry funding alone is often associated with a university laboratory acting as a subcontractor to a company's R&D project, and may produce associated secretive behaviour.
+
+Thus, there is reason to believe that secrecy has increased among academic scientists, but that the focus on commercialization as the cause may be misplaced. Although commercial activity may reduce formal activities such as publication or sharing of materials, it may have fewer negative effects on informal communication among researchers. As this informal communication is significant in transferring information to companies 5 and is at least as important as publication for distributing information among scientists, this is encouraging news.
+
+Although it is right to raise concerns about the negative effects of publication restrictions associated with industry funding, we should not conclude that university–industry linkages per se produce unhealthy levels of secretiveness among academic scientists. Instead, it may be better to focus on alleviating some of the negative consequences of scientific competition.
+
+Recent increases in US government funding for science, if they are sustained, may help to lower the intensity of competition, as well as the dependence on industry funding, and thereby reduce secrecy. Furthermore, although we need to be wary of the strings attached to industry funding, university– industry collaborative research should be encouraged.
+
+We thank Lowell Hargens for providing field-level data from ref. 4.
+
+John P. Walsh*†, Wei Hong†
+
+*Research Center for Advanced Economic Engineering, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan walsh@nee.u-tokyo.ac.jp.
+
+†Department of Sociology, University of Illinois, Chicago, Illinois 60607-7140 USA
+
+1. Blumenthal, D. et al. J. Ass. Med. Assoc. 277, 1224-1228 (1997).
+
+2. Cook-Deegan, R. M. & McCormack, S. J. Science 293, 217 (2001).
+
+3. Campbell, E. G. et al. J. Am. Med. Assoc. 287, 473-480 (2002).
+
+4. Hagstrom, W. O. Am. Sociol. Rev. 39, 1-18 (1974).
+
+5. Cohen, W. M., Nelson, R. R. & Walsh, J. P. Management Sci. 48, 1-23 (2002).
+
+## Pre-genomics training  hinders Indian biotech
+
+Sir — Ashok Parthasarathi in his Commentary “India: a champion of new technologies” (ref. 1) rightly affirms India’s bid to play a leading part in global technology developments. India’s technological independence has come mainly in the physical sciences, such as space research, telecommunications, software, defence systems, energy and supercomputing.
+
+Although Parthasarathi lists biotechnology as a key ‘champion technology’, a tremendous boost is needed for it to make a global impact.
+
+India shares this predicament with other developing countries that have a vast research force but the training and infrastructure of pre-genomics days, and little experience of the research skills required for post-genomic biology. With the genome sequence data for several organisms, including humans, now available, the rules of knowledgebased commercial ventures have changed and an era of ‘omics’ — genomics, proteomics and so on — has emerged 2 , with functional genomics as the new ‘ mantra’ .
+
+In order to become a champion of biotechnology, India needs a paradigm shift in the organization of research and training, priming research institutions and universities to change gear and meet global challenges. Two aspects need urgent attention: establishing functionalgenomics centres for biotechnology and
+
+related basic science research, as is done in China 3 ; and training researchers in state-of-the-art skills, along the lines of initiatives by the European Molecular Biology Laboratory.
+
+All the stakeholders — in the fields of policy, administration, science and industry — have to address this problem and give a directional nudge to research initiatives.
+
+Nature has covered some of these issues recently, for example the low impact factors of Indian biological-science journals 4 ; the urgent need to explore the best scientific options for sustainable development by regional centres of the International Council of Science 5 ; and the effect of post-genomics research on traditional methods of food production 6 . Informal articles on topics such as these serve as a compass for framing policies and making course corrections during implementation.
+
+R. M. Ranganath Department of Botany, Bangalore University, Jnanabharathi Campus, Bangalore 560056, India
+
+L. Parthasarathi, A. Nature 422, 17-18 (2003).
+
+2. Palsson, B. Nature Biotechnol. 20, 649-650 (2002).
+
+3. Cyranoski, D. Nature 410, 10-12 (2001).
+
+4. Vohora, S. B. & Vohora, D. Nature 412, 583 (2001).
+
+5. Clarke, T. Nature 420, 733 (2002).
+
+6. Knight, J. Nature 421, 568-570 (2003).
+
+## Should tobacco ban rule out governments too?
+
+Sir — I was interested to read in your News story “Academics fume as university refuses to reject tobacco dollars ” ( Nature 422, 361; 2003) a scientist at the University of California, San Francisco, quoted as saying it is immoral to accept research funds from tobacco companies because “it is not appropriate to take money from an industry that kills 5 million people worldwide and constantly lies ” .
+
+I guess that means that no research funds should be accepted from the US Department of Defense, nor from its corporate headquarters — the US government. Other national governments would presumably also be banned from supporting research under these new politically correct guidelines.
+
+Richard S. Jope Department of Psychiatry, Sparks Center 1057, 1720 7th Avenue South, University of Alabama at Birmingham, Birmingham, Alabama 35294-0017, USA
+
+### correspondence
+
+Contributions to Correspondence may be submitted to corres@nature.com. They should be no longer than 500 words, and ideally shorter. Published contributions are edited.
+
+802
+
+2003 Nature Publishing Group
+
+NATURE | VOL.422 | 24 APRIL 2003 | www.nature.com/nature
+

processed/00253_W2013293823/metadata.json

@@ -0,0 +1,744 @@
+{
+  "source_file": "/tmp/tmpdet1ob0m/00253_W2013293823_Secrecy_is_increasing_in_step_with_competition.pdf",
+  "total_pages": 2,
+  "pages": [
+    {
+      "page_number": 1,
+      "elements": [
+        {
+          "bbox": [
+            495,
+            22,
+            643,
+            41
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "correspondence"
+        },
+        {
+          "bbox": [
+            54,
+            54,
+            619,
+            87
+          ],
+          "label": "sec_0",
+          "reading_order": 1,
+          "text": "Unrestricted free access works and must continue"
+        },
+        {
+          "bbox": [
+            54,
+            89,
+            610,
+            108
+          ],
+          "label": "para",
+          "reading_order": 2,
+          "text": "Bioinformatics researchers shouldn't need coercion to act responsibly and collegially."
+        },
+        {
+          "bbox": [
+            56,
+            115,
+            239,
+            249
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "Sir — The policy on release of unpublished\ndata from large genome centres has\ngenerated considerable discussion and\nsome confusion, as your Editorial\n“Sacrifice for the greater good?” makes\nplain ( Nature 421, 875; 2003). In our view,\ndata sets from large, centralized, expensive\ngenome data-collection projects should be\nfreely available to the entire scientific\ncommunity, immediately and with no\nrestrictions or conditions."
+        },
+        {
+          "bbox": [
+            56,
+            249,
+            240,
+            430
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "Our position is that pre-publication\nrelease of large genome data sets is a special\ncase, and not a principle that should be\napplied “throughout the world of biology”,\nas was asserted in your editorial. Large\ngenome sequencing has become an\nexpensive, factory-style operation, in\nwhich economies of scale can only be\nrealized by very large centres. Large data\nproduction centres, established and\nsupported by the scientific community,\nrepresent a different model of science from\ntraditional investigator-initiated projects.\nWe argue that they need to operate under\ndifferent rules."
+        },
+        {
+          "bbox": [
+            56,
+            430,
+            239,
+            791
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "The broader scientific community\nsupports the highly centralized model\nrepresented by the US large-scale genome\ncentres (funded via the National Institutes\nof Health and the Department of Energy)\non the condition that everyone in this\ncommunity gets equal access to the data.\nIf this is the case, everyone wins: large data\nsets are generated at the lowest cost and\ngreatest speed, and scientific work\nprogresses on multiple fronts without\ndelay. In contrast, if genome centres\nrestrict their data and get preferential\naccess to it, then some members of the\ncommunity will no longer support\nmonopolistic funding models (in which\nlarge centres sequence one genome after\nanother without peer review of each\nproject). Instead, they will demand the\nright to compete with these empires,\nespecially for the most scientifically\ndesirable genomes. Other scientists,\nespecially bioinformaticians, will seek\nto relocate to the centres to gain the\nadvantage of early data access. Data\nrestrictions will therefore promote\nfactionalization where we should\nbe seeking efficiencies of scale, and\ncentralization where we should be\npromoting diversity."
+        },
+        {
+          "bbox": [
+            56,
+            791,
+            241,
+            864
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "We agree with your editorial that the\nproposed new policy, recently released for\ncomment by the US genome centres (see\nwww.genome.gov/page.cfm?pageID=\n10506537), is ambiguous. It states that\ngenome sequence data “should be available"
+        },
+        {
+          "bbox": [
+            256,
+            115,
+            439,
+            394
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "for all to use without restriction”. This\nstatement, which notably uses the word\n“should” rather than “must”, is qualified\nby a lengthy discussion of conditions,\nincluding a reference to the sequence\nproducers’ interest in “the first peer-\nreviewed published analysis of the results\nof the sequencing project”. This reflects\nthe real concern of the genome centres\nthat prepublication data access may\nput the scientists there at a competitive\ndisadvantage. We understand these\nconcerns, but we believe that the\nqualifying discussion cannot coexist with\nthe principle of “without restriction”. We\npropose that these qualifications are\nsimply dropped to avoid confusion among\ndata consumers, journals and journal\nreviewers. The Human Genome Project\nhas been a spectacularly successful\ndemonstration that the “Bermuda rules”\nof free access without restriction do work,\nfor everyone."
+        },
+        {
+          "bbox": [
+            256,
+            394,
+            439,
+            431
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "As bioinformaticians, we have an\nimportant role in this process. We reaffirm\nour own commitment to respecting the"
+        },
+        {
+          "bbox": [
+            456,
+            116,
+            629,
+            286
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "goals of the scientists at the genome\ncentres, who should be consulted as\npart of any large-scale analysis of\nunpublished genome data, and included\nas collaborators where appropriate. It is\na serious problem that these invaluable\ncentres feel compelled to coerce such\nsimple scientific courtesy from our\ncommunity. We encourage our peers\nin bioinformatics to act responsibly,\ncooperatively and collegially, to help to\nassure open, unrestricted, immediate\nrelease from large community-driven\ndata-collection projects."
+        },
+        {
+          "bbox": [
+            456,
+            286,
+            593,
+            310
+          ],
+          "label": "para",
+          "reading_order": 10,
+          "text": "Steven Salzberg*\n, Ewan Birney†,\nSean Eddy‡, Owen White*"
+        },
+        {
+          "bbox": [
+            456,
+            310,
+            638,
+            430
+          ],
+          "label": "fnote",
+          "reading_order": 11,
+          "text": "*The Institute for Genomic Research, 9712 Medical\nCenter Drive, Rockville, Maryland 20850, USA\nsalzberg@tigr.org\n†European Bioinformatics Institute,\nWellcome Trust Genome Campus, Hinxton,\nCambridge CB10 1SD, UK\n‡Howard Hughes Medical Institute and\nDepartment of Genetics, Washington University\nSchool of Medicine, 4566 Scott Avenue,\nSt Louis, Missouri 63110, USA"
+        },
+        {
+          "bbox": [
+            255,
+            450,
+            402,
+            494
+          ],
+          "label": "sec_1",
+          "reading_order": 12,
+          "text": "Watching an IT icon\nslide into history"
+        },
+        {
+          "bbox": [
+            256,
+            500,
+            433,
+            659
+          ],
+          "label": "para",
+          "reading_order": 13,
+          "text": "Sir — In the famous photographs of\nJames Watson and Francis Crick with\ntheir DNA model (reproduced in Nature\n421, 15 & 417; 2003) a small icon of\ninformation technology can be seen in\nCrick's right hand — a slide rule. (One\nwonders, incidentally, why this was\nshown in the photograph instead of the\n“measuring stick” which, together with\na plumb line, they used “to obtain\nthe relative positions of all atoms”\n, as\ndescribed by Watson in The Double Helix\n— anecdote has it that Crick told the"
+        },
+        {
+          "bbox": [
+            456,
+            452,
+            632,
+            479
+          ],
+          "label": "para",
+          "reading_order": 14,
+          "text": "insistent photographer that the slide rule\nwas irrelevant.)"
+        },
+        {
+          "bbox": [
+            255,
+            659,
+            442,
+            838
+          ],
+          "label": "fig",
+          "reading_order": 15,
+          "text": "![Figure](figures/figure_015.png)",
+          "figure_path": "/tmp/pdf_output/markdown/figures/figure_015.png"
+        },
+        {
+          "bbox": [
+            256,
+            839,
+            440,
+            864
+          ],
+          "label": "cap",
+          "reading_order": 16,
+          "text": "Ruled out: unlike Francis Crick, it seems that the\nslide rule is slowly being forgotten."
+        },
+        {
+          "bbox": [
+            456,
+            478,
+            640,
+            659
+          ],
+          "label": "para",
+          "reading_order": 17,
+          "text": "At Newcastle University we have been\nteaching biology students the use of\ncomputers for nearly three decades,\nbeginning each year with a brief account of\nthe staggering advances in computational\npower since the middle of the twentieth\ncentury. Each year, we show them an\nexample of what Crick held. It seems\nfitting, on the fiftieth anniversary of the\nreconstruction of DNA, and at the\nadvent of the rise of computational\nbiology (C. Surridge Nature 420, 205;\n2002), to report that this year, for the first\ntime, none of our first-year students\nrecognizes a slide rule."
+        },
+        {
+          "bbox": [
+            456,
+            659,
+            624,
+            696
+          ],
+          "label": "para",
+          "reading_order": 18,
+          "text": "W. John Cram\nSchool of Biology, University of Newcastle upon\nTyne, Newcastle upon Tyne NE1 7RU, UK"
+        },
+        {
+          "bbox": [
+            456,
+            718,
+            632,
+            762
+          ],
+          "label": "sec_1",
+          "reading_order": 19,
+          "text": "Secrecy is increasing in\nstep with competition"
+        },
+        {
+          "bbox": [
+            456,
+            766,
+            639,
+            864
+          ],
+          "label": "para",
+          "reading_order": 20,
+          "text": "Sir — Various authors and reports have\nrecently claimed (for example, refs 1–3)\nthat the increasing commercialization of\nacademic science has led to an increase\nin secrecy. However, our comparison of\ntwo surveys of experimental biologists,\nmathematicians and physicists, conducted\nabout 30 years apart, suggests a more"
+        },
+        {
+          "bbox": [
+            56,
+            870,
+            251,
+            883
+          ],
+          "label": "foot",
+          "reading_order": 21,
+          "text": "NATURE | VOL.422|24 APRIL 2003 | www.nature.com/nature"
+        },
+        {
+          "bbox": [
+            291,
+            873,
+            423,
+            886
+          ],
+          "label": "foot",
+          "reading_order": 22,
+          "text": "2003 Nature Publishing Group"
+        },
+        {
+          "bbox": [
+            620,
+            870,
+            640,
+            882
+          ],
+          "label": "foot",
+          "reading_order": 23,
+          "text": "801"
+        },
+        {
+          "bbox": [
+            646,
+            660,
+            656,
+            746
+          ],
+          "label": "watermark",
+          "reading_order": 24,
+          "text": "A. BARRINGTON BROWN/SPL"
+        }
+      ]
+    },
+    {
+      "page_number": 2,
+      "elements": [
+        {
+          "bbox": [
+            52,
+            22,
+            199,
+            42
+          ],
+          "label": "header",
+          "reading_order": 0,
+          "text": "correspondence"
+        },
+        {
+          "bbox": [
+            53,
+            56,
+            208,
+            70
+          ],
+          "label": "half_para",
+          "reading_order": 1,
+          "text": "complicated and interesting picture."
+        },
+        {
+          "bbox": [
+            52,
+            70,
+            233,
+            117
+          ],
+          "label": "para",
+          "reading_order": 2,
+          "text": "As feared, secrecy (measured as\nunwillingness to discuss ongoing research\nwith those outside the research group)\nhas increased."
+        },
+        {
+          "bbox": [
+            52,
+            117,
+            237,
+            346
+          ],
+          "label": "para",
+          "reading_order": 3,
+          "text": "In 1966 (ref. 4), 50% of 1,042\nrespondents reported feeling safe in\ntalking with all others about their current\nresearch, but by 1998, when we surveyed\n202 scientists from the same three fields\n(details of methods and results available\nfrom J.W.), the equivalent number was\n26%. Experimental biologists have become\nparticularly secretive, with only 14%\nwilling to talk openly about their current\nresearch in 1998. Secrecy is strongly\npredicted by scientific competition\n(measured as concern over having one's\nresearch results anticipated). The effects\nof commercial activity, on the other hand,\nare quite mixed. Patenting has no effect;\nindustry funding is associated with greater\nsecrecy; but having industry collaborators\nis associated with less secrecy."
+        },
+        {
+          "bbox": [
+            52,
+            346,
+            235,
+            562
+          ],
+          "label": "para",
+          "reading_order": 4,
+          "text": "These university–industry collabo-\nrations can be viewed as part of a strategy\nto share findings and expertise with the\nwider scientific and technical community.\nFor companies, timeliness and\ncustomization of information are often\nmore important than exclusivity, so they\nare willing to tolerate, even encourage,\ntheir academic collaborators' participation\nin the shared conversation of a scientific\nfield, thereby giving the company access\nto the whole community's expertise.\nIn contrast to these collaborations,\nindustry funding alone is often associated\nwith a university laboratory acting as a\nsubcontractor to a company's R&D\nproject, and may produce associated\nsecretive behaviour."
+        },
+        {
+          "bbox": [
+            52,
+            562,
+            234,
+            744
+          ],
+          "label": "para",
+          "reading_order": 5,
+          "text": "Thus, there is reason to believe that\nsecrecy has increased among academic\nscientists, but that the focus on\ncommercialization as the cause may be\nmisplaced. Although commercial activity\nmay reduce formal activities such as\npublication or sharing of materials, it\nmay have fewer negative effects on\ninformal communication among\nresearchers. As this informal communi-\ncation is significant in transferring\ninformation to companies 5 and is at\nleast as important as publication for\ndistributing information among scientists,\nthis is encouraging news."
+        },
+        {
+          "bbox": [
+            52,
+            744,
+            228,
+            864
+          ],
+          "label": "para",
+          "reading_order": 6,
+          "text": "Although it is right to raise concerns\nabout the negative effects of publication\nrestrictions associated with industry\nfunding, we should not conclude that\nuniversity–industry linkages per se\nproduce unhealthy levels of secretiveness\namong academic scientists. Instead, it\nmay be better to focus on alleviating\nsome of the negative consequences of\nscientific competition."
+        },
+        {
+          "bbox": [
+            253,
+            56,
+            438,
+            166
+          ],
+          "label": "para",
+          "reading_order": 7,
+          "text": "Recent increases in US government\nfunding for science, if they are sustained,\nmay help to lower the intensity of\ncompetition, as well as the dependence\non industry funding, and thereby reduce\nsecrecy. Furthermore, although we need to\nbe wary of the strings attached to industry\nfunding, university– industry collaborative\nresearch should be encouraged."
+        },
+        {
+          "bbox": [
+            253,
+            166,
+            436,
+            189
+          ],
+          "label": "para",
+          "reading_order": 8,
+          "text": "We thank Lowell Hargens for providing\nfield-level data from ref. 4."
+        },
+        {
+          "bbox": [
+            253,
+            189,
+            371,
+            202
+          ],
+          "label": "para",
+          "reading_order": 9,
+          "text": "John P. Walsh*†, Wei Hong†"
+        },
+        {
+          "bbox": [
+            253,
+            202,
+            426,
+            250
+          ],
+          "label": "fnote",
+          "reading_order": 10,
+          "text": "*Research Center for Advanced Economic\nEngineering, University of Tokyo, 4-6-1 Komaba,\nMeguro-ku, Tokyo 153-8904, Japan\nwalsh@nee.u-tokyo.ac.jp."
+        },
+        {
+          "bbox": [
+            253,
+            250,
+            422,
+            273
+          ],
+          "label": "fnote",
+          "reading_order": 11,
+          "text": "†Department of Sociology, University of Illinois,\nChicago, Illinois 60607-7140 USA"
+        },
+        {
+          "bbox": [
+            254,
+            275,
+            383,
+            294
+          ],
+          "label": "reference",
+          "reading_order": 12,
+          "text": "1. Blumenthal, D. et al. J. Ass. Med. Assoc. 277,\n1224-1228 (1997)."
+        },
+        {
+          "bbox": [
+            254,
+            295,
+            408,
+            313
+          ],
+          "label": "reference",
+          "reading_order": 13,
+          "text": "2. Cook-Deegan, R. M. & McCormack, S. J. Science 293,\n217 (2001)."
+        },
+        {
+          "bbox": [
+            254,
+            314,
+            430,
+            323
+          ],
+          "label": "reference",
+          "reading_order": 14,
+          "text": "3. Campbell, E. G. et al. J. Am. Med. Assoc. 287, 473-480 (2002)."
+        },
+        {
+          "bbox": [
+            254,
+            323,
+            399,
+            333
+          ],
+          "label": "reference",
+          "reading_order": 15,
+          "text": "4. Hagstrom, W. O. Am. Sociol. Rev. 39, 1-18 (1974)."
+        },
+        {
+          "bbox": [
+            254,
+            333,
+            424,
+            350
+          ],
+          "label": "reference",
+          "reading_order": 16,
+          "text": "5. Cohen, W. M., Nelson, R. R. & Walsh, J. P. Management Sci.\n48, 1-23 (2002)."
+        },
+        {
+          "bbox": [
+            252,
+            381,
+            419,
+            425
+          ],
+          "label": "sec_1",
+          "reading_order": 17,
+          "text": "Pre-genomics training\n\nhinders Indian biotech"
+        },
+        {
+          "bbox": [
+            253,
+            429,
+            438,
+            551
+          ],
+          "label": "para",
+          "reading_order": 18,
+          "text": "Sir — Ashok Parthasarathi in his\nCommentary “India: a champion of new\ntechnologies” (ref. 1) rightly affirms India’s\nbid to play a leading part in global\ntechnology developments. India’s\ntechnological independence has come\nmainly in the physical sciences, such as\nspace research, telecommunications,\nsoftware, defence systems, energy and\nsupercomputing."
+        },
+        {
+          "bbox": [
+            253,
+            551,
+            433,
+            599
+          ],
+          "label": "para",
+          "reading_order": 19,
+          "text": "Although Parthasarathi lists biotech-\nnology as a key ‘champion technology’, a\ntremendous boost is needed for it to make\na global impact."
+        },
+        {
+          "bbox": [
+            253,
+            599,
+            430,
+            767
+          ],
+          "label": "para",
+          "reading_order": 20,
+          "text": "India shares this predicament with\nother developing countries that have\na vast research force but the training\nand infrastructure of pre-genomics\ndays, and little experience of the research\nskills required for post-genomic biology.\nWith the genome sequence data for\nseveral organisms, including humans,\nnow available, the rules of knowledge-\nbased commercial ventures have changed\nand an era of ‘omics’\n— genomics,\nproteomics and so on — has emerged 2 ,\nwith functional genomics as the new\n‘\nmantra’\n."
+        },
+        {
+          "bbox": [
+            253,
+            767,
+            430,
+            865
+          ],
+          "label": "para",
+          "reading_order": 21,
+          "text": "In order to become a champion of\nbiotechnology, India needs a paradigm\nshift in the organization of research and\ntraining, priming research institutions\nand universities to change gear and meet\nglobal challenges. Two aspects need\nurgent attention: establishing functional-\ngenomics centres for biotechnology and"
+        },
+        {
+          "bbox": [
+            453,
+            56,
+            622,
+            117
+          ],
+          "label": "para",
+          "reading_order": 22,
+          "text": "related basic science research, as is done\nin China 3 ; and training researchers in\nstate-of-the-art skills, along the lines of\ninitiatives by the European Molecular\nBiology Laboratory."
+        },
+        {
+          "bbox": [
+            453,
+            117,
+            627,
+            178
+          ],
+          "label": "para",
+          "reading_order": 23,
+          "text": "All the stakeholders — in the fields\nof policy, administration, science and\nindustry — have to address this problem\nand give a directional nudge to research\ninitiatives."
+        },
+        {
+          "bbox": [
+            453,
+            178,
+            635,
+            335
+          ],
+          "label": "para",
+          "reading_order": 24,
+          "text": "Nature has covered some of these issues\nrecently, for example the low impact\nfactors of Indian biological-science\njournals 4 ; the urgent need to explore the\nbest scientific options for sustainable\ndevelopment by regional centres of the\nInternational Council of Science 5 ; and\nthe effect of post-genomics research on\ntraditional methods of food production 6 .\nInformal articles on topics such as these\nserve as a compass for framing policies\nand making course corrections during\nimplementation."
+        },
+        {
+          "bbox": [
+            453,
+            335,
+            613,
+            382
+          ],
+          "label": "para",
+          "reading_order": 25,
+          "text": "R. M. Ranganath\nDepartment of Botany, Bangalore University,\nJnanabharathi Campus,\nBangalore 560056, India"
+        },
+        {
+          "bbox": [
+            453,
+            383,
+            582,
+            393
+          ],
+          "label": "reference",
+          "reading_order": 26,
+          "text": "L. Parthasarathi, A. Nature 422, 17-18 (2003)."
+        },
+        {
+          "bbox": [
+            453,
+            393,
+            601,
+            403
+          ],
+          "label": "reference",
+          "reading_order": 27,
+          "text": "2. Palsson, B. Nature Biotechnol. 20, 649-650 (2002)."
+        },
+        {
+          "bbox": [
+            453,
+            403,
+            575,
+            411
+          ],
+          "label": "reference",
+          "reading_order": 28,
+          "text": "3. Cyranoski, D. Nature 410, 10-12 (2001)."
+        },
+        {
+          "bbox": [
+            453,
+            411,
+            602,
+            421
+          ],
+          "label": "reference",
+          "reading_order": 29,
+          "text": "4. Vohora, S. B. & Vohora, D. Nature 412, 583 (2001)."
+        },
+        {
+          "bbox": [
+            453,
+            421,
+            557,
+            431
+          ],
+          "label": "reference",
+          "reading_order": 30,
+          "text": "5. Clarke, T. Nature 420, 733 (2002)."
+        },
+        {
+          "bbox": [
+            453,
+            431,
+            571,
+            441
+          ],
+          "label": "reference",
+          "reading_order": 31,
+          "text": "6. Knight, J. Nature 421, 568-570 (2003)."
+        },
+        {
+          "bbox": [
+            452,
+            467,
+            632,
+            512
+          ],
+          "label": "sec_1",
+          "reading_order": 32,
+          "text": "Should tobacco ban rule\nout governments too?"
+        },
+        {
+          "bbox": [
+            453,
+            513,
+            636,
+            635
+          ],
+          "label": "para",
+          "reading_order": 33,
+          "text": "Sir — I was interested to read in your News\nstory\n“Academics fume as university\nrefuses to reject tobacco dollars\n” ( Nature\n422, 361; 2003) a scientist at the University\nof California, San Francisco, quoted as\nsaying it is immoral to accept research\nfunds from tobacco companies because\n“it is not appropriate to take money from\nan industry that kills 5 million people\nworldwide and constantly lies\n”\n."
+        },
+        {
+          "bbox": [
+            453,
+            635,
+            633,
+            732
+          ],
+          "label": "para",
+          "reading_order": 34,
+          "text": "I guess that means that no research\nfunds should be accepted from the US\nDepartment of Defense, nor from its\ncorporate headquarters — the US\ngovernment. Other national governments\nwould presumably also be banned from\nsupporting research under these new\npolitically correct guidelines."
+        },
+        {
+          "bbox": [
+            453,
+            732,
+            611,
+            791
+          ],
+          "label": "para",
+          "reading_order": 35,
+          "text": "Richard S. Jope\nDepartment of Psychiatry,\nSparks Center 1057, 1720 7th Avenue South,\nUniversity of Alabama at Birmingham,\nBirmingham, Alabama 35294-0017, USA"
+        },
+        {
+          "bbox": [
+            453,
+            803,
+            519,
+            815
+          ],
+          "label": "sec_2",
+          "reading_order": 36,
+          "text": "correspondence"
+        },
+        {
+          "bbox": [
+            453,
+            815,
+            629,
+            863
+          ],
+          "label": "para",
+          "reading_order": 37,
+          "text": "Contributions to Correspondence may be submitted\nto corres@nature.com. They should be no longer\nthan 500 words, and ideally shorter. Published\ncontributions are edited."
+        },
+        {
+          "bbox": [
+            52,
+            870,
+            73,
+            882
+          ],
+          "label": "foot",
+          "reading_order": 38,
+          "text": "802"
+        },
+        {
+          "bbox": [
+            291,
+            872,
+            424,
+            886
+          ],
+          "label": "foot",
+          "reading_order": 39,
+          "text": "2003 Nature Publishing Group"
+        },
+        {
+          "bbox": [
+            440,
+            871,
+            638,
+            883
+          ],
+          "label": "foot",
+          "reading_order": 40,
+          "text": "NATURE | VOL.422 | 24 APRIL 2003 | www.nature.com/nature"
+        }
+      ]
+    }
+  ],
+  "metadata": {}
+}