ljx529's picture
Duplicate from chuxuan/REPRO-Bench
c35658f
Raw
History Blame Contribute Delete
372 kB
/*******************************************************************************
This file contains the code to replicate the tables and figures in the appendix.
Contents:
0. Generate or import final dataset
A. DATA APPENDIX
Table A1: Summary statistics
B. ROBUSTNESS
C. IV ESTIMATION FOR MEN
D. ADDITIONAL DESCRIPTIVE EVIDENCE
The file "3 Replication package\programs00_setup.do" installs the required ado
packages sets. It also sets the path to all directories. If you want to run this
do file individually, after installing the required packages unnmark and specify
location of parent directory here:
*global dirroot = "SPECIFY\LOCATION\OF\PARENT\DIRECTORY\HERE"
and then run the following lines for the secondary directories (specified as
downloaded in replication package):
global dirdata = "${dirroot}\data"
global dirdta = "${dirroot}\programs\01_dataprep\dta-temp-files"
global dirresu = "${dirroot}\programs\results"
*******************************************************************************/
* Set-up:
do "${dirroot}\programs\00_setup.do"
* ==============================================================================
* 0. GENERATE OR IMPORT FINAL DATASET:
* ------------------------------------------------------------------------------
clear all
set more off
set matsize 10000
* To generate the final dataset run the following line:
do "${dirroot}\programs\01_dataprep\master-dataprep.do"
* Alternatively, you can import the final dataset data directly by writting
* use "${dirdata}\final-data.dta", clear
* ==============================================================================
********************************************************************************
********************************************************************************
********************************************************************************
* *
* A. DATA APPENDIX *
* *
********************************************************************************
********************************************************************************
********************************************************************************
* ==============================================================================
* TABLE A1:
* Summary statistics
* ------------------------------------------------------------------------------
* Import final dataset
use "${dirdata}\final-data.dta", clear
gen pr5 = (pr4!=1)
label var pr5 "Viscount's/baron's daughter"
global tosum cOut mheir fmissmatch fmissmatch2 fdown celibacy syntheticT mourn pr4 pr5 biorder hengpee hscotpee hiripee
* Summary stats (baseline)
preserve
foreach x in fmissmatch fmissmatch2 fdown celibacy{ // just so that variabels appear in the desired order
replace `x'=.
}
eststo sumbase: qui estpost summ $tosum if base_sample==1
estadd local obs = `e(N)'
restore
* Summary stats (married in landed elite)
preserve
replace celibacy = .
eststo sumlelite: qui estpost summ $tosum if base_sample==1 & fmissmatch!=.
estadd local obs = `e(N)'
restore
* Summary stats (including unmarried)
preserve
foreach x in cOut mheir fmissmatch fmissmatch2 fdown mourn{
replace `x'=.
}
eststo sumunmarr: qui estpost summ $tosum if marital_rates_sample==1
estadd local obs = `e(N)'
restore
* Output
esttab sumbase sumlelite sumunmarr using "${dirresu}\Appendix\TableA1.tex", ///
mlabels("Baseline" "Land elite" "Inc unmarr") ///
cells("mean(pattern(1 1 1) fmt(2)) sd(pattern(1 1 1) fmt(2))") ///
stats(obs, labels("Observations") fmt(%9.0gc)) ///
noobs label replace
* ==============================================================================
* TABLE A2:
* Summary statistics, women with a family seat
* ------------------------------------------------------------------------------
* Import final dataset
use "${dirdata}\final-data.dta", clear
gen pr5 = (pr4!=1)
label var pr5 "Viscount's/baron's daughter"
global tosum cOut mheir fmissmatch fmissmatch2 fdown celibacy syntheticT mourn distlondon pr4 pr5 biorder hengpee hscotpee hiripee
* Sample: women with a family seat
keep if distlondon!=.
* Summary stats (baseline)
preserve
foreach x in fmissmatch fmissmatch2 fdown celibacy{ // just so that variabels appear in the desired order
replace `x'=.
}
eststo sumbase: qui estpost summ $tosum if base_sample==1
estadd local obs = `e(N)'
restore
* Summary stats (married in landed elite)
preserve
replace celibacy = .
eststo sumlelite: qui estpost summ $tosum if base_sample==1 & fmissmatch!=.
estadd local obs = `e(N)'
restore
* Summary stats (including unmarried)
preserve
foreach x in cOut mheir fmissmatch fmissmatch2 fdown mourn{
replace `x'=.
}
eststo sumunmarr: qui estpost summ $tosum if marital_rates_sample==1
estadd local obs = `e(N)'
restore
* Output
esttab sumbase sumlelite sumunmarr using "${dirresu}\Appendix\TableA2.tex", ///
mlabels("Baseline" "Land elite" "Inc unmarr") ///
cells("mean(pattern(1 1 1) fmt(2)) sd(pattern(1 1 1) fmt(2))") ///
stats(obs, labels("Observations") fmt(%9.0gc)) ///
noobs label replace
* ==============================================================================
* TABLE A3:
* Summary statistics for marriage and political power
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}/final-data-sec4.dta", clear
gen pr5 = (pr4!=1)
label var pr5 "Viscount's/baron's daughter"
global tosum d_MPb_aft MPb_years_aft d_MPb_locaft2 MPb_years_locaft d_MPh_aft MPh_years_aft ///
cOut syntheticT pr4 pr5 nsib biorder distlondon ///
d_MPf_bef MPf_years_bef d_MPf_local2 MPf_years_locbef d_MPh_bef MPh_years_bef
* Collapse data such that a woman (and her family) is the unit of observation
collapse $tosum, by(refno)
label var d_MPb_aft "Any brother is MP (after)"
label var MPb_years_aft "All brothers MP years (after)"
label var d_MPb_locaft2 "Any brother is local MP (after)"
label var MPb_years_locaft "All brothers' local MP years (after)"
label var d_MPh_aft "Family head is MP (after)"
label var MPh_years_aft "Family head's MP years (after)"
label var cOut "Married a commoner"
label var syntheticT "Treatment (synthetic prob)"
label var pr4 "Duke, marquis, earl's family"
label var pr5 "Viscount, baron's family"
label var nsib "Number of brothers"
label var biorder "Woman's birth order (excl. heirs)"
label var distlondon "Distance to London"
label var d_MPf_bef "Father is MP (before)"
label var MPf_years_bef "Father's MP years (before)"
label var d_MPf_local2 "Father is local MP (before)"
label var MPf_years_locbef "Father's local MP years (before)"
label var d_MPh_bef "Family head is MP (before)"
label var MPh_years_bef "Family head's MP years (before)"
/* Note: the following measures of political power before marriage are used
only for regressions of brothers' political power. */
foreach x in d_MPb_aft MPb_years_aft d_MPb_locaft2 MPb_years_locaft d_MPf_bef MPf_years_bef d_MPf_local2 MPf_years_locbef{
replace `x' = . if nsib==.
}
* Summary stats
eststo sumA3: qui estpost summ $tosum
* Summary stats for county controls (Hechter)
preserve
use "${dirdata}/final-data-sec4.dta", clear
collapse propman100 logincome propcon propnonc relig, by(countyN) // Collapse data such that a county is the unit of observation
gen propcon100 = propcon*100 // to have them as percentages (%)
gen propnonc100 = propnonc*100
label var propman100 "perc. working in manufacturing"
label var logincome "income pc (in logs)"
label var propcon100 "perc. voting conservative, 1885"
label var propnonc100 "perc. non-conformists"
label var relig "religiosity"
* Summary stats
eststo sumhech: qui estpost summ propman100 logincome propcon100 propnonc100 relig
restore
* Output
esttab sumA3 sumhech using "${dirresu}\Appendix\TableA3.tex", ///
cells("mean(fmt(2)) sd(fmt(2)) min(fmt(2)) max(fmt(2)) count(fmt(0))") ///
noobs label replace
* ==============================================================================
* TABLE A4:
* MP elections, by type of constituency
* ------------------------------------------------------------------------------
* PANEL A: Distance from family seat to closest enfranchised constituency:
* import auxiliary appendix data with all dyads between constituencies' centroids
* and family seats in Section IV's sample. To consider enfranchized constituencies
* only, I exclude rotten boroughs.
import delimited "${dirroot}\programs\03_appendix\aux data appendix\seat_constituency_pairs_same_projection.txt", clear
/* Note: For consistency, in this data file both seats and constituencies
are geolocated using the WGS84 projection. */
* Calculate distance from seats to enfranchised constituencies
forvalues i=1/179{
geodist wgslat_sea wgslon_sea wgslat_cnt`i' wgslon_cnt`i', gen(dist`i') miles
}
egen dist_min = rowmin(dist1 dist2 dist3 dist4 dist5 dist6 dist7 dist8 dist9 dist10 dist11 dist12 dist13 dist14 dist15 dist16 dist17 dist18 dist19 dist20 dist21 dist22 dist23 dist24 dist25 dist26 dist27 dist28 dist29 dist30 dist31 dist32 dist33 dist34 dist35 dist36 dist37 dist38 dist39 dist40 dist41 dist42 dist43 dist44 dist45 dist46 dist47 dist48 dist49 dist50 dist51 dist52 dist53 dist54 dist55 dist56 dist57 dist58 dist59 dist60 dist61 dist62 dist63 dist64 dist65 dist66 dist67 dist68 dist69 dist70 dist71 dist72 dist73 dist74 dist75 dist76 dist77 dist78 dist79 dist80 dist81 dist82 dist83 dist84 dist85 dist86 dist87 dist88 dist89 dist90 dist91 dist92 dist93 dist94 dist95 dist96 dist97 dist98 dist99 dist100 dist101 dist102 dist103 dist104 dist105 dist106 dist107 dist108 dist109 dist110 dist111 dist112 dist113 dist114 dist115 dist116 dist117 dist118 dist119 dist120 dist121 dist122 dist123 dist124 dist125 dist126 dist127 dist128 dist129 dist130 dist131 dist132 dist133 dist134 dist135 dist136 dist137 dist138 dist139 dist140 dist141 dist142 dist143 dist144 dist145 dist146 dist147 dist148 dist149 dist150 dist151 dist152 dist153 dist154 dist155 dist156 dist157 dist158 dist159 dist160 dist161 dist162 dist163 dist164 dist165 dist166 dist167 dist168 dist169 dist170 dist171 dist172 dist173 dist174 dist175 dist176 dist177 dist178 dist179)
label var dist_min "Distance (in miles)"
* Summary stats
eststo sumdist: qui estpost summ dist_min
esttab sumdist using "${dirresu}\Appendix\TableA4_A.tex", ///
cells("mean(fmt(2)) sd(fmt(2)) max(fmt(2))") ///
noobs label replace
* PANEL B: MP elections, by type of constituency:
/* Note: first, reshape peers' biographies dataset to have one entry for
each election (i.e., seat in parliament) */
use "${dirdta}/Peers_biographies.dta", clear
sort refno
* 1. drop sisters
drop if refno==157408
encode fbio1, gen(nfam)
duplicates drop nfam, force
* 2. drop repeated biographies
foreach x in bio1 bio2{
rename f`x' `x'_0
rename h`x' `x'_11
replace `x'_11 = "" if fhead_1872!="other"
}
foreach x in 1 2 3 4 5 6 7 8 9 10{
rename b`x'bio1 bio1_`x'
rename b`x'bio2 bio2_`x'
}
qui sum MPf
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
gen term1_`j'_0 = term1f_`j'
gen cnsty_`j'_0 = cnstyf_`j'
gen cnstycounty_`j'_0 = cnstycountyf_`j'
gen cnstywgs_`j'_0 = cnstywgsf_`j'
}
}
*
qui sum MPh
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
gen term1_`j'_11 = term1h_`j'
gen cnsty_`j'_11 = cnstyh_`j'
gen cnstycounty_`j'_11 = cnstycountyh_`j'
gen cnstywgs_`j'_11 = cnstywgsh_`j'
}
}
*
foreach b in 1 2 3 4 5 6 7 8 9 10{
qui sum MPb`b'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
gen term1_`j'_`b' = term1b`b'_`j'
gen cnsty_`j'_`b' = cnstyb`b'_`j'
gen cnstycounty_`j'_`b' = cnstycountyb`b'_`j'
gen cnstywgs_`j'_`b' = cnstywgsb`b'_`j'
}
}
}
drop firstnames surname title byear dyear myear wom mourn cOut cs fhead_1872 fhead_1878 relation_f relation_b1 heir_b1 note relation_b2 heir_b2 relation_b3 heir_b3 relation_b4 heir_b4 relation_b5 heir_b5 relation_b6 heir_b6 relation_b7 heir_b7 relation_b8 heir_b8 relation_b9 heir_b9 relation_b10 heir_b10 fh b1h b2h b3h b4h b5h b6h b7h b8h b9h b10h relation_b11 heir_b11 b11h byear_b01 byear_b1 byear_b02 byear_b2 byear_b03 byear_b3 byear_b04 byear_b4 byear_b05 byear_b5 byear_b06 byear_b6 byear_b07 byear_b7 byear_b08 byear_b8 byear_b09 byear_b9 byear_b010 byear_b10 myear_f_0 myear_f_alt myear_f myear_h_0 myear_h_alt myear_h myear_b1_0 myear_b1_alt myear_b1 myear_b2_0 myear_b2_alt myear_b2 myear_b3_0 myear_b3_alt myear_b3 myear_b4_0 myear_b4_alt myear_b4 myear_b5_0 myear_b5_alt myear_b5 myear_b6_0 myear_b6_alt myear_b6 myear_b7_0 myear_b7_alt myear_b7 myear_b8_0 myear_b8_alt myear_b8 myear_b9_0 myear_b9_alt myear_b9 myear_b10_0 myear_b10_alt myear_b10 d_MPf MP1fcount MP1fpos1 MP1fpos2 MP1fpos3 MP1fpos4 MP1fpos5 MPf d_MPh MP1hcount MP1hpos1 MP1hpos2 MP1hpos3 MP1hpos4 MPh d_MPb1 MP1b1count MP1b1pos1 MP1b1pos2 MP1b1pos3 MP1b1pos4 MP1b1pos5 MPb1 d_MPb2 MP1b2count MP1b2pos1 MP1b2pos2 MP1b2pos3 MP1b2pos4 MPb2 d_MPb3 MP1b3count MP1b3pos1 MP1b3pos2 MP1b3pos3 MP1b3pos4 MP1b3pos5 MP1b3pos6 MP1b3pos7 MP1b3pos8 MPb3 d_MPb4 MP1b4count MP1b4pos1 MP1b4pos2 MP1b4pos3 MPb4 d_MPb5 MP1b5count MP1b5pos1 MPb5 d_MPb6 MP1b6count MP1b6pos1 MPb6 d_MPb7 MP1b7count MP1b7pos1 MPb7 d_MPb8 MP1b8count MP1b8pos1 MPb8 d_MPb9 MP1b9count MPb9 d_MPb10 MP1b10count MPb10 MPb d_MPb libf toryf MPallf_1 partyf_1 cnstyf_1 term1f_1 term2f_1 MPallf_2 partyf_2 cnstyf_2 term1f_2 term2f_2 MPallf_3 partyf_3 cnstyf_3 term1f_3 term2f_3 MPallf_4 partyf_4 cnstyf_4 term1f_4 term2f_4 MPallf_5 partyf_5 cnstyf_5 term1f_5 term2f_5 libh toryh MPallh_1 partyh_1 cnstyh_1 term1h_1 term2h_1 MPallh_2 partyh_2 cnstyh_2 term1h_2 term2h_2 MPallh_3 partyh_3 cnstyh_3 term1h_3 term2h_3 MPallh_4 partyh_4 cnstyh_4 term1h_4 term2h_4 libb1 toryb1 MPallb1_1 partyb1_1 cnstyb1_1 term1b1_1 term2b1_1 MPallb1_2 partyb1_2 cnstyb1_2 term1b1_2 term2b1_2 MPallb1_3 partyb1_3 cnstyb1_3 term1b1_3 term2b1_3 MPallb1_4 partyb1_4 cnstyb1_4 term1b1_4 term2b1_4 MPallb1_5 partyb1_5 cnstyb1_5 term1b1_5 term2b1_5 libb2 toryb2 MPallb2_1 partyb2_1 cnstyb2_1 term1b2_1 term2b2_1 MPallb2_2 partyb2_2 cnstyb2_2 term1b2_2 term2b2_2 MPallb2_3 partyb2_3 cnstyb2_3 term1b2_3 term2b2_3 MPallb2_4 partyb2_4 cnstyb2_4 term1b2_4 term2b2_4 libb3 toryb3 MPallb3_1 partyb3_1 cnstyb3_1 term1b3_1 term2b3_1 MPallb3_2 partyb3_2 cnstyb3_2 term1b3_2 term2b3_2 MPallb3_3 partyb3_3 cnstyb3_3 term1b3_3 term2b3_3 MPallb3_4 partyb3_4 cnstyb3_4 term1b3_4 term2b3_4 MPallb3_5 partyb3_5 cnstyb3_5 term1b3_5 term2b3_5 MPallb3_6 partyb3_6 cnstyb3_6 term1b3_6 term2b3_6 MPallb3_7 partyb3_7 cnstyb3_7 term1b3_7 term2b3_7 MPallb3_8 partyb3_8 cnstyb3_8 term1b3_8 term2b3_8 libb4 toryb4 MPallb4_1 partyb4_1 cnstyb4_1 term1b4_1 term2b4_1 MPallb4_2 partyb4_2 cnstyb4_2 term1b4_2 term2b4_2 MPallb4_3 partyb4_3 cnstyb4_3 term1b4_3 term2b4_3 libb5 toryb5 MPallb5_1 partyb5_1 cnstyb5_1 term1b5_1 term2b5_1 libb6 toryb6 MPallb6_1 partyb6_1 cnstyb6_1 term1b6_1 term2b6_1 libb7 toryb7 MPallb7_1 partyb7_1 cnstyb7_1 term1b7_1 term2b7_1 libb8 toryb8 MPallb8_1 partyb8_1 cnstyb8_1 term1b8_1 term2b8_1 libb9 toryb9 libb10 toryb10 MPh_years MPh_years_aft d_MPh_bef d_MPh_aft MPf_years MPf_years_aft d_MPf_bef d_MPf_aft MPb1_years MPb1_years_aft d_MPb1_bef d_MPb1_aft MPb2_years MPb2_years_aft d_MPb2_bef d_MPb2_aft MPb3_years MPb3_years_aft d_MPb3_bef d_MPb3_aft MPb4_years MPb4_years_aft d_MPb4_bef d_MPb4_aft MPb5_years MPb5_years_aft d_MPb5_bef d_MPb5_aft MPb6_years MPb6_years_aft d_MPb6_bef d_MPb6_aft MPb7_years MPb7_years_aft d_MPb7_bef d_MPb7_aft MPb8_years MPb8_years_aft d_MPb8_bef d_MPb8_aft MPb9_years MPb9_years_aft d_MPb9_bef d_MPb9_aft MPb10_years MPb10_years_aft d_MPb10_bef d_MPb10_aft d_MPb_aft d_MPb_bef MPb_years MPb_years_aft cnstycountyh_1 cnstywgsh_1 cnstycountyh_2 cnstywgsh_2 cnstycountyh_3 cnstywgsh_3 cnstycountyh_4 cnstywgsh_4 cnstycountyf_1 cnstywgsf_1 cnstycountyf_2 cnstywgsf_2 cnstycountyf_3 cnstywgsf_3 cnstycountyf_4 cnstywgsf_4 cnstycountyf_5 cnstywgsf_5 cnstycountyb1_1 cnstywgsb1_1 cnstycountyb1_2 cnstywgsb1_2 cnstycountyb1_3 cnstywgsb1_3 cnstycountyb1_4 cnstywgsb1_4 cnstycountyb1_5 cnstywgsb1_5 cnstycountyb2_1 cnstywgsb2_1 cnstycountyb2_2 cnstywgsb2_2 cnstycountyb2_3 cnstywgsb2_3 cnstycountyb2_4 cnstywgsb2_4 cnstycountyb3_1 cnstywgsb3_1 cnstycountyb3_2 cnstywgsb3_2 cnstycountyb3_3 cnstywgsb3_3 cnstycountyb3_4 cnstywgsb3_4 cnstycountyb3_5 cnstywgsb3_5 cnstycountyb3_6 cnstywgsb3_6 cnstycountyb3_7 cnstywgsb3_7 cnstycountyb3_8 cnstywgsb3_8 cnstycountyb4_1 cnstywgsb4_1 cnstycountyb4_2 cnstywgsb4_2 cnstycountyb4_3 cnstywgsb4_3 cnstycountyb5_1 cnstywgsb5_1 cnstycountyb6_1 cnstywgsb6_1 cnstycountyb7_1 cnstywgsb7_1 cnstycountyb8_1 cnstywgsb8_1 fhead_1872_2 nfhead_1872 nsib nfam b11bio1 b11bio2
* reshape to have one entry per relative (i.e., fathers and borthers)
reshape long bio1_ bio2_ term1_1_ term1_2_ term1_3_ term1_4_ term1_5_ term1_6_ term1_7_ term1_8_ ///
cnsty_1_ cnsty_2_ cnsty_3_ cnsty_4_ cnsty_5_ cnsty_6_ cnsty_7_ cnsty_8_ ///
cnstycounty_1_ cnstycounty_2_ cnstycounty_3_ cnstycounty_4_ cnstycounty_5_ cnstycounty_6_ cnstycounty_7_ cnstycounty_8_ ///
cnstywgs_1_ cnstywgs_2_ cnstywgs_3_ cnstywgs_4_ cnstywgs_5_ cnstywgs_6_ cnstywgs_7_ cnstywgs_8_ ///
, i(refno) j(nrelative)
drop if bio1_=="" & bio2_==""
sort bio1_ bio2_ refno nrelative
by bio1_ bio2_: gen tokeep = _n
keep if tokeep==1
count // As before, I have 674 unique biographies
* 3. Reshape to have one election per entry
forvalues i=1/8{
rename term1_`i'_ term`i'
rename cnsty_`i'_ pc`i'
rename cnstycounty_`i'_ pccounty`i'
rename cnstywgs_`i'_ pcwgs`i'
}
gen idbio = _n
drop refno bio1_ bio2_ tokeep
reshape long term pc pccounty pcwgs, i(idbio) j(nelection)
sort idbio term nelection
keep if term!=.
rename pc cnsty
replace cnsty = "Egremont" if idbio==90
replace cnsty = "South Ayrshire" if idbio==587
* merge with constituency data (including type of constituency)
preserve
import excel "${dirdata}\constituencies.xlsx", sheet("Sheet1") firstrow clear
keep cnsty note
label var cnsty "Electoral constituency"
label var note "Type of constituency"
save "${dirdta}\constituencies.dta", replace
restore
merge m:1 cnsty using "${dirdta}\constituencies.dta"
* housekeeping
replace note = "County constituency" if cnsty=="Ayrshire" | cnsty=="Caernarvonshire" | cnsty=="Hertfordshire" | cnsty=="Cumberland, Egremont Divisio"
replace note = "Rotten borough" if cnsty=="Newport Cornwall"
replace note = "Borough constituency" if cnsty=="Scarborough" | cnsty=="Brentwood" | cnsty=="Enfield"
* Panel B's sumary stats
capture log close // save results in log file
cd "${dirresu}\Appendix"
log using TableA4_B.log, replace
* All
tab note
* Fathers
tab note if nrelative==0
* Brothers
tab note if nrelative!=0
log close
* ==============================================================================
* TABLE A5:
* Summary statistics for determinants of education provision
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}/final-data-sec4.dta", clear
gen pr5 = (pr4!=1)
label var pr5 "Viscount's/baron's daughter"
* Add total funds raised from rates (pounds)
preserve
import delimited "${dirroot}\programs\03_appendix\aux data appendix\funds_from_rates.txt", encoding(ISO-8859-2) clear
label var refno "Reference number in Hollingsworth"
label var fid_b10sea "FID identifier for family seats of women in baseline sample (ArcGIS)"
label var rate_pound "Funds from rates (pound)"
save "${dirdta}/funds-from-rates.dta", replace
restore
merge 1:1 refno fid_b10sea using "${dirdta}/funds-from-rates.dta", nogen
gen rate_log = log(rate_pound)
* Labels
label var rateperpou "Tax rate average"
label var rate_pound "Funds from rates (pound)"
label var numsb "Number of School Boards"
label var d_MPb_aft "Any brother is MP (after)"
label var MPb_years_aft "All brothers MP years (after)"
label var d_MPb_locaft "Any brother is local MP (after)"
label var MPb_years_locaft "All brothers' local MP years (after)"
label var d_MPh_aft "Family head is MP (after)"
label var MPh_years_aft "Family head's MP years (after)"
label var cOut "Married a commoner"
label var syntheticT "Treatment (synthetic prob)"
label var pr4 "Duke, marquis, earl's family"
label var pr5 "Viscount, baron's family"
label var nsib "Number of brothers"
label var biorder "Woman's birth order (excl. heirs)"
label var distlondon "Distance to London"
label var d_MPf_bef "Father is MP (before)"
label var MPf_years_bef "Father's MP years (before)"
label var d_MPf_local2 "Father is local MP (before)"
label var MPf_years_locbef "Father's local MP years (before)"
label var d_MPh_bef "Family head is MP (before)"
label var MPh_years_bef "Family head's MP years (before)"
global tosum rateperpou rate_pound numsb ///
d_MPb_aft MPb_years_aft d_MPb_locaft MPb_years_locaft d_MPh_aft MPh_years_aft ///
cOut syntheticT pr4 pr5 nsib biorder distlondon ///
d_MPf_bef MPf_years_bef d_MPf_local MPf_years_locbef d_MPh_bef MPh_years_bef
/* Note: the following measures of political power before marriage are used
only for regressions of brothers' political power. */
foreach x in d_MPb_aft MPb_years_aft d_MPb_locaft MPb_years_locaft MPf_years_bef MPf_years_locbef d_MPf_bef d_MPf_local{
replace `x' = . if nsib==.
}
* Summary stats
eststo sumA5: qui estpost summ $tosum
* Output
esttab sumA5 using "${dirresu}\Appendix\TableA5.tex", ///
cells("mean(fmt(2)) sd(fmt(2)) min(fmt(2)) max(fmt(2)) count(fmt(0))") ///
noobs label replace
* ==============================================================================
* FIGURE A7:
* MP elections overtime.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}/Peers_biographies.dta", clear
* Variables for time series figure
gen heir_f = 1
local i = 1
local rel = 1
foreach z in f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
gen year0_`i' = term1`z'_`j'
if `rel'<2{
gen whoel_`i' = 0
}
if `rel'>1{
gen whoel_`i' = .
replace whoel_`i' = 1 if heir_`z'==1
replace whoel_`i' = 2 if heir_`z'==0
}
local i = `i'+1
}
}
local rel = `rel'+1
}
gen MPtot = MPb+MPf
keep refno fbio1 MPtot year0_1 whoel_1 year0_2 whoel_2 year0_3 whoel_3 year0_4 whoel_4 year0_5 whoel_5 year0_6 whoel_6 year0_7 whoel_7 year0_8 whoel_8 year0_9 whoel_9 year0_10 whoel_10 year0_11 whoel_11 year0_12 whoel_12 year0_13 whoel_13 year0_14 whoel_14 year0_15 whoel_15 year0_16 whoel_16 year0_17 whoel_17 year0_18 whoel_18 year0_19 whoel_19 year0_20 whoel_20 year0_21 whoel_21 year0_22 whoel_22 year0_23 whoel_23 year0_24 whoel_24 year0_25 whoel_25 year0_26 whoel_26 year0_27 whoel_27 year0_28 whoel_28 year0_29 whoel_29
bysort fbio1: gen sis = _n
keep if sis==1
reshape long year0_ whoel_, i(refno) j(j)
gen MPi = j
drop j
drop if year0_ == .
gen MPcount = 1
gen MPfcount = 1 if whoel_==0
gen MPhcount = 1 if whoel_==1
gen MPbcount = 1 if whoel_==2
collapse (sum) MPcount (sum) MPfcount (sum) MPhcount (sum) MPbcount, by(year0_)
tsset year0_
drop if year0_>=1889
tssmooth ma MPcount_ma = MPcount, window(2,1,2)
tssmooth ma MPfcount_ma = MPfcount, window(2,1,2)
tssmooth ma MPhcount_ma = MPhcount, window(2,1,2)
tssmooth ma MPbcount_ma = MPbcount, window(2,1,2)
* figure
twoway (scatter MPfcount year0_ if MPfcount>0 & year0_>1820 & year0_<1890, mcolor(black) ms(circle_hollow)) ///
(scatter MPhcount year0_ if MPhcount>0 & year0_>1820 & year0_<1890, mcolor(black)) ///
(scatter MPbcount year0_ if MPbcount>0 & year0_>1820 & year0_<1890, mcolor(black) ms(X)) ///
(line MPcount_ma year0_ if MPcount_ma>0 & year0_>1820 & year0_<1890, lcolor(black)) ///
(scatteri 0.5 1867 0.56 1867 , recast(line) lpattern(dash) lcolor(dkgreen)) ///
(scatteri 0.5 1837 0.56 1837 , recast(line) lpattern(dot) lcolor(gray)), ///
xlabel(1825(10)1885) xtitle("") ytitle("MP appointments") ylabel(0(5)20, nogrid) ///
legend(on order(1 "Fathers" 2 "Brothers (heirs)" 3 "Brothers (non-heirs)" 4 "Moving avg.(2,2), total" 5 "Reform Act" 6 "General Election") rows(6) ring(0) position(1) bmargin(large)) ///
xline(1832, lpattern(dash) lcolor(dkgreen)) xline(1867, lpattern(dash) lcolor(dkgreen)) ///
xline(1884, lpattern(dash) lcolor(dkgreen)) ///
xline(1826, lpattern(dot) lcolor(gray)) xline(1830, lpattern(dot) lcolor(gray)) ///
xline(1831, lpattern(dot) lcolor(gray)) xline(1832, lpattern(dot) lcolor(gray)) ///
xline(1835, lpattern(dot) lcolor(gray)) xline(1837, lpattern(dot) lcolor(gray)) ///
xline(1841, lpattern(dot) lcolor(gray)) xline(1847, lpattern(dot) lcolor(gray)) ///
xline(1852, lpattern(dot) lcolor(gray)) xline(1857, lpattern(dot) lcolor(gray)) ///
xline(1859, lpattern(dot) lcolor(gray)) xline(1865, lpattern(dot) lcolor(gray)) ///
xline(1868, lpattern(dot) lcolor(gray)) xline(1874, lpattern(dot) lcolor(gray)) ///
xline(1880, lpattern(dot) lcolor(gray)) xline(1885, lpattern(dot) lcolor(gray)) ///
xline(1886, lpattern(dot) lcolor(gray)) graphregion(color(white))
graph export "${dirresu}\Appendix\FigureA7.pdf", as(pdf) replace
* ==============================================================================
* FIGURE A8:
* Marriage and political power.
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}/final-data-sec4.dta", clear
* Reshape data to have each brother as a separate data entry
local i = 1
foreach z in b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
gen year0_`i' = term1`z'_`j'-myear
local i = `i'+1
}
}
}
keep refno cOut myear MPb year0_1 year0_2 year0_3 year0_4 year0_5 year0_6 year0_7 year0_8 year0_9 year0_10 year0_11 year0_12 year0_13 year0_14 year0_15 year0_16 year0_17 year0_18 year0_19 year0_20 year0_21 year0_22 year0_23 year0_24
duplicates drop
reshape long year0_ , i(refno) j(j)
* Indicators for MP elections, by years after sister's marriage
gen MPi = j
drop j
drop if year0_ == .
gen MPcount = 1
gen MPout = 1 if cOut==1
gen MPin = 1 if cOut==0
gen year0_five = .
forvalues i = -40(10)40{
replace year0_five = `i' if year0_>=`i' & year0_<`i'+10
}
collapse (sum) MPcount (sum) MPout (sum) MPin, by(cOut year0_five)
drop if year0_five==-40
* Figure
twoway (connect MPcount year0_five if cOut==0 & year0_five<=20, lcolor(blue) ms(smdiamond) mc(blue)) ///
(connect MPcount year0_five if cOut==1 & year0_five<=20, lcolor(red) ms(smdiamond) mc(red) lw(thick)), ///
xtitle("years after sister's marriage") ytitle("Number of brothers elected MP") ///
legend(on order(1 "sister married in the peerage" 2 "sister married a commoner")) ///
xlabels(-30 "-30 to -20" -20 "-20 to -10" -10 "-10 to 0" 0 "0 to 10" 10 "10 to 20" 20 "20 to 30", angle(45)) ///
graphregion(color(white)) xline(-10, lpattern(dash) lcolor(gray))
graph export "${dirresu}\Appendix\FigureA8.pdf", as(pdf) replace
********************************************************************************
********************************************************************************
********************************************************************************
* *
* B. ROBUSTNESS *
* *
********************************************************************************
********************************************************************************
********************************************************************************
* ==============================================================================
* TABLE B1:
* Synthetic probability based on alternative benchmark cohorts
* TABLE B2:
* Synthetic probability based on alternative benchmark cohorts, controlling for distance to London.
* ------------------------------------------------------------------------------
* Compute alternative treatment using different benchmark cohorts
clear
set obs 21
gen mage = _n+14
gen freq = 0
label var mage "Age at marriage"
label var freq "Frequency (=0 as starting point of calculations)"
save "${dirdta}\agessyntheic.dta", replace
use "${dirdata}\final-data.dta", clear
foreach i in 1810 1815 1820 1835 1840 1845{
preserve
local j = `i'+15
local d b
if `i'>1820{
local d m
}
keep if wom==1 & nthismarriage<2 & `d'year<=`j' & `d'year>=`i' & dage>30
replace mage = myear-byear if nthismarr==1 & mage==.
drop if myear>=1861 & myear!=.
replace mage = 999 if nthismarr==0
replace nthismarr = 1 if nthismarr==0
collapse (sum) freq = nthismarr, by(mage)
sort mage
drop if mage==.
replace freq = 0 if freq==.
merge 1:1 mage using "${dirdta}\agessyntheic.dta", nogen
egen tot = sum(freq)
sort mage
gen prob = 100*freq/tot
drop if mage == 999
keep mage prob
sort mage
gen syntheticT_alt = prob+prob[_n+1]+prob[_n+2]
drop prob
rename mage age1861
sort age1861
keep if age1861>=15 & age1861<=35
label var syntheticT_alt "Alternative treatment based on `d' cohort (`i'-`j')"
label var age1861 "Age at the start of the 1861 Season"
save "${dirdta}\syntheticprob`i'_`j'_alt.dta", replace
restore
}
* Regressions
global controls pr4 biorder hengpee
local ite = 1
do "${dirauxp}\cgmreg.do" // bootstrap-t procedure (Cameron, Gelbach and Miller, 2008)
foreach i in 1815 1810 1820 1845 1840 1835{
local panel : word `ite' of `c(ALPHA)'
local j = `i'+15
use "${dirdata}\final-data.dta", clear
merge m:1 age1861 using "${dirdta}\syntheticprob`i'_`j'_alt.dta"
label var syntheticT_alt "Alternative treatment"
label var _merge "Merge results for alternative treatment"
keep if _merge==3
keep if wom==1 & nthismarriage<2 & age1861>=15 & age1861<=35
save "${dirdta}\dataforregs.dta", replace
local q = 1
foreach x in "" distlondon{
use "${dirdta}\dataforregs.dta", clear
keep if nthismarr==1 & myear!=.
drop if prankspo==0 // no foreign spouses (+ drops saxe-coburg)
qui probit cOut syntheticT_alt $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB`q'_`panel'", keep(syntheticT_alt) nocons label dec(3) coefastr ctitle("M common") nor2 tex replace
qui probit mheir syntheticT_alt $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB`q'_`panel'", keep(syntheticT_alt) nocons label dec(3) coefastr ctitle("M heir") nor2 tex
qui reg fmissmatch syntheticT_alt $controls `x', cluster(byear)
outreg2 using "${dirresu}\Appendix\TableB`q'_`panel'", keep(syntheticT_alt) nocons label dec(3) coefastr ctitle("Diff abs") nor2 tex
qui reg fmissmatch2 syntheticT_alt $controls `x', cluster(byear)
outreg2 using "${dirresu}\Appendix\TableB`q'_`panel'", keep(syntheticT_alt) nocons label dec(3) coefastr ctitle("Diff h-w") nor2 tex
qui probit fdown syntheticT_alt $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB`q'_`panel'", keep(syntheticT_alt) nocons label dec(3) coefastr ctitle("M down") nor2 tex
use "${dirdta}\dataforregs.dta", clear
keep if dage>35 & dyear>byear+1 & prank>1
qui probit celibacy syntheticT_alt $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB`q'_`panel'", keep(syntheticT_alt) nocons label dec(3) coefastr ctitle("never M") nor2 tex title("T based on women from `i' to `j'")
* small cluster
/* NOTE 1: IF THE REPLICATOR WISHES TO SKIP THE CALCULATION OF SMALL
CLUSTERS (WHICH TAKES ABOUT 6H), TYPE "0" IN THE FOLLOWING LOCAL: */
local smallcluster = 1
if `smallcluster' == 1{
/* NOTE 2: Results stored in log file in the following locations:
3 Replication package\results\Appendix\...
... TableB1_A_smallcluster.log; ... TableB1_B_smallcluster.log;
... TableB1_C_smallcluster.log; ... TableB1_D_smallcluster.log.
... TableB1_E_smallcluster.log; ... TableB1_F_smallcluster.log.
... TableB2_A_smallcluster.log; ... TableB2_B_smallcluster.log;
... TableB2_C_smallcluster.log; ... TableB2_D_smallcluster.log.
... TableB2_E_smallcluster.log; ... TableB2_F_smallcluster.log.
See also Table 2 for small cluster p-values for benchmark cohorts (panel A
in both tables B1 and B2) */
use "${dirdta}\dataforregs.dta", clear
keep if nthismarr==1 & myear!=.
drop if prankspo==0 // no foreign spouses (+ drops saxe-coburg)
qui probit cOut syntheticT_alt $controls `x', cluster(byear)
local d born
if `i'>1820{
local d marr
}
keep if e(sample)==1
if `ite'>1{
capture log close // save results in log file
cd "${dirresu}\Appendix"
log using TableB`q'_`panel'_smallcluster.log, replace
display "=================================================================="
display "Table B`q', panel `panel', col [1]: `d'. `i' to `j' (M common)"
display "=================================================================="
clustse probit cOut syntheticT_alt $controls `x', cluster(byear) method(pairs) reps(5000)
display "=================================================================="
display "Table B`q', panel `panel', col [2]: `d'. `i' to `j' (M heir) "
display "=================================================================="
clustse probit mheir syntheticT_alt $controls `x', cluster(byear) method(pairs) reps(5000)
display "=================================================================="
display "Table B`q', panel `panel', col [3]: `d'. `i' to `j' (Diff abs) "
display "=================================================================="
****
qui reg fmissmatch syntheticT_alt $controls `x', cluster(byear)
keep if e(sample)==1
****
cgmwildboot fmissmatch syntheticT_alt $controls `x', cluster(byear) bootcluster(byear) reps(5000)
display "=================================================================="
display "Table B`q', panel `panel', col [4]: `d'. `i' to `j' (Diff h-w) "
display "=================================================================="
cgmwildboot fmissmatch2 syntheticT_alt $controls `x', cluster(byear) bootcluster(byear) reps(5000)
display "=================================================================="
display "Table B`q', panel `panel', col [5]: `d'. `i' to `j' (M down) "
display "=================================================================="
clustse probit fdown syntheticT_alt $controls `x', cluster(byear) method(pairs) reps(5000)
use "${dirdta}\dataforregs.dta", clear
keep if dage>35 & dyear>byear+1 & prank>1
qui probit celibacy syntheticT_alt $controls `x', cluster(byear)
keep if e(sample)==1
display "=================================================================="
display "Table B`q', panel `panel', col [6]: `d'. `i' to `j' (never M) "
display "=================================================================="
clustse probit celibacy syntheticT_alt $controls `x', cluster(byear) method(pairs) reps(5000)
log close
}
}
local q = `q'+1
}
local ite = `ite'+1
}
/* Table B1, Panel A and Table B2, Panel A small clusters' p-values are the
same as in Table 2 (both consider benchmark cohorts). Hence, the code is the
same as before: */
if `smallcluster' == 1{
use "${dirdata}\final-data.dta", clear
gsort surname title -nthismarr // to get stable results
***do "${dirauxp}\cgmreg.do" // if not run in setup
capture log close // save results in log file
cd "${dirresu}\Appendix"
log using TableB1_A_smallcluster.log, replace
keep if wom==1 & nthismarriage<2 & age1861>=15 & age1861<=35 // sample
preserve
keep if nthismarr==1 & myear!=. // married
qui probit cOut syntheticT $controls, cluster(byear)
keep if e(sample)==1 // keep sample of non-missing variables
clustse probit cOut syntheticT $controls, cluster(byear) method(pairs) reps(5000)
clustse probit mheir syntheticT $controls, cluster(byear) method(pairs) reps(5000)
cgmwildboot fmissmatch syntheticT $controls, cluster(byear) bootcluster(byear) reps(5000)
cgmwildboot fmissmatch2 syntheticT $controls, cluster(byear) bootcluster(byear) reps(5000)
clustse probit fdown syntheticT $controls, cluster(byear) method(pairs) reps(5000)
restore
keep if dage>35 & dyear>byear+1 & prank>1 // married and unmarried women
qui probit celibacy syntheticT $controls, cluster(byear)
keep if e(sample)==1 // keep sample of non-missing variables
clustse probit celibacy syntheticT $controls, cluster(byear) method(pairs) reps(5000)
log close
*
use "${dirdata}\final-data.dta", clear
gsort surname title -nthismarr // to get stable results
capture log close // save results in log file
cd "${dirresu}\Appendix"
log using TableB2_A_smallcluster.log, replace
keep if wom==1 & nthismarriage<2 & age1861>=15 & age1861<=35 // sample
preserve
keep if nthismarr==1 & myear!=. // married
qui probit cOut syntheticT $controls distlondon, cluster(byear)
keep if e(sample)==1 // keep sample of non-missing variables
clustse probit cOut syntheticT $controls distlondon, cluster(byear) method(pairs) reps(5000)
clustse probit mheir syntheticT $controls distlondon, cluster(byear) method(pairs) reps(5000)
cgmwildboot fmissmatch syntheticT $controls distlondon, cluster(byear) bootcluster(byear) reps(5000)
cgmwildboot fmissmatch2 syntheticT $controls distlondon, cluster(byear) bootcluster(byear) reps(5000)
clustse probit fdown syntheticT $controls distlondon, cluster(byear) method(pairs) reps(5000)
restore
keep if dage>35 & dyear>byear+1 & prank>1 // married and unmarried women
qui probit celibacy syntheticT $controls distlondon, cluster(byear)
keep if e(sample)==1 // keep sample of non-missing variables
clustse probit celibacy syntheticT $controls distlondon, cluster(byear) method(pairs) reps(5000)
log close
}
* Erase temporary dta files
erase "${dirdta}\agessyntheic.dta"
erase "${dirdta}\dataforregs.dta"
erase "${dirdta}\syntheticprob1810_1825_alt.dta"
erase "${dirdta}\syntheticprob1815_1830_alt.dta"
erase "${dirdta}\syntheticprob1820_1835_alt.dta"
erase "${dirdta}\syntheticprob1835_1850_alt.dta"
erase "${dirdta}\syntheticprob1840_1855_alt.dta"
erase "${dirdta}\syntheticprob1845_1860_alt.dta"
* ==============================================================================
* TABLE B3:
* Contingency tables, treatment based on marriage cohorts
* ------------------------------------------------------------------------------
* Import final data
use "${dirdata}\final-data.dta", clear
* Sample
keep if wom==1 & prank>1 & myear>=1858 & myear<=1863 & nthismarr==1 & mage<40
/* Note: The sample is all peers' daughters marrying in 1858-63, excluding
second marriages and women married above 40. */
* Contingency tables
label var grank "Wife"
capture log close // save results in log file
cd "${dirresu}\Appendix"
log using TableB3.log, replace
* Panel A: Control group (Season): peers' daughters marrying in 1858-60
tab grank grankspo if myear>=1858 & myear<=1860 & crankspo!=0 & nthismarr==1, expect
* Panel B: Treatment group (Season): peers' daughters marrying in 1861-63
tab grank grankspo if myear>=1861 & myear<=1863 & crankspo!=0 & nthismarr==1, expect
log close
* ==============================================================================
* TABLE B4:
* Sorting by title, non-parametric tests based on marriage cohorts.
* ------------------------------------------------------------------------------
* Import final data
use "${dirdata}\final-data.dta", clear
* Sample
keep if wom==1 & prank>1 & myear>=1858 & myear<=1863 & nthismarr==1 & mage<40
/* Note: The sample is all peers' daughters marrying in 1858-63, excluding
second marriages and women married above 40. */
* Add results to matrix tab5
mat define tabB4 = (.,.,.\.,.,.\.,.,.\.,.,.\.,.,.\.,.,.\.,.,.)
* Control group, m. in 1858-60
qui tab prank grankspo if myear>=1858 & myear<=1860, expect
qui bootstrap r(taub), reps(1000): tab prank grankspo if mourn==0 & prank>1 & crankspo!=0, taub
matrix taub0 = r(table)
local taub = taub0[1,1]
local taub_p = round(taub0[4,1],0.0001)
qui tab prank grankspo if mourn==0 & prank>1 & crankspo!=0, chi2 lrchi2 taub nofreq
mat tabB4[1,1] = `r(chi2)'
mat tabB4[2,1] = `r(p)'
mat tabB4[3,1] = `r(chi2_lr)'
mat tabB4[4,1] = `r(p_lr)'
mat tabB4[5,1] = `taub'
mat tabB4[6,1] = `taub_p'
mat tabB4[7,1] = `r(N)'
* Treatment group, m. in 1861-63
qui tab prank grankspo if myear>=1861 & myear<=1863, expect
qui bootstrap r(taub), reps(1000): tab prank grankspo if mourn==1 & prank>1 & crankspo!=0, taub
matrix taub0 = r(table)
local taub = taub0[1,1]
local taub_p = round(taub0[4,1],0.0001)
qui tab prank grankspo if mourn==1 & prank>1 & crankspo!=0, chi2 lrchi2 taub nofreq
mat tabB4[1,2] = `r(chi2)'
mat tabB4[2,2] = `r(p)'
mat tabB4[3,2] = `r(chi2_lr)'
mat tabB4[4,2] = `r(p_lr)'
mat tabB4[5,2] = `taub'
mat tabB4[6,2] = `taub_p'
mat tabB4[7,2] = `r(N)'
* Difference
mat tabB4[1,3] = tabB4[1,1]-tabB4[1,2]
mat tabB4[3,3] = tabB4[3,1]-tabB4[3,2]
mat tabB4[5,3] = tabB4[5,1]-tabB4[5,2]
mat tabB4[7,3] = tabB4[7,1]+tabB4[7,2]
forvalues i=0/1{
qui tab prank grankspo if mourn==`i' & prank>1 & crankspo!=0, all
local conv_chi`i' = sqrt(r(chi2)/e(N))
local conv_lrc`i' = sqrt(r(chi2_lr)/e(N))
}
forvalues i=0/1{
qui bootstrap r(taub), reps(500000): tab prank grankspo if mourn==`i' & prank>1 & crankspo!=0, taub
mat b`i' = e(b)
local taub`i' = b`i'[1,1]
local conv`i' = sin(3.141592654*`taub`i''*0.5)
}
* difference: pearson chi2
cortesti `conv_chi0' 95 `conv_chi1' 87
mat tabB4[2,3] = 0.0577
* difference: likelihood chi2
cortesti `conv_lrc0' 95 `conv_lrc1' 87
mat tabB4[4,3] = 0.0459
* difference: taub
cortesti `conv0' 95 `conv1' 87
mat tabB4[6,3] = 0.0017
* Table B4
mat colnames tabB4 = "Control" "Treat" "Diff"
mat rownames tabB4 = "Pearson chi-squared" " p-val" "Likelihood ratio" " p-val" "Kendall rank correlation" " p-val" "N"
mat list tabB4
esttab matrix(tabB4, fmt(3 3)) using "${dirresu}\Appendix\TableB4.tex", mlabel("Table B4") replace
* ==============================================================================
* FIGURE B1:
* Sorting by landholdings, K-S tests based on marriage cohorts
* ------------------------------------------------------------------------------
* Import final data
use "${dirdata}\final-data.dta", clear
* Sample
keep if myear>=1858 & myear<=1866 & prank>1 & mage<40 & fatotal!=. & sfatotal!=.
/* Note: The sample is all peers' daughters marrying in 1858-63, excluding
second marriages and women married above 40. */
* Re-calculate percentiles in new sample
drop sfATot fATot
xtile sfATot = sfatotal, nq(100)
xtile fATot = fatotal, nq(100)
gen fmissmatchA = abs(fATot-sfATot)
gen fmissmatchA2 = sfATot-fATot
* Treatment and control groups
gen prepost = 2 if myear>=1861 & myear<=1863
replace prepost = 1 if myear<1861
keep if prepost!=.
* PANEL A (ABSOLUTE VALUE)
preserve
* cumulative plot
sort fmissmatchA prepost
qui ksmirnov fmissmatchA, by(prepost) exact
local KS1`x' = round(`r(D)',0.1)
local pval1`x' = round(`r(p_exact)',0.01)
cumul fmissmatchA if prepost==1, gen(cum1)
cumul fmissmatchA if prepost==2, gen(cum2)
twoway (line cum1 fmissmatchA, c(J) lpattern(solid) lcolor(blue) lwidth(thin)) ///
(line cum2 fmissmatchA, c(J) lpattern(solid) lcolor(red) lwidth(thick)), ///
xtitle("|wife's percentile rank of acres - husband's percentile rank of acres|", size(med)) ///
ytitle("cumulative", size(med)) ///
legend(on order(1 "Control (Season)" 2 "Treatment (no Season)") rows(2) ring(0) position(5) bmargin(large) region(lstyle(none))) ///
graphregion(color(white)) ///
note("KS = `KS1' (`pval1')") name(A, replace)
restore
* PANEL B (DIFFERENCE)
preserve
* cumulative plot
sort fmissmatchA2 prepost
qui ksmirnov fmissmatchA2, by(prepost) exact
local KS2`x' = round(`r(D)',0.01)
local pval2`x' = round(`r(p_exact)',0.01)
cumul fmissmatchA2 if prepost==1, gen(cum1)
cumul fmissmatchA2 if prepost==2, gen(cum2)
twoway (line cum1 fmissmatchA2, c(J) lpattern(solid) lcolor(blue) lwidth(thin)) ///
(line cum2 fmissmatchA2, c(J) lpattern(solid) lcolor(red) lwidth(thick)) ///
(pcarrowi 0.95 -40 0.95 -60 "Marrying down", color(black)), ///
xtitle("|wife's percentile rank of acres - husband's percentile rank of acres|", size(med)) ///
xline(0, lcolor(black) lwidth(thick)) ytitle("cumulative", size(med)) ///
legend(on order(1 "Control (Season)" 2 "Treat (no Season)") rows(2) ring(0) position(5) bmargin(large) region(lstyle(none))) ///
graphregion(color(white)) ///
note("KS = `KS2' (`pval2')") name(B, replace)
restore
* COMBINE
graph combine A B, graphregion(color(white)) rows(2) altshrink
graph export "${dirresu}\Appendix\FigureB1.pdf", as(pdf) replace
graph drop _all
* ==============================================================================
* TABLE B5:
* Contingency tables separating wives' titles
* ------------------------------------------------------------------------------
* Import data and keep baseline sample
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
* Treatment indicator for non-parametric analysis
_pctile syntheticT, nq(100)
gen treatment = (syntheticT > r(r80))
/* Note: For the non-parametric estimation, I compare a high- vs. a low-treatment
cohort. The high-treatment cohort are women with a synthetic probability to
marry during the interruption above 20 percent. This corresponds to the top
quintile; the 20% of women with the highest synthetic probability to marry in
1861-63. Conversely, the low-treatment cohort are women below the top quintile. */
* Contingency tables
label var ownrank "Wife"
capture log close // save results in log file
cd "${dirresu}\Appendix"
log using TableB5.log, replace
* Panel A: Low-Treatment cohorts (T < 80th percentile)
tab ownrank grankspo if treatment==0, expected matcell(ofreqa)
* Panel B: High-Treatment cohorts (T >= 80th percentile)
tab ownrank grankspo if treatment==1, expected matcell(ofreqa)
log close
* ==============================================================================
* TABLE B6:
* Non-parametric tests for contingency Table B.V
* ------------------------------------------------------------------------------
* Import data and keep baseline sample
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
* Treatment indicator for non-parametric analysis
_pctile syntheticT, nq(100)
gen treatment = (syntheticT > r(r80))
* Non-parametric stats
forvalues i=0/1{
tab ownrank grankspo if treatment==`i', chi2 lrchi2
local chi2_`i' = round(`r(chi2)',0.01)
local chi2LR_`i' = round(`r(chi2_lr)',0.01)
local chi2p_`i' = r(p)
local chi2LRp_`i' = r(p_lr)
local df_`i' = e(df_r)
local N`i' = r(N)
ktau ownrank grankspo if treatment==`i', stats(taub)
local taub_`i' = round(`r(tau_b)',0.01)
local taubp_`i' = r(p)
}
* Add results to matrix tab5
mat define tabB6 = (.,.,.\.,.,.\.,.,.\.,.,.\.,.,.\.,.,.\.,.,.)
forvalues j =0/1{
local col = `j'+1
mat tabB6[1,`col'] = `chi2_`j''
mat tabB6[2,`col'] = `chi2p_`j''
mat tabB6[3,`col'] = `chi2LR_`j''
mat tabB6[4,`col'] = `chi2LRp_`j''
mat tabB6[5,`col'] = `taub_`j''
mat tabB6[6,`col'] = `taubp_`j''
mat tabB6[7,`col'] = `N`j''
}
mat tabB6[5,3] = `taub_0'-`taub_1'
mat tabB6[7,3] = `N0'+`N1'
* Difference chi-squared and LR-chi2
local chi2_diff = `chi2_0'-`chi2_1'
local chi2_diffp = chi2tail(3,`chi2_diff')
local chi2LR_diff = `chi2LR_0'-`chi2LR_1'
local chi2LR_diffp = chi2tail(3,`chi2LR_diff')
mat tabB6[1,3] = `chi2_diff'
mat tabB6[2,3] = `chi2_diffp'
mat tabB6[3,3] = `chi2LR_diff'
mat tabB6[4,3] = `chi2LR_diffp'
* Difference tau-b
qui ktau ownrank grankspo if treatment==0
local convtaub_0 = sin(3.141592654*`r(tau_b)'*0.5)
qui ktau ownrank grankspo if treatment==1
local convtaub_1 = sin(3.141592654*`r(tau_b)'*0.5)
cortesti `convtaub_0' `N0' `convtaub_1' `N1'
mat tabB6[6,3] = 0.1047
* Table B6
mat colnames tabB6 = "Low-treat" "High-treat" "Diff"
mat rownames tabB6 = "Pearson chi-squared" " p-val" "Likelihood ratio" " p-val" "Kendall rank correlation" " p-val" "N"
mat list tabB6
esttab matrix(tabB6, fmt(3 3)) using "${dirresu}\Appendix\TableB6.tex", mlabel("Table B6") replace
* ==============================================================================
* TABLE B7:
* Alternative contingency tables
* ------------------------------------------------------------------------------
* DATA PREPARATION
* ----------------
* Append (i) men and women who did not marry and (ii) men who failed to marry
* peers' daughters for the non-parametric analysis of appendix BIV.
* Sample 1: All women (married and unmarried)
* Import data:
use "${dirdata}\final-data.dta", clear
_pctile syntheticT, nq(100)
gen treatment = (syntheticT > r(r80)) // treatment variable for non-parametric analysis
* Sample:
keep if wom==1 & nthismarriage<2 & age1861>=15 & age1861<=35
drop if nthismarr==0 & dage<35
drop if prankspo==0
* Relabel marriage outcomes (to incorporate unmarried + spouse's rank at age 15):
gen rank_hus = 0 if nthismarr==0 | myear==.
replace rank_hus = grankspo if rank_hus==.
label define marroutcome_label 0 "Unmarried" 1 "Commoner" 2 "Gentry" 3 "Peer son" 4 "Peer heir"
label values rank_hus marroutcome_label
keep grank rank_hus treatment
rename grank rank_wif
gen married = (rank_hus==0)
gen sample = 1
* Save:
label var rank_hus "Unmarried / Husband's rank at age 15"
label var treatment "Treatment: =0 if low (T<80 pctile); =1 if high (T>80 pctile)"
label var married "Ever married (=0 no; =1 yes)"
label var sample "Sample (=1 women; =2 men)"
save "${dirdta}\sampleCT_allwomen.dta", replace
* Sample 2: All married men (incl. those who did not marry peers' daughters')
/* Note: For all women in the baseline sample, I searched the entry (refno)
their corresponding husbands in the hollingsworth dataset. I added husbands
whose wives wer not peers' daughters but who were also aged 15 to 35 in 1861.
The information is stored in the data file:
"${dirroot}\programs\03_appendix\aux data appendix\husbands_contingency_table.txt"
*/
* Import data and append to baseline sample of women
preserve
import delimited "${dirroot}\programs\03_appendix\aux data appendix\husbands_contingency_table.txt", clear
label var refno_hus "Reference number in Hollingsworth (Husbands)"
label var grank "Wife's parental rank"
label define prankspo2_lbl2 1 "Commoner dau." 2 "Gentry dau." 4 "Baron dau." 5 "Duke dau.", replace
label values grank prankspo2_lbl2
label var agewife1861 "Wife's age at the start of the 1861 Season"
label var grankspo "Peer son / Peer heir"
label define grankspohusb_lbl 3 "Peer son" 4 "Peer heir"
label values grankspo grankspohusb_lbl
label var prank_husb "Husband's parental rank"
label var rank_husb "Husband's highest title succeeded to/was heir to at 15/was created for"
label define rnk 0 "Foreign" 1 "Commoner" 2 "Knight" 3 "Baronet" 4 "Baron" 5 "Duke"
label values prank_husb rank_husb rnk
label var crank_husb "Husband's highest rank heir to at 15"
label define crnk 0 "Foreign" 1 "Pure commoner" 2 "Baron younger son" 3 "Duke younger son" 4 "Knight" 5 "Baronet" 6 "Baron" 7 "Duke" // labels
label values crank_husb crnk
label var hrank_husb "Husband's highest rank heir to at 15 (alt. categories)"
label define hnk 0 "Foreign" 1 "Commoner at 15" 2 "Knight" 3 "Baronet" 4 "Baron" 5 "Duke"
label values hrank_husb hnk
save "${dirdta}\husbands_contingency_table.dta", replace
restore
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
gen agewife1861 = age1861
append using "${dirdta}\husbands_contingency_table.dta"
* Re-label and treatment for men
label define prankspo2_lbl2 1 "Commoner dau." 2 "Gentry dau." 4 "Baron dau." 5 "Duke dau.", replace
label values grank prankspo2_lbl2
gen treatment_m = 1 if agewife1861>=19 & agewife1861<=22
replace treatment_m = 0 if agewife1861<19 | agewife1861>22
keep if grankspo>=3
rename grank marroutcome
gen agemen1861 = floor((td(31mar1861)-sbdate)/365)
replace agemen1861 = 1861-sbyear if agemen1861==.
* Save and erase temp files
label var agewife1861 "Wife's age at the start of the 1861 Season"
label var treatment_m "Treatment for men (=0 low; =1 high)"
label var agemen1861 "Men's age at the start of the 1861 Season"
save "${dirdta}\sample_CT_marriedmen.dta", replace
erase "${dirdta}\husbands_contingency_table.dta"
* Sample 3: Unmarried men
* Import data and select unmarried men aged 15-35 in 1861:
use "${dirdata}\final-data.dta", clear
keep if wom==0 & nthismarr==0
gen agemen1861 = floor((td(31mar1861)-bdate)/365)
replace agemen1861 = 1861-byear if agemen1861==.
keep if agemen1861>=15 & agemen1861<=35
replace dage = dyear-byear if dage==.
keep if dage>=35 // for consistency with women's sample
* Re-label marriage outcomes:
gen marroutcome = 0
drop grankspo
gen grankspo = 3 if crank==2 | crank==3 // husband
replace grankspo = 4 if crank==6 | crank==7
drop if prank==1
* Save:
label var agemen1861 "Men's age at the start of the 1861 Season"
label var marroutcome "Marriage outcome (0=unmarried)"
label var grankspo "Peer son / Peer heir"
label define grankspohusb_lbl 3 "Peer son" 4 "Peer heir"
label values grankspo grankspohusb_lbl
save "${dirdta}\sample_CT_unmarriedmen.dta", replace
* Append all samples
* Married and unmarried men
use "${dirdta}\sample_CT_marriedmen.dta", clear
append using "${dirdta}\sample_CT_unmarriedmen.dta"
* Re-label and treatment for men
label define marrout_men_label 0 "Unmarried" 1 "Commoner dau." 2 "Gentry dau." 4 "Baron dau." 5 "Duke dau.", replace
label values marroutcome marrout_men_label
label var marroutcome "Marriage outcome"
replace treatment_m = 1 if agemen1861>=19 & agemen1861<=22 & nthismarr==0
replace treatment_m = 0 if (agemen1861<19 & nthismarr==0) | (agemen1861>22 & nthismarr==0)
keep grankspo marroutcome treatment_m
rename grankspo rank_hus
rename marroutcome rank_wif
rename treatment_m treatment
gen married = (rank_wif==0)
gen sample = 2
drop if rank_wif==4 | rank_wif==5
* Append women, married and unmarried men
append using "${dirdta}\sampleCT_allwomen.dta"
replace rank_wif = rank_wif-1 if rank_wif>=4
label define wom_lbl 0 "No wife" 1 "Commoner dau." 2 "Gentry dau." 3 "Baron dau." 4 "Duke dau."
label values rank_wif wom_lbl
label define men_lbl 0 "No husb." 1 "Commoner" 2 "Gentry" 3 "Peer son" 4 "Peer heir"
label values rank_hus men_lbl
* Save and erase temp files
label var treatment "Treatment (0=low; 1=high)"
label var married "Ever married (0= no; 1= yes)"
label var sample "Sample (=1 women; =2 men)"
save "${dirdta}\sampleCT_allwomen-men.dta", replace
erase "${dirdta}\sample_CT_marriedmen.dta"
erase "${dirdta}\sample_CT_unmarriedmen.dta"
erase "${dirdta}\sampleCT_allwomen.dta"
* CONTINGENCY TABLES
* ------------------
* Import data
use "${dirdta}\sampleCT_allwomen-men.dta", clear
* Contingency table
forvalues T=0/1{
tab rank_wif rank_hus if treatment==`T', matcell(ofreq)
* Re-arrange cells for beautification
local ite = `T'+1
local panel : word `ite' of `c(ALPHA)'
mat define tabB7_`panel' = (.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.)
mat list tabB7_`panel'
* Re-arrange marriages
forvalues i=1/4{
local i2 = `i'+1
mat tabB7_`panel'[`i',1] = ofreq[`i2',2]
mat tabB7_`panel'[`i',2] = ofreq[`i2',3]
mat tabB7_`panel'[`i',3] = ofreq[`i2',4]
mat tabB7_`panel'[`i',4] = ofreq[`i2',5]
}
* Re-arrange N/A values
mat tabB7_`panel'[1,1] = .
mat tabB7_`panel'[2,1] = .
mat tabB7_`panel'[1,2] = .
mat tabB7_`panel'[2,2] = .
* Re-arrange totals
mat tabB7_`panel'[5,1] = tabB7_`panel'[3,1]+tabB7_`panel'[4,1]
mat tabB7_`panel'[5,2] = tabB7_`panel'[3,2]+tabB7_`panel'[4,2]
mat tabB7_`panel'[5,3] = tabB7_`panel'[1,3]+tabB7_`panel'[2,3]+tabB7_`panel'[3,3]+tabB7_`panel'[4,3]
mat tabB7_`panel'[5,4] = tabB7_`panel'[1,4]+tabB7_`panel'[2,4]+tabB7_`panel'[3,4]+tabB7_`panel'[4,4]
mat tabB7_`panel'[1,5] = tabB7_`panel'[1,3]+tabB7_`panel'[1,4]
mat tabB7_`panel'[2,5] = tabB7_`panel'[2,3]+tabB7_`panel'[2,4]
mat tabB7_`panel'[3,5] = tabB7_`panel'[3,1]+tabB7_`panel'[3,2]+tabB7_`panel'[3,3]+tabB7_`panel'[3,4]
mat tabB7_`panel'[4,5] = tabB7_`panel'[4,1]+tabB7_`panel'[4,2]+tabB7_`panel'[4,3]+tabB7_`panel'[4,4]
mat tabB7_`panel'[5,5] = tabB7_`panel'[5,1]+tabB7_`panel'[5,2]+tabB7_`panel'[5,3]+tabB7_`panel'[5,4]
* Re-arrange totals including unmarried
mat tabB7_`panel'[6,1] = tabB7_`panel'[5,1]+ofreq[1,2]
mat tabB7_`panel'[6,2] = tabB7_`panel'[5,2]+ofreq[1,3]
mat tabB7_`panel'[6,3] = tabB7_`panel'[5,3]+ofreq[1,4]
mat tabB7_`panel'[6,4] = tabB7_`panel'[5,4]+ofreq[1,5]
mat tabB7_`panel'[6,5] = tabB7_`panel'[5,5]+ofreq[1,6]
mat tabB7_`panel'[1,6] = tabB7_`panel'[1,5]+ofreq[2,1]
mat tabB7_`panel'[2,6] = tabB7_`panel'[2,5]+ofreq[3,1]
mat tabB7_`panel'[3,6] = tabB7_`panel'[3,5]+ofreq[4,1]
mat tabB7_`panel'[4,6] = tabB7_`panel'[4,5]+ofreq[5,1]
mat tabB7_`panel'[5,6] = tabB7_`panel'[5,5]+ofreq[6,1]
mat tabB7_`panel'[6,6] = `r(N)'
* Row and column labels
mat colnames tabB7_`panel' = "Commoner" "Gentry" "Peer son" "Peer heir" "Total" "Total + unmarr"
mat rownames tabB7_`panel' = "Commoner dau" "Gentry dau" "Baron/Visc dau" "Duke/Earl/Mar dau" "Total" "Total + unmarr"
}
* Table B6
esttab matrix(tabB7_A, fmt(0)) using "${dirresu}\Appendix\TableB7_A.tex", mlabel("Table B7 (Panel A)") replace
esttab matrix(tabB7_B, fmt(0)) using "${dirresu}\Appendix\TableB7_B.tex", mlabel("Table B7 (Panel B)") replace
* ==============================================================================
* TABLE B8:
* Non-parametric tests for contingency Table B7
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\sampleCT_allwomen-men.dta", clear
* Matrix to export results
mat define tabB8_ = (24.6,0.000,3.50,0.320,21.1,0.000\24.9,0.000,3.60,0.315,21.3,0.000\0.20,0.000,0.11,0.191,0.09,0.164\.,.,.,.,.,.\.,.,.,.,.,.\0.20,0.000,0.11,0.191,0.09,0.164\.,.,.,.,.,.\.,.,.,.,.,.\0.20,0.000,0.11,0.191,0.09,0.164\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.\.,.,.,.,.,.)
/* Note: Panel A's results equivalent to those reported in Table 5 in paper
and reproduced in the master-analysis.do do-file. Results on Kendall's tau-b
for Panels B and C also equivalent to those reported in Table 5 in paper
and reproduced in the master-analysis.do do-file. The reason is that:
``Kendall's rank correlation is based on the number of concordant, discordant,
and tied marriages [...] Hence, estimates are not affected whether unmarried
individuals are included or not'' (see appendix B4 for details). */
* Set globals
global rank_wif rank_wif
global rank_hus rank_hus
* Panel B
* ------------------------------------------------------------------------------
global nacells 1 // (=0 for 4x4 tables; =1 if n/a cells)
global sampleM 0 // (=0 baseline; =1 baseline+peer son)
global sampleRM 1 // (=0 baseline; =1 baseline+peer son; =2 baseline+peer son+unmarried)
if $nacells == 0{
global wom 1
global sampleM 999
global sampleRM 999
}
if $sampleM == 0 & $sampleRM == 0{
keep if $rank_wif>2 & $rank_hus>0
}
if $sampleM == 0{
global wom 3
}
if $sampleM == 1{
global wom 1
}
forvalues x = 1(-1)0{
* observed cell counts
forvalues i = 1/4{
forvalues j = 1/4{
count if treatment==`x' & $rank_wif==`i' & $rank_hus==`j'
local O`i'`j' = r(N)
}
}
* expected cell counts
forvalues z = 1/4{
local c`z' = `O1`z''+`O2`z''+`O3`z''+`O4`z''
local r`z' = `O`z'1'+`O`z'2'+`O`z'3'+`O`z'4'
}
if $sampleRM == 2{ // to include unmarried
if `x'==0{
local c3 = `c3' + 75
local c4 = `c4' + 17
local r3 = `r3' + 82
local r4 = `r4' + 59
}
if `x'==1{
local c3 = `c3' + 25
local c4 = `c4' + 3
local r3 = `r3' + 28
local r4 = `r4' + 10
}
}
local Nmen = `c1'+`c2'+`c3'+`c4'
forvalues i = 1/4{
forvalues j = 1/4{
local E`i'`j' = `r`i''*`c`j''/`Nmen'
}
}
if $nacells == 1{
local E11 = 0
local E12 = 0
local E21 = 0
local E22 = 0
}
* concordant and discordant pairs
forvalues j = 1/3{
local y = `j'+1
forvalues i = 1/3{
local C`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=`y'/4{
local C`i'`j' = `C`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
forvalues j = 4(-1)2{
local y = `j'-1
forvalues i = 1/3{
local D`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=1/`y'{
local D`i'`j' = `D`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
local C = 0
local D = 0
forvalues i=3/3{ // rows 3 (and 4, where C D YY are zeros) because sample restricted to women
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
if $sampleM == 1{ // to add all peer's sons cells
local C = `C' +`O13'*(`O24'+`O34'+`O44')+`O23'*(`O34'+`O44')
local D = `D' +`O14'*(`O23'+`O33'+`O43')+`O24'*(`O33'+`O43')
}
if $nacells == 0{ // redo for the 4x4 table
local C = 0
local D = 0
forvalues i=1/3{
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
}
* ties
local T = 0
local S = 0
if $nacells == 0 | $sampleM == 1{
forvalues i=1/4{
local j = `i'
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
local s`j' = `O1`j''+`O2`j''+`O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
if $sampleM == 0{
forvalues i=3/4{
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
}
forvalues j=1/4{
local s`j' = `O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
* total number of pairs
if $nacells == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // for 4x4 tables
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>2 & rank_hus>0 // analysis restricted to women
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 1{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // remove artificial pairs due to n/a
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2 - (`D12'+`D13'+`D14'+`D22'+`D23'+`D24'+`D32'+`D33'+`D34'-`D')
}
* Chi-2 statistics
local chi2_`x' = 0
local chi2LR_`x' = 0
forvalues i=$wom / 4{ // rows 3 and 4 because analysis restricted to women
forvalues j=1/4{
if `E`i'`j''!=0{
local chi2_`x' = `chi2_`x'' + ((`O`i'`j''-`E`i'`j'')^2)/`E`i'`j''
local chi2LR_`x' = `chi2LR_`x'' + 2*`O`i'`j''*log(0.000000000001+`O`i'`j''/`E`i'`j'')
}
}
}
* Chi-2 p-values
local pvchi2_`x' = chi2tail(3,`chi2_`x'')
local pvchi2LR_`x' = chi2tail(3,`chi2LR_`x'')
* Kendall's tau-b
local taub_`x' = (`C'-`D')/(sqrt((`Np_`x''-`T')*(`Np_`x''-`S')))
* Z-score for tau-b significance
foreach xx in vt vs sum1v1 sum2v1 sum1v2 sum2v2{
local `xx' = 0
}
local v0 = 2*`Np_`x''*(2*`n`x''+5)
if $nacells == 0 | $sampleM == 1{
local row2 = 1
}
if $sampleM == 0{
local row2 = 3
}
forvalues i=`row2'/4{
local vt = `vt'+`t`i''*(`t`i''-1)*(2*`t`i''+5)
local sum1v1 = `sum1v1'+`t`i''*(`t`i''-1)
local sum1v2 = `sum1v2'+`t`i''*(`t`i''-1)*(`t`i''-2)
}
forvalues j=1/4{
local vs = `vs'+`s`j''*(`s`j''-1)*(2*`s`j''+5)
local sum2v1 = `sum2v1'+`s`j''*(`s`j''-1)
local sum2v2 = `sum1v2'+`s`j''*(`s`j''-1)*(`s`j''-2)
}
local v1 = `sum1v1'*`sum2v1'/(2*2*`Np_`x'')
local v2 = `sum1v2'*`sum2v2'/(9*2*`Np_`x''*(`n`x''-2))
local v = (`v0'-`vt'-`vs')/18 + `v1' + `v2'
local z_`x' = (`C'-`D')/(sqrt(`v'))
local pv_`x'= 2*(normal(-`z_`x''))
}
* Chi2 difference tests
local chi2_diff = `chi2_0'-`chi2_1'
local pvchi2_diff = chi2tail(3,`chi2_diff')
local chi2LR_diff = `chi2LR_0'-`chi2LR_1'
local pvchi2LR_diff = chi2tail(3,`chi2LR_diff')
* Store results
mat tabB8_[4,1] = `chi2_0'
mat tabB8_[4,2] = `pvchi2_0'
mat tabB8_[4,3] = `chi2_1'
mat tabB8_[4,4] = `pvchi2_1'
mat tabB8_[4,5] = `chi2_0'-`chi2_1'
mat tabB8_[4,6] = `pvchi2_diff'
mat tabB8_[5,1] = `chi2LR_0'
mat tabB8_[5,2] = `pvchi2LR_0'
mat tabB8_[5,3] = `chi2LR_1'
mat tabB8_[5,4] = `pvchi2LR_1'
mat tabB8_[5,5] = `chi2LR_0'-`chi2LR_1'
mat tabB8_[5,6] = `pvchi2LR_diff'
* Panel C
* ------------------------------------------------------------------------------
global nacells 1 // (=0 for 4x4 tables; =1 if n/a cells)
global sampleM 0 // (=0 baseline; =1 baseline+peer son)
global sampleRM 2 // (=0 baseline; =1 baseline+peer son; =2 baseline+peer son+unmarried)
if $nacells == 0{
global wom 1
global sampleM 999
global sampleRM 999
}
if $sampleM == 0 & $sampleRM == 0{
keep if $rank_wif>2 & $rank_hus>0
}
if $sampleM == 0{
global wom 3
}
if $sampleM == 1{
global wom 1
}
forvalues x = 1(-1)0{
* observed cell counts
forvalues i = 1/4{
forvalues j = 1/4{
count if treatment==`x' & $rank_wif==`i' & $rank_hus==`j'
local O`i'`j' = r(N)
}
}
* expected cell counts
forvalues z = 1/4{
local c`z' = `O1`z''+`O2`z''+`O3`z''+`O4`z''
local r`z' = `O`z'1'+`O`z'2'+`O`z'3'+`O`z'4'
}
if $sampleRM == 2{ // to include unmarried
if `x'==0{
local c3 = `c3' + 75
local c4 = `c4' + 17
local r3 = `r3' + 82
local r4 = `r4' + 59
}
if `x'==1{
local c3 = `c3' + 25
local c4 = `c4' + 3
local r3 = `r3' + 28
local r4 = `r4' + 10
}
}
local Nmen = `c1'+`c2'+`c3'+`c4'
forvalues i = 1/4{
forvalues j = 1/4{
local E`i'`j' = `r`i''*`c`j''/`Nmen'
}
}
if $nacells == 1{
local E11 = 0
local E12 = 0
local E21 = 0
local E22 = 0
}
* concordant and discordant pairs
forvalues j = 1/3{
local y = `j'+1
forvalues i = 1/3{
local C`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=`y'/4{
local C`i'`j' = `C`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
forvalues j = 4(-1)2{
local y = `j'-1
forvalues i = 1/3{
local D`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=1/`y'{
local D`i'`j' = `D`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
local C = 0
local D = 0
forvalues i=3/3{ // rows 3 (and 4, where C D YY are zeros) because sample restricted to women
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
if $sampleM == 1{ // to add all peer's sons cells
local C = `C' +`O13'*(`O24'+`O34'+`O44')+`O23'*(`O34'+`O44')
local D = `D' +`O14'*(`O23'+`O33'+`O43')+`O24'*(`O33'+`O43')
}
if $nacells == 0{ // redo for the 4x4 table
local C = 0
local D = 0
forvalues i=1/3{
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
}
* ties
local T = 0
local S = 0
if $nacells == 0 | $sampleM == 1{
forvalues i=1/4{
local j = `i'
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
local s`j' = `O1`j''+`O2`j''+`O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
if $sampleM == 0{
forvalues i=3/4{
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
}
forvalues j=1/4{
local s`j' = `O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
* total number of pairs
if $nacells == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // for 4x4 tables
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>2 & rank_hus>0 // analysis restricted to women
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 1{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // remove artificial pairs due to n/a
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2 - (`D12'+`D13'+`D14'+`D22'+`D23'+`D24'+`D32'+`D33'+`D34'-`D')
}
* Chi-2 statistics
local chi2_`x' = 0
local chi2LR_`x' = 0
forvalues i=$wom / 4{ // rows 3 and 4 because analysis restricted to women
forvalues j=1/4{
if `E`i'`j''!=0{
local chi2_`x' = `chi2_`x'' + ((`O`i'`j''-`E`i'`j'')^2)/`E`i'`j''
local chi2LR_`x' = `chi2LR_`x'' + 2*`O`i'`j''*log(0.000000000001+`O`i'`j''/`E`i'`j'')
}
}
}
* Chi-2 p-values
local pvchi2_`x' = chi2tail(3,`chi2_`x'')
local pvchi2LR_`x' = chi2tail(3,`chi2LR_`x'')
* Kendall's tau-b
local taub_`x' = (`C'-`D')/(sqrt((`Np_`x''-`T')*(`Np_`x''-`S')))
* Z-score for tau-b significance
foreach xx in vt vs sum1v1 sum2v1 sum1v2 sum2v2{
local `xx' = 0
}
local v0 = 2*`Np_`x''*(2*`n`x''+5)
if $nacells == 0 | $sampleM == 1{
local row2 = 1
}
if $sampleM == 0{
local row2 = 3
}
forvalues i=`row2'/4{
local vt = `vt'+`t`i''*(`t`i''-1)*(2*`t`i''+5)
local sum1v1 = `sum1v1'+`t`i''*(`t`i''-1)
local sum1v2 = `sum1v2'+`t`i''*(`t`i''-1)*(`t`i''-2)
}
forvalues j=1/4{
local vs = `vs'+`s`j''*(`s`j''-1)*(2*`s`j''+5)
local sum2v1 = `sum2v1'+`s`j''*(`s`j''-1)
local sum2v2 = `sum1v2'+`s`j''*(`s`j''-1)*(`s`j''-2)
}
local v1 = `sum1v1'*`sum2v1'/(2*2*`Np_`x'')
local v2 = `sum1v2'*`sum2v2'/(9*2*`Np_`x''*(`n`x''-2))
local v = (`v0'-`vt'-`vs')/18 + `v1' + `v2'
local z_`x' = (`C'-`D')/(sqrt(`v'))
local pv_`x'= 2*(normal(-`z_`x''))
}
* Chi2 difference tests
local chi2_diff = `chi2_0'-`chi2_1'
local pvchi2_diff = chi2tail(3,`chi2_diff')
local chi2LR_diff = `chi2LR_0'-`chi2LR_1'
local pvchi2LR_diff = chi2tail(3,`chi2LR_diff')
* Store results
mat tabB8_[7,1] = `chi2_0'
mat tabB8_[7,2] = `pvchi2_0'
mat tabB8_[7,3] = `chi2_1'
mat tabB8_[7,4] = `pvchi2_1'
mat tabB8_[7,5] = `chi2_0'-`chi2_1'
mat tabB8_[7,6] = `pvchi2_diff'
mat tabB8_[8,1] = `chi2LR_0'
mat tabB8_[8,2] = `pvchi2LR_0'
mat tabB8_[8,3] = `chi2LR_1'
mat tabB8_[8,4] = `pvchi2LR_1'
mat tabB8_[8,5] = `chi2LR_0'-`chi2LR_1'
mat tabB8_[8,6] = `pvchi2LR_diff'
esttab matrix(tabB8_, fmt(3)), mlabel("Table B8 (Panel B)")
* Panel D
* ------------------------------------------------------------------------------
global nacells 1 // (=0 for 4x4 tables; =1 if n/a cells)
global sampleM 1 // (=0 baseline; =1 baseline+peer son)
global sampleRM 1 // (=0 baseline; =1 baseline+peer son; =2 baseline+peer son+unmarried)
if $nacells == 0{
global wom 1
global sampleM 999
global sampleRM 999
}
if $sampleM == 0 & $sampleRM == 0{
keep if $rank_wif>2 & $rank_hus>0
}
if $sampleM == 0{
global wom 3
}
if $sampleM == 1{
global wom 1
}
forvalues x = 1(-1)0{
* observed cell counts
forvalues i = 1/4{
forvalues j = 1/4{
count if treatment==`x' & $rank_wif==`i' & $rank_hus==`j'
local O`i'`j' = r(N)
}
}
* expected cell counts
forvalues z = 1/4{
local c`z' = `O1`z''+`O2`z''+`O3`z''+`O4`z''
local r`z' = `O`z'1'+`O`z'2'+`O`z'3'+`O`z'4'
}
if $sampleRM == 2{ // to include unmarried
if `x'==0{
local c3 = `c3' + 75
local c4 = `c4' + 17
local r3 = `r3' + 82
local r4 = `r4' + 59
}
if `x'==1{
local c3 = `c3' + 25
local c4 = `c4' + 3
local r3 = `r3' + 28
local r4 = `r4' + 10
}
}
local Nmen = `c1'+`c2'+`c3'+`c4'
forvalues i = 1/4{
forvalues j = 1/4{
local E`i'`j' = `r`i''*`c`j''/`Nmen'
}
}
if $nacells == 1{
local E11 = 0
local E12 = 0
local E21 = 0
local E22 = 0
}
* concordant and discordant pairs
forvalues j = 1/3{
local y = `j'+1
forvalues i = 1/3{
local C`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=`y'/4{
local C`i'`j' = `C`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
forvalues j = 4(-1)2{
local y = `j'-1
forvalues i = 1/3{
local D`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=1/`y'{
local D`i'`j' = `D`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
local C = 0
local D = 0
forvalues i=3/3{ // rows 3 (and 4, where C D YY are zeros) because sample restricted to women
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
if $sampleM == 1{ // to add all peer's sons cells
local C = `C' +`O13'*(`O24'+`O34'+`O44')+`O23'*(`O34'+`O44')
local D = `D' +`O14'*(`O23'+`O33'+`O43')+`O24'*(`O33'+`O43')
}
if $nacells == 0{ // redo for the 4x4 table
local C = 0
local D = 0
forvalues i=1/3{
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
}
* ties
local T = 0
local S = 0
if $nacells == 0 | $sampleM == 1{
forvalues i=1/4{
local j = `i'
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
local s`j' = `O1`j''+`O2`j''+`O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
if $sampleM == 0{
forvalues i=3/4{
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
}
forvalues j=1/4{
local s`j' = `O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
* total number of pairs
if $nacells == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // for 4x4 tables
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>2 & rank_hus>0 // analysis restricted to women
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 1{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // remove artificial pairs due to n/a
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2 - (`D12'+`D13'+`D14'+`D22'+`D23'+`D24'+`D32'+`D33'+`D34'-`D')
}
* Chi-2 statistics
local chi2_`x' = 0
local chi2LR_`x' = 0
forvalues i=$wom / 4{ // rows 3 and 4 because analysis restricted to women
forvalues j=1/4{
if `E`i'`j''!=0{
local chi2_`x' = `chi2_`x'' + ((`O`i'`j''-`E`i'`j'')^2)/`E`i'`j''
local chi2LR_`x' = `chi2LR_`x'' + 2*`O`i'`j''*log(0.000000000001+`O`i'`j''/`E`i'`j'')
}
}
}
* Chi-2 p-values
local pvchi2_`x' = chi2tail(3,`chi2_`x'')
local pvchi2LR_`x' = chi2tail(3,`chi2LR_`x'')
* Kendall's tau-b
local taub_`x' = (`C'-`D')/(sqrt((`Np_`x''-`T')*(`Np_`x''-`S')))
* Z-score for tau-b significance
foreach xx in vt vs sum1v1 sum2v1 sum1v2 sum2v2{
local `xx' = 0
}
local v0 = 2*`Np_`x''*(2*`n`x''+5)
if $nacells == 0 | $sampleM == 1{
local row2 = 1
}
if $sampleM == 0{
local row2 = 3
}
forvalues i=`row2'/4{
local vt = `vt'+`t`i''*(`t`i''-1)*(2*`t`i''+5)
local sum1v1 = `sum1v1'+`t`i''*(`t`i''-1)
local sum1v2 = `sum1v2'+`t`i''*(`t`i''-1)*(`t`i''-2)
}
forvalues j=1/4{
local vs = `vs'+`s`j''*(`s`j''-1)*(2*`s`j''+5)
local sum2v1 = `sum2v1'+`s`j''*(`s`j''-1)
local sum2v2 = `sum1v2'+`s`j''*(`s`j''-1)*(`s`j''-2)
}
local v1 = `sum1v1'*`sum2v1'/(2*2*`Np_`x'')
local v2 = `sum1v2'*`sum2v2'/(9*2*`Np_`x''*(`n`x''-2))
local v = (`v0'-`vt'-`vs')/18 + `v1' + `v2'
local z_`x' = (`C'-`D')/(sqrt(`v'))
local pv_`x'= 2*(normal(-`z_`x''))
}
* Chi2 difference tests
local chi2_diff = `chi2_0'-`chi2_1'
local pvchi2_diff = chi2tail(3,`chi2_diff')
local chi2LR_diff = `chi2LR_0'-`chi2LR_1'
local pvchi2LR_diff = chi2tail(3,`chi2LR_diff')
* Tau-b
* step 1: convert tau-b into Pearson's r (Kendall 1970 Rank correlation methods: p.126)
local r_0 = sin(3.1416*`taub_0'*0.5)
local r_1 = sin(3.1416*`taub_1'*0.5)
* step 2: compute Spearman’s rho (Gilpin 1993 "Table for conversion of Kendall’s tau to Spearman’s rho": p. 91)
local rho_0 = 3* `taub_0' * asin(`r_0'/2) / asin(`r_0')
local rho_1 = 3* `taub_1' * asin(`r_1'/2) / asin(`r_1')
* step 2: convert r into Fisher Zr (Rosenthal 1994 "Parametric measures of effect size": p.237)
local Zr_0 = 0.5*log((1+`r_0')/(1-`r_0'))
local Zr_1 = 0.5*log((1+`r_1')/(1-`r_1'))
local zobs = (`Zr_0'-`Zr_1') / sqrt(1/(`n0'-3) + 1/(`n1'-3))
* step 3: calculate p-value
local pvtaub_diff = 2*(1-normal(abs(`zobs')))
* Store results
mat tabB8_[10,1] = `chi2_0'
mat tabB8_[10,2] = `pvchi2_0'
mat tabB8_[10,3] = `chi2_1'
mat tabB8_[10,4] = `pvchi2_1'
mat tabB8_[10,5] = `chi2_0'-`chi2_1'
mat tabB8_[10,6] = `pvchi2_diff'
mat tabB8_[11,1] = `chi2LR_0'
mat tabB8_[11,2] = `pvchi2LR_0'
mat tabB8_[11,3] = `chi2LR_1'
mat tabB8_[11,4] = `pvchi2LR_1'
mat tabB8_[11,5] = `chi2LR_0'-`chi2LR_1'
mat tabB8_[11,6] = `pvchi2LR_diff'
mat tabB8_[12,1] = `taub_0'
mat tabB8_[12,2] = `pv_0'
mat tabB8_[12,3] = `taub_1'
mat tabB8_[12,4] = `pv_1'
mat tabB8_[12,5] = `taub_0'-`taub_1'
mat tabB8_[12,6] = `pvtaub_diff'
esttab matrix(tabB8_, fmt(3)), mlabel("Table B8 (Panel B)")
* Panel E
* ------------------------------------------------------------------------------
global nacells 1 // (=0 for 4x4 tables; =1 if n/a cells)
global sampleM 1 // (=0 baseline; =1 baseline+peer son)
global sampleRM 2 // (=0 baseline; =1 baseline+peer son; =2 baseline+peer son+unmarried)
if $nacells == 0{
global wom 1
global sampleM 999
global sampleRM 999
}
if $sampleM == 0 & $sampleRM == 0{
keep if $rank_wif>2 & $rank_hus>0
}
if $sampleM == 0{
global wom 3
}
if $sampleM == 1{
global wom 1
}
forvalues x = 1(-1)0{
* observed cell counts
forvalues i = 1/4{
forvalues j = 1/4{
count if treatment==`x' & $rank_wif==`i' & $rank_hus==`j'
local O`i'`j' = r(N)
}
}
* expected cell counts
forvalues z = 1/4{
local c`z' = `O1`z''+`O2`z''+`O3`z''+`O4`z''
local r`z' = `O`z'1'+`O`z'2'+`O`z'3'+`O`z'4'
}
if $sampleRM == 2{ // to include unmarried
if `x'==0{
local c3 = `c3' + 75
local c4 = `c4' + 17
local r3 = `r3' + 82
local r4 = `r4' + 59
}
if `x'==1{
local c3 = `c3' + 25
local c4 = `c4' + 3
local r3 = `r3' + 28
local r4 = `r4' + 10
}
}
local Nmen = `c1'+`c2'+`c3'+`c4'
forvalues i = 1/4{
forvalues j = 1/4{
local E`i'`j' = `r`i''*`c`j''/`Nmen'
}
}
if $nacells == 1{
local E11 = 0
local E12 = 0
local E21 = 0
local E22 = 0
}
* concordant and discordant pairs
forvalues j = 1/3{
local y = `j'+1
forvalues i = 1/3{
local C`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=`y'/4{
local C`i'`j' = `C`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
forvalues j = 4(-1)2{
local y = `j'-1
forvalues i = 1/3{
local D`i'`j' = 0
local z = `i'+1
forvalues i0=`z'/4{
forvalues j0=1/`y'{
local D`i'`j' = `D`i'`j''+`O`i'`j''*`O`i0'`j0''
}
}
}
}
local C = 0
local D = 0
forvalues i=3/3{ // rows 3 (and 4, where C D YY are zeros) because sample restricted to women
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
if $sampleM == 1{ // to add all peer's sons cells
local C = `C' +`O13'*(`O24'+`O34'+`O44')+`O23'*(`O34'+`O44')
local D = `D' +`O14'*(`O23'+`O33'+`O43')+`O24'*(`O33'+`O43')
}
if $nacells == 0{ // redo for the 4x4 table
local C = 0
local D = 0
forvalues i=1/3{
local C = `C' + `C`i'1'+`C`i'2'+`C`i'3'
local D = `D' + `D`i'2'+`D`i'3'+`D`i'4'
}
}
* ties
local T = 0
local S = 0
if $nacells == 0 | $sampleM == 1{
forvalues i=1/4{
local j = `i'
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
local s`j' = `O1`j''+`O2`j''+`O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
if $sampleM == 0{
forvalues i=3/4{
local t`i' = `O`i'1'+`O`i'2'+`O`i'3'+`O`i'4'
local T = `T' + `t`i''*(`t`i''-1)/2
}
forvalues j=1/4{
local s`j' = `O3`j''+`O4`j''
local S = `S' + `s`j''*(`s`j''-1)/2
}
}
* total number of pairs
if $nacells == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // for 4x4 tables
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 0{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>2 & rank_hus>0 // analysis restricted to women
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2
}
if $sampleM == 1{
qui tab $rank_wif $rank_hus if treatment==`x' & rank_wif>0 & rank_hus>0 // remove artificial pairs due to n/a
local n`x'= r(N)
local Np_`x' = `n`x''*(`n`x''-1)/2 - (`D12'+`D13'+`D14'+`D22'+`D23'+`D24'+`D32'+`D33'+`D34'-`D')
}
* Chi-2 statistics
local chi2_`x' = 0
local chi2LR_`x' = 0
forvalues i=$wom / 4{ // rows 3 and 4 because analysis restricted to women
forvalues j=1/4{
if `E`i'`j''!=0{
local chi2_`x' = `chi2_`x'' + ((`O`i'`j''-`E`i'`j'')^2)/`E`i'`j''
local chi2LR_`x' = `chi2LR_`x'' + 2*`O`i'`j''*log(0.000000000001+`O`i'`j''/`E`i'`j'')
}
}
}
* Chi-2 p-values
local pvchi2_`x' = chi2tail(3,`chi2_`x'')
local pvchi2LR_`x' = chi2tail(3,`chi2LR_`x'')
* Kendall's tau-b
local taub_`x' = (`C'-`D')/(sqrt((`Np_`x''-`T')*(`Np_`x''-`S')))
* Z-score for tau-b significance
foreach xx in vt vs sum1v1 sum2v1 sum1v2 sum2v2{
local `xx' = 0
}
local v0 = 2*`Np_`x''*(2*`n`x''+5)
if $nacells == 0 | $sampleM == 1{
local row2 = 1
}
if $sampleM == 0{
local row2 = 3
}
forvalues i=`row2'/4{
local vt = `vt'+`t`i''*(`t`i''-1)*(2*`t`i''+5)
local sum1v1 = `sum1v1'+`t`i''*(`t`i''-1)
local sum1v2 = `sum1v2'+`t`i''*(`t`i''-1)*(`t`i''-2)
}
forvalues j=1/4{
local vs = `vs'+`s`j''*(`s`j''-1)*(2*`s`j''+5)
local sum2v1 = `sum2v1'+`s`j''*(`s`j''-1)
local sum2v2 = `sum1v2'+`s`j''*(`s`j''-1)*(`s`j''-2)
}
local v1 = `sum1v1'*`sum2v1'/(2*2*`Np_`x'')
local v2 = `sum1v2'*`sum2v2'/(9*2*`Np_`x''*(`n`x''-2))
local v = (`v0'-`vt'-`vs')/18 + `v1' + `v2'
local z_`x' = (`C'-`D')/(sqrt(`v'))
local pv_`x'= 2*(normal(-`z_`x''))
}
* Chi2 difference tests
local chi2_diff = `chi2_0'-`chi2_1'
local pvchi2_diff = chi2tail(3,`chi2_diff')
local chi2LR_diff = `chi2LR_0'-`chi2LR_1'
local pvchi2LR_diff = chi2tail(3,`chi2LR_diff')
* Tau-b
* step 1: convert tau-b into Pearson's r (Kendall 1970 Rank correlation methods: p.126)
local r_0 = sin(3.1416*`taub_0'*0.5)
local r_1 = sin(3.1416*`taub_1'*0.5)
* step 2: compute Spearman’s rho (Gilpin 1993 "Table for conversion of Kendall’s tau to Spearman’s rho": p. 91)
local rho_0 = 3* `taub_0' * asin(`r_0'/2) / asin(`r_0')
local rho_1 = 3* `taub_1' * asin(`r_1'/2) / asin(`r_1')
* step 2: convert r into Fisher Zr (Rosenthal 1994 "Parametric measures of effect size": p.237)
local Zr_0 = 0.5*log((1+`r_0')/(1-`r_0'))
local Zr_1 = 0.5*log((1+`r_1')/(1-`r_1'))
local zobs = (`Zr_0'-`Zr_1') / sqrt(1/(`n0'-3) + 1/(`n1'-3))
* step 3: calculate p-value
local pvtaub_diff = 2*(1-normal(abs(`zobs')))
* Store results
mat tabB8_[13,1] = `chi2_0'
mat tabB8_[13,2] = `pvchi2_0'
mat tabB8_[13,3] = `chi2_1'
mat tabB8_[13,4] = `pvchi2_1'
mat tabB8_[13,5] = `chi2_0'-`chi2_1'
mat tabB8_[13,6] = `pvchi2_diff'
mat tabB8_[14,1] = `chi2LR_0'
mat tabB8_[14,2] = `pvchi2LR_0'
mat tabB8_[14,3] = `chi2LR_1'
mat tabB8_[14,4] = `pvchi2LR_1'
mat tabB8_[14,5] = `chi2LR_0'-`chi2LR_1'
mat tabB8_[14,6] = `pvchi2LR_diff'
mat tabB8_[15,1] = `taub_0'
mat tabB8_[15,2] = `pv_0'
mat tabB8_[15,3] = `taub_1'
mat tabB8_[15,4] = `pv_1'
mat tabB8_[15,5] = `taub_0'-`taub_1'
mat tabB8_[15,6] = `pvtaub_diff'
* TABLE B8 RESULTS
* ----------------
mat colnames tabB8_ = "Low-treat" "(pval)" "High-treat" "(pval)" "Diff" "(pval)"
mat rownames tabB8_ = "Panel A: Pearson chi2" "Panel A: LR chi2" "Panel A: Kendall taub" "Panel B: Pearson chi2" "Panel B: LR chi2" "Panel B: Kendall taub" "Panel C: Pearson chi2" "Panel C: LR chi2" "Panel C: Kendall taub" "Panel D: Pearson chi2" "Panel D: LR chi2" "Panel D: Kendall taub" "Panel E: Pearson chi2" "Panel E: LR chi2" "Panel E: Kendall taub"
esttab matrix(tabB8_, fmt(3)) using "${dirresu}\Appendix\TableB8.tex", mlabel("Table B8") replace
* erase temp files
erase "${dirdta}\sampleCT_allwomen-men.dta"
* ==============================================================================
* TABLE B9:
* Alternative outcome variables based on age at marriage
* ------------------------------------------------------------------------------
* Import data and keep baseline sample
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
* Age difference between spouses
replace mage = myear-byear if mage==.
replace smage = myear-sbyear if smage==.
gen agediff0 = smage-mage
gen agediff1 = abs(smage-mage)
label var distlondon "Distance to London"
* Regressions
global controls pr4 biorder hengpee
qui reg mage syntheticT $controls, cluster(byear)
qui summ mage if e(sample)==1
local mdepvar = round(r(mean),0.01)
outreg2 using "${dirresu}\Appendix\TableB9", keep(syntheticT distlondon) nocons label dec(3) coefastr ctitle("age at marr") addtext(Mean DP, `mdepvar', Controls, YES) tex replace
qui reg mage syntheticT $controls distlondon, cluster(byear)
qui summ mage if e(sample)==1
local mdepvar = round(r(mean),0.01)
outreg2 using "${dirresu}\Appendix\TableB9", keep(syntheticT distlondon) nocons label dec(3) coefastr ctitle("age at marr") addtext(Mean DP, `mdepvar', Controls, YES) tex
local ct = "h-w"
forvalues j=0/1{
if `j'==1{
local ct = "abs"
}
foreach x in "" distlondon{
qui reg agediff`j' syntheticT $controls `x', cluster(byear)
qui summ agediff`j' if e(sample)==1
local mdepvar`j'`x' = round(r(mean),0.01)
outreg2 using "${dirresu}\Appendix\TableB9", keep(syntheticT distlondon) nocons label dec(3) coefastr ctitle("age diff `ct'") addtext(Mean DP, `mdepvar`j'`x'', Controls, YES) tex
}
}
* ==============================================================================
* FIGURE B2:
* Distance to London, marrying a commoner, and family landholdings
* ------------------------------------------------------------------------------
* Project england map (albers projection)
cd "${dirdta}"
shp2dta using "${dirdata}\shapefiles\eng_dissolved", database(ukmap1) coordinates(ukcoord1) genid(id) replace
use ukcoord1, clear
drop if (_X>-2.63 & _X<0 & _Y<53.4937 & _Y>51.0959) | (_X>0 & _X<0.505281 & _Y>50.9604 & _Y<52.7362) | (_X>-3.3 & _X<-2 & _Y>54.3 & _Y<54.8) | (_X>-2.6 & _X<-0.47 & _Y>50.9 & _Y<54.3) | (_X>-4 & _X<-1.5 & _Y>50.8 & _Y<51.1) | (_X>-1 & _X<-0.7 & _Y>50.8 & _Y<51)
geo2xy _Y _X, replace proj(albers)
save ukcoord1, replace
* Project seats (albers projection)
use "${dirdata}/final-data-sec4.dta", clear
rename wgslat WGSlat
rename wgslon WGSlon
set obs `=_N+1' // Add London marker
replace WGSlat = 51.507222 if WGSlat==.
replace WGSlon = -0.1275 if WGSlon==.
gen london = (_n==_N)
replace fatotal = 10000 if london==1
geo2xy WGSlat WGSlon, gen(y_lat x_lon) proj(albers)
gen id_point=_n
* Map married a commoner
preserve
duplicates drop countryseat distlondon WGSlat WGSlon cOut, force
replace cOut = 3 if countryseat=="" // missing values for cOut
graph drop _all
spmap using ukcoord1, id(id_point) ///
point(xcoord(x_lon) ycoord(y_lat) sh(smdiamond smdiamond_hollow square) oc(blue%30 red%30 red) fc(blue%30 red%30 red) by(cOut)) ///
oc(black) fc(white) os(thin) ///
legend(on order(3 "Woman married in peerage" 4 "Woman married a commoner" 5 "London") rows(3)) name(gA, replace)
restore
preserve
use "${dirdata}/final-data-sec4.dta", clear
collapse cOut distlondon, by(refno)
spearman cOut distlondon
local corr = round(r(rho),0.001)
local pval = round(r(p),0.1)
twoway (scatter distlondon cOut, msymbol(circle_hollow)) ///
(lfit distlondon cOut), ///
xlabel(-1 " " 0 "M. peerage" 1 "M. commoner" 2 " ") ///
ytitle("Distance from family seat to London (miles)") ///
xtitle(" ") ///
ylabel(0(50)300) ///
text(290 1.25 "corr. 0`corr'", size(medium)) ///
text(275 1.25 "(p-value = 0`pval')", size(medium)) ///
title(" ") legend(off) name(gB, replace) graphregion(color(white))
restore
graph combine gA gB, graphregion(color(white))
graph export "${dirresu}\Appendix\FigureB2_A.pdf", as(pdf) replace
* Map landholdings
preserve
duplicates drop countryseat distlondon WGSlat WGSlon fatotal, force
spmap using ukcoord1, id(id_point) ///
point(xcoord(x_lon) ycoord(y_lat) prop(fatotal) sh(smcircle square) oc(blue%30 red) fc(blue%30 red) by(london)) ///
oc(black) fc(white) os(thin) legend(on order(3 "Family landholdings" 4 "London") rows(1)) name(gC, replace)
pwcorr fatotal distlondon, sig
local corr = round(r(rho),0.001)
mat define C = r(sig)
local pval = round(C[2,1],0.001)
twoway (scatter fatotal distlondon, msymbol(circle_hollow)) ///
(lfit fatotal distlondon), ///
xtitle("Distance from family seat to London (miles)") ///
ytitle("Family landholdings (acres)") ///
text(270000 225 "corr. `corr'", size(medium)) ///
text(250000 225 "(p-value = 0`pval')", size(medium)) ///
title(" ") legend(off) name(gD, replace) graphregion(color(white))
restore
graph combine gC gD, graphregion(color(white))
graph export "${dirresu}\Appendix\FigureB2_B.pdf", as(pdf) replace
graph drop _all
erase "${dirdta}\ukcoord1.dta"
erase "${dirdta}\ukmap1.dta"
* ==============================================================================
* TABLE B10:
* [Appendix: Political power and sorting by landholdings]
* Dep. variable is a family's political power after woman's marriage.
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}/final-data-sec4.dta", clear
* Control variables:
gen nsib2 = nsib^2 // Number of siblings squared
global controlsb distlondon i.prank nsib nsib2 // controls for regressions with brothers
global controlsh distlondon i.prank // controls for regressions with family heads
global hechter propman100 logincome propcon propnonc relig // county controls from Hechter
* Collapse data such that a woman (and her family) is the unit of observation
encode fbio1, gen(nfam) // clusters at the family level
collapse cOut fmissmatch fmissmatch2 fdown ///
syntheticT byear nfam d_MPb_aft d_MPf_bef MPb_years_aft MPf_years_bef d_MPb_locaft2 d_MPf_local2 MPb_years_locaft MPf_years_locbef d_MPh_aft d_MPh_bef MPh_years_aft MPh_years_bef ///
biorder distlondon prank nsib nsib2 $hechter, by(refno)
* Regressions
gen nfmissmatch2 = -fmissmatch2 // such that difference in landholdings is wom - husband (Panel C)
/* Note: Panel A's results equivalent to those reported in Table 6 in paper
and reproduced in the master-analysis.do do-file. */
label var cOut "W married a commoner"
global polit d_MPb_aft
global polbefore d_MPf_bef
* Col. [1]
qui ivprobit $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first twostep
weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB10_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("MP (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex replace
* Col [2]
global polit MPb_years_aft
global polbefore MPf_years_bef
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB10_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("Years (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [3]
global polit d_MPb_locaft2
global polbefore d_MPf_local2
qui ivprobit $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first twostep
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB10_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("local MP (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [4]
global polit MPb_years_locaft
global polbefore MPf_years_locbef
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB10_A", keep(cOut) nocons label nose dec(3) noas noobs ///
ctitle("lyears (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [5]
global polit d_MPh_aft
global polbefore d_MPh_bef
qui ivprobit $polit (cOut = syntheticT biorder) $hechter $controlsh $polbefore, first twostep
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsh $polbefore, vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg cOut syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = cOut $hechter $controlsh $polbefore) (cOut = syntheticT biorder $hechter $controlsh $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB10_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("MP (h)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [6]
global polit MPh_years_aft
global polbefore MPh_years_bef
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsh $polbefore, first //
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsh $polbefore, vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg cOut syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB10_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("Years (h)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
local ite = 2
foreach x in fmissmatch nfmissmatch2 fdown{
local panel : word `ite' of `c(ALPHA)'
global Mvar `x'
label var $Mvar "coef"
* Col [1]: Any brother is MP after woman's marriage
global polit d_MPb_aft
global polbefore d_MPf_bef
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg $Mvar syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = $Mvar $hechter $controlsb $polbefore) ($Mvar = syntheticT biorder $hechter $controlsb $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB10_`panel'", keep($Mvar) nocons label nose dec(3) noas noobs ///
ctitle("MP (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex replace
* Col [2]: Years MP after woman's marriage (brothers)
global polit MPb_years_aft
global polbefore MPf_years_bef
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg $Mvar syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = $Mvar $hechter $controlsb $polbefore) ($Mvar = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB10_`panel'", keep($Mvar) nocons label nose dec(3) noas noobs ///
ctitle("Years (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [3]: Any brother is local MP after woman's marriage
global polit d_MPb_locaft2
global polbefore d_MPf_local2
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg $Mvar syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = $Mvar $hechter $controlsb $polbefore) ($Mvar = syntheticT biorder $hechter $controlsb $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB10_`panel'", keep($Mvar) nocons label nose dec(3) noas noobs ///
ctitle("local MP (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [4]: Years local MP after woman's marriage (brothers)
global polit MPb_years_locaft
global polbefore MPf_years_locbef
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsb $polbefore, vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg $Mvar syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = $Mvar $hechter $controlsb $polbefore) ($Mvar = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB10_`panel'", keep($Mvar) nocons label nose dec(3) noas noobs ///
ctitle("lyears (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [5]: Family head is MP after woman's marriage
global polit d_MPh_aft
global polbefore d_MPh_bef
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsh $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsh $polbefore, vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg $Mvar syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = $Mvar $hechter $controlsh $polbefore) ($Mvar = syntheticT biorder $hechter $controlsh $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB10_`panel'", keep($Mvar) nocons label nose dec(3) noas noobs ///
ctitle("MP (h)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
* Col [6]: Years MP after woman's marriage (family head)
global polit MPh_years_aft
global polbefore MPh_years_bef
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsh $polbefore, first //
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui ivregress liml $polit ($Mvar = syntheticT biorder) $hechter $controlsh $polbefore, vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
qui reg $Mvar syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.1)
qui cmp ($polit = $Mvar $hechter $controlsb $polbefore) ($Mvar = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB10_`panel'", keep($Mvar) nocons label nose dec(3) noas noobs ///
ctitle("years (h)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', F-stat, `F1', Nb observations, `Nbobs') tex
local ite = `ite'+1
}
* ==============================================================================
* TABLE B11:
* Marriage and political power: Families where women were exposed to the
* interruption but brothers married before 1861.
* ------------------------------------------------------------------------------
* 1. DATA EXCLUDING BROTHERS MARRIED BEFORE 1861
* ----------------------------------------------
import excel "${dirdata}\biographies", sheet("Sheet1") firstrow clear
* Brother's birth year
forvalues i=1/10{
qui gen byear_b0`i' = substr(b`i'bio1,strpos(b`i'bio1,"b."),.)
qui gen byear_b`i' = substr(byear_b0`i',1,strpos(byear_b0`i',", ")-1) if regexm(byear_b0`i',", ")==1
qui replace byear_b`i' = substr(byear_b0`i',1,strpos(byear_b0`i',"Last")-1) if byear_b`i'==""
qui replace byear_b`i' = substr(byear_b`i',length(byear_b`i')-4,.)
qui replace byear_b`i' = strrtrim(byear_b`i')
qui replace byear_b`i' = strltrim(byear_b`i')
destring byear_b`i', replace
}
* Marriage year
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
qui gen myear_`x'_0 = substr(`x'bio2,strpos(`x'bio2,"married"),.) if regexm(`x'bio2,"unmarried")==0
gen myear_`x'_alt = regexs(1) if regexm(myear_`x'_0, "([0-9]?[0-9]? [A-Z]*[a-z]* [0-9][0-9][0-9][0-9])")
replace myear_`x'_alt=strtrim(myear_`x'_alt)
gen myear_`x' = substr(myear_`x'_alt,length(myear_`x'_alt)-4,.)
destring myear_`x', replace
}
* Robustness: to drop brothers who married after 1861
forvalues i=1/10{
replace b`i'bio2=" " if myear_b`i'>1861 & myear_b`i'!=.
}
drop if myear_b1>1861 & myear_b2>1861 & myear_b3>1861 & myear_b4>1861 & myear_b5>1861 & myear_b6>1861 & myear_b7>1861 & myear_b8>1861 & myear_b9>1861 & myear_b10>1861
* MP count(s) and indicator(s)
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
gen d_MP`x' = regexm(`x'bio2,"(M.P.)")
moss `x'bio2, match("(M.P.)") prefix(MP1`x')
gen MP`x' = MP1`x'count
replace MP`x' = 0 if MP`x'==.
}
gen MPb = MPb1+MPb2+MPb3+MPb4+MPb5+MPb6+MPb7+MPb8+MPb9+MPb10
gen d_MPb = (MPb>0)
* MP information (party, terms, constituency)
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
gen `x'bio2_1 = `x'bio2
gen lib`x' = 0
gen tory`x' = 0
qui sum MP`x'
local y=r(max)
forvalues j=1/`y'{
* all
gen MPall`x'_`j' = regexs(1) if regexm(`x'bio2_`j',"[Member of Parliament ]+[\(]+[M]+[\.]+[P]+[\.]+[\)]+[1]*[ ]*[for ]*(.+)")
replace MPall`x'_`j' = subinstr(MPall`x'_`j', "St.", "St",.)
replace MPall`x'_`j' = substr(MPall`x'_`j',1,strpos(MPall`x'_`j',"."))
local i = `j' + 1
gen `x'bio2_`i' = substr(`x'bio2_`j',strpos(`x'bio2_`j',"Member of Parliament")+28,.)
drop `x'bio2_`j'
* party
gen party`x'_`j' = substr(MPall`x'_`j',strpos(MPall`x'_`j',"(")+1,strpos(MPall`x'_`j',")")-2)
replace lib`x' = 1 if party`x'_`j'=="Lib" | party`x'_`j'=="Liberal" | party`x'_`j'=="Whig" | party`x'_`j'=="Liberal Union"
replace tory`x' = 1 if party`x'_`j'=="C" | party`x'_`j'=="Conservative" | party`x'_`j'=="Tory" | party`x'_`j'=="U" | party`x'_`j'=="Unionist"
* constituency
gen cnsty`x'_`j' = regexs(1) if regexm(MPall`x'_`j',"[the ]*(.+)") & party`x'_`j'==""
replace cnsty`x'_`j' = regexs(1) if regexm(MPall`x'_`j',"[\(]+[a-zA-Z]+[\)]+[ ]+[for ]*[between ]*[the ]*(.+)") & party`x'_`j'!=""
foreach endword in "between " "from " "1"{
replace cnsty`x'_`j' = strrtrim(substr(cnsty`x'_`j',1,strpos(cnsty`x'_`j',"`endword'")-1)) if regexm(cnsty`x'_`j',"`endword'")
}
replace cnsty`x'_`j' = strrtrim(substr(cnsty`x'_`j',1,length(cnsty`x'_`j')-1)) if regexm(cnsty`x'_`j',"\.") | regexm(cnsty`x'_`j',",")
replace cnsty`x'_`j' = subinstr(cnsty`x'_`j' , " in", "",.)
* term
gen term1`x'_`j' = substr(MPall`x'_`j',strpos(MPall`x'_`j',"1"),4)
gen term2`x'_`j' = substr(MPall`x'_`j',strpos(MPall`x'_`j',"1"),.)
replace term2`x'_`j' = regexs(1) if regexm(term2`x'_`j',"[0-9]*[ ]*[and ]*[to ]*[–]*[a-zA-Z]*[ ]*[a-zA-Z]*[ ]*([0-9]+)")
replace term2`x'_`j' = term1`x'_`j' if length(term2`x'_`j')<2
replace term2`x'_`j' = substr(term1`x'_`j',1,2) + substr(term2`x'_`j',1,2) if length(term2`x'_`j')==2
destring term1`x'_`j' term2`x'_`j', replace
}
}
* Housekeeping
qui sum MPb1
local y=r(max)
if `y'>0{
* Arthur Edward Holland Grey Egerton, 3rd Earl of Wilton
replace term1b1_2 = 1859 if (refno==263156 & cnstyb1_2=="Weymouth") | (refno==263157 & cnstyb1_2=="Weymouth") | (refno==263159 & cnstyb1_2=="Weymouth")
replace term2b1_2 = 1865 if (refno==263156 & cnstyb1_2=="Weymouth") | (refno==263157 & cnstyb1_2=="Weymouth") | (refno==263159 & cnstyb1_2=="Weymouth")
* Arthur Edward Holland Grey Egerton, 3rd Earl of Wilton
replace term1b1_1 = 1873 if (refno==263156 & cnstyb1_1=="Bath") | (refno==263157 & cnstyb1_1=="Bath") | (refno==263159 & cnstyb1_1=="Bath")
replace term2b1_1 = 1874 if (refno==263156 & cnstyb1_1=="Bath") | (refno==263157 & cnstyb1_1=="Bath") | (refno==263159 & cnstyb1_1=="Bath")
* George Augustus Frederick Louis Curzon-Howe, 2nd Earl Howe
replace term1b1_1 = 1857 if (refno==136403 & cnstyb1_1=="South Lancashire") | (refno==136407 & cnstyb1_1=="South Lancashire") | (refno==136408 & cnstyb1_1=="South Lancashire")
replace term2b1_1 = 1870 if (refno==136403 & cnstyb1_1=="South Lancashire") | (refno==136407 & cnstyb1_1=="South Lancashire") | (refno==136408 & cnstyb1_1=="South Lancashire")
}
qui sum MPb2
local y=r(max)
if `y'>0{
* Lord Augustus Henry Charles Hervey
replace term1b2_1 = 1864 if refno==35306 & cnstyb2_1=="West Suffolk"
replace term2b2_1 = 1875 if refno==35306 & cnstyb2_1=="West Suffolk"
}
qui sum MPb3
local y=r(max)
if `y'>0{
* Rt. Hon. Lord Frederick Charles Cavendish
replace term1b3_1 = 1865 if refno==79852 & cnstyb3_1=="North and West Ridings of Yorkshire"
replace term2b3_1 = 1882 if refno==79852 & cnstyb3_1=="North and West Ridings of Yorkshire"
* Commander Hon. Frederick Walpole
replace term1b3_1 = 1868 if refno==196005 & cnstyb3_1=="North Norfolk"
replace term2b3_1 = 1885 if refno==196005 & cnstyb3_1=="North Norfolk"
* Lord Arthur Pelham-Clinton
replace term1b3_1 = 1865 if refno==189002 & cnstyb3_1=="Newark"
replace term2b3_1 = 1868 if refno==189002 & cnstyb3_1=="Newark"
}
qui sum MPb4
local y=r(max)
if `y'>0{
* Hon. Elliot Constantine Yorke
replace term1b4_1 = 1874 if (refno==126701 & cnstyb4_1=="Cambridgeshire") | (refno==126702 & cnstyb4_1=="Cambridgeshire") | (refno==126703 & cnstyb4_1=="Cambridgeshire")
replace term2b4_1 = 1878 if (refno==126701 & cnstyb4_1=="Cambridgeshire") | (refno==126702 & cnstyb4_1=="Cambridgeshire") | (refno==126703 & cnstyb4_1=="Cambridgeshire")
}
qui sum MPb5
local y=r(max)
if `y'>0{
* Lord George Charles Gordon-Lennox
replace term1b5_1 = 1860 if (refno==209856 & cnstyb5_1=="Lymington") | (refno==209860 & cnstyb5_1=="Lymington")
replace term2b5_1 = 1874 if (refno==209856 & cnstyb5_1=="Lymington") | (refno==209860 & cnstyb5_1=="Lymington")
}
qui sum MPh
local y=r(max)
if `y'>0{
* George Augustus Frederick Louis Curzon-Howe, 2nd Earl Howe
replace term1h_1 = 1857 if (refno==136403 & cnstyh_1=="South Lancashire") | (refno==136407 & cnstyh_1=="South Lancashire") | (refno==136408 & cnstyh_1=="South Lancashire")
replace term2h_1 = 1870 if (refno==136403 & cnstyh_1=="South Lancashire") | (refno==136407 & cnstyh_1=="South Lancashire") | (refno==136408 & cnstyh_1=="South Lancashire")
}
qui sum MPf
local y=r(max)
if `y'>0{
* Hon. Robert Henry Clive
replace term1f_1 = 1832 if refno==264509 & cnstyf_1=="South Shropshire"
replace term2f_1 = 1854 if refno==264509 & cnstyf_1=="South Shropshire"
}
* MPs before and after marriage (and total years)
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
local i=1
gen MP`z'_aft = 0
gen MP`z'_bef = 0
gen MP`z'_years = 0
gen MP`z'_years_aft = 0
gen d_MP`z'_bef = 0
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
summ term1`z'_`j' term2`z'_`j'
replace MP`z'_aft = 1 if (term1`z'_`j'>=myear & term1`z'_`j'!=.) | (term2`z'_`j'>=myear & term2`z'_`j'!=.)
replace MP`z'_bef = 1 if term2`z'_`j'< myear & term2`z'_`j'!=.
replace MP`z'_years = MP`z'_years + term2`z'_`j'-term1`z'_`j' if term1`z'_`j'!=.
replace MP`z'_years_aft = MP`z'_years_aft + term2`z'_`j'-term1`z'_`j' if myear<=term1`z'_`j' & term1`z'_`j'!=.
replace MP`z'_years_aft = MP`z'_years_aft + term2`z'_`j'-myear if myear>term1`z'_`j' & myear<=term2`z'_`j' & term1`z'_`j'!=.
}
}
gen d_MP`z'_aft = (MP`z'_aft>0 & MP`z'_aft!=.)
if `y'>0{
replace d_MP`z'_bef = (term2`z'_1<myear & term2`z'_1!=.)
}
}
foreach x in aft bef years years_aft{
gen MPb_`x' = MPb1_`x'+MPb2_`x'+MPb3_`x'+MPb4_`x'+MPb5_`x'+MPb6_`x'+MPb7_`x'+MPb8_`x'+MPb9_`x'+MPb10_`x'
gen d_MPb_`x' = (MPb_`x'>0)
}
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues i=1/`y'{
* Constituencies where elected
preserve
import excel "${dirdata}\constituencies.xlsx", sheet("Sheet1") firstrow clear
rename cnsty cnsty`z'_`i'
rename countyN cnstycounty`z'_`i'
rename wgs cnstywgs`z'_`i'
label var cnsty`z'_`i' "`z''s electoral constituency (`i' M.P. occurrance)"
label var cnstycounty`z'_`i' "`z''s county of electoral constituency (`i' M.P. occurrance)"
label var cnstywgs`z'_`i' "`z''s lat & lon of electoral constituency (`i' M.P. occurrance)"
label var note "`z''s type of electoral constituency (`i' M.P. occurrance)"
label var history "`z''s electoral constituency (`i' M.P. occurrance) active from . to ."
save "${dirdta}\constituencies.dta", replace
restore
merge m:1 cnsty`z'_`i' using "${dirdta}\constituencies.dta", gen(cnsty`z'_merge)
drop if cnsty`z'_merge==2
drop cnsty`z'_merge
}
}
}
* Relation to family head
gen fhead_1872_2 = fhead_1872
replace fhead_1872_2 = "brother" if fhead_1872=="brother (1)" | fhead_1872=="brother (2)" | fhead_1872=="brother (3)" | fhead_1872=="brother (4)"
encode fhead_1872_2, gen(nfhead_1872)
* Number of brothers
gen nsib = .
forvalues i=1/10{
replace nsib = `i' if b`i'bio1!=""
}
* Save temporary data
label var refno "Reference number in Hollingsworth"
label var firstnames "First Names"
label var surname "Surname"
label var title "Title"
label var byear "Birth/baptism Year"
label var dyear "Death/burial Year"
label var myear "Marriage year"
label var wom "Woman (1=yes;0=no)"
label var mourn "Married in 1861-63"
label var cOut "Married a commoner"
label var cs "Recorded family seat indicator (1=yes;0=no)"
label var fhead_1872 "Family member who was family head in 1872 (brother 1;brother 2... father;other)"
label var fhead_1878 "Family member who was family head in 1878 (brother 1;brother 2... father;other)"
label var relation_f "Woman's family relation to f variables"
label var fh "Father is family head in 1872 (=0 no, =1 yes)"
forvalues i=1/10{
label var relation_b`i' "Woman's family relation to b`i' variables"
label var heir_b`i' "Brother `i' is heir (=0 no; =1 yes)"
label var b`i'h "Brother `i' is family head in 1872 (=0 no; =1 yes)"
}
label var relation_b11 "Woman's relation to B11 (ie, to 'other' relative who is family head in 1872)"
label var heir_b11 "B11 is heir (=0 no; =1 yes)"
label var b11bio1 "B11's short biography in thepeerage.com"
label var b11bio2 "B11's long biography in thepeerage.com"
label var b11h "B11 is family head in 1872 (=0 no; =1 yes)"
label var note "Notes on thepeerage.com biographies"
local j = -1
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
if `j'==-1{
local xlab = "Father"
}
if `j'==0{
local xlab = "Family head in 1872"
}
if `j'>0{
local xlab = "Brother `j'"
}
label var `x'bio1 "`xlab''s short biography in thepeerage.com"
label var `x'bio2 "`xlab''s long biography in thepeerage.com"
label var myear_`x'_0 "`xlab''s marriage info in thepeerage.com biography"
label var myear_`x'_alt "`xlab''s marriage date (day month year)"
label var myear_`x' "`xlab''s marriage year"
label var d_MP`x' "`xlab' was ever MP (=0 no; 1=yes)"
label var MP1`x'count "N occurrances of 'M.P.' string in long biography for `xlab'"
label var MP`x' "N occurrances of 'M.P.' string for `xlab' (incl. zeros)"
label var lib`x' "`xlab' was ever Liberal/Whig/Liberal Union MP (=0 no; 1=yes)"
label var tory`x' "`xlab' was ever Conservative/Unionist MP (=0 no; 1=yes)"
label var MP`x'_years "Total years MP (`xlab')"
label var MP`x'_years_aft "Years MP after a woman's marriage (`xlab')"
label var d_MP`x'_bef "`xlab' is MP before woman's marriage"
label var d_MP`x'_aft "`xlab' is MP after woman's marriage"
qui summ MP1`x'count
if `r(max)'>0{
forvalues k=1/`r(max)'{
local kord : word `k' of 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th
label var MP1`x'pos`k' "Position in biography of `kord' occurrance of 'M.P.' for `xlab'"
label var MPall`x'_`k' "`xlab''s MP info in thepeerage.com biography (`kord' M.P. occurrance)"
label var party`x'_`k' "`xlab''s political party (`kord' M.P. occurrance)"
label var cnsty`x'_`k' "`xlab''s electoral constituency (`kord' M.P. occurrance)"
label var term1`x'_`k' "`xlab''s term start year (`kord' M.P. occurrance)"
label var term2`x'_`k' "`xlab''s term end year (`kord' M.P. occurrance)"
label var cnstycounty`x'_`k' "`xlab''s county of electoral constituency (`kord' M.P. occurrance)"
label var cnstywgs`x'_`k' "`xlab''s lat & lon of electoral constituency (`kord' M.P. occurrance)"
}
}
local j = `j'+1
}
forvalues i=1/10{
label var byear_b0`i' "Notes on Brother `i''s birth & death info in thepeerage.com"
label var byear_b`i' "Brother `i''s birth year"
}
label var MPb "N occurrances of 'M.P.' string in long biography for all brothers"
label var d_MPb "Any brother was ever MP (=0 no; 1=yes)"
label var d_MPb_aft "Any brother is MP after woman's marriage"
label var d_MPb_bef "Any brother is MP before woman's marriage"
label var MPb_years "Total years MP (all brothers)"
label var MPb_years_aft "Years MP after a woman's marriage (all brothers)"
label var fhead_1872_2 "Family member who was family head in 1872 (brother; father; other)"
label var nfhead_1872 "Encoded family member who was family head in 1872 (brother; father; other)"
label var nsib "Number of brothers"
drop fbio2_6 hbio2_5 b1bio2_6 b3bio2_9 b6bio2_2 b7bio2_2 b9bio2_1 b10bio2_1 b2bio2_3 b4bio2_3 b5bio2_1 b8bio2_1 MPh_aft MPh_bef MPf_aft MPf_bef MPb1_aft MPb1_bef MPb2_bef MPb2_aft MPb3_aft MPb3_bef MPb4_aft MPb4_bef MPb5_aft MPb5_bef MPb6_aft MPb6_bef MPb7_aft MPb7_bef MPb8_aft MPb8_bef MPb9_aft MPb9_bef MPb10_aft MPb10_bef MPb_aft MPb_bef d_MPb_years d_MPb_years_aft
save "${dirdta}/Peers_biographies_appB11_exclbro.dta", replace
* 2. DATA EXCLUDING FAMILY HEADS MARRIED BEFORE 1861
* --------------------------------------------------
import excel "${dirdata}\biographies", sheet("Sheet1") firstrow clear
* Brother's birth year
forvalues i=1/10{
qui gen byear_b0`i' = substr(b`i'bio1,strpos(b`i'bio1,"b."),.)
qui gen byear_b`i' = substr(byear_b0`i',1,strpos(byear_b0`i',", ")-1) if regexm(byear_b0`i',", ")==1
qui replace byear_b`i' = substr(byear_b0`i',1,strpos(byear_b0`i',"Last")-1) if byear_b`i'==""
qui replace byear_b`i' = substr(byear_b`i',length(byear_b`i')-4,.)
qui replace byear_b`i' = strrtrim(byear_b`i')
qui replace byear_b`i' = strltrim(byear_b`i')
destring byear_b`i', replace
}
* Marriage year
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
qui gen myear_`x'_0 = substr(`x'bio2,strpos(`x'bio2,"married"),.) if regexm(`x'bio2,"unmarried")==0
gen myear_`x'_alt = regexs(1) if regexm(myear_`x'_0, "([0-9]?[0-9]? [A-Z]*[a-z]* [0-9][0-9][0-9][0-9])")
replace myear_`x'_alt=strtrim(myear_`x'_alt)
gen myear_`x' = substr(myear_`x'_alt,length(myear_`x'_alt)-4,.)
destring myear_`x', replace
}
* Robustness: to drop family heads marrying after 1861
replace hbio2=" " if myear_h>1861 & myear_h!=.
* MP count(s) and indicator(s)
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
gen d_MP`x' = regexm(`x'bio2,"(M.P.)")
moss `x'bio2, match("(M.P.)") prefix(MP1`x')
gen MP`x' = MP1`x'count
replace MP`x' = 0 if MP`x'==.
}
gen MPb = MPb1+MPb2+MPb3+MPb4+MPb5+MPb6+MPb7+MPb8+MPb9+MPb10
gen d_MPb = (MPb>0)
* MP information (party, terms, constituency)
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
gen `x'bio2_1 = `x'bio2
gen lib`x' = 0
gen tory`x' = 0
qui sum MP`x'
local y=r(max)
forvalues j=1/`y'{
* all
gen MPall`x'_`j' = regexs(1) if regexm(`x'bio2_`j',"[Member of Parliament ]+[\(]+[M]+[\.]+[P]+[\.]+[\)]+[1]*[ ]*[for ]*(.+)")
replace MPall`x'_`j' = subinstr(MPall`x'_`j', "St.", "St",.)
replace MPall`x'_`j' = substr(MPall`x'_`j',1,strpos(MPall`x'_`j',"."))
local i = `j' + 1
gen `x'bio2_`i' = substr(`x'bio2_`j',strpos(`x'bio2_`j',"Member of Parliament")+28,.)
drop `x'bio2_`j'
* party
gen party`x'_`j' = substr(MPall`x'_`j',strpos(MPall`x'_`j',"(")+1,strpos(MPall`x'_`j',")")-2)
replace lib`x' = 1 if party`x'_`j'=="Lib" | party`x'_`j'=="Liberal" | party`x'_`j'=="Whig" | party`x'_`j'=="Liberal Union"
replace tory`x' = 1 if party`x'_`j'=="C" | party`x'_`j'=="Conservative" | party`x'_`j'=="Tory" | party`x'_`j'=="U" | party`x'_`j'=="Unionist"
* constituency
gen cnsty`x'_`j' = regexs(1) if regexm(MPall`x'_`j',"[the ]*(.+)") & party`x'_`j'==""
replace cnsty`x'_`j' = regexs(1) if regexm(MPall`x'_`j',"[\(]+[a-zA-Z]+[\)]+[ ]+[for ]*[between ]*[the ]*(.+)") & party`x'_`j'!=""
foreach endword in "between " "from " "1"{
replace cnsty`x'_`j' = strrtrim(substr(cnsty`x'_`j',1,strpos(cnsty`x'_`j',"`endword'")-1)) if regexm(cnsty`x'_`j',"`endword'")
}
replace cnsty`x'_`j' = strrtrim(substr(cnsty`x'_`j',1,length(cnsty`x'_`j')-1)) if regexm(cnsty`x'_`j',"\.") | regexm(cnsty`x'_`j',",")
replace cnsty`x'_`j' = subinstr(cnsty`x'_`j' , " in", "",.)
* term
gen term1`x'_`j' = substr(MPall`x'_`j',strpos(MPall`x'_`j',"1"),4)
gen term2`x'_`j' = substr(MPall`x'_`j',strpos(MPall`x'_`j',"1"),.)
replace term2`x'_`j' = regexs(1) if regexm(term2`x'_`j',"[0-9]*[ ]*[and ]*[to ]*[–]*[a-zA-Z]*[ ]*[a-zA-Z]*[ ]*([0-9]+)")
replace term2`x'_`j' = term1`x'_`j' if length(term2`x'_`j')<2
replace term2`x'_`j' = substr(term1`x'_`j',1,2) + substr(term2`x'_`j',1,2) if length(term2`x'_`j')==2
destring term1`x'_`j' term2`x'_`j', replace
}
}
* Housekeeping
qui sum MPb1
local y=r(max)
if `y'>0{
* Arthur Edward Holland Grey Egerton, 3rd Earl of Wilton
replace term1b1_2 = 1859 if (refno==263156 & cnstyb1_2=="Weymouth") | (refno==263157 & cnstyb1_2=="Weymouth") | (refno==263159 & cnstyb1_2=="Weymouth")
replace term2b1_2 = 1865 if (refno==263156 & cnstyb1_2=="Weymouth") | (refno==263157 & cnstyb1_2=="Weymouth") | (refno==263159 & cnstyb1_2=="Weymouth")
* Arthur Edward Holland Grey Egerton, 3rd Earl of Wilton
replace term1b1_1 = 1873 if (refno==263156 & cnstyb1_1=="Bath") | (refno==263157 & cnstyb1_1=="Bath") | (refno==263159 & cnstyb1_1=="Bath")
replace term2b1_1 = 1874 if (refno==263156 & cnstyb1_1=="Bath") | (refno==263157 & cnstyb1_1=="Bath") | (refno==263159 & cnstyb1_1=="Bath")
* George Augustus Frederick Louis Curzon-Howe, 2nd Earl Howe
replace term1b1_1 = 1857 if (refno==136403 & cnstyb1_1=="South Lancashire") | (refno==136407 & cnstyb1_1=="South Lancashire") | (refno==136408 & cnstyb1_1=="South Lancashire")
replace term2b1_1 = 1870 if (refno==136403 & cnstyb1_1=="South Lancashire") | (refno==136407 & cnstyb1_1=="South Lancashire") | (refno==136408 & cnstyb1_1=="South Lancashire")
}
qui sum MPb2
local y=r(max)
if `y'>0{
* Lord Augustus Henry Charles Hervey
replace term1b2_1 = 1864 if refno==35306 & cnstyb2_1=="West Suffolk"
replace term2b2_1 = 1875 if refno==35306 & cnstyb2_1=="West Suffolk"
}
qui sum MPb3
local y=r(max)
if `y'>0{
* Rt. Hon. Lord Frederick Charles Cavendish
replace term1b3_1 = 1865 if refno==79852 & cnstyb3_1=="North and West Ridings of Yorkshire"
replace term2b3_1 = 1882 if refno==79852 & cnstyb3_1=="North and West Ridings of Yorkshire"
* Commander Hon. Frederick Walpole
replace term1b3_1 = 1868 if refno==196005 & cnstyb3_1=="North Norfolk"
replace term2b3_1 = 1885 if refno==196005 & cnstyb3_1=="North Norfolk"
* Lord Arthur Pelham-Clinton
replace term1b3_1 = 1865 if refno==189002 & cnstyb3_1=="Newark"
replace term2b3_1 = 1868 if refno==189002 & cnstyb3_1=="Newark"
}
qui sum MPb4
local y=r(max)
if `y'>0{
* Hon. Elliot Constantine Yorke
replace term1b4_1 = 1874 if (refno==126701 & cnstyb4_1=="Cambridgeshire") | (refno==126702 & cnstyb4_1=="Cambridgeshire") | (refno==126703 & cnstyb4_1=="Cambridgeshire")
replace term2b4_1 = 1878 if (refno==126701 & cnstyb4_1=="Cambridgeshire") | (refno==126702 & cnstyb4_1=="Cambridgeshire") | (refno==126703 & cnstyb4_1=="Cambridgeshire")
}
qui sum MPb5
local y=r(max)
if `y'>0{
* Lord George Charles Gordon-Lennox
replace term1b5_1 = 1860 if (refno==209856 & cnstyb5_1=="Lymington") | (refno==209860 & cnstyb5_1=="Lymington")
replace term2b5_1 = 1874 if (refno==209856 & cnstyb5_1=="Lymington") | (refno==209860 & cnstyb5_1=="Lymington")
}
qui sum MPh
local y=r(max)
if `y'>0{
* George Augustus Frederick Louis Curzon-Howe, 2nd Earl Howe
replace term1h_1 = 1857 if (refno==136403 & cnstyh_1=="South Lancashire") | (refno==136407 & cnstyh_1=="South Lancashire") | (refno==136408 & cnstyh_1=="South Lancashire")
replace term2h_1 = 1870 if (refno==136403 & cnstyh_1=="South Lancashire") | (refno==136407 & cnstyh_1=="South Lancashire") | (refno==136408 & cnstyh_1=="South Lancashire")
}
qui sum MPf
local y=r(max)
if `y'>0{
* Hon. Robert Henry Clive
replace term1f_1 = 1832 if refno==264509 & cnstyf_1=="South Shropshire"
replace term2f_1 = 1854 if refno==264509 & cnstyf_1=="South Shropshire"
}
* MPs before and after marriage (and total years)
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
local i=1
gen MP`z'_aft = 0
gen MP`z'_bef = 0
gen MP`z'_years = 0
gen MP`z'_years_aft = 0
gen d_MP`z'_bef = 0
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
summ term1`z'_`j' term2`z'_`j'
replace MP`z'_aft = 1 if (term1`z'_`j'>=myear & term1`z'_`j'!=.) | (term2`z'_`j'>=myear & term2`z'_`j'!=.)
replace MP`z'_bef = 1 if term2`z'_`j'< myear & term2`z'_`j'!=.
replace MP`z'_years = MP`z'_years + term2`z'_`j'-term1`z'_`j' if term1`z'_`j'!=.
replace MP`z'_years_aft = MP`z'_years_aft + term2`z'_`j'-term1`z'_`j' if myear<=term1`z'_`j' & term1`z'_`j'!=.
replace MP`z'_years_aft = MP`z'_years_aft + term2`z'_`j'-myear if myear>term1`z'_`j' & myear<=term2`z'_`j' & term1`z'_`j'!=.
}
}
gen d_MP`z'_aft = (MP`z'_aft>0 & MP`z'_aft!=.)
if `y'>0{
replace d_MP`z'_bef = (term2`z'_1<myear & term2`z'_1!=.)
}
}
foreach x in aft bef years years_aft{
gen MPb_`x' = MPb1_`x'+MPb2_`x'+MPb3_`x'+MPb4_`x'+MPb5_`x'+MPb6_`x'+MPb7_`x'+MPb8_`x'+MPb9_`x'+MPb10_`x'
gen d_MPb_`x' = (MPb_`x'>0)
}
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues i=1/`y'{
* Constituencies where elected
preserve
import excel "${dirdata}\constituencies.xlsx", sheet("Sheet1") firstrow clear
rename cnsty cnsty`z'_`i'
rename countyN cnstycounty`z'_`i'
rename wgs cnstywgs`z'_`i'
label var cnsty`z'_`i' "`z''s electoral constituency (`i' M.P. occurrance)"
label var cnstycounty`z'_`i' "`z''s county of electoral constituency (`i' M.P. occurrance)"
label var cnstywgs`z'_`i' "`z''s lat & lon of electoral constituency (`i' M.P. occurrance)"
label var note "`z''s type of electoral constituency (`i' M.P. occurrance)"
label var history "`z''s electoral constituency (`i' M.P. occurrance) active from . to ."
save "${dirdta}\constituencies.dta", replace
restore
merge m:1 cnsty`z'_`i' using "${dirdta}\constituencies.dta", gen(cnsty`z'_merge)
drop if cnsty`z'_merge==2
drop cnsty`z'_merge
}
}
}
* Relation to family head
gen fhead_1872_2 = fhead_1872
replace fhead_1872_2 = "brother" if fhead_1872=="brother (1)" | fhead_1872=="brother (2)" | fhead_1872=="brother (3)" | fhead_1872=="brother (4)"
encode fhead_1872_2, gen(nfhead_1872)
* Number of brothers
gen nsib = .
forvalues i=1/10{
replace nsib = `i' if b`i'bio1!=""
}
* Labels
label var refno "Reference number in Hollingsworth"
label var firstnames "First Names"
label var surname "Surname"
label var title "Title"
label var byear "Birth/baptism Year"
label var dyear "Death/burial Year"
label var myear "Marriage year"
label var wom "Woman (1=yes;0=no)"
label var mourn "Married in 1861-63"
label var cOut "Married a commoner"
label var cs "Recorded family seat indicator (1=yes;0=no)"
label var fhead_1872 "Family member who was family head in 1872 (brother 1;brother 2... father;other)"
label var fhead_1878 "Family member who was family head in 1878 (brother 1;brother 2... father;other)"
label var relation_f "Woman's family relation to f variables"
label var fh "Father is family head in 1872 (=0 no, =1 yes)"
forvalues i=1/10{
label var relation_b`i' "Woman's family relation to b`i' variables"
label var heir_b`i' "Brother `i' is heir (=0 no; =1 yes)"
label var b`i'h "Brother `i' is family head in 1872 (=0 no; =1 yes)"
}
label var relation_b11 "Woman's relation to B11 (ie, to 'other' relative who is family head in 1872)"
label var heir_b11 "B11 is heir (=0 no; =1 yes)"
label var b11bio1 "B11's short biography in thepeerage.com"
label var b11bio2 "B11's long biography in thepeerage.com"
label var b11h "B11 is family head in 1872 (=0 no; =1 yes)"
label var note "Notes on thepeerage.com biographies"
local j = -1
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
if `j'==-1{
local xlab = "Father"
}
if `j'==0{
local xlab = "Family head in 1872"
}
if `j'>0{
local xlab = "Brother `j'"
}
label var `x'bio1 "`xlab''s short biography in thepeerage.com"
label var `x'bio2 "`xlab''s long biography in thepeerage.com"
label var myear_`x'_0 "`xlab''s marriage info in thepeerage.com biography"
label var myear_`x'_alt "`xlab''s marriage date (day month year)"
label var myear_`x' "`xlab''s marriage year"
label var d_MP`x' "`xlab' was ever MP (=0 no; 1=yes)"
label var MP1`x'count "N occurrances of 'M.P.' string in long biography for `xlab'"
label var MP`x' "N occurrances of 'M.P.' string for `xlab' (incl. zeros)"
label var lib`x' "`xlab' was ever Liberal/Whig/Liberal Union MP (=0 no; 1=yes)"
label var tory`x' "`xlab' was ever Conservative/Unionist MP (=0 no; 1=yes)"
label var MP`x'_years "Total years MP (`xlab')"
label var MP`x'_years_aft "Years MP after a woman's marriage (`xlab')"
label var d_MP`x'_bef "`xlab' is MP before woman's marriage"
label var d_MP`x'_aft "`xlab' is MP after woman's marriage"
if `j'<10{
qui summ MP1`x'count
if `r(max)'>0{
forvalues k=1/`r(max)'{
local kord : word `k' of 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th
label var MP1`x'pos`k' "Position in biography of `kord' occurrance of 'M.P.' for `xlab'"
label var MPall`x'_`k' "`xlab''s MP info in thepeerage.com biography (`kord' M.P. occurrance)"
label var party`x'_`k' "`xlab''s political party (`kord' M.P. occurrance)"
label var cnsty`x'_`k' "`xlab''s electoral constituency (`kord' M.P. occurrance)"
label var term1`x'_`k' "`xlab''s term start year (`kord' M.P. occurrance)"
label var term2`x'_`k' "`xlab''s term end year (`kord' M.P. occurrance)"
label var cnstycounty`x'_`k' "`xlab''s county of electoral constituency (`kord' M.P. occurrance)"
label var cnstywgs`x'_`k' "`xlab''s lat & lon of electoral constituency (`kord' M.P. occurrance)"
}
}
}
local j = `j'+1
}
forvalues i=1/10{
label var byear_b0`i' "Notes on Brother `i''s birth & death info in thepeerage.com"
label var byear_b`i' "Brother `i''s birth year"
}
label var MPb "N occurrances of 'M.P.' string in long biography for all brothers"
label var d_MPb "Any brother was ever MP (=0 no; 1=yes)"
label var d_MPb_aft "Any brother is MP after woman's marriage"
label var d_MPb_bef "Any brother is MP before woman's marriage"
label var MPb_years "Total years MP (all brothers)"
label var MPb_years_aft "Years MP after a woman's marriage (all brothers)"
label var fhead_1872_2 "Family member who was family head in 1872 (brother; father; other)"
label var nfhead_1872 "Encoded family member who was family head in 1872 (brother; father; other)"
label var nsib "Number of brothers"
drop b8bio2_2 b5bio2_2 b4bio2_4 b2bio2_5 fbio2_6 hbio2_5 b1bio2_6 b3bio2_9 b6bio2_2 b7bio2_2 b9bio2_1 b10bio2_1 MPh_aft MPh_bef MPf_aft MPf_bef MPb1_aft MPb1_bef MPb2_bef MPb2_aft MPb3_aft MPb3_bef MPb4_aft MPb4_bef MPb5_aft MPb5_bef MPb6_aft MPb6_bef MPb7_aft MPb7_bef MPb8_aft MPb8_bef MPb9_aft MPb9_bef MPb10_aft MPb10_bef MPb_aft MPb_bef d_MPb_years d_MPb_years_aft
* Robustness: to drop family heads marrying after 1861
drop if myear_h>1861
* Save temporary data
save "${dirdta}/Peers_biographies_appB11_exclheads.dta", replace
erase "${dirdta}/constituencies.dta"
* 3. DATA FOR REGS EXCLUDING BROTHERS M BEFORE 1861
* --------------------------------------------------
* Import data with geo-located family seats
use "${dirdta}\eduXseatsaverages.dta", clear
* Merge with Hollingsworth final data
gen nthismarriage = 1 // by construction, all women in this sample correspond to first-marriages
merge m:1 refno nthismarriage using "${dirdata}\final-data.dta", keep(mas mat) nogen
* Merge with Peers' biographies data (excluding brothers)
merge m:1 refno using "${dirdta}/Peers_biographies_appB11_exclbro.dta", nogen
drop distlondon // use distance from each seat to London
rename distlondon2 distlondon
* Local political power
/* Note: This is defined here to use geo-located family seats and check
whether individuals were elected MP in the county where the family seat is
located */
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8{
gen MP`z'_local = 0
gen MP`z'_locaft = 0
qui sum MP`z'
local y = r(max)
forvalues i=1/`y'{
replace MP`z'_local = 1 if cnstycounty`z'_`i'==countyN
replace MP`z'_locaft = 1 if (cnstycounty`z'_`i'==countyN & term2`z'_`i'>=myear) | (cnstycounty`z'_`i'==countyN & term1`z'_`i'>=myear)
}
gen d_MP`z'_local = (MP`z'_local>0)
gen d_MP`z'_locaft = (MP`z'_locaft>0)
}
* Years local political power
/* Note: idem */
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
gen MP`z'_years_loc = 0
gen MP`z'_years_locaft = 0
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
summ term1`z'_`j' term2`z'_`j'
replace MP`z'_years_loc = MP`z'_years_loc+term2`z'_`j'-term1`z'_`j' if term1`z'_`j'!=. & cnstycounty`z'_`j'==countyN
replace MP`z'_years_locaft = MP`z'_years_locaft + term2`z'_`j'-term1`z'_`j' if myear<=term1`z'_`j' & term1`z'_`j'!=. & cnstycounty`z'_`j'==countyN
replace MP`z'_years_locaft = MP`z'_years_locaft + term2`z'_`j'-myear if myear>term1`z'_`j' & myear<=term2`z'_`j' & term1`z'_`j'!=. & cnstycounty`z'_`j'==countyN
}
}
}
foreach x in years_loc years_locaft{
gen MPb_`x' = MPb1_`x'+MPb2_`x'+MPb3_`x'+MPb4_`x'+MPb5_`x'+MPb6_`x'+MPb7_`x'+MPb8_`x'+MPb9_`x'+MPb10_`x'
}
gen MPb_local = MPb1_local+MPb2_local+MPb3_local+MPb4_local+MPb5_local+MPb6_local+MPb7_local+MPb8_local
gen d_MPb_local = (MPb_local>0)
gen MPb_locaft = MPb1_locaft+MPb2_locaft+MPb3_locaft+MPb4_locaft+MPb5_locaft+MPb6_locaft+MPb7_locaft+MPb8_locaft
gen d_MPb_locaft = (MPb_locaft>0)
drop MPb1_locaft d_MPb1_locaft MPb2_locaft d_MPb2_locaft MPb3_locaft d_MPb3_locaft MPb4_locaft d_MPb4_locaft MPb5_locaft d_MPb5_locaft MPb6_locaft d_MPb6_locaft MPb7_locaft d_MPb7_locaft MPb8_locaft d_MPb8_locaft MPb1_local MPb2_local MPb3_local MPb4_local MPb5_local MPb6_local MPb7_local MPb8_local d_MPb1_local d_MPb2_local d_MPb3_local d_MPb4_local d_MPb5_local d_MPb6_local d_MPb7_local d_MPb8_local
bysort refno: egen d_MPb_locaft2 = max(d_MPb_locaft) // for women with more than one seat, take the max local political power (this variable is then collapsed)
bysort refno: egen d_MPf_local2 = max(d_MPf_local)
* Labels
label var d_MPb_aft "Any brother is MP after woman's marriage"
label var d_MPf_bef "MP indicator before woman's marriage"
label var MPb_years_aft "Years MP after a woman's marriage (all brothers)"
gen MPf_years_bef = MPf_years-MPf_years_aft
label var MPf_years_bef "Years MP before a woman's marriage"
label var d_MPb_locaft "Any brother is local MP after woman's marriage"
label var d_MPb_locaft2 "Any brother is local MP after woman's marriage"
label var d_MPf_local2 "Local MP indicator before woman's marriage"
label var MPb_years_locaft "Years local MP after a woman's marriage (all brothers)"
gen MPf_years_locbef = MPf_years_loc-MPf_years_locaft
label var MPf_years_locbef "Years local MP before a woman's marriage (all brothers)"
label var d_MPh_aft "Family head is MP after woman's marriage"
label var d_MPh_bef "Family head is MP before woman's marriage"
label var MPh_years_aft "Years MP after a woman's marriage (family heads)"
gen MPh_years_bef = MPh_years-MPh_years_aft
label var MPh_years_bef "Years MP before a woman's marriage (family heads)"
label var d_MPf_local "Local MP indicator before woman's marriage, by family seat"
label var d_MPf_local2 "Local MP indicator before woman's marriage, by woman"
label var MPb_years_loc "Total years local MP (all brothers)"
label var MPb_local "N occurrances of 'M.P.' string in same county as family seat (all brothers)"
label var d_MPb_local "Any brother was ever local MP (=0 no; 1=yes)"
label var MPb_locaft "N occurrances of 'M.P.' string in same county as family seat (all brothers)"
label var d_MPb_locaft "Any brother is local MP after woman's marriage, by family seat"
label var d_MPb_locaft2 "Any brother is local MP after woman's marriage, by woman"
label var MPf_years_bef "Years MP before a woman's marriage"
label var MPf_years_locbef "Years local MP before a woman's marriage (father)"
label var MPh_years_bef "Years MP before a woman's marriage (family heads)"
local j = -1
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
if `j'==-1{
local xlab = "Father"
}
if `j'==0{
local xlab = "Family head in 1872"
}
if `j'>0{
local xlab = "Brother `j'"
}
label var MP`x'_years_loc "Total years local MP (`xlab'), by family seat"
label var MP`x'_years_locaft "Years local MP after woman's marriage (`xlab'), by family seat"
local j = `j'+1
}
label var nthismarriage "Number this marriage"
drop d_MPh_local MPh_local d_MPh_locaft MPh_locaft MPf_local MPf_locaft d_MPf_locaft
save "${dirdta}/final-data-sec4-robust1.dta", replace
* 4. DATA FOR REGS EXCLUDING FAMILY HEADS M BEFORE 1861
* -----------------------------------------------------
* Import data with geo-located family seats
use "${dirdta}\eduXseatsaverages.dta", clear
* Merge with Hollingsworth final data
gen nthismarriage = 1 // by construction, all women in this sample correspond to first-marriages
merge m:1 refno nthismarriage using "${dirdata}\final-data.dta", keep(mas mat) nogen
* Merge with Peers' biographies data (excluding brothers)
merge m:1 refno using "${dirdta}/Peers_biographies_appB11_exclheads.dta", nogen
drop distlondon // use distance from each seat to London
rename distlondon2 distlondon
* Local political power
/* Note: This is defined here to use geo-located family seats and check
whether individuals were elected MP in the county where the family seat is
located */
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8{
gen MP`z'_local = 0
gen MP`z'_locaft = 0
qui sum MP`z'
local y = r(max)
forvalues i=1/`y'{
replace MP`z'_local = 1 if cnstycounty`z'_`i'==countyN
replace MP`z'_locaft = 1 if (cnstycounty`z'_`i'==countyN & term2`z'_`i'>=myear) | (cnstycounty`z'_`i'==countyN & term1`z'_`i'>=myear)
}
gen d_MP`z'_local = (MP`z'_local>0)
gen d_MP`z'_locaft = (MP`z'_locaft>0)
}
* Years local political power
/* Note: idem */
foreach z in h f b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
gen MP`z'_years_loc = 0
gen MP`z'_years_locaft = 0
qui sum MP`z'
local y = r(max)
if `y'>0{
forvalues j = 1/`y'{
summ term1`z'_`j' term2`z'_`j'
replace MP`z'_years_loc = MP`z'_years_loc+term2`z'_`j'-term1`z'_`j' if term1`z'_`j'!=. & cnstycounty`z'_`j'==countyN
replace MP`z'_years_locaft = MP`z'_years_locaft + term2`z'_`j'-term1`z'_`j' if myear<=term1`z'_`j' & term1`z'_`j'!=. & cnstycounty`z'_`j'==countyN
replace MP`z'_years_locaft = MP`z'_years_locaft + term2`z'_`j'-myear if myear>term1`z'_`j' & myear<=term2`z'_`j' & term1`z'_`j'!=. & cnstycounty`z'_`j'==countyN
}
}
}
foreach x in years_loc years_locaft{
gen MPb_`x' = MPb1_`x'+MPb2_`x'+MPb3_`x'+MPb4_`x'+MPb5_`x'+MPb6_`x'+MPb7_`x'+MPb8_`x'+MPb9_`x'+MPb10_`x'
}
gen MPb_local = MPb1_local+MPb2_local+MPb3_local+MPb4_local+MPb5_local+MPb6_local+MPb7_local+MPb8_local
gen d_MPb_local = (MPb_local>0)
gen MPb_locaft = MPb1_locaft+MPb2_locaft+MPb3_locaft+MPb4_locaft+MPb5_locaft+MPb6_locaft+MPb7_locaft+MPb8_locaft
gen d_MPb_locaft = (MPb_locaft>0)
drop MPb1_locaft d_MPb1_locaft MPb2_locaft d_MPb2_locaft MPb3_locaft d_MPb3_locaft MPb4_locaft d_MPb4_locaft MPb5_locaft d_MPb5_locaft MPb6_locaft d_MPb6_locaft MPb7_locaft d_MPb7_locaft MPb8_locaft d_MPb8_locaft MPb1_local MPb2_local MPb3_local MPb4_local MPb5_local MPb6_local MPb7_local MPb8_local d_MPb1_local d_MPb2_local d_MPb3_local d_MPb4_local d_MPb5_local d_MPb6_local d_MPb7_local d_MPb8_local
bysort refno: egen d_MPb_locaft2 = max(d_MPb_locaft) // for women with more than one seat, take the max local political power (this variable is then collapsed)
bysort refno: egen d_MPf_local2 = max(d_MPf_local)
* Labels
* Labels
label var d_MPb_aft "Any brother is MP after woman's marriage"
label var d_MPf_bef "MP indicator before woman's marriage"
label var MPb_years_aft "Years MP after a woman's marriage (all brothers)"
gen MPf_years_bef = MPf_years-MPf_years_aft
label var MPf_years_bef "Years MP before a woman's marriage"
label var d_MPb_locaft "Any brother is local MP after woman's marriage"
label var d_MPb_locaft2 "Any brother is local MP after woman's marriage"
label var d_MPf_local2 "Local MP indicator before woman's marriage"
label var MPb_years_locaft "Years local MP after a woman's marriage (all brothers)"
gen MPf_years_locbef = MPf_years_loc-MPf_years_locaft
label var MPf_years_locbef "Years local MP before a woman's marriage (all brothers)"
label var d_MPh_aft "Family head is MP after woman's marriage"
label var d_MPh_bef "Family head is MP before woman's marriage"
label var MPh_years_aft "Years MP after a woman's marriage (family heads)"
gen MPh_years_bef = MPh_years-MPh_years_aft
label var MPh_years_bef "Years MP before a woman's marriage (family heads)"
label var d_MPf_local "Local MP indicator before woman's marriage, by family seat"
label var d_MPf_local2 "Local MP indicator before woman's marriage, by woman"
label var MPb_years_loc "Total years local MP (all brothers)"
label var MPb_local "N occurrances of 'M.P.' string in same county as family seat (all brothers)"
label var d_MPb_local "Any brother was ever local MP (=0 no; 1=yes)"
label var MPb_locaft "N occurrances of 'M.P.' string in same county as family seat (all brothers)"
label var d_MPb_locaft "Any brother is local MP after woman's marriage, by family seat"
label var d_MPb_locaft2 "Any brother is local MP after woman's marriage, by woman"
label var MPf_years_bef "Years MP before a woman's marriage"
label var MPf_years_locbef "Years local MP before a woman's marriage (father)"
label var MPh_years_bef "Years MP before a woman's marriage (family heads)"
local j = -1
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9 b10{
if `j'==-1{
local xlab = "Father"
}
if `j'==0{
local xlab = "Family head in 1872"
}
if `j'>0{
local xlab = "Brother `j'"
}
label var MP`x'_years_loc "Total years local MP (`xlab'), by family seat"
label var MP`x'_years_locaft "Years local MP after woman's marriage (`xlab'), by family seat"
local j = `j'+1
}
label var nthismarriage "Number this marriage"
drop d_MPh_local MPh_local d_MPh_locaft MPh_locaft MPf_local MPf_locaft d_MPf_locaft
save "${dirdta}/final-data-sec4-robust2.dta", replace
* 5. REGRESSIONS (BROTHERS)
* ----------------------------------------------
* Import data
use "${dirdta}/final-data-sec4-robust1.dta", clear
* Control variables:
gen nsib2 = nsib^2 // Number of siblings squared
global controlsb distlondon i.prank nsib nsib2 // controls for regressions with brothers
global controlsh distlondon i.prank // controls for regressions with family heads
global hechter propman100 logincome propcon propnonc relig // county controls from Hechter
* Collapse data such that a woman (and her family) is the unit of observation
encode fbio1, gen(nfam) // clusters at the family level
collapse cOut syntheticT byear nfam d_MPb_aft d_MPf_bef MPb_years_aft MPf_years_bef d_MPb_locaft2 d_MPf_local2 MPb_years_locaft MPf_years_locbef d_MPh_aft d_MPh_bef MPh_years_aft MPh_years_bef ///
biorder distlondon prank nsib nsib2 $hechter, by(refno)
label var cOut "W married a commoner"
label var syntheticT "Treatment"
* Col [1]: Any brother is MP after woman's marriage
global polit d_MPb_aft
global polbefore d_MPf_bef
* CLR p-values
qui ivprobit $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first twostep
weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
summ $polit if e(sample)==1
local mdepvar = round(r(mean),0.01)
* Second stage coefficients
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB11_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("MP (b)") addtext(CLR-pval, `CLR1', Mean DP, `mdepvar', Nb observations, `Nbobs') tex replace
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
outreg2 using "${dirresu}\Appendix\TableB11_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex replace
* Col [2]: Years MP after woman's marriage (brothers)
global polit MPb_years_aft
global polbefore MPf_years_bef
* CLR p-values
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
summ $polit if e(sample)==1
local mdepvar = round(r(mean),0.01)
* Second stage coefficients
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB11_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("Years (b)") addtext(CLR-pval, `CLR1', Mean DP, `mdepvar', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
outreg2 using "${dirresu}\Appendix\TableB11_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* Col [3]: Any brother is local MP after woman's marriage
global polit d_MPb_locaft2
global polbefore d_MPf_local2
* CLR p-values
qui ivprobit $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first twostep
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
summ $polit if e(sample)==1
local mdepvar = round(r(mean),0.01)
* Second stage coefficients
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB11_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("local MP (b)") addtext(CLR-pval, `CLR1', Mean DP, `mdepvar', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
outreg2 using "${dirresu}\Appendix\TableB11_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* Col [4]: Years local MP after woman's marriage (brothers)
global polit MPb_years_locaft
global polbefore MPf_years_locbef
* CLR p-values
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
summ $polit if e(sample)==1
local mdepvar = round(r(mean),0.01)
* Second stage coefficients
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB11_A", keep(cOut) nocons label nose dec(3) noas noobs ///
ctitle("loc years (b)") addtext(CLR-pval, `CLR1', Mean DP, `mdepvar', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.1)
outreg2 using "${dirresu}\Appendix\TableB11_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* 5. REGRESSIONS (FAMILY HEADS)
* ----------------------------------------------
* Import data
use "${dirdta}/final-data-sec4-robust2.dta", clear
* Control variables:
gen nsib2 = nsib^2 // Number of siblings squared
global controlsb distlondon i.prank nsib nsib2 // controls for regressions with brothers
global controlsh distlondon i.prank // controls for regressions with family heads
global hechter propman100 logincome propcon propnonc relig // county controls from Hechter
* Collapse data such that a woman (and her family) is the unit of observation
encode fbio1, gen(nfam) // clusters at the family level
collapse cOut syntheticT byear nfam d_MPb_aft d_MPf_bef MPb_years_aft MPf_years_bef d_MPb_locaft2 d_MPf_local2 MPb_years_locaft MPf_years_locbef d_MPh_aft d_MPh_bef MPh_years_aft MPh_years_bef ///
biorder distlondon prank nsib nsib2 $hechter, by(refno)
label var cOut "W married a commoner"
label var syntheticT "Treatment"
* Col [5]: Family head is MP after woman's marriage
global polit d_MPh_aft
global polbefore d_MPh_bef
* CLR p-values
qui ivprobit $polit (cOut = syntheticT biorder) $hechter $controlsh $polbefore, first twostep
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
summ $polit if e(sample)==1
local mdepvar = round(r(mean),0.01)
* Second stage coefficients
qui cmp ($polit = cOut $hechter $controlsh $polbefore) (cOut = syntheticT biorder $hechter $controlsh $polbefore), ind(4 1) quietly cluster(byear)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Appendix\TableB11_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("MP (h)") addtext(CLR-pval, `CLR1', Mean DP, `mdepvar', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.1)
outreg2 using "${dirresu}\Appendix\TableB11_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* Col [6]: Years MP after woman's marriage (family head)
global polit MPh_years_aft
global polbefore MPh_years_bef
* CLR p-values
qui ivregress liml $polit (cOut = syntheticT biorder) $hechter $controlsh $polbefore, first //
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
* Second stage coefficients
qui cmp ($polit = cOut $hechter $controlsb $polbefore) (cOut = syntheticT biorder $hechter $controlsb $polbefore), ind(1 1) quietly cluster(byear)
summ $polit if e(sample)==1
local mdepvar = round(r(mean),0.01)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableB11_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("Years (h)") addtext(CLR-pval, `CLR1', Mean DP, `mdepvar', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.1)
outreg2 using "${dirresu}\Appendix\TableB11_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* erase temp files
erase "${dirdta}\Peers_biographies_appB11_exclbro.dta"
erase "${dirdta}\Peers_biographies_appB11_exclheads.dta"
erase "${dirdta}\eduXseatsaverages.dta"
* ==============================================================================
* FIGURE B3:
* Age difference between brothers and sisters.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}/Peers_biographies.dta", clear
* Generate age difference between women and her brothers
local i = 1
forvalues i=1/10{
gen year0_`i' = byear-byear_b`i'
}
* Reshape to have one entry for each woman-brother pair
keep refno cOut myear MPb year0_1 year0_2 year0_3 year0_4 year0_5 year0_6 year0_7 year0_8 year0_9 year0_10
reshape long year0_ , i(refno) j(j)
gen MPi = j
drop j
drop if year0_ == .
* Figure
label var year0_ "age difference with brothers"
kdensity year0_ if cOut==0, lcolor(blue) ///
addplot(kdensity year0_ if cOut==1, lcolor(red)) ///
graphregion(color(white)) title("") xtitle("") legend(on order(1 "sister married in the peerage" 2 "sister married a commoner") rows(1)) name(p1, replace)
gen year_0 = year0_ if cOut==0
gen year_1 = year0_ if cOut==1
graph hbox year_0 year_1, box(1, fcolor(white) lcolor(blue)) box(2, fcolor(white) lcolor(red)) ///
noout graphregion(color(white)) name(p2, replace)
grc1leg p1 p2, legendfrom(p1) graphregion(color(white)) title("Age difference with brothers", color(black))
graph export "${dirresu}\Appendix\FigureB3.pdf", as(pdf) replace
erase "${dirdta}/Peers_biographies.dta" // not used any more
* ==============================================================================
* FIGURE B4:
* Loss of political power, men
* ------------------------------------------------------------------------------
* Import data
import excel "${dirroot}\programs\03_appendix\aux data appendix\biographies_men_sample_appB10", first clear
/* Note: This file contains the data used in Appendix B10: the biographies
of 389 men marrying during the interruption of the Season (1861-63) and
their brothers (offspring universe, N=288) and their fathers (fathers
universe, N=101). */
* Was the father an heir?
set more off
drop if fbio1==""
gen heir_f = 0
replace heir_f = 1 if regexm(fbio1,"Baron")==1 | regexm(fbio1,"Earl")==1 | regexm(fbio1,"Viscount")==1 | regexm(fbio1,"Marquess")==1 | regexm(fbio1,"Duke")==1 | regexm(fbio1,"Lord")==1
replace heir_f = 1 if refno==335100
* Age at death of brothers
foreach x in b1 b2 b3 b4 b5 b6 b7 b8 b9{
foreach dd in b d{
gen `x'`dd'year = regexs(1) if regexm(`x'bio1,"[`dd']+[\.]+[ ]+[0-9]?[0-9]?[ ]*[a-zA-Z]*[ ]*([0-9]+)")
destring `x'`dd'year, replace
}
replace `x'byear = 1800 + `x'byear if `x'byear<100
replace `x'dyear = 1800 + `x'dyear if `x'dyear<100 & `x'dyear>=`x'byear-1800
replace `x'dyear = 1900 + `x'dyear if `x'dyear<100 & `x'dyear<`x'byear-1800
gen `x'dage = `x'dyear-`x'byear
}
replace b1dage = 1000 if refno==20203 | refno==32050
replace b2dage = 1000 if refno==153950 | refno==160900 | refno==301700
replace b3dage = 1000 if refno==229657 | refno==357000
replace b4dage = 1000 if refno==162150 | refno==357000
foreach x in b1 b2 b3 b4 b5 b6 b7 b8 b9{
replace `x'bio2 = "" if `x'dage<21 | `x'dage==.
}
* Positions in the administration
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9{
* Academic
moss `x'bio2, match("(F.R.S.)") prefix(`x'acd1)
moss `x'bio2, match("(F.S.A.)") prefix(`x'acd2)
moss `x'bio2, match("President of the British Archaeological Association") prefix(`x'acd3)
moss `x'bio2, match("Trustee of the British Museum") prefix(`x'acd4)
moss `x'bio2, match("Trustee of the National Gallery") prefix(`x'acd5)
gen `x'acd = `x'acd1count+`x'acd2count+`x'acd3count+`x'acd4count+`x'acd5count
* Colonial
moss `x'bio2, match("Colonial Secretary") prefix(`x'col1)
moss `x'bio2, match("Governor and Commander-in-Chief of Madras") prefix(`x'col2)
moss `x'bio2, match("Parliamentary Under-Secretary, Colonies") prefix(`x'col3)
moss `x'bio2, match("Under-Secretary of the Colonies") prefix(`x'col4)
moss `x'bio2, match("U-Sec Colonies") prefix(`x'col5)
gen `x'col = `x'col1count+`x'col2count+`x'col3count+`x'col4count+`x'col5count
* Diplomatic
moss `x'bio2, match("Ambassador") prefix(`x'dip1)
moss `x'bio2, match("Ambassador Extraordinary to the Ottoman Empire") prefix(`x'dip2)
moss `x'bio2, match("British Minister at") prefix(`x'dip3)
moss `x'bio2, match("Envoy") prefix(`x'dip4)
moss `x'bio2, match("Envoy Extraordinary and Minister Plenipotentiary") prefix(`x'dip5)
moss `x'bio2, match("Secretary of Embassy") prefix(`x'dip6)
moss `x'bio2, match("Secretary of Legation") prefix(`x'dip7)
moss `x'bio2, match("Under-Secretary of Foreign Affairs") prefix(`x'dip8)
moss `x'bio2, match("U-Sec For Affrs") prefix(`x'dip9)
gen `x'dip = `x'dip1count+`x'dip2count+`x'dip3count+`x'dip4count+`x'dip5count+`x'dip6count+`x'dip7count+`x'dip8count+`x'dip9count
* Great Officer of State
moss `x'bio2, match("High Steward") prefix(`x'gos1)
moss `x'bio2, match("Lord Chancellor") prefix(`x'gos2)
moss `x'bio2, match("Lord of the Treasury") prefix(`x'gos3)
moss `x'bio2, match("Lord Privy Seal") prefix(`x'gos4)
moss `x'bio2, match("First Lord of the Admiralty") prefix(`x'gos5)
gen `x'gos = `x'gos1count+`x'gos2count+`x'gos3count+`x'gos4count+`x'gos5count
moss `x'bio2, match("Lord President of the Council") prefix(w`x'gos1)
moss `x'bio2, match("Lord Great Chamberlain") prefix(w`x'gos2)
moss `x'bio2, match("Earl Marshal") prefix(w`x'gos3)
moss `x'bio2, match("Lord High Constable") prefix(w`x'gos4)
replace `x'gos = `x'gos+w`x'gos1count+w`x'gos2count+w`x'gos3count+w`x'gos4count
* Judicial
moss `x'bio2, match("(J.P.)") prefix(`x'jud1)
moss `x'bio2, match("(K.C.)") prefix(`x'jud2)
moss `x'bio2, match("(Q.C.)") prefix(`x'jud3)
moss `x'bio2, match("Attorney-General") prefix(`x'jud4)
moss `x'bio2, match("Barrister") prefix(`x'jud5)
moss `x'bio2, match("Crown Prosecutor") prefix(`x'jud6)
moss `x'bio2, match("Master in Chancery") prefix(`x'jud7)
moss `x'bio2, match("Solicitor-General") prefix(`x'jud8)
moss `x'bio2, match("Treasurer of the Inner Temple") prefix(`x'jud9)
gen `x'jud = `x'jud1count+`x'jud2count+`x'jud3count+`x'jud4count+`x'jud5count+`x'jud6count+`x'jud7count+`x'jud8count+`x'jud9count
* Military
moss `x'bio2, match("Aide-de-Camp") prefix(`x'mil1)
moss `x'bio2, match("Assistant Military Secretary") prefix(`x'mil2)
moss `x'bio2, match("Brevet Lieutenant-Colonel") prefix(`x'mil3)
moss `x'bio2, match("Brevet Major") prefix(`x'mil4)
moss `x'bio2, match("Captain") prefix(`x'mil5)
moss `x'bio2, match("Colonel") prefix(`x'mil6)
moss `x'bio2, match("commissioned") prefix(`x'mil7)
moss `x'bio2, match("Commodore") prefix(`x'mil8)
moss `x'bio2, match("Ensign") prefix(`x'mil9)
moss `x'bio2, match("Extra Aide-de-Camp") prefix(`x'mil10)
moss `x'bio2, match("Extra Militia Aide-de-Camp") prefix(`x'mil11)
moss `x'bio2, match("Field Marshal") prefix(`x'mil12)
moss `x'bio2, match("General") prefix(`x'mil13)
moss `x'bio2, match("Lieutenant-Colonel") prefix(`x'mil14)
moss `x'bio2, match("Lieutenant-General") prefix(`x'mil15)
moss `x'bio2, match("Lieutenant") prefix(`x'mil16)
moss `x'bio2, match("Major") prefix(`x'mil17)
moss `x'bio2, match("Major-General") prefix(`x'mil18)
moss `x'bio2, match("Militia Aide-de-Camp") prefix(`x'mil19)
moss `x'bio2, match("Midshipman") prefix(`x'mil20)
moss `x'bio2, match("Officer") prefix(`x'mil21)
moss `x'bio2, match("Vicar") prefix(`x'mil22)
moss `x'bio2, match("Vice-Admiral") prefix(`x'mil23)
moss `x'bio2, match("Vice-Admiral of the Yorkshire Coast") prefix(`x'mil24)
moss `x'bio2, match("Secretary of State for War") prefix(`x'mil25)
gen `x'mil = `x'mil1count+`x'mil2count+`x'mil3count+`x'mil4count+`x'mil5count+`x'mil6count+`x'mil7count+`x'mil8count+`x'mil9count+`x'mil10count+`x'mil11count+`x'mil12count+`x'mil13count+`x'mil14count+`x'mil15count+`x'mil16count+`x'mil17count+`x'mil18count+`x'mil19count+`x'mil20count+`x'mil21count+`x'mil22count+`x'mil23count+`x'mil24count+`x'mil25count
* Order
moss `x'bio2, match("(K.C.B.)") prefix(`x'ord1)
moss `x'bio2, match("(K.C.H.)") prefix(`x'ord2)
moss `x'bio2, match("(K.G.)") prefix(`x'ord3)
moss `x'bio2, match("(C.B.)") prefix(`x'ord4)
moss `x'bio2, match("(G.C.B.)") prefix(`x'ord5)
moss `x'bio2, match("(K.P.)") prefix(`x'ord6)
moss `x'bio2, match("(K.T.)") prefix(`x'ord7)
gen `x'ord = `x'ord1count+`x'ord2count+`x'ord3count+`x'ord4count+`x'ord5count+`x'ord6count+`x'ord7count
* Cabinet
moss `x'bio2, match("(P.C.)") prefix(`x'cab1)
moss `x'bio2, match("Chairman of the Customs Board") prefix(`x'cab2)
moss `x'bio2, match("Commissioner of National Education") prefix(`x'cab3)
moss `x'bio2, match("Deputy Chairman of the Customs Board") prefix(`x'cab4)
moss `x'bio2, match("Home Dept") prefix(`x'cab5)
moss `x'bio2, match("Home Secretary") prefix(`x'cab6)
moss `x'bio2, match("Paymaster General") prefix(`x'cab7)
moss `x'bio2, match("Paymaster-Gen") prefix(`x'cab8)
moss `x'bio2, match("PMG with Cabinet seat") prefix(`x'cab9)
moss `x'bio2, match("Postmaster General") prefix(`x'cab10)
moss `x'bio2, match("Pres BOT") prefix(`x'cab11)
moss `x'bio2, match("President of the Board of Trade") prefix(`x'cab12)
moss `x'bio2, match("Vice-President of the Board of Trade") prefix(`x'cab13)
moss `x'bio2, match("V-Pres BOT") prefix(`x'cab14)
moss `x'bio2, match("Sec Treasury") prefix(`x'cab15)
moss `x'bio2, match("Secretary of the Treasury") prefix(`x'cab16)
moss `x'bio2, match("Prime Minister") prefix(`x'cab17)
gen `x'cab = `x'cab1count+`x'cab2count+`x'cab3count+`x'cab4count+`x'cab5count+`x'cab6count+`x'cab7count+`x'cab8count+`x'cab9count+`x'cab10count+`x'cab11count+`x'cab12count+`x'cab13count+`x'cab14count+`x'cab15count+`x'cab16count+`x'cab17count
* Local
moss `x'bio2, match("(D.L.)") prefix(`x'loc1)
moss `x'bio2, match("Burgess of Glasgow") prefix(`x'loc2)
moss `x'bio2, match("Constable") prefix(`x'loc3)
moss `x'bio2, match("Custos Rotulorum") prefix(`x'loc4)
moss `x'bio2, match("Governor") prefix(`x'loc5)
moss `x'bio2, match("Hereditary Sheriff") prefix(`x'loc6)
moss `x'bio2, match("High Sheriff") prefix(`x'loc7)
moss `x'bio2, match("Lord Lieutenant") prefix(`x'loc8)
moss `x'bio2, match("Lord Warden of the Stannaries") prefix(`x'loc9)
moss `x'bio2, match("Sheriff") prefix(`x'loc10)
moss `x'bio2, match("Chief Secretary of Ireland") prefix(`x'loc11)
gen `x'loc = `x'loc1count+`x'loc2count+`x'loc3count+`x'loc4count+`x'loc5count+`x'loc6count+`x'loc7count+`x'loc8count+`x'loc9count+`x'loc10count+`x'loc11count
* Royal Household
moss `x'bio2, match("Gold Stick in Waiting") prefix(`x'hou1)
moss `x'bio2, match("Lord in Waiting") prefix(`x'hou2)
moss `x'bio2, match("Lord of the Bedchamber") prefix(`x'hou3)
moss `x'bio2, match("Treasurer of the Household") prefix(`x'hou4)
moss `x'bio2, match("Equerry") prefix(e`x'hou1)
moss `x'bio2, match("Groom-in-Waiting") prefix(e`x'hou2)
gen `x'hou = `x'hou1count+`x'hou2count+`x'hou3count+`x'hou4count+e`x'hou1count+e`x'hou2count
* Religious
moss `x'bio2, match("Bishop") prefix(`x'rel1)
moss `x'bio2, match("Canon Residentiary") prefix(`x'rel2)
moss `x'bio2, match("Chaplain") prefix(`x'rel3)
moss `x'bio2, match("Dean") prefix(`x'rel4)
moss `x'bio2, match("Ecclesiastical Commissioner") prefix(`x'rel5)
moss `x'bio2, match("Lord High Commissioner of the General Assembly of the Church of Scotland") prefix(`x'rel6)
moss `x'bio2, match("Minister") prefix(`x'rel7)
moss `x'bio2, match("Rector") prefix(`x'rel8)
moss `x'bio2, match("Rector Poole") prefix(`x'rel9)
gen `x'rel = `x'rel1count+`x'rel2count+`x'rel3count+`x'rel4count+`x'rel5count+`x'rel6count+`x'rel7count+`x'rel8count+`x'rel9count
* Other
moss `x'bio2, match("president of the Highland Agricultural Society") prefix(`x'other1)
moss `x'bio2, match("Chief Commissioner of Woods and Forests") prefix(`x'other2)
moss `x'bio2, match("First Commissioner of Woods and Forests") prefix(`x'other3)
gen `x'other = `x'other1count+`x'other2count+`x'other3count
* Parliament
moss `x'bio2, match("(M.P.)") prefix(`x'par1)
gen `x'par = `x'par1count
}
*
keep refno firstnames surname title wom byear dyear myear mourn cOut cs ///
relation_f fbio1 fbio2 relation_h heir_h hbio1 hbio2 ///
relation_b1 heir_b1 b1bio1 b1bio2 relation_b2 heir_b2 b2bio1 b2bio2 ///
relation_b3 heir_b3 b3bio1 b3bio2 relation_b4 heir_b4 b4bio1 b4bio2 ///
relation_b5 heir_b5 b5bio1 b5bio2 relation_b6 heir_b6 b6bio1 b6bio2 ///
relation_b7 heir_b7 b7bio1 b7bio2 relation_b8 heir_b8 b8bio1 b8bio2 ///
relation_b9 heir_b9 b9bio1 b9bio2 heir_f ///
b1byear b1dyear b1dage b2byear b2dyear b2dage b3byear b3dyear b3dage ///
b4byear b4dyear b4dage b5byear b5dyear b5dage b6byear b6dyear b6dage ///
b7byear b7dyear b7dage b8byear b8dyear b8dage b9byear b9dyear b9dage ///
facd fcol fdip fgos fjud fmil ford fcab floc fhou frel fother fpar ///
hacd hcol hdip hgos hjud hmil hord hcab hloc hhou hrel hother hpar ///
b1acd b1col b1dip b1gos b1jud b1mil b1ord b1cab b1loc b1hou b1rel b1other b1par ///
b2acd b2col b2dip b2gos b2jud b2mil b2ord b2cab b2loc b2hou b2rel b2other b2par ///
b3acd b3col b3dip b3gos b3jud b3mil b3ord b3cab b3loc b3hou b3rel b3other b3par ///
b4acd b4col b4dip b4gos b4jud b4mil b4ord b4cab b4loc b4hou b4rel b4other b4par ///
b5acd b5col b5dip b5gos b5jud b5mil b5ord b5cab b5loc b5hou b5rel b5other b5par ///
b6acd b6col b6dip b6gos b6jud b6mil b6ord b6cab b6loc b6hou b6rel b6other b6par ///
b7acd b7col b7dip b7gos b7jud b7mil b7ord b7cab b7loc b7hou b7rel b7other b7par ///
b8acd b8col b8dip b8gos b8jud b8mil b8ord b8cab b8loc b8hou b8rel b8other b8par ///
b9acd b9col b9dip b9gos b9jud b9mil b9ord b9cab b9loc b9hou b9rel b9other b9par
* Replace for missing value if no brother & generate dummies
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9{
foreach y in acd col dip gos jud mil ord par cab loc hou rel other{
replace `x'`y' = . if `x'bio2==""
gen d_`x'`y' = 1 if `x'`y'>0 & `x'`y'!=.
replace d_`x'`y' = 0 if `x'`y'==0
}
replace relation_`x' = "" if `x'bio2==""
}
* Globals
global all_vars loc office rel jud forg hou other par
global d_all_vars d_loc d_office d_rel d_jud d_forg d_hou d_other d_par
forvalues i=1/10{
global all_vars`i' loc`i' office`i' rel`i' jud`i' forg`i' hou`i' other`i' par`i'
global d_all_vars`i' d_loc`i' d_office`i' d_rel`i' d_jud`i' d_forg`i' d_hou`i' d_other`i' d_par`i'
}
global allrel_vars relloc reloffice relrel reljud relforg relhou relother relpar
global d_allrel_vars d_relloc d_reloffice d_relrel d_reljud d_relforg d_relhou d_relother d_relpar
label define lpos 1 "Total" 2 "local" 3 "office" 4 "relig." 5 "judicial" 6 "foreign" 7 "house." 8 "other" 9 "MP"
global num_vars pos1 pos2 pos3 pos4 pos5 pos6 pos7 pos8 pos9
global d_num_vars d_pos1 d_pos2 d_pos3 d_pos4 d_pos5 d_pos6 d_pos7 d_pos8 d_pos9
* Reshape data
local i = 0
foreach x in f h b1 b2 b3 b4 b5 b6 b7 b8 b9{
rename heir_`x' `x'heir
*
gen `x'forg = `x'col+`x'dip
gen d_`x'forg = d_`x'col+d_`x'dip
drop `x'col `x'dip d_`x'col d_`x'dip
*
gen `x'office = `x'cab+`x'gos
gen d_`x'office = d_`x'cab+d_`x'gos
drop `x'cab `x'gos d_`x'cab d_`x'gos
*
local i = `i'+1
rename `x'heir heir`i'
foreach y in $all_vars{
rename `x'`y' `y'`i'
rename d_`x'`y' d_`y'`i'
}
}
keep refno cOut heir1 heir2 heir3 heir4 heir5 heir6 heir7 heir8 heir9 heir10 ///
$all_vars1 $all_vars2 $all_vars3 $all_vars4 $all_vars5 $all_vars6 $all_vars7 ///
$all_vars8 $all_vars9 $all_vars10 $d_all_vars1 $d_all_vars2 $d_all_vars3 ///
$d_all_vars4 $d_all_vars5 $d_all_vars6 $d_all_vars7 $d_all_vars8 $d_all_vars9 $d_all_vars10
gen id = _n
reshape long heir $all_vars $d_all_vars, i(id) j(j)
keep if heir!=.
drop if forg==. & office==. & jud==. & par==. & loc==. & hou==. & rel==. & other==.
replace heir = 2 if j==1
* Panels A and B: ALL OFFSPRING
preserve
replace heir = 1 if heir==0
collapse j cOut $all_vars $d_all_vars, by(refno heir)
egen tot = rowtotal($all_vars)
egen d_tot = rowtotal($d_all_vars)
gsort refno heir
foreach x in tot $all_vars{
by refno: gen rel`x' = `x'-`x'[_N]
by refno: gen d_rel`x' = d_`x'-d_`x'[_N]
}
drop if j==1
* reshape
local j = 0
foreach x in tot $all_vars{
local j = `j'+1
rename rel`x' pos`j'
rename d_rel`x' d_pos`j'
}
gen idi = _n
sort idi id heir cOut $num_vars $d_num_vars
keep idi id heir cOut $num_vars $d_num_vars
qui reshape long pos d_pos, i(idi) j(ptype)
label values ptype lpos
* collapse for bar graph
collapse pos d_pos, by(ptype cOut)
sort ptype cOut
drop if ptype==8
replace ptype = 8 if ptype==9
gen out_type = cOut + 3*(ptype-1)
* confidence intervals
gen ciL = -.2725088 if cOut==0 & ptype==1
gen ciH = .8795029 if cOut==0 & ptype==1
replace ciL = -1.740913 if cOut==1 & ptype==1
replace ciH = -.4484183 if cOut==1 & ptype==1
gen d_ciL = -0.1107418 if cOut==0 & ptype==1
gen d_ciH = 0.4650573 if cOut==0 & ptype==1
replace d_ciL = -0.7535773 if cOut==1 & ptype==1
replace d_ciH = -0.1454582 if cOut==1 & ptype==1
* figures
global ylab -1.8(0.6)1.8
global ylabd_ -1(0.5)1
global title0 Other offspring
global title1 Heirs
global stitle total number of appointments
global stitled_ appointments indicator
foreach d in "" d_{
local l = "ylab`d'"
local CD = "stitle`d'"
twoway (bar `d'pos out_type if cOut==0) ///
(bar `d'pos out_type if cOut==1, fi(30)) ///
(rcap `d'ciH `d'ciL out_type, lcolor(black)), ///
ylabel(#3, labsize(small)) ///
xlabel(0.5 "total" 3.5 "local" 6.5 "office" 9.5 "relig." 12.5 "judicial" ///
15.5 "foreign" 18.5 "house." 21.5 "MP", labs(small) angle(vertical) notick) ///
xline(2, lcolor(gray) lpattern(dot)) xline(5, lcolor(gray) lpattern(dot)) ///
xline(8, lcolor(gray) lpattern(dot)) xline(11, lcolor(gray) lpattern(dot)) ///
xline(14, lcolor(gray) lpattern(dot)) xline(17, lcolor(gray) lpattern(dot)) ///
xline(20, lcolor(gray) lpattern(dot)) ///
legend( order(1 "families marrying in" 2 "families marrying out" 3 "90 c.i.") col(3) size(small)) ///
title("All offspring", color(black) size(medlarge)) ///
subtitle("$`CD'", color(black) size(small)) ///
ytitle("son vs. father") ///
xtitle("") ///
graphregion(color(white)) name(`d'panel_all)
}
restore
* Panels C to F: HEIRS AND OTHER OFFSPRING
collapse j refno cOut $all_vars $d_all_vars, by(id heir)
egen tot = rowtotal($all_vars)
egen d_tot = rowtotal($d_all_vars)
gsort id heir
foreach x in tot $all_vars{
by id: gen rel`x' = `x'-`x'[_N]
by id: gen d_rel`x' = d_`x'-d_`x'[_N]
}
drop if j==1
* reshape
local j = 0
foreach x in tot $all_vars{
local j = `j'+1
rename rel`x' pos`j'
rename d_rel`x' d_pos`j'
}
gen idi = _n
sort idi id heir cOut $num_vars $d_num_vars
keep idi id heir cOut $num_vars $d_num_vars
qui reshape long pos d_pos, i(idi) j(ptype)
label values ptype lpos
* t-tests
ttest pos if ptype==1 & heir==0, by(cOut) level(90)
ttest pos if ptype==1 & heir==1, by(cOut) level(90)
ttest d_pos if ptype==1 & heir==0, by(cOut) level(90)
ttest d_pos if ptype==1 & heir==1, by(cOut) level(90)
* collapse for bar graph
collapse pos d_pos, by(ptype cOut heir)
sort ptype cOut
drop if ptype==8
replace ptype = 8 if ptype==9
gen out_type = cOut + 3*(ptype-1)
* figures
forvalues i=0/1{
foreach d in "" d_{
local l = "ylab`d'"
local L = "title`i'"
local CD = "stitle`d'"
twoway (bar `d'pos out_type if cOut==0 & heir==`i') ///
(bar `d'pos out_type if cOut==1 & heir==`i', fi(30)), ///
ylabel(#3, labsize(small)) ///
xlabel(0.5 "total" 3.5 "local" 6.5 "office" 9.5 "relig." 12.5 "judicial" ///
15.5 "foreign" 18.5 "house." 21.5 "MP", labs(small) angle(vertical) notick) ///
xline(2, lcolor(gray) lpattern(dot)) xline(5, lcolor(gray) lpattern(dot)) ///
xline(8, lcolor(gray) lpattern(dot)) xline(11, lcolor(gray) lpattern(dot)) ///
xline(14, lcolor(gray) lpattern(dot)) xline(17, lcolor(gray) lpattern(dot)) ///
xline(20, lcolor(gray) lpattern(dot)) ///
legend( order(1 "families marrying in" 2 "families marrying out" 3 "90 c.i.") col(3) size(small)) ///
title("$`L'", color(black) size(medlarge)) ///
subtitle("$`CD'", color(black) size(small)) ///
ytitle("son vs. father") ///
xtitle("") ///
graphregion(color(white)) name(`d'panel_`i')
}
}
* Combine into Figure B4
grc1leg panel_all panel_1 panel_0 d_panel_all d_panel_1 d_panel_0, graphregion(color(white))
graph drop panel_all panel_1 panel_0 d_panel_all d_panel_1 d_panel_0
graph export "${dirresu}\Appendix\FigureB4.pdf", as(pdf) replace
* ==============================================================================
* Table B13:
* Determinants of investments in state education, robustness checks.
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}/final-data-sec4.dta", clear
* Control variables:
gen nsib2 = nsib^2 // Number of siblings squared
global controlsrf distlondon i.prank // controls for reduced-form effect
global controlsb distlondon i.prank nsib nsib2 // controls for regressions with brothers
global controlsh distlondon i.prank // controls for regressions with family heads
global hechter propman100 logincome propcon propnonc relig // county controls from Hechter
encode fbio1, gen(nfam) // clusters at the family level
label var cOut "W married a commoner"
* Rename political variables to use in loop
rename d_MPb_aft politaft_col2
rename d_MPf_bef politbef_col2
rename MPb_years_aft politaft_col3
rename MPf_years_bef politbef_col3
rename d_MPb_locaft politaft_col4
rename d_MPf_local politbef_col4
rename MPb_years_locaft politaft_col5
rename MPf_years_locbef politbef_col5
rename d_MPh_aft politaft_col6
rename d_MPh_bef politbef_col6
rename MPh_years_aft politaft_col7
rename MPh_years_bef politbef_col7
* Panel A: BASELINE
* -----------------
* Col. [1]: Reduced form effect
gen effect = cOut
label var effect "coef"
ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsrf , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsrf
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsrf
outreg2 using "${dirresu}\Appendix\TableB13_A", keep(effect) stats(coef) noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, M common) tex replace
* Cols [2] to [5]
local listxvars "Mcommon MP MPyears localMP LocalMPyears HeadMP HeadMPyears"
forvalues i=2/5{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsb $polbefore
outreg2 using "${dirresu}\Appendix\TableB13_A", keep(effect) stats(coef) noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex
}
* Cols [6] and [7]
forvalues i=6/7{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsh $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsh $polbefore
outreg2 using "${dirresu}\Appendix\TableB13_A", keep(effect) stats(coef) noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex title("Panel A")
}
* Panel B: 50x50 MILES GRID FIXED EFFECT:
* ---------------------------------------
* Import grid cell of 50x50 miles
/* Note: this is generated with ArcGis over the rovided map of England and
exported as a text file in 03_appendix\aux data appendix\grid50x50miles.txt" */
preserve
import delimited "${dirroot}\programs\03_appendix\aux data appendix\grid50x50miles.txt", clear
label var fid_b10sea "FID identifier for family seats of women in baseline sample (ArcGIS)"
label var fid_grid50 "FID identifier for 50x50 miles grid"
save "${dirdta}\grid50x50miles.dta", replace
restore
merge 1:1 fid_b10sea using "${dirdta}\grid50x50miles.dta", nogen
* Geographical controls
gen inter = wgslat*wgslon
global geocontrols i.fid_grid wgslat wgslon inter
global hechter propman100 logincome propcon propnonc relig
global controlsrf i.prank
global controlsb i.prank nsib nsib2
global controlsh i.prank
* Col. [1]: Reduced form effect
replace effect = cOut
label var effect "coef"
ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsrf $geocontrols, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsrf $geocontrols
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsrf $geocontrols
outreg2 using "${dirresu}\Appendix\TableB13_B", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, M common) tex replace
* Cols [2] to [5]
local listxvars "Mcommon MP MPyears localMP LocalMPyears HeadMP HeadMPyears"
forvalues i=2/5{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsb $polbefore $geocontrols, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsb $polbefore $geocontrols
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsb $polbefore $geocontrols
outreg2 using "${dirresu}\Appendix\TableB13_B", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex
}
* Cols [6] and [7]
forvalues i=6/7{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsh $polbefore $geocontrols, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsh $polbefore $geocontrols
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou (effect = syntheticT biorder) $hechter $controlsh $polbefore $geocontrols
outreg2 using "${dirresu}\Appendix\TableB13_B", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex title("Panel A")
}
* Panel C: excluding cities
* -------------------------
* Calculate average tax rate excluding School Boards in cities:
preserve
import delimited "${dirdata}\shapefiles\IseatXsb_10mi.txt", clear // Import all family seat-School Board dyads within 10 miles.
foreach x in refno wgslat wgslon wgs1_lat wgs1_lon wgs2_lat wgs2_lon wgs3_lat wgs3_lon wgs4_lat wgs4_lon rateperpou{
replace `x'=subinstr(`x',",",".",.) // Replace , for . for decimals
destring `x', replace
}
rename fid_seats_ fid_b10sea
duplicates tag refno countrysea wgslon wgslat schoolboar wgs1_lat wgs1_lon, gen(dup1)
drop if dup1 & countyn!=county
* Exclude School Boards in cities:
gen city = 0
foreach x in "Reading" "Macclesfield" "Stalybridge" "Stockport" "Carlisle" "Chesterfield" "Derby" "Exeter" "Plymouth" "Darlington" "Hartlepool" "South Shields" "Stockton-on-Tees" "Sunderland" "Bristol" "Gloucester" "Aldershot" "Portsmouth" "Southampton" "Canterbury" "Folkestone" "Gravesend" "Maidstone" "Ashton-under-Lyne" "Blackburn" "Bolton" "Burnley" "Liverpool" "Manchester" "Salford" "Oldham" "Rochdale" "Wigan" "Leicester" "London" "Great Yarmouth" "Norwich" "Northampton" "Newcastle-on-Tyne" "Tynemouth" "Nottingham" "Oxford" "Bath" "Burslem" "Burton-upon-Trent" "Hanley" "Longton" "Rowley Regis" "Tipton" "Walsall" "Wednesbury" "West Bromwich" "Wolverhampton" "Ipswich" "Brighton" "Hove" "Hastings" "Birmingham" "Coventry" "Dudley" "Worcester" "Kingston-on-Hull" "Middlesborough" "Scarborough" "Barnsley" "Batley" "Bradford" "Dewsbury" "Halifax" "Huddersfield" "Leeds" "Rotherham" "Sheffield" "Todmorden" "Wakefield"{
qui replace city = 1 if regexm(schoolboar,"`x'")
}
keep if city==0
* Collapse by refno family seat to get the average tax rate:
collapse rateperpound = rateperpou, by(refno countyn fid_b10sea)
rename rateperpound rate_exclcities
keep refno fid_b10sea rate_exclcities
label var fid_b10sea "FID identifier for family seats of women in baseline sample (ArcGIS)"
label var refno "Reference number in Hollingsworth"
label var rate_exclcities "Av tax rate % 1872-78 by School Boards <10mi from seat (Excluding SB in cities)"
save "${dirdta}\rate_exclcities.dta", replace
restore
* Merge average tax rate excluding School Boards in cities
merge 1:1 refno fid_b10sea using "${dirdta}\rate_exclcities.dta", nogen
* Globals
global controlsrf distlondon i.prank
global controlsb distlondon i.prank nsib nsib2
global controlsh distlondon i.prank
global hechter propman100 logincome propcon propnonc relig
* Col. [1]: Reduced form effect
replace effect = cOut
label var effect "coef"
ivregress liml rate_exclcities (effect = syntheticT biorder) $hechter $controlsrf, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsrf
local F1 = round(e(F),0.01)
qui ivregress liml rate_exclcities (effect = syntheticT biorder) $hechter $controlsrf
outreg2 using "${dirresu}\Appendix\TableB13_C", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, M common) tex replace
* Cols [2] to [5]
local listxvars "Mcommon MP MPyears localMP LocalMPyears HeadMP HeadMPyears"
forvalues i=2/5{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
ivregress liml rate_exclcities (effect = syntheticT biorder) $hechter $controlsb $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivregress liml rate_exclcities (effect = syntheticT biorder) $hechter $controlsb $polbefore
outreg2 using "${dirresu}\Appendix\TableB13_C", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex
}
* Cols [6] and [7]
forvalues i=6/7{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
ivregress liml rate_exclcities (effect = syntheticT biorder) $hechter $controlsh $polbefore, first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.01)
qui ivregress liml rate_exclcities (effect = syntheticT biorder) $hechter $controlsh $polbefore
outreg2 using "${dirresu}\Appendix\TableB13_C", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex title("Panel A")
}
* Panel D
* ----------
* Use auxiliary data with all seat - school board pairs:
import delimited "${dirroot}\programs\03_appendix\aux data appendix\sbXseats_allpairs.txt", clear
sort fid_b10sea
geodist wgs1_lat wgs1_lon wgslat wgslon, miles gen(distSB_S)
sort fid_b10sea distSB_S
/* Note: To calculate the weighted average tax rate, I use weights that
decay exponentially by the distance between each School Board and the
corresponding seat. Speciffcally, weights are exp(-distance/14:43), where
14.43 is set such that School Boards 10 miles away from a seat receive a
weight of 0.5. */
gen prox_per = exp(-distSB_S/14.4269504089)
collapse (mean) rateperpou [iweight = prox_per], by(fid_b10sea)
rename rateperpou rateperpou_weight
label var fid_b10sea "FID identifier for family seats of women in baseline sample (ArcGIS)"
label var rateperpou_weight "Weighted average tax rate for all seat-SB pairs, weights are exp(-dist/14:43)"
save "${dirdta}\peersdau_weights.dta", replace
* Import data and add weighted average tax rate
use "${dirdata}/final-data-sec4.dta", clear
merge m:1 fid_b10sea using "${dirdta}\peersdau_weights.dta", nogen
* Control variables:
gen nsib2 = nsib^2 // Number of siblings squared
global controlsrf distlondon i.prank // controls for reduced-form effect
global controlsb distlondon i.prank nsib nsib2 // controls for regressions with brothers
global controlsh distlondon i.prank // controls for regressions with family heads
global hechter propman100 logincome propcon propnonc relig // county controls from Hechter
encode fbio1, gen(nfam) // clusters at the family level
label var cOut "W married a commoner"
* Rename political variables to use in loop
rename d_MPb_aft politaft_col2
rename d_MPf_bef politbef_col2
rename MPb_years_aft politaft_col3
rename MPf_years_bef politbef_col3
rename d_MPb_locaft politaft_col4
rename d_MPf_local politbef_col4
rename MPb_years_locaft politaft_col5
rename MPf_years_locbef politbef_col5
rename d_MPh_aft politaft_col6
rename d_MPh_bef politbef_col6
rename MPh_years_aft politaft_col7
rename MPh_years_bef politbef_col7
* Panel A: BASELINE
* -----------------
* Col. [1]: Reduced form effect
gen effect = cOut
label var effect "coef"
qui ivregress liml rateperpou_weight (effect = syntheticT biorder) $hechter $controlsrf , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsrf
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou_weight (effect = syntheticT biorder) $hechter $controlsrf
outreg2 using "${dirresu}\Appendix\TableB13_D", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, M common) tex replace
* Cols [2] to [5]
local listxvars "Mcommon MP MPyears localMP LocalMPyears HeadMP HeadMPyears"
forvalues i=2/5{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
qui ivregress liml rateperpou_weight (effect = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou_weight (effect = syntheticT biorder) $hechter $controlsb $polbefore
outreg2 using "${dirresu}\Appendix\TableB13_D", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex
}
* Cols [6] and [7]
forvalues i=6/7{
local effectof : word `i' of `listxvars'
replace effect = politaft_col`i'
global polbefore politbef_col`i'
qui ivregress liml rateperpou_weight (effect = syntheticT biorder) $hechter $controlsh $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
qui reg effect syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.01)
qui ivregress liml rateperpou_weight (effect = syntheticT biorder) $hechter $controlsh $polbefore
outreg2 using "${dirresu}\Appendix\TableB13_D", keep(effect) stats(coef) nor2 noobs nocons label dec(2) noas ///
ctitle("tax") addtext(CLR-pval, `CLR1', F stage 1, `F1', Nb observations, `e(N)', Effect of, `effectof') tex title("Panel A")
}
* erase temp files
erase "${dirdta}\rate_exclcities.dta"
erase "${dirdta}\grid50x50miles.dta"
erase "${dirdta}\peersdau_weights.dta"
* ==============================================================================
* Table B14:
* Determinants of state education, measured with total funds raised from taxes.
* ------------------------------------------------------------------------------
* Import data and add total education funds raised from taxes
use "${dirdata}/final-data-sec4.dta", clear
merge 1:1 refno fid_b10sea using "${dirdta}/funds-from-rates.dta", nogen
/* Note: This file contains the total education funds raised from taxes
raised by the average School Board within 10 miles of each family seat. */
gen rate_log = log(rate_pound) // dep. var. in logs
* Control variables:
gen nsib2 = nsib^2 // Number of siblings squared
global controlsrf distlondon i.prank // controls for reduced-form effect
global controlsb distlondon i.prank nsib nsib2 // controls for regressions with brothers
global controlsh distlondon i.prank // controls for regressions with family heads
global hechter propman100 logincome propcon propnonc relig // county controls from Hechter
encode fbio1, gen(nfam) // clusters at the family level
label var cOut "W married a commoner"
* Col [1]: Reduced-form effect
* CLR p-values
qui ivregress liml rate_log (cOut = syntheticT biorder) $hechter $controlsrf , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log (cOut = syntheticT biorder) $hechter $controlsrf , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log (cOut = syntheticT biorder) $hechter $controlsrf , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep(cOut) stats(coef) nocons label noobs dec(2) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex replace
* First stage coefs in Panel B
qui reg cOut syntheticT biorder $hechter $controlsrf
local F1 = round(e(F),0.01)
ivreg2 rate_log (cOut = syntheticT biorder) $hechter $controlsrf , first cluster(nfam) savefirst
est restore _ivreg2_cOut
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log (cOut = syntheticT biorder) $hechter $controlsrf , first savefirst
est restore _ivreg2_cOut
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("M common") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs replace
* Col [2]: Any brother is MP after marriage
global polit d_MPb_aft
global polbefore d_MPf_bef
* CLR p-values
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg $polit syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_d_MPb_aft
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_d_MPb_aft
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("MP (b)") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs
* Col [3]: Years brother is MP after marriage
global polit MPb_years_aft
global polbefore MPf_years_bef
* CLR p-values
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg $polit syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_MPb_years_aft
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_MPb_years_aft
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("Years MP (b)") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs
* Col [4]: Any brother is local MP after marriage
global polit d_MPb_locaft
global polbefore d_MPf_local
label var d_MPb_locaft "Any brother is local MP after woman's marriage"
* CLR p-values
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg $polit syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_d_MPb_locaft
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_d_MPb_locaft
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("local MP (b)") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs
* Col [5]: Years brother is local MP after marriage
global polit MPb_years_locaft
global polbefore MPf_years_locbef
* CLR p-values
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep($polit) stats(coef) nocons label noobs dec(3) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg $polit syntheticT biorder $hechter $controlsb $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_MPb_years_locaft
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_MPb_years_locaft
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("local years (b)") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs
* Col [6]: Family head is local MP after marriage
global polit d_MPh_aft
global polbefore d_MPh_bef
* CLR p-values
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg $polit syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_d_MPh_aft
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first savefirst
est restore _ivreg2_d_MPh_aft
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("MP (h)") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs
* Col [7]: Years family head is local MP after marriage
global polit MPh_years_aft
global polbefore MPh_years_bef
* CLR p-values
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
outreg2 using "${dirresu}\Appendix\TableB14_A", keep($polit) stats(coef) nocons label noobs dec(3) noas ///
ctitle("log funds") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', Nb observations, `Nbobs') tex
* First stage coefs in Panel B
qui reg $polit syntheticT biorder $hechter $controlsh $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_MPh_years_aft
local Nobs = `e(N)'
local pval2 = round(2 * ttail(e(df_r), abs(_b[syntheticT]/_se[syntheticT])),0.001)
qui ivreg2 rate_log ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first savefirst
est restore _ivreg2_MPh_years_aft
outreg2 using "${dirresu}\Appendix\TableB14_B", keep(syntheticT) tex ctitle("Years MP (h)") addtext(pfc, [`pval2'], Birth order, "YES", F-stat, `F1', Obs, `Nobs') label dec(3) noaster nocons nor2 stats(coef pval) noobs
* erase temp files:
erase "${dirdta}/funds-from-rates.dta"
* ==============================================================================
* Figure B5:
* Funds raised from taxes and other measures of state education.
* ------------------------------------------------------------------------------
* Import data on other education measures at the county level.
import delimited "${dirroot}\programs\03_appendix\aux data appendix\other_edu.txt", clear
/* Note: This file contains different measures of education for 38
historic counties in England (excluding the London area). Source is
Goni (2017) "Landed elites and education provision in England. Evidence
from School Boards, 1870-99." */
label var county "county"
label var countyabb "county (abbreviation)"
label var ratepc_d "funds raised from taxes pence per child)"
label var expenpc_d "total expenditure in state schools (pence per child)"
label var statepriv "state to private schools (ratio)"
label var salary "teachers' salary (pence per teacher)"
label var arith "% pass in national arithmetics exam"
* Figure
pwcorr expenpc_d ratepc_d, star(0.01)
local corr1 = round(`r(rho)', 0.01)
twoway (scatter expenpc_d ratepc_d, mlabel(countyabb) mlabp(0) msymbol(i) mlabc(black) mlabsize(vsmall)) ///
(lfit expenpc_d ratepc_d, lwidth(thin)), ///
graphregion(color(white)) xtitle("funds raised from taxes" "(pence per child)", size(small)) ///
xlabel(10(30)170, labsize(vsmall)) ylabel(0(100)300, labsize(vsmall)) ytitle("total expend. in state schools" "(pence per child)", size(small)) legend(off) ///
note("correlation = 0`corr1'***") name(pA, replace)
pwcorr statepriv ratepc_d, star(0.01)
local corr1 = round(`r(rho)', 0.01)
twoway (scatter statepriv ratepc_d if countyabb!="CON", mlabel(countyabb) mlabp(0) msymbol(i) mlabc(black) mlabsize(vsmall)) ///
(lfit statepriv ratepc_d, lwidth(thin)), ///
graphregion(color(white)) xtitle("funds raised from taxes" "(pence per child)", size(small)) ///
xlabel(10(30)170, labsize(vsmall)) ylabel(0.10(0.10)0.70, labsize(vsmall)) ytitle("state / private schools" "(ratio)", size(small)) legend(off) ///
note("correlation = 0`corr1'***") name(pB, replace)
pwcorr salary ratepc_d, star(0.01)
local corr1 = round(`r(rho)', 0.01)
twoway (scatter salary ratepc_d, mlabel(countyabb) mlabp(0) msymbol(i) mlabc(black) mlabsize(vsmall)) ///
(lfit salary ratepc_d, lwidth(thin)), ///
graphregion(color(white)) xtitle("funds raised from taxes" "(pence per child)", size(small)) ///
xlabel(10(30)170, labsize(vsmall)) ylabel(0(2000)6000, labsize(vsmall)) ytitle("teachers' salary" "(pence per teacher)", size(small)) legend(off) ///
note("correlation = 0`corr1'***") name(pC, replace)
pwcorr arithm ratepc_d, star(0.01)
local corr1 = round(`r(rho)', 0.01)
twoway (scatter arithm ratepc_d, mlabel(countyabb) mlabp(0) msymbol(i) mlabc(black) mlabsize(vsmall)) ///
(lfit arithm ratepc_d, lwidth(thin)), ///
graphregion(color(white)) xtitle("funds raised from taxes" "(pence per child)", size(small)) ///
xlabel(10(30)170, labsize(vsmall)) ylabel(70(5)85, labsize(vsmall)) ytitle("national arithm. exam" "(% pass)", size(small)) legend(off) ///
note("correlation = 0`corr1'***") name(pD, replace)
graph combine pA pB pC pD, graphregion(color(white))
* Export figure
graph export "${dirresu}\Appendix\FigureB5.pdf", as(pdf) replace
graph drop _all
********************************************************************************
********************************************************************************
********************************************************************************
* *
* C. IV ESTIMATION FOR MEN *
* *
********************************************************************************
********************************************************************************
********************************************************************************
* ==============================================================================
* DATA PREPARATION FOR MEN'S ANALYSIS
* ------------------------------------------------------------------------------
use "${dirdata}\final-data.dta", clear
* COHORT VARIABLES
* ----------------
* Relative number of marriageable age in one's class
foreach x in "" spo{
gen brank`x' = crank`x'
qui replace brank`x' = 6 if brank`x'==2
qui replace brank`x' = 7 if brank`x'==3
forvalues i=0/7{
gen b`i'`x' = (brank`x'==`i')
}
}
gen rel1 = .
forvalues i=1527/1945 {
foreach var of varlist b0 b1 b2 b3 b4 b5 b6 b7 {
qui summ `var' if byear>`i'-3 & byear<`i'+3 & nthismarriage<2 & dage>18
qui replace rel1 = r(mean) if byear==`i' & `var'==1 & dage>18 & nthismarriage<2
}
}
drop brank brankspo b0 b1 b2 b3 b4 b5 b6 b7 b0spo b1spo b2spo b3spo b4spo b5spo b6spo b7spo
* Sex ratio
gen all = 1
gen mrisk = . // Men at risk of marriage
gen wrisk2 = . // Women at risk of marriage using primary sex ratio (ratio of 105)
forvalues i=1527/1959{
qui summ all if nthismarr<2 & `i'-byear>=19 & `i'-byear<=24 & dyear>=`i' & wom==0 // # boys aged 18-23 when the LS starts
scalar define a`i' = r(N)*0.952380952 // sex ratio conversion
qui replace wrisk2 = a`i' if myear==`i' // estimated number of girls between 18-23
}
forvalues i=1527/1959{
qui summ all if nthismarr<2 & `i'-byear>=23 & `i'-byear<=28 & dyear>=`i' & wom==0 // # boys aged 19-24 when the LS starts
qui replace mrisk = r(N) if myear==`i'
}
gen riskMin = min(wrisk, mrisk) // cohort size (min)
gen sexr = mrisk/wrisk2 // sex ratio
* Ratio of heirs to peers' daughters
sort refno
gen id_h = 0
replace id_h = 1 if refno[_n]!=refno[_n-1]+1
/* Note: The reference number (refno) of non-heirs is a consecutive number
of the refno of their fathers (who are peers or peers' heirs). The reference
number of heirs is not a consecutive number of their fathers (it is their
fathers +20, +50, or +more). Here I use the consecutiveness of the reference
number to distinguish these two groups. */
gen heirstatus = (wom==0 & id_h==1 & crank>=4)
gen heirrisk = .
forvalues i=1527/1959{
qui summ all if nthismarr<2 & `i'-byear>=23 & `i'-byear<=28 & dyear>=`i' & heirstatus==1 // # heirs aged 19-24 when the LS starts
qui replace heirrisk = r(N) if myear==`i'
}
gen heirratio = heirrisk/wrisk2
drop heirrisk all id_h
* Decade indicators, trend, and other Crystal Palace fair indicator (1851)
gen dec50 = (myear>1850 & myear<=1860)
gen dec60 = (myear>1860 & myear<=1870)
gen dec70 = (myear>1850 & myear<=1880)
egen t=group(myear) if myear>=1851 & myear<=1875
gen d1851 = (myear==1851) // Crystal Palace fair (1851)
* GEOGRAPHIC ENDOGAMY VARIABLES
* -----------------------------
* spouse's seat with 100 miles of one's seat
gen disthom = (dist<=100)
* relative size of the marriageable cohort in one's division (the same first-level NUTS in England and Wales, electoral region in Scotland, or province in Ireland)
preserve
import delimited "${dirroot}\programs\03_appendix\aux data appendix\seat_division.txt", clear
label var refno "Reference number in Hollingsworth"
label var nthismarriage "Number this Marriage in Hollingsworth dataset"
local j = 1
foreach x in "" 2 3 4{
label var division`x' "NUTS1, electoral region (SCO), or province (IRE) of family seat `j'/n"
label var division`x'spo "NUTS1, electoral region (SCO), or province (IRE) of spouse's family seat `j'/n"
local j = `j'+1
}
label var division4 "NUTS1, electoral region (SCO), or province (IRE) of family seat 4/4"
label var division5spo "NUTS1, electoral region (SCO), or province (IRE) of spouse's family seat 5/5"
label var numseat "Number of seats in entry"
save "${dirdta}\seat_division.dta", replace
restore
/* Note: This file contains the division in which each country seat is
located. Divisions are first-level NUTS (if seat is in England and Wales),
electoral regions (if seat is in Scotland), or province (if seat is in
Ireland)*/
merge 1:1 refno nthismarr using "${dirdta}\seat_division.dta", nogen
foreach k in "" 2 3 4{
encode(division`k'), gen(edivision`k')
}
qui gen reldivision = 0 if numseat>0
qui gen divisiontot = 0 if numseat>0
forvalues i=1845/1880{
display `i'
foreach k in "" 2 3 4{
qui summ edivision`k' if byear>=`i'-2 & byear<=`i'+2 & numseat>0
qui replace divisiontot = r(N)+ divisiontot if byear==`i'
}
}
forvalues i=1845/1880{
display `i'
forvalues j=1/100{
foreach k in "" 2 3 4{
qui summ edivision`k' if edivision`k'==`j' & byear>=`i'-2 & byear<=`i'+2
qui replace reldivision = r(N)/divisiontot + reldivision if byear==`i' & edivision`k'==`j'
}
}
}
gen reldiv = reldivision*100
* Same division
gen divhom = 0 if countryseat!="" & countryseatspo!="" // Division level homogamy
foreach y in "" 2 3 4{
foreach z in "" 2 3 4 5{
replace divhom = 1 if division`y'==division`z'spo & division`y'!=""
}
}
drop division division2 division3 division4 divisionspo division2spo division3spo division4spo division5spo numseat edivision edivision2 edivision3 edivision4 reldivision divisiontot
erase "${dirdta}\seat_division.dta"
* LANDHOLDINGS' VARIABLES
* -----------------------
* Rename acres and gross annual rents' variables
rename acrestotal ATotal
rename valuetotal VTotal
rename sacrestotal sATotal
rename svaluetotal sVTotal
* Bateman's classes (acres)
foreach x in "" s{
gen `x'landclassA = .
replace `x'landclassA = 6 if `x'ATotal>=100000 & `x'ATotal!=.
replace `x'landclassA = 5 if `x'ATotal< 100000 & `x'ATotal>=50000 & `x'ATotal!=.
replace `x'landclassA = 4 if `x'ATotal< 50000 & `x'ATotal>=20000 & `x'ATotal!=.
replace `x'landclassA = 3 if `x'ATotal< 20000 & `x'ATotal>=10000 & `x'ATotal!=.
replace `x'landclassA = 2 if `x'ATotal< 10000 & `x'ATotal>=6000 & `x'ATotal!=.
replace `x'landclassA = 1 if `x'ATotal< 6000 & `x'ATotal>=1000 & `x'ATotal!=.
replace `x'landclassA = 0 if `x'ATotal==.
}
label define bat 0 "Not GL" 1 "2,000-6,000" 2 "6,000-10,000" 3 "10,000-20,000" 4 "20,000-50,000" 5 "50,000-100,000" 6 ">100,000"
label values landclassA slandclassA
* percentiles and deciles for land rents' classes
foreach x in "" s{
xtile `x'landclassV = `x'VTotal, nq(10)
foreach z in A V{
xtile `x'`z'Tot = `x'`z'Total if myear>=1851 & myear<=1875 & nthismarr==1 & wom==0 & `x'`z'Total!=., nq(100)
}
}
* sorting by land variables
gen landhoA = .
replace landhoA = 0 if landclassA!=slandclassA & landclassA!=. & slandclassA!=.
replace landhoA = 1 if landclassA==slandclassA & landclassA!=. & slandclassA!=.
gen landhoV = (landclassV == slandclassV | landclassV == slandclassV+1 | landclassV == slandclassV-1)
replace landhoV = 0 if sVTotal==.
foreach x in A V{
gen miss`x' = abs(`x'Tot-s`x'Tot) if `x'Tot!=. & s`x'Tot!=.
}
* relative size of land class
gen relandA = .
forvalues t=1800/1875{
qui summ sATotal if sbyear<`t'-18 & sbyear>=`t'-24 & nthismarr==1 & wom==0
gen num_`t'A = r(N)
forvalues i=1/6{
qui summ sATotal if slandclassA==`i' & byear<`t'-18 & byear>=`t'-24 & nthismarr==1 & wom==0
gen num`i'_`t'A = r(N)
gen reland`i'_`t'A = num`i'_`t'A/num_`t'A
replace relandA = reland`i'_`t'A if myear==`t' & landclassA==`i'
drop num`i'_`t'A reland`i'_`t'A
}
drop num_`t'A
}
gen relandV = .
forvalues t=1800/1875{
qui summ sVTotal if sbyear<`t'-18 & sbyear>=`t'-24 & nthismarr==1 & wom==0
gen num_`t'V = r(N)
forvalues i=1/10{
qui summ sVTotal if slandclassV==`i' & byear<`t'-18 & byear>=`t'-24 & nthismarr==1 & wom==0 | slandclassV==`i'-1 & byear<`t'-18 & byear>=`t'-24 & nthismarr==1 & wom==0 | slandclassV==`i'+1 & byear<`t'-18 & byear>=`t'-24 & nthismarr==1 & wom==0
gen num`i'_`t'V = r(N)
gen q20rel`i'_`t'V = num`i'_`t'V/num_`t'V
replace relandV = q20rel`i'_`t'V if myear==`t' & landclassV==`i'
drop num`i'_`t'V q20rel`i'_`t'V
}
drop num_`t'V
}
* heirness status of wives
gen heirspo = .
replace heirspo = 1 if heir==3
replace heirspo = 2 if heir==2
replace heirspo = 3 if heir==1
replace heirspo = 4 if heir==5
replace heirspo = 5 if heir==4
label define hhh 1 "Wife not" 2 "Wife coheiress" 3 "Wife sole heiress" 4 "Husband not" 5 "Husband heir"
label values heirspo hhh
* OTHER CONTROLS
* --------------
* Railway length
/* Note: This file contains the number of miles of railway in Britain and
Ireland each year. The source is Mitchell (1988). */
preserve
import delimited "${dirroot}\programs\03_appendix\aux data appendix\mitchell_railway.txt", clear
label var myear "Year"
label var raillength "Length of railway (miles) in England and Ireland"
save "${dirdta}\mitchell_railway.dta", replace
restore
merge m:1 myear using "${dirdta}\mitchell_railway.dta", nogen
gen railLength100 = raillength/100 // in hundreds of mi
* SAMPLE
* ------
/* Note: The sample for appendix C is all peers and peers' offspring
marrying in 1851-75. I exclude second marriages. */
keep if myear>=1851 & myear<=1875 & mage!=. & nthismarriage<2
* labels
label var rel1 "Relative no. of own marriageble age in one's class"
label var mrisk "Size of the marriageble cohort (men)"
label var wrisk2 "Size of the marriageble cohort calculated using birth ratios (women)"
label var riskMin "Min size of marriageable cohort (men vs. women)"
label var sexr "Sex ratio"
label var heirstatus "Heir to Baronet, Baron/Visc., Earl/Marquis/Duke (based on refno)"
label var heirratio "Ratio of heirs to peers' daughters"
label var dec50 "Married in 1850s"
label var dec60 "Married in 1860s"
label var dec70 "Married in 1870s"
label var t "Time trend (1851-75)"
label var d1851 "Married during Crystal Palace fair (1851)"
label var disthom "Wife's seat < 100 miles further than husband's seat'"
label var reldiv "Relative size of marriageable cohort in one's division (the same first-level NUTS in England and Wales, electoral region in Scotland, or province in Ireland)"
label var reldiv "Relative size of marr cohort in NUTS1, electoral region (SCO), or province (IRE)"
label var divhom "Married in same NUTS1, electoral region (SCO), or province (IRE)"
label var landclassA "Acreage class (Bateman)"
label var slandclassA "Spouse's acreage class (Bateman)"
label var landclassV "Land value class (deciles)"
label var ATot "Acreage, percentile (1851-75)"
label var VTot "Gross annual rents from land, percentile (1851-75)"
label var slandclassV "Spouse's land value (Bateman)"
label var sATot "Spouses' family acreage, percentil (1851-75)"
label var sVTot "Spouses' family gross annual rents from land, percentil (1851-75)"
label var landhoA "Spouses' from same acreage class (Bateman)"
label var landhoV "Spouses' from same land value class (deciles)"
label var missA "Absolute value of the difference in spouses' acreage (percentiles)"
label var missV "Absolute value of the difference in spouses' land value (percentiles)"
label var relandA "Relative size of marriage cohort in own land class (Bateman acreage classes)"
label var relandV "Relative size of marriage cohort in own land class (land value deciles)"
label var heirspo "Wife is a heir or co-heir"
label var raillength "Railway length (mi)"
label var railLength100 "Railway length (100s mi)"
label var ATotal "Acreage"
label var VTotal "Gross annual rents from land"
label var sATotal "Spouse's family acreage"
label var sVTotal "Spouse's family gross annual rents from land"
label var woods "Estates include woods"
label var oldfamily "Family holds land since the time of Henry VIII"
label var soldfamily "Spouse's family holds land since the time of Henry VIII"
label var swoods "Spouse's family has woods"
* save final data for men
save "${dirdta}\final-data-men.dta", replace
erase "${dirdta}\mitchell_railway.dta"
* ==============================================================================
* Figure C1:
* No selection for peers' sons.
* ------------------------------------------------------------------------------
* Import final data
use "${dirdata}\final-data.dta", clear
keep if nthismarriage<2
* Add father's reference number to identify brothers
preserve
import delimited "${dirroot}\programs\03_appendix\aux data appendix\father_son_refno.txt", clear
label var refno "Reference number in Hollingsworth"
label var refno_p "Father's reference number in Hollingsworth"
save "${dirdta}\father_son_refno.dta", replace
restore
/* Note: This file contains an individual's reference number (refno) and
their fathers' reference number (refno_p) in the Hollingsworth dataset.
Source is Gobbi and Goni (2020) "Childless aristocrats." */
merge 1:1 refno using "${dirdta}\father_son_refno.dta"
keep if _merge==3
* Sample of men married during, before, and after the interruption and their brothers
forvalues y=1858(3)1864{
local j = `y'+2
gen i`y' = (wom==0 & nthismarriage==1 & myear>=`y' & myear<=`j')
bysort refno_p: egen f`y' = max(i`y')
}
keep if wom==0 & refno_p!=. & (f1858==1 | f1861==1 | f1864==1)
* Birth order for ALL brothers
sort refno_p byear refno
by refno_p: gen birthorder = _n // birth order among brothers (including heirs)
* Panel A
preserve
replace birthorder = birthorder-0.2 if i1858==1 // just for visual purpuses
replace birthorder = birthorder+0.2 if i1864==1
twoway (histogram birthorder if i1864==1, width(0.2) discrete frac color(eltblue) lcolor(black) lwidth(thin) lpattern(dot)) ///
(histogram birthorder if i1858==1, width(0.2) discrete frac color(grey*.1) lcolor(black) lwidth(thin) lpattern(dash)) ///
(histogram birthorder if i1861==1, width(0.2) discrete frac color(black) lcolor(black) lwidth(vvvthin)), ///
legend(order(3 "Married during interruption (1861-63)" 2 "Married before interruption (1858-60)" 1 "Married after interruption (1864-66)" ) rows(3)) ///
xtitle("Birth order (among brothers)") xlabel(1(1)10) ///
title("Panel A", color(black)) graphregion(color(white)) name(gbirthorder, replace)
restore
* Panel B
gen rank_hus = prank
replace rank_hus = rank if prank<4
preserve
replace rank_hus = 10*rank_hus
replace rank_hus = rank_hus-1 if i1858==1 // just for visual purpuses
replace rank_hus = rank_hus+1 if i1864==1
twoway (histogram rank_hus if i1864==1, width(1) frac color(eltblue) lcolor(black) lwidth(thin) lpattern(dot)) ///
(histogram rank_hus if i1858==1, width(1) frac color(grey*.1) lcolor(black) lwidth(thin) lpattern(dash)) ///
(histogram rank_hus if i1861==1, width(1) frac color(black) lcolor(black) lwidth(vvvthin)), ///
legend(off) ///
xlabel(35 " " 40.5 "Baron/Viscount" 49.5 "Duke/Earl/Marquis" 55 " ") ///
title("Panel B", color(black)) graphregion(color(white)) xtitle(" ") name(gtitles, replace)
restore
* Combine and export
grc1leg gbirthorder gtitles, graphregion(color(white)) legendfrom(gbirthorder)
graph export "${dirresu}\Appendix\FigureC1.pdf", as(pdf) replace
graph drop _all
* ==============================================================================
* Table C1:
* No selection for peers' sons.
* ------------------------------------------------------------------------------
* Import final data
use "${dirdata}\final-data.dta", clear
keep if nthismarriage<2
* Add father's reference number to identify brothers
merge 1:1 refno using "${dirdta}\father_son_refno.dta"
keep if _merge==3
* Sample of men married during, before, and after the interruption and their brothers
forvalues y=1858(3)1864{
local j = `y'+2
gen i`y' = (wom==0 & nthismarriage==1 & myear>=`y' & myear<=`j')
bysort refno_p: egen f`y' = max(i`y')
}
keep if wom==0 & refno_p!=. & (f1858==1 | f1861==1 | f1864==1)
* Birth order for ALL brothers
sort refno_p byear refno
by refno_p: gen birthorder = _n // birth order among brothers (including heirs)
* Age at marriage regressions
replace mage = myear-byear if mage==.
qui reg mage i.i1861 i.birthorder i.refno_p if wom==0 & f1861==1
outreg2 using "${dirresu}\Appendix\TableC1", keep(i.i1861) nocons label dec(3) ///
ctitle("M age") addtext(% correct, `e(r2)', Birth order indicators, YES, Family FE, YES, Cluster, no) tex replace
qui reg mage i.i1861 i.birthorder i.refno_p if wom==0 & f1861==1, cluster(refno_p)
outreg2 using "${dirresu}\Appendix\TableC1", keep(i.i1861) nocons label dec(3) ///
ctitle("M age") addtext(% correct, `e(r2)', Birth order indicators, YES, Family FE, YES, Cluster, by family) tex
* Erase temp files
erase "${dirdta}\father_son_refno.dta"
* ==============================================================================
* FIGURE C2:
* Cohort size instrument and Season attendance.
* ------------------------------------------------------------------------------
* Import final data
use "${dirdata}\final-data.dta", clear
* Collapse attendees and cohort size by year, between 1851-75
keep if myear>=1851 & myear<=1875
collapse wrisk attend100, by(myear)
* Detrend cohort size
qui reg wrisk myear
predict wrisk_detrend, resid
reg attend100 myear
predict attend100_detrend, resid
* Jitter myear for visual purposes (shaded areas)
set obs `=_N+1'
replace myear = 1860.5 in l
set obs `=_N+1'
replace myear = 1863.5 in l
set obs `=_N+1'
replace myear = 1850.75 in l
set obs `=_N+1'
replace myear = 1851.25 in l
sort myear
gen low = -48 if myear>1860 & myear<1864
gen high = 37 if myear>1860 & myear<1864
gen Glow = -48 if myear>1850 & myear<1852
gen Ghigh = 37 if myear>1850 & myear<1852
* Figure
twoway (rarea low high myear, color(gs14)) ///
(rarea Glow Ghigh myear, color(gs14)) ///
(bar attend100_detrend myear, yaxis(1) barw(0.3) color(sand) ylabel(-40(10)40, nogrid)) ///
(connect wrisk_detrend myear, yaxis(2) lcolor(black) lwidth(thin) msize(small) mcolor(black) msymbol(smdiamond) ylabel(-10(5)10, axis(2) nogrid)), ///
legend(order(3 "Attendees Season (detrended)" 4 "Cohort size (detrended)")) ///
text(40 1850.5 "Great Exhibition", place(e) size(small)) ///
text(40 1860.35 "Interruption", place(e) size(small)) ///
xlabel(1851(1)1875, angle(90)) ///
graphregion(color(white)) ytitle("attendees", axis(1)) ytitle("cohort", axis(2)) xtitle("")
graph export "${dirresu}\Appendix\FigureC2.pdf", as(pdf) replace
* ==============================================================================
* TABLE C2:
* Attendance to the Season and peer-commoner intermarriage, IV estimation.
* ------------------------------------------------------------------------------
* Import data for men's analysis
use "${dirdta}\final-data-men.dta", clear
* Globals
global controls1 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60
global controls2 i.crank biorder rel1 mage hengpee dec50 t
global controlsMM rel1 sexr railLength100
global controlsTIME t dec50 dec60
label var attend100 "Attendees Season"
* Cols. [1] to [3] (women)
* probit
preserve
qui probit cOut attend100 $controls1 if wom==1 & cOut!=., cluster(myear)
keep if e(sample)==1
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC2_A", keep(attend100) nocons label dec(3) coefastr ctitle("probit") addtext(% correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Cohort controls, YES, Decade FE, YES, Trend, YES, Sample years, 1851-75) tex replace
* IV probit (1858-66)
qui ivregress 2sls cOut $controlsTIME (attend100 = mourn) if myear>=1858 & myear<=1866 & wom==1 & cOut!=., robust cluster(myear) first
qui estat firststage
matrix fist1 = r(singleresults)
local F1 = round(fist1[1,4],1)
cmp (cOut= attend100 $controls2) (attend100= mourn $controlsTIME) if myear>=1858 & myear<=1866 & wom==1 & cOut!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==cOut & p1!=.
replace correct = 0 if p2!=cOut & p1!=.
qui summ correct if myear>=1858 & myear<=1866 & wom==1 & cOut!=.
local pcp1 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC2_A", keep(attend100) nocons label dec(3) coefastr ctitle("IV-probit") addtext(% correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Cohort controls, YES, Decade FE, YES, Trend, YES, Sample years, 1858-66) tex
cmp (cOut = attend100 $controls2) (attend100 = mourn $controlsTIME) if myear>=1858 & myear<=1866 & wom==1 & cOut!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Appendix\TableC2_B", keep(mourn wrisk) nocons label dec(1) coefastr ctitle("attendees") addtext(F-stat, `F1', Cohort controls, NO, Decade FE and trend, YES, Indicator Great Exhibition, YES) tex replace
* IV probit (1851-75)
ivreg2 cOut (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMM heirratio if wom==1 & cOut!=., liml robust cluster(myear) partial(t)
local F1 = round(e(widstat),1)
local J1 = round(e(j),0.01)
local Jp1 = round(e(jp),0.01)
cmp (cOut= attend100 $controls1 heirratio) (attend100= mourn wrisk d1851 $controlsTIME $controlsMM heirratio) if wom==1 & cOut!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==cOut & p1!=.
replace correct = 0 if p2!=cOut & p1!=.
qui summ correct if wom==1 & cOut!=.
local pcp1 = round(100*r(mean),1)
drop p1 p2 correct
margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC2_A", keep(attend100) nocons label dec(3) coefastr ctitle("IV-probit") addtext(% correctly predicted, `pcp1', Sargan test, `J1', ., p=0`Jp1', Individual controls, YES, Cohort controls, YES, Decade FE, YES, Trend, YES, Sample years, 1851-75) tex
cmp (cOut= attend100 $controls1 heirratio) (attend100 = mourn wrisk d1851 $controlsTIME $controlsMM heirratio) if wom==1 & cOut!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Appendix\TableC2_B", keep(mourn wrisk) nocons label dec(1) coefastr ctitle("attendees") addtext(F-stat, `F1', Cohort controls, YES, Decade FE and trend, YES, Indicator Great Exhibition, YES) tex
restore
* Cols. [4] to [6] (men)
* probit (1851-75)
preserve
qui probit cOut attend100 $controls1 if wom==0, cluster(myear)
keep if e(sample)==1
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC2_A", keep(attend100) nocons label dec(3) coefastr ctitle("probit") addtext(% correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Cohort controls, YES, Decade FE, YES, Trend, YES, Sample years, 1851-75) tex
* IV probit (1858-66)
qui ivregress 2sls cOut $controlsTIME (attend100 = mourn) if myear>=1858 & myear<=1866 & wom==0 & cOut!=., robust cluster(myear) first
qui estat firststage
matrix fist1 = r(singleresults)
local F1 = round(fist1[1,4],1)
cmp (cOut= attend100 $controls2) (attend100 = mourn $controlsTIME) if myear>=1858 & myear<=1866 & wom==0 & cOut!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==cOut & p1!=.
replace correct = 0 if p2!=cOut & p1!=.
qui summ correct if myear>=1858 & myear<=1866 & wom==0 & cOut!=.
local pcp1 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC2_A", keep(attend100) nocons label dec(3) coefastr ctitle("IV-probit") addtext(% correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Cohort controls, YES, Decade FE, YES, Trend, YES, Sample years, 1858-66) tex
cmp (cOut= attend100 $controls2) (attend100 = mourn $controlsTIME) if myear>=1858 & myear<=1866 & wom==0 & cOut!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Appendix\TableC2_B", keep(mourn wrisk) nocons label dec(1) coefastr ctitle("attendees") addtext(F-stat, `F1', Cohort controls, YES, Decade FE and trend, YES, Indicator Great Exhibition, YES) tex
* IV probit (1851-75)
qui ivreg2 cOut (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMM if wom==0, liml robust cluster(myear) partial(t)
local F1 = round(e(widstat),1)
local J1 = round(e(j),0.01)
local Jp1 = round(e(jp),0.01)
cmp (cOut= attend100 $controls1) (attend100 = mourn wrisk d1851 $controlsTIME $controlsMM) if wom==0 & cOut!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==cOut & p1!=.
replace correct = 0 if p2!=cOut & p1!=.
qui summ correct if wom==0 & cOut!=.
local pcp1 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC2_A", keep(attend100) nocons label dec(3) coefastr ctitle("IV-probit") addtext(% correctly predicted, `pcp1', Sargan test, `J1', ., p=0`Jp1', Individual controls, YES, Cohort controls, YES, Decade FE, YES, Trend, YES, Sample years, 1851-75) tex
cmp (cOut= attend100 $controls1) (attend100 = mourn wrisk d1851 $controlsTIME $controlsMM) if wom==0 & cOut!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Appendix\TableC2_B", keep(mourn wrisk) nocons label dec(1) coefastr ctitle("attendees") addtext(F-stat, `F1', Cohort controls, YES, Decade FE and trend, YES, Indicator Great Exhibition, YES) tex
restore
* ==============================================================================
* TABLE C3:
* Attendance to the Season and other sorting patterns, IV estimation for men.
* ------------------------------------------------------------------------------
* Import data for men's analysis
use "${dirdta}\final-data-men.dta", clear
keep if wom==0
* Globals
global controlsA1 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60
global controlsA2 ATot VTot relandA mage hengpee dec50 t
global controlsV1 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsV2 ATot VTot relandV mage hengpee dec50 t
global controlsMMa relandA sexr railLength100
global controlsMMv relandV sexr railLength100
global controlsTIME t dec50 dec60
label var attend100 "Attendees at the Season (100's)"
* Panel A: Sorting by acres
* -------------------------
* probit (1851-75)
qui probit landhoA attend100 $controlsA1 if landhoA!=., cluster(myear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_A", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(% correctly predicted, `pcp1', F-stat from first stage, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) ///
tex replace
* IV probit (1858-66)
qui ivregress 2sls landhoA $controlsTIME (attend100 = mourn) if myear>=1858 & myear<=1866 & landhoA!=., robust cluster(myear) first
qui estat firststage
matrix fist5 = r(singleresults)
local F5 = round(fist5[1,4],1)
qui cmp (landhoA= attend100 $controlsA2) (attend100 = mourn $controlsTIME) ///
if myear>=1858 & myear<=1866 & landhoA!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==landhoA & p1!=.
replace correct = 0 if p2!=landhoA & p1!=.
summ correct if myear>=1858 & myear<=1866 & landhoA!=.
local pcp7 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_A", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(% correctly predicted, `pcp7', F-stat from first stage, `F5', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, NO, Sample years, 1858-66) tex
* IV probit (1851-75)
qui ivreg2 landhoA (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMMa if landhoA!=., liml robust cluster(myear) partial(t)
local F6 = round(e(widstat),1)
local J6 = round(e(j),0.01)
local Jp6 = round(e(jp),0.01)
qui ivprobit landhoA $controlsA1 (attend100 = wrisk d1851 mourn) ///
if landhoA!=., vce(cluster myear)
qui cmp (landhoA= attend100 $controlsA1) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMa) ///
if landhoA!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==landhoA & p1!=.
replace correct = 0 if p2!=landhoA & p1!=.
qui summ correct if landhoA!=.
local pcp8 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_A", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(% correctly predicted, `pcp8', F-stat from first stage, `F6', Sargan test pval, `Jp6', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* ols (1851-75)
qui reg missA attend100 $controlsA1 if missA!=., vce(cluster myear)
outreg2 using "${dirresu}\Appendix\TableC3_A", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., % correctly predicted, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) tex
* IV (1858-66)
qui ivregress 2sls missA (attend100 = mourn) $controlsTIME ///
if myear>=1858 & myear<=1866 & missA!=., vce(cluster myear)
qui estat firststage
matrix fist7 = r(singleresults)
local F7 = round(fist7[1,4],1)
qui cmp (missA= attend100 $controlsA2) (attend100 = mourn $controlsTIME) ///
if myear>=1858 & myear<=1866 & missA!=., ind(1 1) cluster(myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC3_A", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(% correctly predicted, ., F-stat from first stage, `F7', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, NO, Sample years, 1858-66) tex
* IV (1851-75)
qui ivreg2 missA (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMMa if missA!=., liml robust cluster(myear) partial(t)
local F8 = round(e(widstat),1)
local J8 = round(e(j),0.01)
local Jp8 = round(e(jp),0.01)
qui cmp (missA= attend100 $controlsA1) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMa) ///
if missA!=., ind(1 1) cluster(myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC3_A", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(% correctly predicted, ., F-stat from first stage, `F8', Sargan test pval, `Jp8', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* Panel B: Sorting by land rents
* -------------------------
* probit (1851-75)
qui probit landhoV attend100 $controlsV1 if landhoA!=., cluster(myear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_B", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(% correctly predicted, `pcp1', F-stat from first stage, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) ///
tex replace
* IV probit (1858-66)
qui ivregress 2sls landhoV $controlsTIME (attend100 = mourn) if myear>=1858 & myear<=1866 & landhoA!=., robust cluster(myear)
qui estat firststage
matrix fist9 = r(singleresults)
local F9 = round(fist9[1,4],1)
qui cmp (landhoV= attend100 $controlsV2) (attend100 = mourn $controlsTIME) ///
if myear>=1858 & myear<=1866 & landhoA!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==landhoV & p1!=.
replace correct = 0 if p2!=landhoV & p1!=.
qui summ correct if myear>=1858 & myear<=1866 & landhoA!=.
local pcp9 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_B", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(% correctly predicted, `pcp9', F-stat from first stage, `F9', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, NO, Sample years, 1858-66) tex
* IV probit (1851-75)
qui ivreg2 landhoV $controlsTIME $controlsMMv (attend100 = wrisk d1851 mourn) if landhoV!=., liml robust cluster(myear) partial(t)
local F10 = round(e(widstat),1)
local J10 = round(e(j),0.01)
local Jp10 = round(e(jp),0.01)
qui cmp (landhoV= attend100 $controlsV1) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMv) ///
if landhoV!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==landhoV & p1!=.
replace correct = 0 if p2!=landhoV & p1!=.
qui summ correct if landhoV!=.
local pcp10 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_B", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(% correctly predicted, `pcp10', F-stat from first stage, `F10', Sargan test pval, `Jp10', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* ols (1851-75)
qui reg missV attend100 $controlsV1 if missA!=., vce(cluster myear)
outreg2 using "${dirresu}\Appendix\TableC3_B", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(% correctly predicted, ., F-stat from first stage, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) tex
* IV (1858-66)
qui ivregress 2sls missV (attend100 = mourn) $controlsTIME ///
if myear>=1858 & myear<=1866 & missA!=., vce(cluster myear)
qui estat firststage
matrix fist11 = r(singleresults)
local F11 = round(fist11[1,4],1)
qui cmp (missV= attend100 $controlsV2) (attend100 = mourn $controlsTIME) ///
if myear>=1858 & myear<=1866 & missA!=., ind(1 1) cluster(myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC3_B", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(% correctly predicted, ., F-stat from first stage, `F11', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, NO, Sample years, 1858-66) tex
* IV (1851-75)
qui ivreg2 missV (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMMv if missA!=., liml robust cluster(myear) partial(t)
local F12 = round(e(widstat),1)
local J12 = round(e(j),0.01)
local Jp12 = round(e(jp),0.01)
qui cmp (missV= attend100 $controlsV1) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMv) ///
if missV!=., ind(1 1) cluster(myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC3_B", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(% correctly predicted, ., F-stat from first stage, `F12', Sargan test pval, `Jp12', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* Panel C: PCA
* -------------------------
* Data and variables for PCA
use "${dirdta}\final-data-men.dta", clear
keep if ATotal!=. & wom==0 // Keep men with landholdings
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
* Globals
global controlsPCA1 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsPCA2 i.crank ATot VTot wood2 old relandV mage hengpee t dec50 dec60
global controlsMM relandV sexr railLength100
global controlsTIME t dec50 dec60
label var attend100 "Attendees at the Season (100's)"
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
* PCA
gsort crank ATotal VTotal hengpee hscotpee hiripee old wood2 -refno
qui pca $ses
qui pca $ses, mineigen(1)
qui predict ses ses2 ses3 ses4 ses5 ses6 ses7, score
gsort prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2 -refno
qui pca $sesppo0
qui pca $sespo, mineigen(1)
qui predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7, score
* PCA sorting (missmatch by SES rank, based on PCA)
/* Note: refno included in the sorting to have stable results. */
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking)
* OLS (1851-75)
qui reg pcasort attend100 $controlsPCA1 if wom==0 & pcasort!=., cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC3_C", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) ///
tex replace
* IV (1858-66)
ivregress 2sls pcasort (attend100 = mourn) $controlsTIME ///
if myear>=1858 & myear<=1866 & wom==0 & pcasort!=., vce(cluster myear)
qui estat firststage
matrix fist13 = r(singleresults)
local F13 = round(fist13[1,4],1)
qui cmp (pcasort = attend100 $controlsPCA2) (attend100 = mourn $controlsTIME) ///
if myear>=1858 & myear<=1866 & wom==0 & pcasort!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC3_C", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F13', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, NO, Sample years, 1858-66) tex
* IV (1851-75)
qui ivreg2 pcasort (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMM if wom==0 & pcasort!=., liml robust cluster(myear) partial(t)
local F14 = round(e(widstat),1)
local J14 = round(e(j),0.01)
local Jp14 = round(e(jp),0.01)
qui cmp (pcasort = attend100 $controlsPCA1) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMM) ///
if wom==0 & pcasort!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC3_C", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F14', Sargan test pval,`Jp14', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* Panel D: Geographic endogamy
* ----------------------------
* Data
use "${dirdta}\final-data-men.dta", clear
keep if dist!=. & wom==0 // Keep spouses with seats
* Globals and labels
global controlsG1 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60
global controlsG2 i.crank reldiv mage wom hengpee t dec50 dec60
global controlsTIME t dec50 dec60
global controlsMM reldiv sexr railLength100
label var attend100 "Attendees at the Season (100's)"
* Wife < 100 miles
* Probit (1851-75)
qui probit disthom attend100 $controlsG1, vce(cluster myear)
qui estat classification
local pcp15 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_D", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(% correctly predicted, `pcp15', F-stat from first stage, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) ///
tex replace
* IV probit (1858-66)
qui ivregress 2sls disthom $controlsTIME (attend100 = mourn) if myear>=1858 & myear<=1866, robust cluster(myear)
qui estat firststage
matrix fist16 = r(singleresults)
local F16 = round(fist16[1,4],1)
qui cmp (disthom = attend100 $controlsG2) (attendLC = mourn $controlsTIME) ///
if myear>=1858 & myear<=1866, ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==disthom & p1!=.
replace correct = 0 if p2!=disthom & p1!=.
qui summ correct if myear>=1858 & myear<=1866
local pcp16 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_D", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(% correctly predicted, `pcp16', F-stat from first stage, `F16', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, no, Sample years, 1858-66) tex
* IV probit (1851-75)
qui ivreg2 disthom $controlsTIME $controlsMM (attend100 = wrisk mourn d1851), liml robust cluster(myear) partial(t)
local F17 = round(e(widstat),1)
local J17 = round(e(j),0.01)
local Jp17 = round(e(jp),0.01)
qui cmp (disthom = attend100 $controlsG1) (attendLC = wrisk mourn d1851 $controlsTIME $controlsMM) ///
, ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==disthom & p1!=.
replace correct = 0 if p2!=disthom & p1!=.
qui summ correct
local pcp17 = round(100*r(mean),1)
drop p1 p2 correct
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC3_D", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(% correctly predicted, `pcp17', F-stat from first stage, `F17', Sargan test pval, `Jp17', Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* Distance between spouses' seats
* OLS (1851-75)
qui reg dist attend100 $controlsG1, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC3_D", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., Sargan test pval, ., Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, ., Cohort size IV, ., Sample years, 1851-75) tex
* IV (1858-66)
ivregress 2sls dist (attend100 = mourn) $controlsTIME ///
if myear>=1858 & myear<=1866, vce(cluster myear)
qui estat firststage
matrix fist18 = r(singleresults)
local F18 = round(fist18[1,4],1)
qui cmp (dist=attend100 $controlsG2) (attendLC= mourn $controlsTIME) ///
if myear>=1858 & myear<=1866, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC3_D", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F18', Sargan test pval, ., Individual controls, YES, Cohort controls, NO, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, NO, Sample years, 1858-66) tex
* IV (1851-75)
qui ivreg2 dist (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMM, liml robust cluster(myear) partial(t)
local F19 = round(e(widstat),1)
local J19 = round(e(j),0.01)
local Jp19 = round(e(jp),0.01)
cmp (dist=attend100 $controlsG1) (attendLC= wrisk d1851 mourn $controlsTIME $controlsMM) ///
, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC3_D", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F19', Sargan test pval, `Jp19', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Interruption IV, YES, Cohort size IV, YES, Sample years, 1851-75) tex
* ==============================================================================
* TABLE C4:
* Attendance to the Season and sorting patterns, IV estimation controlling for distance to London.
* ------------------------------------------------------------------------------
* Import data for men's analysis
use "${dirdta}\final-data-men.dta", clear
* Globals
global controls1 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60
global controlsMM rel1 sexr railLength100
global controlsTIME t dec50 dec60
global controlsA1 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60
global controlsMMa relandA sexr railLength100
global controlsPCA1 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsG2 i.crank reldiv mage wom hengpee t dec50 dec60
label var attend100 "Attendees Season"
* Col. [1]: Peer-commoner intermarriage (women)
preserve
qui probit cOut attend100 $controls1 distlondon if wom==1 & cOut!=., cluster(myear)
keep if e(sample)==1
qui ivreg2 cOut (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMM heirratio distlondon if wom==1 & cOut!=., liml robust cluster(myear) partial(t)
local F2 = round(e(widstat),1)
local J2 = round(e(j),0.01)
local Jp2 = round(e(jp),0.01)
qui cmp (cOut= attend100 $controls1 heirratio distlondon) (attend100= wrisk d1851 mourn $controlsTIME $controlsMM heirratio distlondon) if wom==1 & cOut!=., ind(4 1) cluster(myear)
predict p1, pr
gen p2 = round(p1)
gen correct = .
replace correct = 1 if p2==cOut & p1!=.
replace correct = 0 if p2!=cOut & p1!=.
qui summ correct if wom==1 & cOut!=.
local pcp3 = round(100*r(mean),1)
drop p1 p2 correct
margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("cOut (wom)") addtext(F-stat from first stage, `F2', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, women, Sample years, 1851-75, Model, IV probit) tex replace
restore
* distance to london for men
drop distlondon
gsort surname title wgslat -nthismarr // distance London
foreach x in wgslat wgslon wgslat2 wgslon2 wgslat3 wgslon3 wgslat4 wgslon4{
by surname title: gen `x'0 = `x'[1]
}
foreach i in "" 2 3 4{
geodist wgslat`i'0 wgslon`i'0 51.507222 -0.1275, miles gen(distlnd`i')
}
egen distlondon = rowmin(distlnd distlnd2 distlnd3 distlnd4)
* Col. [2]: Peer-commoner intermarriage (men)
qui probit cOut attend100 $controls1 distlondon if wom==0, cluster(myear)
keep if e(sample)==1
qui ivreg2 cOut (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMM distlondon if wom==0, liml robust cluster(myear) partial(t)
local F4 = round(e(widstat),1)
qui cmp (cOut= attend100 $controls1 distlondon) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMM distlondon) if wom==0 & cOut!=., ind(4 1) cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("cOut (men)") addtext(F-stat from first stage, `F4', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, men, Sample years, 1851-75, Model, IV probit) tex
* Col. [3]: Acres homogamy
use "${dirdta}\final-data-men.dta", clear
label var attend100 "Attendees Season"
* distance to london for men
drop distlondon
gsort surname title wgslat -nthismarr // distance London
foreach x in wgslat wgslon wgslat2 wgslon2 wgslat3 wgslon3 wgslat4 wgslon4{
by surname title: gen `x'0 = `x'[1]
}
foreach i in "" 2 3 4{
geodist wgslat`i'0 wgslon`i'0 51.507222 -0.1275, miles gen(distlnd`i')
}
egen distlondon = rowmin(distlnd distlnd2 distlnd3 distlnd4)
keep if myear>=1851 & myear<=1875 & mage!=. & nthismarriage<2 & ATotal!=. & wom==0
qui ivreg2 landhoA (attend100 = wrisk d1851 mourn) distlondon $controlsTIME $controlsMMa if landhoA!=., liml robust cluster(myear) partial(t)
local F6 = round(e(widstat),1)
qui cmp (landhoA= attend100 $controlsA1 distlondon) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMa distlondon) if landhoA!=., ind(4 1) cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("acres Ho") addtext(F-stat from first stage, `F6', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, men, Sample years, 1851-75, Model, IV probit) tex
* Col. [4]: Acres missmatch
qui ivreg2 missA (attend100 = wrisk d1851 mourn) $controlsTIME $controlsMMa distlondon if missA!=., liml robust cluster(myear) partial(t)
local F8 = round(e(widstat),1)
qui cmp (missA= attend100 $controlsA1 distlondon) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMa distlondon) if missA!=., ind(1 1) cluster(myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("acres miss") addtext(F-stat from first stage, `F8', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, men, Sample years, 1851-75, Model, IV) tex
* Col. [5]: Land rents homogamy
qui ivreg2 landhoV $controlsTIME $controlsMMv distlondon (attend100 = wrisk d1851 mourn) if landhoV!=., liml robust cluster(myear) partial(t)
local F10 = round(e(widstat),1)
qui cmp (landhoV= attend100 $controlsV1 distlondon) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMv distlondon) if landhoV!=., ind(4 1) cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("rent ho") addtext(F-stat from first stage, `F10', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, men, Sample years, 1851-75, Model, IV probit) tex
* Col. [6]: Land rents missmatch
qui ivreg2 missV (attend100 = wrisk d1851 mourn) distlondon $controlsTIME $controlsMMv if missA!=., liml robust cluster(myear) partial(t)
local F12 = round(e(widstat),1)
qui cmp (missV= attend100 $controlsV1 distlondon) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMMv distlondon) if missV!=., ind(1 1) cluster(myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("rent miss") addtext(F-stat from first stage, `F12', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, men, Sample years, 1851-75, Model, IV) tex
* Col. [7]: PCA
use "${dirdta}\final-data-men.dta", clear
label var attend100 "Attendees Season"
* distance to london for men
drop distlondon
gsort surname title wgslat -nthismarr // distance London
foreach x in wgslat wgslon wgslat2 wgslon2 wgslat3 wgslon3 wgslat4 wgslon4{
by surname title: gen `x'0 = `x'[1]
}
foreach i in "" 2 3 4{
geodist wgslat`i'0 wgslon`i'0 51.507222 -0.1275, miles gen(distlnd`i')
}
egen distlondon = rowmin(distlnd distlnd2 distlnd3 distlnd4)
* sample
keep if ATotal!=. & wom==0
* PCA variables
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global controlsPCA1 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsPCA2 i.crank ATot VTot wood2 old relandV mage hengpee t dec50 dec60
global controlsMM relandV sexr railLength100
global controlsTIME t dec50 dec60
label var attend100 "Attendees at the Season (100's)"
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort crank ATotal VTotal hengpee hscotpee hiripee old wood2 -refno
qui pca $ses
qui pca $ses, mineigen(1)
qui predict ses ses2 ses3 ses4 ses5 ses6 ses7, score
gsort prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2 -refno
qui pca $sesppo0
qui pca $sespo, mineigen(1)
qui predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7, score
/* Note: refno included in the sorting to have stable results. */
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking)
* regression
qui ivreg2 pcasort (attend100 = wrisk d1851 mourn) distlondon $controlsTIME $controlsMM if wom==0 & pcasort!=., liml robust cluster(myear) partial(t)
local F14 = round(e(widstat),1)
qui cmp (pcasort = attend100 $controlsPCA1 distlondon) (attend100 = wrisk d1851 mourn $controlsTIME $controlsMM distlondon) if wom==0 & pcasort!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC4", keep(attend100 distlondon) nocons label dec(3) coefastr ///
ctitle("PCA") addtext(F-stat from first stage, `F14', Individual controls, YES, Cohort controls, YES, Decade FE and trend, YES, Sample, men, Sample years, 1851-75, Model, IV) tex
* ==============================================================================
* FIGURE C3:
* Increasing returns to scale, Probit estimation.
* ------------------------------------------------------------------------------
* Import data for men's analysis
use "${dirdta}\final-data-men.dta", clear
* Dep. Variable: Wife in same acres class
cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) ///
(attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoA!=., ind(4 1) cluster(myear)
margins, dydx(attend100) predict(eq(#1) pr) at(attend100=(0(1)60)) level(90) post
marginsplot, ///
recast(line) plotopts(lcolor(dknavy)) recastci(rline) ciopts(lcolor(dknavy) lpattern(dash)) ///
title("Dep. Var.: Wife in same class (acres)", color(black) size(medlarge)) ///
xtitle("Attendees at the Season", size(medlarge)) ytitle("marginal effect", size(medlarge)) ///
xlabel(0 "0" 10 "1,000" 20 "2,000" 30 "3,000" 40 "4,000" 50 "5,000" 60 "6,000") yline(0) ///
graphregion(color(white)) name(panelA_IRS, replace)
* Dep. Variable: Wife in same rents class
cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) ///
(attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoV!=., ind(4 1) cluster(myear)
margins, dydx(attend100) predict(eq(#1) pr) at(attend100=(0(1)60)) level(90) post
marginsplot, ///
recast(line) plotopts(lcolor(dknavy)) recastci(rline) ciopts(lcolor(dknavy) lpattern(dash)) ///
yscale(range(0/0.012)) ///
title("Dep. Var.: Wife in same class (land rents)", color(black) size(medlarge)) ///
xtitle("Attendees at the Season", size(medlarge)) ytitle("marginal effect", size(medlarge)) ///
xlabel(0 "0" 10 "1,000" 20 "2,000" 30 "3,000" 40 "4,000" 50 "5,000" 60 "6,000") yline(0) ///
graphregion(color(white)) name(panelB_IRS, replace)
* Combine and export
graph combine panelA_IRS panelB_IRS, graphregion(color(white)) altshrink
graph export "${dirresu}\Appendix\FigureC3.pdf", replace
graph drop _all
* ==============================================================================
* TABLE C5:
* Sample stratification.
* ------------------------------------------------------------------------------
* Import data for men's analysis
use "${dirdta}\final-data-men.dta", clear
* SES ranking, using pca
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort $ses -refno
pca $ses if wom==0, mineigen(1)
predict ses ses2 ses3 ses4 ses5 ses6 ses7 if wom==0, score
gsort $sespo -refno
pca $sespo if wom==0, mineigen(1)
predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7 if wom==0, score
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking) if wom==0
* Results "wife in acres class" (row 1)
preserve
keep if ATotal!=. & wom==0
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B11 = _b[attend100]
scalar define S11 = _se[attend100]
scalar define N11 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & crank>=6, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B12 = _b[attend100]
scalar define S12 = _se[attend100]
scalar define N12 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & crank<6, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B13 = _b[attend100]
scalar define S13 = _se[attend100]
scalar define N13 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & ATotal>=15270.5, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B14 = _b[attend100]
scalar define S14 = _se[attend100]
scalar define N14 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & ATotal<15270.5, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B15 = _b[attend100]
scalar define S15 = _se[attend100]
scalar define N15 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & VTotal>=15157, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B16 = _b[attend100]
scalar define S16 = _se[attend100]
scalar define N16 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & VTotal<15157, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B17 = _b[attend100]
scalar define S17 = _se[attend100]
scalar define N17 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & ses>=-.669556, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B18 = _b[attend100]
scalar define S18 = _se[attend100]
scalar define N18 = r(N)
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoA!=. & ses<-.669556, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B19 = _b[attend100]
scalar define S19 = _se[attend100]
scalar define N19 = r(N)
restore
* Results "wife in rents class" (row 2)
preserve
keep if ATotal!=. & wom==0
label var attend100 "Wife in rents class"
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B21 = _b[attend100]
scalar define S21 = _se[attend100]
scalar define N21 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & crank>=6, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B22 = _b[attend100]
scalar define S22 = _se[attend100]
scalar define N22 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & crank<6, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B23 = _b[attend100]
scalar define S23 = _se[attend100]
scalar define N23 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & ATotal>=15270.5, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B24 = _b[attend100]
scalar define S24 = _se[attend100]
scalar define N24 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & ATotal<15270.5, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B25 = _b[attend100]
scalar define S25 = _se[attend100]
scalar define N25 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & VTotal>=15157, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B26 = _b[attend100]
scalar define S26 = _se[attend100]
scalar define N26 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & VTotal<15157, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B27 = _b[attend100]
scalar define S27 = _se[attend100]
scalar define N27 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & ses>=-.669556, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B28 = _b[attend100]
scalar define S28 = _se[attend100]
scalar define N28 = r(N)
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if landhoV!=. & ses<-.669556, ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B29 = _b[attend100]
scalar define S29 = _se[attend100]
scalar define N29 = r(N)
restore
* Results "missmatch in acres" (row 3)
preserve
keep if ATotal!=. & wom==0
label var attend100 "Missmatch in acres"
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B31 = _b[attend100]
scalar define S31 = _se[attend100]
scalar define N31 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & crank>=6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B32 = _b[attend100]
scalar define S32 = _se[attend100]
scalar define N32 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & crank<6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B33 = _b[attend100]
scalar define S33 = _se[attend100]
scalar define N33 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & ATotal>=15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B34 = _b[attend100]
scalar define S34 = _se[attend100]
scalar define N34 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & ATotal<15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B35 = _b[attend100]
scalar define S35 = _se[attend100]
scalar define N35 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & VTotal>= 15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B36 = _b[attend100]
scalar define S36 = _se[attend100]
scalar define N36 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & VTotal<15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B37 = _b[attend100]
scalar define S37 = _se[attend100]
scalar define N37 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & ses>=-.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B38 = _b[attend100]
scalar define S38 = _se[attend100]
scalar define N38 = r(N)
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missA!=. & ses<-.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B39 = _b[attend100]
scalar define S39 = _se[attend100]
scalar define N39 = r(N)
restore
* Results "missmatch in rents" (row 4)
preserve
keep if ATotal!=. & wom==0
label var attend100 "Missmatch in rents"
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B41 = _b[attend100]
scalar define S41 = _se[attend100]
scalar define N41 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & crank>=6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B42 = _b[attend100]
scalar define S42 = _se[attend100]
scalar define N42 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & crank<6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B43 = _b[attend100]
scalar define S43 = _se[attend100]
scalar define N43 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & ATotal>= 15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B44 = _b[attend100]
scalar define S44 = _se[attend100]
scalar define N44 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & ATotal<15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B45 = _b[attend100]
scalar define S45 = _se[attend100]
scalar define N45 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & VTotal>= 15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B46 = _b[attend100]
scalar define S46 = _se[attend100]
scalar define N46 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & VTotal<15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B47 = _b[attend100]
scalar define S47 = _se[attend100]
scalar define N47 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & ses>=-.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B48 = _b[attend100]
scalar define S48 = _se[attend100]
scalar define N48 = r(N)
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if missV!=. & ses<-.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B49 = _b[attend100]
scalar define S49 = _se[attend100]
scalar define N49 = r(N)
restore
* Results "missmatch in SES" (row 5)
preserve
keep if wom==0
label var attend100 "Missmatch in SES"
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B51 = _b[attend100]
scalar define S51 = _se[attend100]
scalar define N51 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & crank>=6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B52 = _b[attend100]
scalar define S52 = _se[attend100]
scalar define N52 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & crank<6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B53 = _b[attend100]
scalar define S53 = _se[attend100]
scalar define N53 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & ATotal>= 15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B54 = _b[attend100]
scalar define S54 = _se[attend100]
scalar define N54 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & ATotal<15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B55 = _b[attend100]
scalar define S55 = _se[attend100]
scalar define N55 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & VTotal>= 15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B56 = _b[attend100]
scalar define S56 = _se[attend100]
scalar define N56 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & VTotal<15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B57 = _b[attend100]
scalar define S57 = _se[attend100]
scalar define N57 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & pcasort!=. & ses>=-.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B58 = _b[attend100]
scalar define S58 = _se[attend100]
scalar define N58 = r(N)
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if pcasort!=. & ses<-.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B59 = _b[attend100]
scalar define S59 = _se[attend100]
scalar define N59 = r(N)
restore
* Results "distance btw seats" (row 6)
preserve
keep if dist!=. & wom==0
label var attend100 "Distance between spouses' seats"
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100), ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B61 = _b[attend100]
scalar define S61 = _se[attend100]
scalar define N61 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if crank>=6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B62 = _b[attend100]
scalar define S62 = _se[attend100]
scalar define N62 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if crank<6, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B63 = _b[attend100]
scalar define S63 = _se[attend100]
scalar define N63 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if ATotal>= 15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B64 = _b[attend100]
scalar define S64 = _se[attend100]
scalar define N64 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if ATotal<15270.5, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B65 = _b[attend100]
scalar define S65 = _se[attend100]
scalar define N65 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if VTotal>= 15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B66 = _b[attend100]
scalar define S66 = _se[attend100]
scalar define N66 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if VTotal<15157, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B67 = _b[attend100]
scalar define S67 = _se[attend100]
scalar define N67 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if ses>= -.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B68 = _b[attend100]
scalar define S68 = _se[attend100]
scalar define N68 = r(N)
qui cmp (dist=attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attendLC=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if ses< -.669556, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B69 = _b[attend100]
scalar define S69 = _se[attend100]
scalar define N69 = r(N)
restore
* Table:
matrix strat = (B11, B12, B13, B14, B15, B16, B17, B18, B19\ ///
S11, S12, S13, S14, S15, S16, S17, S18, S19\ ///
N11, N12, N13, N14, N15, N16, N17, N18, N19\ ///
B21, B22, B23, B24, B25, B26, B27, B28, B29\ ///
S21, S22, S23, S24, S25, S26, S27, S28, S29\ ///
N21, N22, N23, N24, N25, N26, N27, N28, N29\ ///
B31, B32, B33, B34, B35, B36, B37, B38, B39\ ///
S31, S32, S33, S34, S35, S36, S37, S38, S39\ ///
N31, N32, N33, N34, N35, N36, N37, N38, N39\ ///
B41, B42, B43, B44, B45, B46, B47, B48, B49\ ///
S41, S42, S43, S44, S45, S46, S47, S48, S49\ ///
N41, N42, N43, N44, N45, N46, N47, N48, N49\ ///
B51, B52, B53, B54, B55, B56, B57, B58, B59\ ///
S51, S52, S53, S54, S55, S56, S57, S58, S59\ ///
N51, N52, N53, N54, N55, N56, N57, N58, N59\ ///
B61, B62, B63, B64, B65, B66, B67, B68, B69\ ///
S61, S62, S63, S64, S65, S66, S67, S68, S69\ ///
N61, N62, N63, N64, N65, N66, N67, N68, N69 ///
)
mat colnames strat = "Baseline" "Heir" "No-heir" "Acres above med" "Acres below med" "Rents above med" "Rents below med" "PCA above med" "PCA below med"
mat rownames strat = "Wife in acres class" "se" "N" "Wife in rents class" "se" "N" "Missmatch in acres" "se" "N" "Missmatch in rents" "se" "N" "Missmatch in SES" "se" "N" "Distance btw seats" "se" "N"
esttab matrix(strat, fmt(3)) using "${dirresu}\Appendix\TableC5.tex", ///
mlabel("Table C5") replace
* ==============================================================================
* TABLE C6:
* The Season and the quality of the match (1851-75).
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\final-data-men.dta", clear
label var attend100 "Attendees Season"
* Code number of births variable as a numeric variable
foreach x in livethismarr{
replace `x' = "." if `x'=="yy"
replace `x' = "0" if `x'=="xx"
destring `x', replace
}
* Total number of births (cols. [1] and [2])
qui reg livethismarr attend100 i.crank biorder mage hengpee sexr railLength100 t dec50 dec60 if wom==1 & mage<=30 & livethismarr>1 & crank!=1, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC6", keep(attend100) nocons label dec(3) coefastr ///
ctitle("OLS(tot)") ///
addtext(Sargan test, ., ., ., Individual controls, YES, Marriage market co., YES, Decade FE, YES, Year trend, YES, Interruption IV, .,Cohort size IV, .,F-stat from first stage, .) ///
tex replace
preserve
collapse attend100 riskMin d1851 mourn sexr t dec50 dec60 railLength100, by(myear)
reg attend100 riskMin d1851 mourn sexr t dec50 dec60 railLength100
local F1 = round(`e(F)',0.1)
restore
qui ivregress 2sls livethismarr (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) i.crank biorder mage hengpee sexr railLength100 t dec50 dec60 if wom==1 & mage<=30 & livethismarr>1 & crank!=1
estat overid
local S1 = round(`r(sargan)',0.01)
local P1 = round(`r(p_sargan)',0.1)
cmp (livethismarr = attend100 i.crank biorder mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==1 & mage<=30 & livethismarr>1 & crank!=1, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC6", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV(tot)") ///
addtext(Sargan test, `S1', ., p=0`P1', Individual controls, YES, Marriage market co., YES, Decade FE, YES, Year trend, YES, Interruption IV, YES, Cohort size IV, YES,F-stat from first stage, `F1') ///
tex
* Births over fertile marriage lifetime (cols. [3] and [4])
gen fert_years = 40-mage // fertile years after marriage
gen wage0 = sddate-bdate
gen wage = floor(wage0/365)
drop wage0
replace wage = sdyear - byear if sddate==. | bdate==.
egen dagemin = rowmin(dage wage)
replace fert_years = dagemin-mage if dage<40 // fertile years married (i.e., considering widowhood)
gen births_fert = livethismarr/fert_years // Births / fertile marr. lifetime
replace births_fert = . if fert_years<=0
*
qui reg births_fert attend100 i.crank biorder mage hengpee sexr railLength100 t dec50 dec60 if wom==1 & mage<=30 & livethismarr>1 & crank!=1, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC6", keep(attend100) nocons label dec(3) coefastr ///
ctitle("OLS(p.year)") ///
addtext(Sargan test, ., ., ., Individual controls, YES, Marriage market co., YES, Decade FE, YES, Year trend, YES, Interruption IV, ., Cohort size IV, .,F-stat from first stage, .) ///
tex
preserve
collapse attend100 riskMin d1851 mourn sexr t dec50 dec60 railLength100, by(myear)
reg attend100 riskMin d1851 mourn sexr t dec50 dec60 railLength100
local F1 = round(`e(F)',0.1)
restore
qui ivregress 2sls births_fert (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) i.crank biorder mage hengpee sexr railLength100 t dec50 dec60 if wom==1 & mage<=30 & livethismarr>1 & crank!=1
estat overid
local S1 = round(`r(sargan)',0.01)
local P1 = round(`r(p_sargan)',0.1)
cmp (births_fert = attend100 i.crank biorder mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==1 & mage<=30 & livethismarr>1 & crank!=1, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC6", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV(p.year)") ///
addtext(Sargan test, `S1', ., p=0`P1', Individual controls, YES, Marriage market co., YES, Decade FE, YES, Year trend, YES, Interruption IV, YES, Cohort size IV, YES,F-stat from first stage, `F1') ///
tex
* Births relative to hutterites (cols. [5] and [6])
label define bmonth1 1 "x" 2 "y" 3 "0" 4 "1" 5 "2" 6 "3" 7 "4" 8 "5" 9 "6" 10 "7" 11 "8" 12 "9"
encode bmonth, generate(bmonth1)
gen a10date = mdy(bmonth1, bday, byear+15) // years married at each age group (20-24; 25-29; 30-34; 35-39; 40-44)
forvalues t=15(5)50{
gen a`t'date = mdy(bmonth1, bday, byear+`t'+5)
format a`t'date %td
gen yb`t' = `t'+5-mage if mage<`t'+5 & mage>=`t' & sddate>a`t'date
replace yb`t' = 0 if mage>=`t'+5 & sddate>a`t'date
replace yb`t' = 5 if mage<`t' & sddate>a`t'date
replace yb`t' = sdyear-mage if mage<`t'+5 & mage>=`t' & sddate<a`t'date & sddate>a`t-5'date
gen a`t'years0 = sddate-a`t'date
gen a`t'years = floor(a`t'years0/365)
replace yb`t' = 5 - a`t'years if mage<`t' & sddate<a`t'date & sddate>a`t-5'date
replace yb`t' = . if wage>=dage & ddate<a`t'date & ddate<a`t-5'date
gen a`t'years02 = ddate-a`t'date
gen a`t'years2 = floor(a`t'years02/365)
replace yb`t' = 5 - a`t'years2 if wage>=dage & ddate<a`t'date & ddate>a`t-5'date
}
gen huttyb20 = yb20*0.55 // Calculate the number of births a Hutterite couple would have had at each age group
gen huttyb25 = yb25*0.502
gen huttyb30 = yb30*0.447
gen huttyb35 = yb35*0.406
gen huttyb40 = yb40*0.222
forvalues t=20(5)40{
replace huttyb`t' = 0 if yb`t'==.
replace yb`t' = 0 if yb`t'==.
}
gen hutt = huttyb20 + huttyb25 + huttyb30 + huttyb35 + huttyb40 // Total number of births a Hutterite couple would have had
gen births_hutt = livethismarr/hutt // Births relative to a Hutterite couple
replace births_hutt = . if hutt==0
label var births_hutt "Number of births rel. to Hutterities"
*
qui reg births_hutt attend100 i.crank biorder mage hengpee sexr railLength100 t dec50 dec60 if wom==1 & mage<=30 & livethismarr>1 & crank!=1 & hutt>1, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC6", keep(attend100) nocons label dec(3) coefastr ///
ctitle("OLS(Hutt.)") ///
addtext(Sargan test, ., ., ., Individual controls, YES, Marriage market co., YES, Decade FE, YES, Year trend, YES, Interruption IV, ., Cohort size IV, .,F-stat from first stage, .) ///
tex
preserve
collapse attend100 riskMin d1851 mourn sexr t dec50 dec60 railLength100, by(myear)
reg attend100 riskMin d1851 mourn sexr t dec50 dec60 railLength100
local F1 = round(`e(F)',0.1)
restore
qui ivregress 2sls births_hutt (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) i.crank biorder mage hengpee sexr railLength100 t dec50 dec60 if wom==1 & mage<=30 & livethismarr>1 & crank!=1 & hutt>1
estat overid
local S1 = round(`r(sargan)',0.01)
local P1 = round(`r(p_sargan)',0.1)
cmp (births_hutt = attend100 i.crank biorder mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==1 & mage<=30 & livethismarr>1 & crank!=1 & hutt>1, ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
outreg2 using "${dirresu}\Appendix\TableC6", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV(Hutt.)") title("Dep. Variable: Number of births") ///
addtext(Sargan test, `S1', ., p=0`P1', Individual controls, YES, Marriage market co., YES, Decade FE, YES, Year trend, YES, Interruption IV, YES, Cohort size IV, YES,F-stat from first stage, `F1') ///
tex
* ==============================================================================
* TABLE C7:
* Marriage and political preferences (1817-75).
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}\final-data.dta", clear
keep if myear<=1875
keep if nthismarriage<2
* Political clubs
/* Note: This is based on the clubs reported in Bateman's data. Specifically,
Brook's, Reform, and Devonshire are liberal Clubs; Carlton, Junior Carlton,
Conservative, or St. Stephen's are conservative Clubs (Bateman 1883: 497). */
* Club names housekeeping
replace club2 = "St. James'" if club2=="St. Jas."
replace club3 = "St. James'" if club3=="St. Jas."
replace club3 = "St. Stephen's" if club3=="Stephen's"
replace club4 = "Beaconsfield" if club4=="Beacsfd."
replace club7 = "United Service" if club7=="U. Serv."
replace sclub = "Carlton" if sclub=="Carlton'"
replace sclub = "Marlborough" if sclub=="Marlborough'"
replace sclub2 = "Army and Navy'" if sclub2=="Army and Navy'"
replace sclub2 = "Brooks's" if sclub2=="Brooks'"
replace sclub2 = "Conservative" if sclub2=="Conservative'"
replace sclub2 = "Naval and Military" if sclub2=="N. And M."
replace sclub2 = "St. James'" if sclub2=="St. Jas."
replace sclub2 = "White's" if sclub2=="White's'"
replace sclub3 = "Dev." if sclub3=="Devonshire"
replace sclub3 = "Guards" if sclub3=="Guard"
replace sclub3 = "St. James's" if sclub3=="St. Jas."
replace sclub3 = "Turf" if sclub3=="Turf'"
replace sclub4 = "Beaconsfield" if sclub4=="Beacsfd."
foreach x in "" 2 3 4 5 6 7 {
replace sclub`x' = "Army and Navy" if sclub`x'=="Army and Navy'"
replace sclub`x' = "Devonshire" if sclub`x'=="Dev."
replace sclub`x' = "Gun." if sclub`x'=="Gun"
replace sclub`x' = "New Edinburgh" if sclub`x'=="New Edinborough"
replace sclub`x' = "St. James'" if sclub`x'=="St. James's"
}
* Conservative vs liberal clubs
foreach s in "" s{
gen str60 `s'polit = ""
foreach x in "" 2 3 4 5 6 7{
replace `s'polit = "Tory" if `s'club`x'=="Carlton" | `s'club`x'=="Jun. Carlton" | `s'club`x'=="Conservative" | `s'club`x'=="St. Stephen's"
replace `s'polit = "Liberal" if `s'club`x'=="Brooks's" | `s'club`x'=="Devonshire" | `s'club`x'=="Reform"
}
}
* Results matrix
mat define tabC7 = (.,.,.,.,.\.,.,.,.,.\.,.,.,.,.\.,.,.,.,.\.,.,.,.,.\.,.,.,.,.\.,.,.,.,.\.,.,.,.,.\.,.,.,.,.)
* observed frequencies
tab polit spolit, matcell(ofreq)
mat tabC7[1,3] = ofreq[1,1]+ofreq[1,2]
mat tabC7[4,3] = ofreq[2,1]+ofreq[2,2]
mat tabC7[7,1] = ofreq[1,1]+ofreq[2,1]
mat tabC7[7,2] = ofreq[1,2]+ofreq[2,2]
mat tabC7[7,3] = tabC7[7,1]+tabC7[7,2]
* observed percentages
mat tabC7[1,1] = round(100*ofreq[1,1]/tabC7[1,3],0.1)
mat tabC7[1,2] = round(100*ofreq[1,2]/tabC7[1,3],0.1)
mat tabC7[4,1] = round(100*ofreq[2,1]/tabC7[4,3],0.1)
mat tabC7[4,2] = round(100*ofreq[2,2]/tabC7[4,3],0.1)
* observed percentages
tab polit spolit, expected
mat tabC7[2,1] = round(100*12.6/tabC7[1,3],0.1)
mat tabC7[2,2] = round(100*30.4/tabC7[1,3],0.1)
mat tabC7[5,1] = round(100*28.4/tabC7[4,3],0.1)
mat tabC7[5,2] = round(100*68.6/tabC7[4,3],0.1)
* differences
mat tabC7[3,1] = round(tabC7[1,1]-tabC7[2,1],0.1)
mat tabC7[3,2] = round(tabC7[1,2]-tabC7[2,2],0.1)
mat tabC7[6,1] = round(tabC7[4,1]-tabC7[5,1],0.1)
mat tabC7[6,2] = round(tabC7[4,2]-tabC7[5,2],0.1)
* chi-squared test of association
tab polit spolit, all cchi
mat tabC7[8,4] = round(r(chi2),0.001)
mat tabC7[8,5] = round(r(p),0.01)
mat tabC7[9,4] = round(r(gamma),0.01)
mat tabC7[9,5] = round(r(ase_gam),0.01)
* Export results
mat colnames tabC7 = "Liberal club" "Tory club" "N" " . " " . "
mat rownames tabC7 = "Liberal club: observed" "Liberal club: expected" "Liberal club: diff" "Tory club: observed" "Tory club: expected" "Tory club: diff" "Total: N" "Tests: Chi2" "Tests: Gamma"
esttab matrix(tabC7, fmt(2)) using "${dirresu}\Appendix\TableC7.tex", mlabel("Table C7") replace
* ==============================================================================
* FIGURE C4:
* Political endogamy over time.
* ------------------------------------------------------------------------------
* Import data
use "${dirdata}\final-data.dta", clear
keep if myear<=1875
keep if nthismarriage<2
* Political clubs
/* Note: This is based on the clubs reported in Bateman's data. Specifically,
Brook's, Reform, and Devonshire are liberal Clubs; Carlton, Junior Carlton,
Conservative, or St. Stephen's are conservative Clubs (Bateman 1883: 497). */
* Club names housekeeping
replace club2 = "St. James'" if club2=="St. Jas."
replace club3 = "St. James'" if club3=="St. Jas."
replace club3 = "St. Stephen's" if club3=="Stephen's"
replace club4 = "Beaconsfield" if club4=="Beacsfd."
replace club7 = "United Service" if club7=="U. Serv."
replace sclub = "Carlton" if sclub=="Carlton'"
replace sclub = "Marlborough" if sclub=="Marlborough'"
replace sclub2 = "Army and Navy'" if sclub2=="Army and Navy'"
replace sclub2 = "Brooks's" if sclub2=="Brooks'"
replace sclub2 = "Conservative" if sclub2=="Conservative'"
replace sclub2 = "Naval and Military" if sclub2=="N. And M."
replace sclub2 = "St. James'" if sclub2=="St. Jas."
replace sclub2 = "White's" if sclub2=="White's'"
replace sclub3 = "Dev." if sclub3=="Devonshire"
replace sclub3 = "Guards" if sclub3=="Guard"
replace sclub3 = "St. James's" if sclub3=="St. Jas."
replace sclub3 = "Turf" if sclub3=="Turf'"
replace sclub4 = "Beaconsfield" if sclub4=="Beacsfd."
foreach x in "" 2 3 4 5 6 7 {
replace sclub`x' = "Army and Navy" if sclub`x'=="Army and Navy'"
replace sclub`x' = "Devonshire" if sclub`x'=="Dev."
replace sclub`x' = "Gun." if sclub`x'=="Gun"
replace sclub`x' = "New Edinburgh" if sclub`x'=="New Edinborough"
replace sclub`x' = "St. James'" if sclub`x'=="St. James's"
}
* Conservative vs liberal clubs
foreach s in "" s{
gen str60 `s'polit = ""
foreach x in "" 2 3 4 5 6 7{
replace `s'polit = "Tory" if `s'club`x'=="Carlton" | `s'club`x'=="Jun. Carlton" | `s'club`x'=="Conservative" | `s'club`x'=="St. Stephen's"
replace `s'polit = "Liberal" if `s'club`x'=="Brooks's" | `s'club`x'=="Devonshire" | `s'club`x'=="Reform"
}
}
* Political homogamy
gen polhom = (polit==spolit & polit!="")
* other
gen decadeclub = .
forvalues i=1851(5)1875{
replace decadeclub = `i' if myear>=`i' & myear<`i'+5
}
* Figure C4
keep if polit!="" & spolit!="" & attendLC!=.
collapse (sum) attendLC (sd) sd1= attendLC (count) n1=attendLC (mean) polhom (sd) sd2=polhom (count) n2=polhom, by(decadeclub)
gen hi2 = polhom + invttail(n2-1,0.025)*(sd2 / sqrt(n2))
gen lo2 = polhom - invttail(n2-1,0.025)*(sd2 / sqrt(n2))
drop if decadeclub<1850
replace hi2=1 if hi2>1
replace lo2=0 if lo2<0
twoway (bar attendLC decadeclub, yaxis(2) color(sand)) ///
(connect polhom decadeclub, yaxis(1) mcolor(black) lcolor(black) msymbol(diamond)) ///
(connect hi2 decadeclub, yaxis(1) mcolor(black) lcolor(black) msymbol(none) lpattern(dash)) ///
(connect lo2 decadeclub, yaxis(1) mcolor(black) lcolor(black) msymbol(none) lpattern(dash)), ///
ylabel(0(40000)130000, axis(2)) ylabel(0.2 "20" 0.4 "40" 0.6 "60" 0.8 "80" 1 "100", axis(1)) ///
xlabel(1851 "1851-55" 1856 "1856-60" 1861 "1861-65" 1866 "1866-70" 1871 "1871-75") ///
xtitle("") ytitle("%", axis(1)) ytitle("attendees", axis(2)) graphregion(color(white)) ///
legend(on order(1 "Royal parties" 2 "Political endogamy" 3 "c.i. (95%)"))
graph export "${dirresu}\Appendix\FigureC4.pdf", as(pdf) replace
* ==============================================================================
* TABLE C8:
* Reduced-form effects of the interruption of the Season.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\final-data-men.dta", clear
* Globals
global controls1 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60
global controls2 i.crank biorder rel1 mage hengpee dec50 t
global controlsA1 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60
global controlsA2 ATot VTot relandA mage hengpee dec50 t
global controlsV1 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsV2 ATot VTot relandV mage hengpee dec50 t
global controlsPCA1 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsPCA2 i.crank ATot VTot wood2 old relandV mage hengpee t dec50 dec60
global controlsG1 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60
global controlsG2 i.crank reldiv mage wom hengpee t dec50 dec60
* Peer-comoner intermarriage (women)
* (1858-66)
probit cOut mourn $controls2 if myear>=1858 & myear<=1866 & wom==1 & cOut!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Mcommon (wom)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) ///
tex replace
* (1851-75)
qui probit cOut mourn $controls1 if wom==1 & cOut!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Mcommon (wom)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) ///
tex replace
* Peer-comoner intermarriage (men)
* (1858-66)
probit cOut mourn $controls2 if myear>=1858 & myear<=1866 & wom==0 & cOut!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Mcommon (men)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* (1851-75)
qui probit cOut mourn $controls1 if wom==0, cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Mcommon (men)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* Acres homogamy
use "${dirdta}\final-data-men.dta", clear
keep if ATotal!=. & wom==0
* (1858-66)
probit landhoA mourn $controlsA2 if myear>=1858 & myear<=1866 & landhoA!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Same class (acres)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* (1851-75)
qui probit landhoA mourn $controlsA1 if landhoA!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Same class (acres)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* Acres missmatch
* (1858-66)
qui reg missA mourn $controlsA2 if myear>=1858 & myear<=1866 & missA!=., cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("Miss (acres)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* (1851-75)
qui reg missA mourn $controlsA1 if missA!=., vce(cluster myear)
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("Miss (acres)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* Land rents homogamy
* (1858-66)
probit landhoV mourn $controlsV2 if myear>=1858 & myear<=1866 & landhoV!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Same class (rent)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* (1851-75)
qui probit landhoV mourn $controlsV1 if landhoV!=., cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("Same class (rent)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* Land rents missmatch
* (1858-66)
reg missV mourn $controlsV2 if myear>=1858 & myear<=1866 & missV!=., cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("Miss (rent)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* (1851-75)
qui reg missV mourn $controlsV1 if missV!=., vce(cluster myear)
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("Miss (rent)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* PCA
use "${dirdta}\final-data-men.dta", clear
gsort surname title wgslat -nthismarr
keep if ATotal!=. & wom==0
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort crank ATotal VTotal hengpee hscotpee hiripee old wood2 -refno
qui pca $ses
qui pca $ses, mineigen(1)
qui predict ses ses2 ses3 ses4 ses5 ses6 ses7, score
gsort prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2 -refno
qui pca $sesppo0
qui pca $sespo, mineigen(1)
qui predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7, score
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking)
* (1858-66)
reg pcasort mourn $controlsPCA2 if myear>=1858 & myear<=1866 & wom==0 & pcasort!=., cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("Miss (pca)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* (1851-75)
qui reg pcasort mourn $controlsPCA1 if wom==0 & pcasort!=., cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("Miss (pca)") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* region homogamy
use "${dirdta}\final-data-men.dta", clear
keep if dist!=. & wom==0
* (1858-66)
probit disthom mourn $controlsG2 if myear>=1858 & myear<=1866, cluster(myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("<100mi") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* (1851-75)
qui probit disthom mourn $controlsG1, vce(cluster myear)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr ///
ctitle("<100mi") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, probit) tex
* distance
* (1858-66)
reg dist mourn $controlsG2 if myear>=1858 & myear<=1866, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC8_A.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("dist") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* (1851-75)
qui reg dist mourn $controlsG1, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC8_B.tex", keep(mourn) nocons label dec(3) coefastr nor2 ///
ctitle("dist") addtext(Individual controls, YES, Decade FE, YES, Trend, YES, model, OLS) tex
* ==============================================================================
* TABLE C9:
* Alternative measures of attendance to the Season.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\final-data-men.dta", clear
* SES ranking, using pca
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort $ses -refno
pca $ses if wom==0, mineigen(1)
predict ses ses2 ses3 ses4 ses5 ses6 ses7 if wom==0, score
gsort $sespo -refno
pca $sespo if wom==0, mineigen(1)
predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7 if wom==0, score
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking) if wom==0
* Regressions
foreach x in attend100 invited100 attended_bc100 invited_bc100{
qui cmp (cOut= `x' i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60) (`x'=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==1 & cOut!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B1`x' = _b[`x']
scalar define S1`x' = _se[`x']
*
qui cmp (landhoA = `x' ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (`x' = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoA!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B2`x' = _b[`x']
scalar define S2`x' = _se[`x']
*
qui cmp (missA = `x' ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (`x' = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missA!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B3`x' = _b[`x']
scalar define S3`x' = _se[`x']
*
qui cmp (landhoV = `x' ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (`x' = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoV!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B4`x' = _b[`x']
scalar define S4`x' = _se[`x']
*
qui cmp (missV = `x' ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (`x' = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missV!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B5`x' = _b[`x']
scalar define S5`x' = _se[`x']
*
qui cmp (pcasort = `x' i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60) (`x'=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==0 & pcasort!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B6`x' = _b[`x']
scalar define S6`x' = _se[`x']
*
qui cmp (divhom = `x' i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (`x'=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(4 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1) pr) post
scalar define B7`x' = _b[`x']
scalar define S7`x' = _se[`x']
*
qui cmp (dist=`x' i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (`x'=riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(1 1) cluster(myear)
qui margins, dydx(*) predict(eq(#1)) post
scalar define B8`x' = _b[`x']
scalar define S8`x' = _se[`x']
}
matrix alternative = (B1attend100, B1invited100, B1attended_bc100, B1invited_bc100\ ///
S1attend100, S1invited100, S1attended_bc100, S1invited_bc100\ ///
B2attend100, B2invited100, B2attended_bc100, B2invited_bc100\ ///
S2attend100, S2invited100, S2attended_bc100, S2invited_bc100\ ///
B4attend100, B4invited100, B4attended_bc100, B4invited_bc100\ ///
S4attend100, S4invited100, S4attended_bc100, S4invited_bc100\ ///
B3attend100, B3invited100, B3attended_bc100, B3invited_bc100\ ///
S3attend100, S3invited100, S3attended_bc100, S3invited_bc100\ ///
B5attend100, B5invited100, B5attended_bc100, B5invited_bc100\ ///
S5attend100, S5invited100, S5attended_bc100, S5invited_bc100\ ///
B6attend100, B6invited100, B6attended_bc100, B6invited_bc100\ ///
S6attend100, S6invited100, S6attended_bc100, S6invited_bc100\ ///
B7attend100, B7invited100, B7attended_bc100, B7invited_bc100\ ///
S7attend100, S7invited100, S7attended_bc100, S7invited_bc100\ ///
B8attend100, B8invited100, B8attended_bc100, B8invited_bc100\ ///
S8attend100, S8invited100, S8attended_bc100, S8invited_bc100)
mat rown alternative = "Husb is commoner" "se" "Wife in acres class" "se" "Wife in rents class" "se" "Missmatch (acres)" "se" "Missmatch (rents)" "se" "Missmatch (pca)" "se" "Wife <100 mi" "se" "Distance btw seats" "se"
mat coln alternative = "baseline" "invited" "attend balls concerts" "invited balls concenrts"
esttab matrix(alternative, fmt(4)) using "${dirresu}\Appendix\TableC9.tex", mlabel("Table C9") replace
* ==============================================================================
* TABLE C10:
* IV estimates for plausibly exogenous cohort size instrument.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\final-data-men.dta", clear
* SES ranking, using pca
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort $ses -refno
pca $ses if wom==0, mineigen(1)
predict ses ses2 ses3 ses4 ses5 ses6 ses7 if wom==0, score
gsort $sespo -refno
pca $sespo if wom==0, mineigen(1)
predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7 if wom==0, score
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking) if wom==0
* Regressions
qui cmp (cOut = attend100 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==1 & cOut!=., ind(4 1) cluster(myear)
scalar define B11 = _b[attend100]
scalar define S11 = _se[attend100]
nlcom pla: [cOut]_b[attend100]+0.1 *[cOut]_b[attend100]/[attend100]_b[riskMin], post
scalar define B21 = _b[pla]
scalar define S21 = _se[pla]
qui cmp (cOut = attend100 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==1 & cOut!=., ind(4 1) cluster(myear)
nlcom pla: [cOut]_b[attend100]+0.25*[cOut]_b[attend100]/[attend100]_b[riskMin], post
scalar define B31 = _b[pla]
scalar define S31 = _se[pla]
qui cmp (cOut = attend100 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==1 & cOut!=., ind(4 1) cluster(myear)
nlcom pla: [cOut]_b[attend100]+0.5 *[cOut]_b[attend100]/[attend100]_b[riskMin], post
scalar define B41 = _b[pla]
scalar define S41 = _se[pla]
qui cmp (cOut = attend100 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==1 & cOut!=., ind(4 1) cluster(myear)
nlcom pla: [cOut]_b[attend100]+0.75*[cOut]_b[attend100]/[attend100]_b[riskMin], post
scalar define B51 = _b[pla]
scalar define S51 = _se[pla]
*
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoA!=., ind(4 1) cluster(myear)
scalar define B12 = _b[attend100]
scalar define S12 = _se[attend100]
nlcom pla: [landhoA]_b[attend100]+0.1 *[landhoA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B22 = _b[pla]
scalar define S22 = _se[pla]
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoA!=., ind(4 1) cluster(myear)
nlcom pla: [landhoA]_b[attend100]+0.25*[landhoA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B32 = _b[pla]
scalar define S32 = _se[pla]
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoA!=., ind(4 1) cluster(myear)
nlcom pla: [landhoA]_b[attend100]+0.5 *[landhoA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B42 = _b[pla]
scalar define S42 = _se[pla]
qui cmp (landhoA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoA!=., ind(4 1) cluster(myear)
nlcom pla: [landhoA]_b[attend100]+0.75*[landhoA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B52 = _b[pla]
scalar define S52 = _se[pla]
*
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoV!=., ind(4 1) cluster(myear)
scalar define B13 = _b[attend100]
scalar define S13 = _se[attend100]
nlcom pla: [landhoV]_b[attend100]+0.1 *[landhoV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B23 = _b[pla]
scalar define S23 = _se[pla]
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoV!=., ind(4 1) cluster(myear)
nlcom pla: [landhoV]_b[attend100]+0.25*[landhoV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B33 = _b[pla]
scalar define S33 = _se[pla]
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoV!=., ind(4 1) cluster(myear)
nlcom pla: [landhoV]_b[attend100]+0.5 *[landhoV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B43 = _b[pla]
scalar define S43 = _se[pla]
qui cmp (landhoV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & landhoV!=., ind(4 1) cluster(myear)
nlcom pla: [landhoV]_b[attend100]+0.75*[landhoV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B53 = _b[pla]
scalar define S53 = _se[pla]
*
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missA!=., ind(1 1) cluster(myear)
scalar define B14 = _b[attend100]
scalar define S14 = _se[attend100]
nlcom pla: [missA]_b[attend100]+0.1 *[missA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B24 = _b[pla]
scalar define S24 = _se[pla]
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missA!=., ind(1 1) cluster(myear)
nlcom pla: [missA]_b[attend100]+0.25*[missA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B34 = _b[pla]
scalar define S34 = _se[pla]
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missA!=., ind(1 1) cluster(myear)
nlcom pla: [missA]_b[attend100]+0.5 *[missA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B44 = _b[pla]
scalar define S44 = _se[pla]
qui cmp (missA = attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missA!=., ind(1 1) cluster(myear)
nlcom pla: [missA]_b[attend100]+0.75*[missA]_b[attend100]/[attend100]_b[riskMin], post
scalar define B54 = _b[pla]
scalar define S54 = _se[pla]
*
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missV!=., ind(1 1) cluster(myear)
scalar define B15 = _b[attend100]
scalar define S15 = _se[attend100]
nlcom pla: [missV]_b[attend100]+0.1 *[missV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B25 = _b[pla]
scalar define S25 = _se[pla]
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missV!=., ind(1 1) cluster(myear)
nlcom pla: [missV]_b[attend100]+0.25*[missV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B35 = _b[pla]
scalar define S35 = _se[pla]
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missV!=., ind(1 1) cluster(myear)
nlcom pla: [missV]_b[attend100]+0.5 *[missV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B45 = _b[pla]
scalar define S45 = _se[pla]
qui cmp (missV = attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100 = riskMin d1851 mourn sexr t dec50 dec60 railLength100) if wom==0 & missV!=., ind(1 1) cluster(myear)
nlcom pla: [missV]_b[attend100]+0.75*[missV]_b[attend100]/[attend100]_b[riskMin], post
scalar define B55 = _b[pla]
scalar define S55 = _se[pla]
*
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==0 & pcasort!=., ind(1 1) cluster(myear)
scalar define B16 = _b[attend100]
scalar define S16 = _se[attend100]
nlcom pla: [pcasort]_b[attend100]+0.1 *[pcasort]_b[attend100]/[attend100]_b[riskMin], post
scalar define B26 = _b[pla]
scalar define S26 = _se[pla]
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==0 & pcasort!=., ind(1 1) cluster(myear)
nlcom pla: [pcasort]_b[attend100]+0.25*[pcasort]_b[attend100]/[attend100]_b[riskMin], post
scalar define B36 = _b[pla]
scalar define S36 = _se[pla]
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==0 & pcasort!=., ind(1 1) cluster(myear)
nlcom pla: [pcasort]_b[attend100]+0.5 *[pcasort]_b[attend100]/[attend100]_b[riskMin], post
scalar define B46 = _b[pla]
scalar define S46 = _se[pla]
qui cmp (pcasort = attend100 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60) (attend100=riskMin mourn d1851 sexr railLength100 t dec50 dec60) if wom==0 & pcasort!=., ind(1 1) cluster(myear)
nlcom pla: [pcasort]_b[attend100]+0.75*[pcasort]_b[attend100]/[attend100]_b[riskMin], post
scalar define B56 = _b[pla]
scalar define S56 = _se[pla]
*
qui cmp (divhom = attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(4 1) cluster(myear)
scalar define B17 = _b[attend100]
scalar define S17 = _se[attend100]
nlcom pla: [divhom]_b[attend100]+0.1 *[divhom]_b[attend100]/[attend100]_b[riskMin], post
scalar define B27 = _b[pla]
scalar define S27 = _se[pla]
qui cmp (divhom = attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(4 1) cluster(myear)
nlcom pla: [divhom]_b[attend100]+0.25*[divhom]_b[attend100]/[attend100]_b[riskMin], post
scalar define B37 = _b[pla]
scalar define S37 = _se[pla]
qui cmp (divhom = attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(4 1) cluster(myear)
nlcom pla: [divhom]_b[attend100]+0.5 *[divhom]_b[attend100]/[attend100]_b[riskMin], post
scalar define B47 = _b[pla]
scalar define S47 = _se[pla]
qui cmp (divhom = attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(4 1) cluster(myear)
nlcom pla: [divhom]_b[attend100]+0.75*[divhom]_b[attend100]/[attend100]_b[riskMin], post
scalar define B57 = _b[pla]
scalar define S57 = _se[pla]
*
qui cmp (dist= attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(1 1) cluster(myear)
scalar define B18 = _b[attend100]
scalar define S18 = _se[attend100]
nlcom pla: [dist]_b[attend100]+0.1 *[dist]_b[attend100]/[attend100]_b[riskMin], post
scalar define B28 = _b[pla]
scalar define S28 = _se[pla]
qui cmp (dist= attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(1 1) cluster(myear)
nlcom pla: [dist]_b[attend100]+0.25*[dist]_b[attend100]/[attend100]_b[riskMin], post
scalar define B38 = _b[pla]
scalar define S38 = _se[pla]
qui cmp (dist= attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(1 1) cluster(myear)
nlcom pla: [dist]_b[attend100]+0.5 *[dist]_b[attend100]/[attend100]_b[riskMin], post
scalar define B48 = _b[pla]
scalar define S48 = _se[pla]
qui cmp (dist= attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60) (attend100 =riskMin mourn d1851 sexr t dec50 dec60 railLength100) if wom==0 & dist!=., ind(1 1) cluster(myear)
nlcom pla: [dist]_b[attend100]+0.75*[dist]_b[attend100]/[attend100]_b[riskMin], post
scalar define B58 = _b[pla]
scalar define S58 = _se[pla]
matrix plau = (B11, B21, B31, B41, B51\ ///
S11, S21, S31, S41, S51\ ///
B12, B22, B32, B42, B52\ ///
S12, S22, S32, S42, S52\ ///
B13, B23, B33, B43, B53\ ///
S13, S23, S33, S43, S53\ ///
B14, B24, B34, B44, B54\ ///
S14, S24, S34, S44, S54\ ///
B15, B25, B35, B45, B55\ ///
S15, S25, S35, S45, S55\ ///
B16, B26, B36, B46, B56\ ///
S16, S26, S36, S46, S56\ ///
B17, B27, B37, B47, B57\ ///
S17, S27, S37, S47, S57\ ///
B18, B28, B38, B48, B58\ ///
S18, S28, S38, S48, S58 ///
)
mat coln plau = "baseline" "0.1 X Beta" "0.25 X Beta" "0.5 X Beta" "0.75 X Beta"
mat rown plau = "Husb is commoner" "se" "Wife in acres class" "se" "Wife in rents class" "se" "Missmatch (acres)" "se" "Missmatch (rents)" "se" "Missmatch (pca)" "se" "Wife <100 mi" "se" "Distance btw seats" "se"
esttab matrix(plau, fmt(3)) using "${dirresu}\Appendix\TableC10.tex", mlabel("Table C10") replace
* ==============================================================================
* TABLE C11:
* Using selection from observables to assess the selection on unobservables.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\final-data-men.dta", clear
* SES ranking, using pca
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort $ses -refno
pca $ses if wom==0, mineigen(1)
predict ses ses2 ses3 ses4 ses5 ses6 ses7 if wom==0, score
gsort $sespo -refno
pca $sespo if wom==0, mineigen(1)
predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7 if wom==0, score
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking) if wom==0
* Full set of controls
qui reg cOut attend100 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60 if wom==1, cluster(myear)
scalar define Bf1 = _b[attend100]
*
preserve
keep if ATotal!=.
qui reg landhoA attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Bf2 = _b[attend100]
qui reg missA attend100 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Bf3 = _b[attend100]
restore
*
preserve
keep if VTotal!=.
qui reg landhoV attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Bf4 = _b[attend100]
qui reg missV attend100 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Bf5 = _b[attend100]
restore
*
preserve
keep if ses!=.
qui reg pcasort attend100 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Bf6 = _b[attend100]
restore
*
preserve
keep if dist!=.
qui reg divhom attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60 , cluster(myear)
scalar define Bf7 = _b[attend100]
qui reg dist attend100 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60 , cluster(myear)
scalar define Bf8 = _b[attend100]
restore
* Restricted set of controls
* (none)
qui reg cOut attend100 if wom==1, cluster(myear)
scalar define Br11 = _b[attend100]
preserve
keep if ATotal!=.
qui reg landhoA attend100 if wom==0, cluster(myear)
scalar define Br12 = _b[attend100]
qui reg missA attend100 if wom==0, cluster(myear)
scalar define Br13 = _b[attend100]
restore
*
preserve
keep if VTotal!=.
qui reg landhoV attend100 if wom==0, cluster(myear)
scalar define Br14 = _b[attend100]
qui reg missV attend100 if wom==0, cluster(myear)
scalar define Br15 = _b[attend100]
restore
*
preserve
keep if ses!=.
qui reg pcasort attend100 if wom==0, cluster(myear)
scalar define Br16 = _b[attend100]
restore
preserve
keep if dist!=.
qui reg divhom attend100 , cluster(myear)
scalar define Br17 = _b[attend100]
qui reg dist attend100 , cluster(myear)
scalar define Br18 = _b[attend100]
restore
* (time effects)
reg cOut attend100 t dec50 dec60 if wom==1, cluster(myear)
scalar define Br21= _b[attend100]
*
preserve
keep if ATotal!=.
qui reg landhoA attend100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Br22= _b[attend100]
qui reg missA attend100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Br23= _b[attend100]
restore
*
preserve
keep if VTotal!=.
qui reg landhoV attend100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Br24= _b[attend100]
qui reg missV attend100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Br25= _b[attend100]
restore
*
preserve
keep if ses!=.
qui reg pcasort attend100 t dec50 dec60 if wom==0, cluster(myear)
scalar define Br26= _b[attend100]
restore
*
preserve
keep if dist!=.
qui reg divhom attend100 t dec50 dec60 , cluster(myear)
scalar define Br27= _b[attend100]
qui reg dist attend100 t dec50 dec60 , cluster(myear)
scalar define Br28= _b[attend100]
restore
* (time effects + cohort controls)
qui reg cOut attend100 rel1 sexr t dec50 dec60 if wom==1, cluster(myear)
scalar define Br31= _b[attend100]
*
preserve
keep if ATotal!=.
qui reg landhoA attend100 relandA sexr t dec50 dec60 if wom==0, cluster(myear)
scalar define Br32= _b[attend100]
qui reg missA attend100 relandA sexr t dec50 dec60 if wom==0, cluster(myear)
scalar define Br33= _b[attend100]
restore
*
preserve
keep if VTotal!=.
qui reg landhoV attend100 relandV sexr t dec50 dec60 if wom==0, cluster(myear)
scalar define Br34= _b[attend100]
qui reg missV attend100 relandV sexr t dec50 dec60 if wom==0, cluster(myear)
scalar define Br35= _b[attend100]
restore
*
preserve
keep if ses!=.
qui reg pcasort attend100 relandV sexr t dec50 dec60 if wom==0, cluster(myear)
scalar define Br36= _b[attend100]
restore
*
preserve
keep if dist!=.
qui reg divhom attend100 reldiv sexr t dec50 dec60 , cluster(myear)
scalar define Br37= _b[attend100]
qui reg dist attend100 reldiv sexr t dec50 dec60 , cluster(myear)
scalar define Br38= _b[attend100]
restore
* (time effects + class controls)
qui reg cOut attend100 i.crank rel1 t dec50 dec60 if wom==1, cluster(myear)
scalar define Br41= _b[attend100]
*
preserve
keep if ATotal!=.
qui reg landhoA attend100 ATot VTot relandA t dec50 dec60 if wom==0, cluster(myear)
scalar define Br42= _b[attend100]
qui reg missA attend100 ATot VTot relandA t dec50 dec60 if wom==0, cluster(myear)
scalar define Br43= _b[attend100]
restore
*
preserve
keep if VTotal!=.
qui reg landhoV attend100 ATot VTot relandV t dec50 dec60 if wom==0, cluster(myear)
scalar define Br44= _b[attend100]
qui reg missV attend100 ATot VTot relandV t dec50 dec60 if wom==0, cluster(myear)
scalar define Br45= _b[attend100]
restore
*
preserve
keep if ses!=.
qui reg pcasort attend100 i.crank ATot VTot wood2 old relandV t dec50 dec60 if wom==0, cluster(myear)
scalar define Br46= _b[attend100]
restore
*
preserve
keep if dist!=.
qui reg divhom attend100 i.crank reldiv t dec50 dec60 , cluster(myear)
scalar define Br47= _b[attend100]
qui reg dist attend100 i.crank reldiv t dec50 dec60 , cluster(myear)
scalar define Br48= _b[attend100]
restore
matrix altonji = J(4,6,0)
mat rown altonji = none time timecohort timeclass
mat coln altonji = out landhoA missA landhoV missV pcasort
forvalues i=1/6{
forvalues j=1/4{
matrix altonji[`j',`i'] = abs(Bf`i'/(Br`j'`i' - Bf`i'))
}
}
mat list altonji
mat coln altonji = "H is common" "Wife acres class" "Wife rent class" "Miss acres" "Miss rent" "PCA"
mat rown altonji = "None" "Time effects" "Time effects + cohort co" "Time effects + class co"
esttab matrix(altonji, fmt(1)) using "${dirresu}\Appendix\TableC11.tex", mlabel("Table C11") replace
* ==============================================================================
* TABLE C12:
* Robustness using standard IV.
* ------------------------------------------------------------------------------
* Import data
use "${dirdta}\final-data-men.dta", clear
* Golbals
global controls1 i.crank biorder rel1 mage hengpee sexr railLength100 t dec50 dec60
global controls2 i.crank biorder rel1 mage hengpee dec50 t
label var attend100 "Attendees Season"
* Panel A. Spouse is a commoner
* ---------------------------------
* women, probit
qui probit cOut attend100 $controls1 if wom==1 & cOut!=., cluster(myear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_A.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(F-stat from first stage, ., % correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE, YES, Year trend, YES, years, 1851-75) ///
tex replace
* women, IV probit (1858-66)
qui ivregress 2sls cOut $controls2 (attend100 = mourn) if myear>=1858 & myear<=1866 & wom==1 & cOut!=., robust cluster(myear) first
qui estat firststage
matrix fist1 = r(singleresults)
local F1 = round(fist1[1,4],1)
qui ivprobit cOut $controls2 (attend100 = mourn) ///
if myear>=1858 & myear<=1866 & wom==1 & cOut!=., vce(cluster myear) first
qui estat classification
local pcp2 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_A.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F1', % correctly predicted, `pcp2', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE, YES, Year trend, YES, years, 1858-66) tex
* women, IV probit (1851-75)
qui ivreg2 cOut (attend100 = riskMin d1851 mourn) $controls1 if wom==1 & cOut!=., liml robust cluster(myear) partial(t)
local F2 = round(e(widstat),1)
local J2 = round(e(j),0.01)
local Jp2 = round(e(jp),0.01)
qui ivregress 2sls cOut $controls1 (attend100 = riskMin d1851 mourn) if wom==1 & cOut!=.
estat endog
qui ivprobit cOut $controls1 (attend100 = riskMin d1851 mourn) ///
if wom==1 & cOut!=., vce(cluster myear)
qui estat classification
local pcp3 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_A.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F2', % correctly predicted, `pcp3', Sargan test, `J2', ., p=0`Jp2', Individual controls, YES, Marriage market controls, YES, Decade FE, YES, Year trend, YES, years, 1851-75) tex
* men, probit (1851-75)
qui probit cOut attend100 $controls1 if wom==0, cluster(myear)
qui estat classification
local pcp4 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_A.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(F-stat from first stage, ., % correctly predicted, `pcp4', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE, YES, Year trend, YES, years, 1851-75) tex
* men, IV probit (1858-66)
qui ivregress 2sls cOut $controls2 (attend100 = mourn) if myear>=1858 & myear<=1866 & wom==0 & cOut!=., robust cluster(myear) first
qui estat firststage
matrix fist3 = r(singleresults)
local F3 = round(fist3[1,4],1)
qui ivprobit cOut $controls2 (attend100 = mourn) ///
if myear>=1858 & myear<=1866 & wom==0 & cOut!=., vce(cluster myear)
qui estat classification
local pcp5 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_A.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F3', % correctly predicted, `pcp5', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE, YES, Year trend, YES, years, 1858-66) tex
* men, IV probit (1851-75)
qui ivreg2 cOut (attend100 = riskMin d1851 mourn) $controls1 if wom==0, liml robust cluster(myear) partial(t)
local F4 = round(e(widstat),1)
local J4 = round(e(j),0.01)
local Jp4 = round(e(jp),0.01)
qui ivprobit cOut $controls1 (attend100 = riskMin d1851 mourn) ///
if wom==0, vce(cluster myear)
qui estat classification
local pcp6 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_A.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F4', % correctly predicted, `pcp6', Sargan test, `J4', ., p=0`Jp4', Individual controls, YES, Marriage market controls, YES, Decade FE, YES, Year trend, YES, years, 1851-75) tex
* Panel B. Acreage
* ---------------------------------
use "${dirdta}\final-data-men.dta", clear
keep if ATotal!=. & wom==0
global controlsA1 ATot VTot relandA mage hengpee sexr railLength100 t dec50 dec60
global controlsA2 ATot VTot relandA mage hengpee dec50 t
global controlsV1 ATot VTot relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsV2 ATot VTot relandV mage hengpee dec50 t
label var attend100 "Attendees Season"
* wife in land class (acres), probit (1851-75)
qui probit landhoA attend100 $controlsA1 if landhoA!=., cluster(myear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_B.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(F-stat from first stage, ., % correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) ///
tex replace
* wife in land class (acres), IV probit (1858-66)
qui ivregress 2sls landhoA $controlsA2 (attend100 = mourn) if myear>=1858 & myear<=1866 & landhoA!=., robust cluster(myear) first
qui estat firststage
matrix fist5 = r(singleresults)
local F5 = round(fist5[1,4],1)
qui ivprobit landhoA $controlsA2 (attend100 = mourn) ///
if myear>=1858 & myear<=1866 & landhoA!=., vce(cluster myear)
qui estat classification
local pcp2 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_B.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F5', % correctly predicted, `pcp2', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* wife in land class (acres), (1851-75)
qui ivreg2 landhoA (attend100 = riskMin d1851 mourn) $controlsA1 if landhoA!=., liml robust cluster(myear) partial(t)
local F6 = round(e(widstat),1)
local J6 = round(e(j),0.01)
local Jp6 = round(e(jp),0.01)
qui ivprobit landhoA $controlsA1 (attend100 = riskMin d1851 mourn) ///
if landhoA!=., vce(cluster myear)
qui estat classification
local pcp3 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_B.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F6', % correctly predicted, `pcp3', Sargan test, `J6', ., p=0`Jp6', Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1851-75) tex
* missmatch (acres), ols (1851-75)
qui reg missA attend100 $controlsA1 if missA!=., vce(cluster myear)
outreg2 using "${dirresu}\Appendix\TableC12_B.tex", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., % correctly predicted, ., Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* missmatch (acres), IV (1858-66)
qui ivregress 2sls missA (attend100 = mourn) $controlsA2 ///
if myear>=1858 & myear<=1866 & missA!=., vce(cluster myear)
qui estat firststage
matrix fist7 = r(singleresults)
local F7 = round(fist7[1,4],1)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC12_B.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F7', % correctly predicted, ., Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* missmatch (acres), IV (1851-75)
qui ivreg2 missA (attend100 = riskMin d1851 mourn) $controlsA1 if missA!=., liml robust cluster(myear) partial(t)
local F8 = round(e(widstat),1)
local J8 = round(e(j),0.01)
local Jp8 = round(e(jp),0.01)
qui ivregress 2sls missA (attend100 = riskMin d1851 mourn) $controlsA1 ///
if missA!=., vce(cluster myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC12_B.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F8', % correctly predicted, ., Sargan test, `J8', ., p=0`Jp8', Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* Panel C: Land rents
* ---------------------------------
* wife in land class (land rents), probit (1851-75)
qui probit landhoV attend100 $controlsV1 if landhoA!=., cluster(myear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_C.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(F-stat from first stage, ., % correctly predicted, `pcp1', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) ///
tex replace
* wife in land class (land rents), IV probit (1858-66)
qui ivregress 2sls landhoV $controlsV2 (attend100 = mourn) if myear>=1858 & myear<=1866 & landhoA!=., robust cluster(myear)
qui estat firststage
matrix fist9 = r(singleresults)
local F9 = round(fist9[1,4],1)
qui ivprobit landhoV $controlsV2 (attend100 = mourn) ///
if myear>=1858 & myear<=1866 & landhoA!=., vce(cluster myear)
qui estat classification
local pcp2 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_C.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F9', % correctly predicted, `pcp2', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* wife in land class (land rents), IV probit (1851-75)
qui ivreg2 landhoV $controlsV1 (attend100 = riskMin d1851 mourn) if landhoV!=., liml robust cluster(myear) partial(t)
local F10 = round(e(widstat),1)
local J10 = round(e(j),0.01)
local Jp10 = round(e(jp),0.01)
qui ivprobit landhoV $controlsV1 (attend100 = riskMin d1851 mourn) ///
if landhoV!=., vce(cluster myear)
qui estat classification
local pcp3 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_C.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F10', % correctly predicted, `pcp3', Sargan test, `J10', ., p=0`Jp10', Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* missmatch (land rents), ols (1851-75)
qui reg missV attend100 $controlsV1 if missA!=., vce(cluster myear)
outreg2 using "${dirresu}\Appendix\TableC12_C.tex", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., % correctly predicted, ., Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* missmatch (land rents), IV (1858-66)
qui ivregress 2sls missV (attend100 = mourn) $controlsV2 ///
if myear>=1858 & myear<=1866 & missA!=., vce(cluster myear)
qui estat firststage
matrix fist11 = r(singleresults)
local F11 = round(fist11[1,4],1)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC12_C.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F11', % correctly predicted, ., Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* missmatch (land rents), IV (1851-75)
qui ivreg2 missV (attend100 = riskMin d1851 mourn) $controlsV1 if missA!=., liml robust cluster(myear) partial(t)
local F12 = round(e(widstat),1)
local J12 = round(e(j),0.01)
local Jp12 = round(e(jp),0.01)
qui ivregress 2sls missV (attend100 = riskMin d1851 mourn) $controlsV1 ///
if missA!=., vce(cluster myear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Appendix\TableC12_C.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F12', % correctly predicted, ., Sargan test, `J12', ., p=0`Jp12', Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* Panel D: Principal components analysis (PCA)
* --------------------------------------------
use "${dirdta}\final-data-men.dta", clear
keep if ATotal!=. & wom==0
global controlsPCA1 i.crank ATot VTot wood2 old relandV mage hengpee sexr railLength100 t dec50 dec60
global controlsPCA2 i.crank ATot VTot wood2 old relandV mage hengpee t dec50 dec60
label var attend100 "Attendees Season"
foreach x in "" s{
gen `x'old = 1 if `x'oldfamily=="yes"
replace `x'old = 0 if `x'oldfamily=="no"
gen `x'wood2 = 1 if `x'woods=="yes"
replace `x'wood2 = 0 if `x'woods=="no"
}
gen hengpeespo = (hpeeragespo==2)
gen hscopeespo = (hpeeragespo==3)
gen hiripeespo = (hpeeragespo==4)
global ses crank ATotal VTotal hengpee hscotpee hiripee old wood2
global sespo prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2
gsort crank ATotal VTotal hengpee hscotpee hiripee old wood2 -refno
qui pca $ses
qui pca $ses, mineigen(1)
qui predict ses ses2 ses3 ses4 ses5 ses6 ses7, score
gsort prankspo sATotal sVTotal hengpeespo hscopeespo hiripeespo sold heirspo swood2 -refno
qui pca $sesppo
qui pca $sespo, mineigen(1)
qui predict sespo sespo2 sespo3 sespo4 sespo5 sespo6 sespo7, score
gsort ses -refno
gen ranking = _n if wom==0 & ses!=.
gsort sespo -refno
gen sranking = _n if wom==0 & sespo!=.
gen pcasort = abs(ranking-sranking)
* PCA missmatch, OLS (1851-75)
qui reg pcasort attend100 $controlsPCA1 if wom==0 & pcasort!=., cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC12_D.tex", keep(attend100) nocons label dec(2) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) ///
tex replace
* PCA missmatch, IV (1858-66)
ivregress 2sls pcasort (attend100 = mourn) $controlsPCA2 ///
if myear>=1858 & myear<=1866 & wom==0 & pcasort!=., vce(cluster myear)
qui estat firststage
matrix fist13 = r(singleresults)
local F13 = round(fist13[1,4],1)
outreg2 using "${dirresu}\Appendix\TableC12_D.tex", keep(attend100) nocons label dec(2) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F13', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* PCA missmatch, IV (1851-75)
qui ivreg2 pcasort (attend100 = riskMin d1851 mourn) $controlsPCA1 if wom==0 & pcasort!=., liml robust cluster(myear) partial(t)
local F14 = round(e(widstat),1)
local J14 = round(e(j),0.01)
local Jp14 = round(e(jp),0.01)
outreg2 using "${dirresu}\Appendix\TableC12_D.tex", keep(attend100) nocons label dec(2) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F14', Sargan test, `J14', ., p=0`Jp14', Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* Panel E: Geographic endogamy
* --------------------------------------------
use "${dirdta}\final-data-men.dta", clear
keep if dist!=. & wom==0
global controlsG1 i.crank reldiv mage wom hengpee sexr railLength100 t dec50 dec60
global controlsG2 i.crank reldiv mage wom hengpee t dec50 dec60
label var attend100 "Attendees Season"
* Wife < 100 miles, probit (1851-75)
qui probit disthom attend100 $controlsG1, vce(cluster myear)
qui estat classification
local pcp15 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_E.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("probit") addtext(F-stat from first stage, ., % correctly predicted, `pcp15', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) ///
tex replace
* Wife < 100 miles, IV probit (1858-66)
qui ivregress 2sls disthom $controlsG2 (attend100 = mourn) if myear>=1858 & myear<=1866, robust cluster(myear)
qui estat firststage
matrix fist16 = r(singleresults)
local F16 = round(fist16[1,4],1)
qui ivprobit disthom $controlsG2 (attend100 = mourn) ///
if myear>=1858 & myear<=1866, vce(cluster myear)
qui estat classification
local pcp16 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_E.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F16', % correctly predicted, `pcp16', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1851-75) tex
* Wife < 100 miles, IV probit (1851-75)
qui ivreg2 disthom $controlsG1 (attend100 = riskMin mourn d1851), liml robust cluster(myear) partial(t)
local F17 = round(e(widstat),1)
local J17 = round(e(j),0.01)
local Jp17 = round(e(jp),0.01)
qui ivprobit disthom $controlsG1 (attend100 = riskMin mourn d1851) ///
, vce(cluster myear)
qui estat classification
local pcp17 = round(r(P_corr),1)
qui margins, dydx(*) pred(pr) post
outreg2 using "${dirresu}\Appendix\TableC12_E.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV-probit") addtext(F-stat from first stage, `F17', % correctly predicted, `pcp17', Sargan test, `J17', ., p=0`Jp17', Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* Distance, OLS (1851-75)
qui reg dist attend100 $controlsG1, cluster(myear)
outreg2 using "${dirresu}\Appendix\TableC12_E.tex", keep(attend100) nocons label dec(3) coefastr nor2 ///
ctitle("OLS") addtext(F-stat from first stage, ., Sargan test, ., ., ., Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
* Distance, IV (1858-66)
ivregress 2sls dist (attend100 = mourn) $controlsG2 ///
if myear>=1858 & myear<=1866, vce(cluster myear)
qui estat firststage
matrix fist18 = r(singleresults)
local F18 = round(fist18[1,4],1)
outreg2 using "${dirresu}\Appendix\TableC12_E.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F18', Sargan test, ., ., ., Individual controls, YES, Marriage market controls, NO, Decade FE and trend, YES, years, 1858-66) tex
* Distance, IV (1851-75)
qui ivreg2 dist (attend100 = riskMin d1851 mourn) $controlsG1, liml robust cluster(myear) partial(t)
local F19 = round(e(widstat),1)
local J19 = round(e(j),0.01)
local Jp19 = round(e(jp),0.01)
outreg2 using "${dirresu}\Appendix\TableC12_E.tex", keep(attend100) nocons label dec(3) coefastr ///
ctitle("IV") addtext(F-stat from first stage, `F19', Sargan test, `J19', ., p=0`Jp19', Individual controls, YES, Marriage market controls, YES, Decade FE and trend, YES, years, 1851-75) tex
********************************************************************************
********************************************************************************
********************************************************************************
* *
* D. ADDITIONAL DESCRIPTIVE EVIDENCE *
* *
********************************************************************************
********************************************************************************
********************************************************************************
* ==============================================================================
* FIGURE D1:
* Top wealth-holders by source of wealth.
* ------------------------------------------------------------------------------
* Rubinstein data
/* Note: This data is from Rubinstein, W. D. 1977. "Wealth, Elites and the
Class Structure of Modern Britain." Past and Present, no. 76: pp. 99-126. */
clear
input str7 period t manu nmanu food nfood comm ncomm prof nprof land nland wealth
1809-58 1 2.631578947 5 0 0 1.578947368 3 0.526315789 1 95.26315789 181 1
1858-79 2 8.843537415 13 0.680272109 1 10.88435374 16 0 0 79.59183673 117 1
1880-99 3 22.68041237 22 14.43298969 14 23.71134021 23 0 0 39.17525773 38 1
1900-14 4 20 20 14 14 38 38 1 1 27 27 1
1809-58 1 2.770780856 11 0.251889169 1 7.052896725 28 2.01511335 8 87.9093199 349 2
1858-79 2 12.03007519 32 0.751879699 2 22.55639098 60 2.631578947 7 62.03007519 165 2
1880-99 3 20.33898305 60 7.796610169 23 22.37288136 66 3.050847458 9 46.44067797 137 2
1900-14 4 22.60536398 59 8.429118774 22 34.86590038 91 3.448275862 9 30.651341 80 2
1809-58 1 2.72572402 16 0.170357751 1 5.28109029 31 1.533219761 9 90.28960818 530 3
1858-79 2 10.89588378 45 0.726392252 3 18.40193705 76 1.694915254 7 68.28087167 282 3
1880-99 3 20.91836735 82 9.43877551 37 22.70408163 89 2.295918367 9 44.64285714 175 3
1900-14 4 21.88365651 79 9.972299169 36 35.73407202 129 2.770083102 10 29.6398892 107 3
end
gen manufood = manu+food
gen nmanufood = nmanu+nfood
label var manu "Manufacture"
label var nmanu "Manufacture (#)"
label var manufood "Manufacture, food, drink, tobacco"
label var nmanufood "Manufacture, food, drink, tobacco (#)"
label var food "Food, drink, tobacco"
label var nfood "Food, drink, tobacco (#)"
label var comm "Commercial"
label var ncomm "Commercial (#)"
label var prof "Professional, public administration, defence"
label var nprof "Professional, public administration, defence (#)"
label var land "Land"
label var nland "Land (#)"
label var wealth "Probate wealth category"
label define wlab 1 "millionaire" 2 "half-millionaire" 3 "total"
label values wealth wlab
label define tlab 1 "1809-58" 2 "1858-79" 3 "1880-99" 4 "1900-14"
label values t tlab
label var t "Time period"
drop period
* Figure
replace manufood = 4 if t==1 // for the sake of clarity
replace manufood = 8 if t==2
twoway (connected manufood t if wealth==1, mcolor(black) lcolor(black) msymbol(diamond_hollow) jitter(0 0)) ///
(connected comm t if wealth==1, mcolor(black) lcolor(black) msymbol(circle) jitter(1 1)) ///
(connected prof t if wealth==1, mcolor(black) lcolor(black) msymbol(diamond) jitter(1 1)) ///
(connected land t if wealth==1, mcolor(black) lcolor(black) msymbol(triangle)), ///
xlabel(1 "1809-58" 2 "1858-79" 3 "1880-99" 4 "1900-14") ylabel(0(20)100) ///
ytitle(% of Millionaires) xtitle("") graphregion(color(white)) legend(size(small))
graph export "${dirresu}\Appendix\FigureD1.pdf", as(pdf) replace
* ==============================================================================
* TABLE D1:
* Top wealth-holders by source of wealth.
* ------------------------------------------------------------------------------
/* Note: For the US top 1 percent, see references in the table notes.
Specifically, for rows 1 and 2 see the Statistics of Income division of the
Internal Revenue Center in 2005 (Bakija et al. 2012: Table 4). Rows 3 and 4
consider households surveyed in the Gallup polls in 2009-11 with incomes of
at least $516,633. Row 4 uses households in the top 1 percent from the 2010
Survey of Consumer Finances (Wolff 2013). */
local UStop1row1 = 12.5
local UStop1row2 = 24
local UStop1row3 = 20
local UStop1row4 = 39
local UStop1row5 = 1300
* Import data
use "${dirdata}\final-data.dta", clear
* Political clubs
replace club2 = "St. James'" if club2=="St. Jas."
replace club3 = "St. James'" if club3=="St. Jas."
replace club3 = "St. Stephen's" if club3=="Stephen's"
replace club4 = "Beaconsfield" if club4=="Beacsfd."
replace club7 = "United Service" if club7=="U. Serv."
replace sclub = "Carlton" if sclub=="Carlton'"
replace sclub = "Marlborough" if sclub=="Marlborough'"
replace sclub2 = "Army and Navy'" if sclub2=="Army and Navy'"
replace sclub2 = "Brooks's" if sclub2=="Brooks'"
replace sclub2 = "Conservative" if sclub2=="Conservative'"
replace sclub2 = "Naval and Military" if sclub2=="N. And M."
replace sclub2 = "St. James'" if sclub2=="St. Jas."
replace sclub2 = "White's" if sclub2=="White's'"
replace sclub3 = "Dev." if sclub3=="Devonshire"
replace sclub3 = "Guards" if sclub3=="Guard"
replace sclub3 = "St. James's" if sclub3=="St. Jas."
replace sclub3 = "Turf" if sclub3=="Turf'"
replace sclub4 = "Beaconsfield" if sclub4=="Beacsfd."
foreach x in "" 2 3 4 5 6 7 {
replace sclub`x' = "Army and Navy" if sclub`x'=="Army and Navy'"
replace sclub`x' = "Devonshire" if sclub`x'=="Dev."
replace sclub`x' = "Gun." if sclub`x'=="Gun"
replace sclub`x' = "New Edinburgh" if sclub`x'=="New Edinborough"
replace sclub`x' = "St. James'" if sclub`x'=="St. James's"
}
* Conservative vs liberal clubs
foreach s in "" s{
gen str60 `s'polit = ""
foreach x in "" 2 3 4 5 6 7{
replace `s'polit = "Tory" if `s'club`x'=="Carlton" | `s'club`x'=="Jun. Carlton" | `s'club`x'=="Conservative" | `s'club`x'=="St. Stephen's"
replace `s'polit = "Liberal" if `s'club`x'=="Brooks's" | `s'club`x'=="Devonshire" | `s'club`x'=="Reform"
}
}
replace polit = "none" if polit==""
* Results matrix
mat def tabD1 = (`UStop1row1' , . \ `UStop1row2' , . \ `UStop1row3' , . \ `UStop1row4' , . \ `UStop1row5' , . \ . , . )
mat l tabD1
/* Note: For the US top 1 percent, see references in the table notes:
Statistics of Income division of the Internal Revenue Center in 2005 (Bakija
et al. 2012: Table 4); Gallup polls in 2009-11; 2010 Survey of Consumer
Finances (Wolff 2013). */
* British peerage (% not married)
preserve
keep if byear>=1820 & byear<=1845 & dage>35 & nthismarr<2 & wom==0
tab nthismarr
local tot = `r(N)'
tab nthismarr if nthismarr==0
local nm = `r(N)'
mat tabD1[1,2]=round(100*`nm'/`tot',0.1)
restore
* British peerage (% spouse is a peer's daughter')
preserve
keep if myear>=1851 & myear<=1875 & nthismarr==1 & wom==0
tab cOut
local tot = `r(N)'
tab cOut if cOut==0
local mp = `r(N)'
mat tabD1[2,2]=round(100*`mp'/`tot',0.1)
restore
* British peerage (% member of a liberal club)
preserve
keep if myear>=1851 & myear<=1875 & acrestotal!=. & nthismarr==1 & wom==0
tab polit
local tot = `r(N)'
tab polit if polit=="Liberal"
local lib = `r(N)'
mat tabD1[3,2]=round(100*`lib'/`tot',0.1)
restore
* British peerage (% member of a tory club)
preserve
keep if myear>=1851 & myear<=1875 & acrestotal!=. & nthismarr==1 & wom==0
tab polit
local tot = `r(N)'
tab polit if polit=="Tory"
local tor = `r(N)'
mat tabD1[4,2]=round(100*`tor'/`tot',0.1)
restore
* British peerage (average annual income)
preserve
keep if myear>=1851 & myear<=1875 & acrestotal!=. & nthismarr==1 & wom==0
gen VTotalk = valuetotal/1000
summ VTotalk
mat tabD1[5,2] = round(`r(mean)',0.1)
mat tabD1[6,2] = 2500
restore
mat coln tabD1 = "US top 1 percent" "British peerage"
mat rown tabD1 = "Not married (\%)" "Spouse in top-notch job | is peer daughter (\%)" "Liberal | Member of a liberal club (\%)" "Conservative | Member of a tory club (\%)" "Av annual income dollars | Rents from land 1870 pounds (k)" "Rents from land (2008 dollar, k)"
*
esttab matrix(tabD1, fmt(1)) using "${dirresu}\Appendix\TableD1.tex", mlabel("Table D1") replace
* clear
erase "${dirdta}/final-data-sec4-robust1.dta"
erase "${dirdta}/final-data-sec4-robust2.dta"
***erase "${dirdta}\final-data-men.dta"
********************************************************************************
********************************************************************************
********************************************************************************
* *
* E. CONCEPTUAL FRAMEWORK APPENDIX *
* *
********************************************************************************
********************************************************************************
********************************************************************************
* ==============================================================================
* FIGURE E1:
* Comparative statics on search costs.
* ------------------------------------------------------------------------------
/* Note: This figure is based on simulating the model described in Appendix
E. Specifically, the equilibrium class bounds (i.e., the reservation
strategies) are calculated using Proposition 1, equations E8 and E9. */
clear
set obs 1000
gen x = _n/1000
* Parameters
* ----------
* Baseline:
local beta_A = 0.8
local lambda_A = 1
local alpha_A = 0.5
* High alpha:
local beta_B = 0.8
local lambda_B = 1
local alpha_B = 1.0
* High lambda:
local beta_C = 0.8
local lambda_C = 1.5
local alpha_C = 0.5
* Equilibrium class bounds:
* -------------------------
* Class bounds (i.e., reservation strategies) calculated using Proposition 1
* Baseline equilibrium:
local A0 = 1
local delta_A = (`beta_A'/(1-`beta_A'))*`alpha_A'*(`lambda_A^2')
local A1 = (1+`delta_A'*`A0'-sqrt(1+2*`delta_A'*`A0'))/`delta_A'
local A2 = (1+`delta_A'*`A1'-sqrt(1+2*`delta_A'*`A1'))/`delta_A'
local A3 = (1+`delta_A'*`A2'-sqrt(1+2*`delta_A'*`A2'))/`delta_A'
local A4 = (1+`delta_A'*`A3'-sqrt(1+2*`delta_A'*`A3'))/`delta_A'
* High alpha equilibrium:
local B0 = 1
local delta_B = (`beta_B'/(1-`beta_B'))*`alpha_B'*(`lambda_B^2')
local B1 = (1+`delta_B'*`B0'-sqrt(1+2*`delta_B'*`B0'))/`delta_B'
local B2 = (1+`delta_B'*`B1'-sqrt(1+2*`delta_B'*`B1'))/`delta_B'
local B3 = (1+`delta_B'*`B2'-sqrt(1+2*`delta_B'*`B2'))/`delta_B'
local B4 = (1+`delta_B'*`B3'-sqrt(1+2*`delta_B'*`B3'))/`delta_B'
* High lambda equilibrium:
local C0 = 1
local delta_C = (`beta_C'/(1-`beta_C'))*`alpha_C'*(`lambda_C^2')
local C1 = (1+`delta_C'*`C0'-sqrt(1+2*`delta_C'*`C0'))/`delta_C'
local C2 = (1+`delta_C'*`C1'-sqrt(1+2*`delta_C'*`C1'))/`delta_C'
local C3 = (1+`delta_C'*`C2'-sqrt(1+2*`delta_C'*`C2'))/`delta_C'
local C4 = (1+`delta_C'*`C3'-sqrt(1+2*`delta_C'*`C3'))/`delta_C'
* Figure:
* -------
foreach z in A B C{
gen hi_`z' = ``z'0' if x>=``z'1'
replace hi_`z' = ``z'1' if x>=``z'2' & x<``z'1'
replace hi_`z' = ``z'2' if x>=``z'3' & x<``z'2'
replace hi_`z' = ``z'3' if x>=0 & x<``z'3'
gen lo_`z' = ``z'1' if x>=``z'1'
replace lo_`z' = ``z'2' if x>=``z'2' & x<``z'1'
replace lo_`z' = ``z'3' if x>=``z'3' & x<``z'2'
replace lo_`z' = ``z'4' if x>=0 & x<``z'3'
}
twoway (rarea hi_A lo_A x, fc(none) lc(black) lp(dash)) ///
(rarea hi_B lo_B x, fc(none) lc(blue)), ///
legend(order(1 "beta = `beta_A'; lambda = `lambda_A'; alpha = `alpha_A'" 2 "beta = `beta_B'; lambda = `lambda_B'; alpha = `alpha_B'") row(2) lc(none)) ///
title("Panel A: Encounter speed (alpha)", size(medsmall) color(black)) aspect(1) ///
xtitle("x") ytitle("y") graphregion(color(white)) name(PanelA, replace)
twoway (rarea hi_A lo_A x, fc(none) lc(black) lp(dash)) ///
(rarea hi_C lo_C x, fc(none) lc(red)), ///
legend(order(1 "beta = `beta_A'; lambda = `lambda_A'.0; alpha = `alpha_A'" 2 "beta = `beta_C'; lambda = `lambda_C'; alpha = `alpha_C'") row(2) lc(none)) ///
title("Panel B: Num of attendees (lambda)", size(medsmall) color(black)) ///
aspect(1) xtitle("x") ytitle("y") graphregion(color(white)) name(PanelB, replace)
graph combine PanelA PanelB, graphregion(color(white))
graph export "${dirresu}\Appendix\FigureE1.pdf", as(pdf) replace
* ==============================================================================
* FIGURE E2:
* Comparative statics on market segmentation.
* ------------------------------------------------------------------------------
/* Note: This figure is based on simulating the model described in Appendix
E. Specifically, the equilibrium class bounds (i.e., the reservation
strategies) are calculated using Proposition 1, equations E8 and E9. For the
model with market segregation, the equilibrium classes are derived in page
86 of the online appendix */
clear
set obs 1000
gen x = _n/1000
* Parameters
* ----------
* Baseline:
local beta_A = 0.8
local lambda_A = 1
local alpha_A = 1
local z_A = 0
* Market segmentation (z=0.24):
local beta_B = 0.8
local lambda_B = 1
local alpha_B = 1
local z_B = 0.24
* Equilibrium class bounds:
* -------------------------
* Class bounds (i.e., reservation strategies) calculated using Proposition 1
* and the eq. reservation strategies derived in page 86 of the online appendix.
* Baseline equilibrium:
local A0 = 1
local coef1 = (((1-`z_A')^2) * sqrt(1-`beta_A')) / (`beta_A'*`alpha_A'*(1-`z_A')*(`lambda_A'^1.1))
local coef2 = `beta_A'*`alpha_A'*(1-`z_A')*(`lambda_A'^(1.1)) / ((1-`z_A')^2)
local A1 = `A0' - `coef1' * (sqrt(1-`beta_A'+2*`A0'*`coef2') - sqrt(1-`beta_A'))
local A2 = `A1' - `coef1' * (sqrt(1-`beta_A'+2*`A1'*`coef2') - sqrt(1-`beta_A'))
local A3 = `A2' - `coef1' * (sqrt(1-`beta_A'+2*`A2'*`coef2') - sqrt(1-`beta_A'))
local A4 = `A3' - `coef1' * (sqrt(1-`beta_A'+2*`A3'*`coef2') - sqrt(1-`beta_A'))
* Market segmentation equilibrium:
local B0 = 1
local coef1 = (((1-`z_B')^2) * sqrt(1-`beta_B')) / (`beta_B'*`alpha_B'*(1-`z_B')*(`lambda_B'^1.1))
local coef2 = `beta_B'*`alpha_B'*(1-`z_B')*(`lambda_B'^(1.1)) / ((1-`z_B')^2)
local B1 = `B0' - `coef1' * (sqrt(1-`beta_B'+2*`B0'*`coef2') - sqrt(1-`beta_B'))
local B2 = `B1' - `coef1' * (sqrt(1-`beta_B'+2*`B1'*`coef2') - sqrt(1-`beta_B'))
local B3 = `B2' - `coef1' * (sqrt(1-`beta_B'+2*`B2'*`coef2') - sqrt(1-`beta_B'))
local B4 = `B3' - `coef1' * (sqrt(1-`beta_B'+2*`B3'*`coef2') - sqrt(1-`beta_B'))
* Figure:
* -------
foreach q in A B{
gen hi_`q' = ``q'0' if x>=``q'1'
replace hi_`q' = ``q'1' if x>=``q'2' & x<``q'1'
replace hi_`q' = ``q'2' if x>=``q'3' & x<``q'2'
replace hi_`q' = ``q'3' if x>=0 & x<``q'3'
gen lo_`q' = ``q'1' if x>=``q'1'
replace lo_`q' = ``q'2' if x>=``q'2' & x<``q'1'
replace lo_`q' = ``q'3' if x>=``q'3' & x<``q'2'
replace lo_`q' = ``q'4' if x>=0 & x<``q'3'
* mark those excluded from the market as missing values:
replace hi_`q' = . if x<`z_`q''
replace lo_`q' = . if x<`z_`q''
}
twoway (rarea hi_A lo_A x, fc(none) lc(black) lp(dash)) ///
(rarea hi_B lo_B x, fc(none) lc(blue)), ///
xline(`z_B', lpattern(dot) lc(blue)) yline(`z_B', lpattern(dot) lc(blue)) ///
legend(order(1 "beta = `beta_A'; lambda = `lambda_A'; alpha = `alpha_A'; z = `z_A'" 2 "beta = `beta_B'; lambda = `lambda_B'; alpha = `alpha_B'; z = `z_B'") row(2) lc(none)) ///
aspect(1) xtitle("x") ytitle("y") graphregion(color(white))
graph export "${dirresu}\Appendix\FigureE2.pdf", as(pdf) replace