anonymous-submission-acl2025's picture
add 3 4
f81d4f9
/*******************************************************************************
This file contains the code to replicate the tables and figures in the paper.
Contents:
0. Generate or import final dataset
Figure 1: The Season in 1841
Figure 2: Pressure marry young.
Figure 3: Attendees at royal parties, by type of event.
Figure 4: Synthetic probability to marry during the Season's interruption (1861-63).
Table 1: Marriage market before, during, and after the Season's interruption.
Figure 5: The interruption of the Season and distance between spouses' seats.
Table 2: The Season's interruption and marriage outcomes, probit and OLS estimation.
Figure 6: Placebo tests. Dependent variable: married a commoner.
Figure 7: Peer-commoner intermarriage, Probit estimation with age dummies.
Table 3: The Season's interruption and marriage outcomes, IV estimation.
Table 4: Contingency tables
Table 5: The interruption and sorting by title, non-parametric estimates.
Figure 8: Sorting by landholdings, non-parametric estimation.
Table 6: Women's marriages to commoners and her family's political power, IV estimation.
Table 7: Determinants of investments in state education, IV estimation.
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
* ==============================================================================
* ==============================================================================
* FIGURE 1:
* The Season in 1841.
* ------------------------------------------------------------------------------
* Import Sheppard (1977)'s data on 4,000 movements into and out of London of
* families participating in the Season of 1841, as reported in The Morning Post.
import delimited "${dirdata}\sheppard.txt", clear
label var week "Week"
label var arrivals "Arrivals to London of members of Fashionable World"
label var departures "Departures from London of members of Fashionable World"
label var cumulative "Cumulative members of Fashionable World in London"
* Figure 1:
twoway (bar arrivals week, fcolor(dknavy) fintensity(inten50) lcolor(dknavy) barw(1)) ///
(bar departures week, fcolor(maroon) fintensity(inten50) lcolor(maroon) barw(1)) ///
(line cumulative week, lcolor(black)) ///
(scatter cumulative week if week==4 | week==15 | week==20 | week==32), ///
ytitle("Number of families") xtitle("") ///
legend(col(3) on order(1 "Arrivals" 2 "Departures" 3 "Cummulative")) graphregion(color(white)) ///
xlabel(1 "Jan" 5 "Feb" 9 "Mar" 13 "Apr" 17 "May" 22 "Jun" 26 "Jul" 31 "Ago" 35 "Sep" 39 "Oct" 44 "Nov" 48 "Dec") ///
text(310 4 "Opening of Parliament", place(e) size(small)) ///
text(440 17 "Queen arrives to London", place(s) size(small)) ///
text(850 20 "Royal ball", place(n) size(small)) ///
text(420 32 "End of the Season", place(e) size(small))
graph export "${dirresu}\Figure1.pdf", as(pdf) replace
* ==============================================================================
* FIGURE 2:
* Pressure marry young.
* ------------------------------------------------------------------------------
* Import final dataset
use "${dirdata}\final-data.dta", clear
* Sample: all 796 peers' daughters first marrying in 1851-75
keep if wom==1 & nthismarriage==1 & myear>=1851 & myear<=1875 & mage!=.
* Variables for Figure
gen agegroup = . // age groups in figure
forvalues i=18(3)35{
replace agegroup = `i' if mage>=`i' & mage<=`i'+2
}
replace agegroup = 30 if mage>=30
keep if agegroup!=. & agegroup<=30
gen duk = (rankspo==5) // marrying a dukes' heir
* Figure 2, first panel
preserve
collapse (mean) duk (sd) sd_duk=duk (count) n_duk=duk, by(agegroup)
gen hi_duk = duk + invttail(n_duk-1,0.025)*(sd_duk / sqrt(n_duk))
gen lo_duk = duk - invttail(n_duk-1,0.025)*(sd_duk / sqrt(n_duk))
label var agegroup "Age at marriage"
label var duk "% marrying duke, marquis, earl heir"
label var hi_duk "CI high, % marrying duke, marquis, earl heir"
label var lo_duk "CI low, % marrying duke, marquis, earl heir"
twoway (scatter duk agegroup, lcolor(dknavy) mcolor(black) msymbol(circle)) ///
(rcap hi_duk lo_duk agegroup, lcolor(black)), ///
xlabel(18 "18-20" 21 "21-23" 24 "24-26" 27 "27-29" 30 "{&ge}30", labsize(medlarge) angle(45)) ///
ylabel(0 "0" 0.1 "10" 0.2 "20" 0.3 "30", labsize(medlarge)) ///
legend(on order(1 "All peers' daughters") ring(0) position(1) nobox region(lstyle(none))) ///
ytitle("% marrying duke, marquis, earl heir", size(medlarge)) ///
xtitle("Age at marriage", size(medlarge)) ///
name(fig2panela, replace) graphregion(color(white))
restore
* Figure 2, second panel
preserve
collapse (mean) duk (sd) sd_duk=duk (count) n_duk=duk, by(agegroup prank)
gen hi_duk = duk + invttail(n_duk-1,0.025)*(sd_duk / sqrt(n_duk))
gen lo_duk = duk - invttail(n_duk-1,0.025)*(sd_duk / sqrt(n_duk))
replace lo_duk = 0 if lo_duk<0 // restrict lower bound of c.i. to zero for visual purposes
label var duk "% marrying duke, marquis, earl heir"
label var hi_duk "CI high, % marrying duke, marquis, earl heir"
label var lo_duk "CI low, % marrying duke, marquis, earl heir"
gen agegroup1 = agegroup-0.3 // separate estimates for the two groups for visibility
gen agegroup0 = agegroup+0.3
label var agegroup "Age at marriage"
label var agegroup0 "Age at marriage"
label var agegroup1 "Age at marriage"
twoway (rcap hi_duk lo_duk agegroup1 if prank==5, lcolor(blue)) ///
(rcap hi_duk lo_duk agegroup0 if prank==4, lcolor(red)) ///
(scatter duk agegroup1 if prank==5, lcolor(blue) mcolor(blue) msymbol(diamond)) ///
(scatter duk agegroup0 if prank==4, lcolor(red) msymbol(diamond) mlc(red) mfc(white)), ///
ylabel(0 "0" 0.1 "10" 0.2 "20" 0.3 "30" 0.4 "40", labsize(medlarge)) ///
xlabel(18 "18-20" 21 "21-23" 24 "24-26" 27 "27-29" 30 "{&ge}30", labsize(medlarge) angle(45)) ///
legend(on order(3 "Dukes', Marquis', Earls' daughters" 4 "Barons', Viscounts' daughters") rows(2) ring(0) position(1) nobox region(lstyle(none))) ///
ytitle(" ", size(medlarge)) ///
xtitle("Age at marriage", size(medlarge)) ///
name(fig2panelb, replace) graphregion(color(white))
restore
* Figure 2, combined panels
graph combine fig2panela fig2panelb, graphregion(color(white))
graph export "${dirresu}\Figure2.pdf", replace
graph drop fig2panela fig2panelb
* ==============================================================================
* FIGURE 3:
* Attendees at royal parties, by type of event.
* ------------------------------------------------------------------------------
* Import data with the invitations to royal parties, the number of attendees
* at each party, and the numbers presented at court in the Seasons of 1851-75.
import delimited "${dirdata}\seasonattend.txt", clear
* Calculate attendance to royal parties, by type of event
gen str15 partytype2 = partytype
replace partytype2 = "Concert and Evening party" if partytype=="Concert" | partytype=="Evening Party"
replace partytype2 = "Breakfast and Afternoon party" if partytype=="Breakfast" | partytype=="Afternoon"
replace partytype2 = "Children's Ball" if partytype2=="Child's Ball"
local j = 1
foreach x in "Ball" "Concert and Evening party" "Breakfast and Afternoon party" "Children's Ball" "Court"{
bysort year: egen tattend`j' = sum(attended) if partytype2=="`x'"
gsort year -tattend`j'
by year: replace tattend`j' = tattend`j'[1] if tattend`j'==.
replace tattend`j' = 0 if tattend`j'==.
local j = `j'+1
}
gen tAttend2 = tattend1+tattend2
gen tAttend3 = tAttend2+tattend3
gen tAttend4 = tAttend3+tattend4
gen tAttend5 = tAttend4+tattend5
collapse tattend1 tAttend2 tAttend3 tAttend4 tAttend5 court, by(year)
* Figure 3
twoway ///
(bar tattend1 year, barw(0.7) bcolor(dknavy) yaxis(1)) ///
(rbar tattend1 tAttend2 year, bcolor(orange) barw(0.7) yaxis(1)) ///
(rbar tAttend2 tAttend3 year, bcolor(forest_green) barw(0.7) yaxis(1)) ///
(rbar tAttend3 tAttend4 year, bcolor(maroon) barw(0.7) yaxis(1)) ///
(rbar tAttend4 tAttend5 year, bcolor(teal) barw(0.7) yaxis(1)) ///
(line court year, yaxis(2) lcolor(black)), ///
ytitle("attendees", axis(1)) ytitle("debutantes", axis(2)) ///
legend(on order(1 "Ball" 2 "Concert and Evening party" 3 "Children's Ball" ///
4 "Breakfast and Afternoon party" 5 "Court reception" 6 "Debutantes presented at court (Ellenberger 1990)") size(small) nobox region(lstyle(none))) ///
xtitle("") xlabel(1851(1)1875, angle(90)) graphregion(color(white))
graph export "${dirresu}\Figure3.pdf", replace
* ==============================================================================
* FIGURE 4:
* Synthetic probability to marry during the Season's interruption (1861-63).
* ------------------------------------------------------------------------------
* Import final dataset
use "${dirdata}\final-data.dta", clear
* Figure 4
keep if age1861>=15 & age1861<=35
collapse syntheticT, by(age1861)
twoway (connect syntheticT age1861, sort lcolor(dknavy) mcolor(dknavy) msymbol(diamond)), ///
ytitle("synthetic probability to marry in 1861-63 (%)") xtitle("age in 1861") ///
legend(off) graphregion(color(white))
graph export "${dirresu}\Figure4.pdf", replace
* ==============================================================================
* TABLE 1:
* Marriage market before, during, and after the Season's interruption.
* ------------------------------------------------------------------------------
* Import final dataset
clear all
use "${dirdata}\final-data.dta", clear
* Sample and labels
keep if wom==1 & nthismarriage==1 & myear>=1858 & myear<=1866
xtile fATot_tab1 = fatotal, nq(100) // percentiles for Table 1's sample
gen normal = 1-mourn // to have Interruption - Normal
global balancelist mage dage biorder pr4 hengpee fATot_tab1
* Results
mat define tab1 = (.,.,.,.,.,.,.,.,.,.\.,.,.,.,.,.,.,.,.,.\.,.,.,.,.,.,.,.,.,.\.,.,.,.,.,.,.,.,.,.\.,.,.,.,.,.,.,.,.,.\.,.,.,.,.,.,.,.,.,.\.,.,.,.,.,.,.,.,.,.)
* Panel A, mean
qui estpost sum $balancelist if normal==0
qui matrix mu=e(mean)
forvalues j=1/7{
mat tab1[`j',1] = mu[1,`j']
}
mat drop mu
qui estpost sum $balancelist if normal==1 & myear<=1863
qui matrix mu=e(mean)
forvalues j=1/7{
mat tab1[`j',3] = mu[1,`j']
}
mat drop mu
qui estpost sum $balancelist if normal==1 & myear>1863
qui matrix mu=e(mean)
forvalues j=1/7{
mat tab1[`j',7] = mu[1,`j']
}
mat drop mu
* Panel A, ttest
eststo tstbef: qui estpost ttest $balancelist if myear<=1863, by(normal)
qui matrix di1=e(b)
qui matrix di2=e(se)
forvalues j=1/7{
mat tab1[`j',5] = di1[1,`j']
mat tab1[`j',6] = di2[1,`j']
}
mat drop di1 di2
eststo tstaft: qui estpost ttest $balancelist if myear>=1861, by(normal)
qui matrix di1=e(b)
qui matrix di2=e(se)
forvalues j=1/7{
mat tab1[`j',9] = di1[1,`j']
mat tab1[`j',10] = di2[1,`j']
}
mat drop di1 di2
* Panel A, se of means
local j = 1
foreach x in mage dage biorder pr4 hengpee fATot_tab1{
qui ttest `x' if myear<=1863, by(normal)
local se1 = `r(sd_1)'/sqrt(`r(N_1)')
local se2 = `r(sd_2)'/sqrt(`r(N_2)')
mat tab1[`j',2] = `se1'
mat tab1[`j',4] = `se2'
qui ttest `x' if myear>=1861, by(normal)
local se2 = `r(sd_2)'/sqrt(`r(N_2)')
mat tab1[`j',8] = `se2'
local j = `j'+1
}
* Panel B, mean
collapse wrisk mourn normal, by(myear)
qui estpost sum wrisk if normal==0
qui matrix mu=e(mean)
mat tab1[7,1] = mu[1,1]
mat drop mu
qui estpost sum wrisk if normal==1 & myear<=1863
qui matrix mu=e(mean)
mat tab1[7,3] = mu[1,1]
mat drop mu
qui estpost sum wrisk if normal==1 & myear>1863
qui matrix mu=e(mean)
mat tab1[7,7] = mu[1,1]
mat drop mu
* Panel B, ttest
eststo tstbef: qui estpost ttest wrisk if myear<=1863, by(normal)
qui matrix di1=e(b)
qui matrix di2=e(se)
mat tab1[7,5] = di1[1,1]
mat tab1[7,6] = di2[1,1]
mat drop di1 di2
eststo tstaft: qui estpost ttest wrisk if myear>=1861, by(normal)
qui matrix di1=e(b)
qui matrix di2=e(se)
mat tab1[7,9] = di1[1,1]
mat tab1[7,10] = di2[1,1]
mat drop di1 di2
* Panel B, se of means
qui ttest wrisk if myear<=1863, by(normal)
local se1 = `r(sd_1)'/sqrt(`r(N_1)')
local se2 = `r(sd_2)'/sqrt(`r(N_2)')
mat tab1[7,2] = `se1'
mat tab1[7,4] = `se2'
qui ttest wrisk if myear>=1861, by(normal)
local se2 = `r(sd_2)'/sqrt(`r(N_2)')
mat tab1[7,8] = `se2'
* Table 1
mat list tab1
mat colnames tab1 = "Int:mean" "Int:se" "Bef:mean" "Bef:se" "Diff:mean" "Diff:se" "Aft:mean" "Aft:se" "Diff:mean" "Diff:se"
mat rownames tab1 = "Panel A: Age at first marriage" "Panel A: Life expectancy" "Panel A: Birth order (excl heirs)" "Panel A: Duke/Earl/Marquis daughter" "Panel A: Peerage of England" "Panel A: Family acreage (percentile)" "Panel B: Female cohort size"
esttab matrix(tab1, fmt(1 1 1 1 1 1 1)) using "${dirresu}\Table1.tex", ///
mlabel("Table 1") replace
* ==============================================================================
* FIGURE 5:
* The interruption of the Season and distance between spouses' seats.
* ------------------------------------------------------------------------------
* Import final dataset
use "${dirdata}\final-data.dta", clear
* Figure 5
keep if wom==1 & nthismarriage<2 & myear>=1858 & myear<=1866
collapse (mean) dist (mean) attendLC, by(myear)
twoway (bar attendLC myear, barw(0.3) yaxis(2) color(sand) xlabel(1858(1)1866)) ///
(connect dist myear, yaxis(1) mcolor(black) lcolor(black) msymbol(diamond)), ///
ytitle("miles", axis(1)) ytitle("attendees", axis(2)) xtitle("year") ///
legend(on order(1 "Royal parties" 2 "Distance between spouses' seats")) ///
graphregion(color(white))
graph export "${dirresu}\Figure5.pdf", replace
* ==============================================================================
* TABLE 2:
* The Season's interruption and marriage outcomes, probit and OLS estimation.
* ------------------------------------------------------------------------------
* Table 2, columns [1] to [5]
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
/* Note: Baseline sample is 644 peers' daughters aged 15-35 in 1861 who ever
married, excluding second-marriages, women married to foreigners, and
members of the royal family. */
global controls pr4 biorder hengpee
label var distlondon "Distance"
local j = 1
foreach x in "" distlondon{
local panel "A"
if `j'==2{
local panel "B"
}
qui probit cOut syntheticT $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui summ cOut if e(sample)==1
local mdepvar = round(r(mean),0.01)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table2_`panel'", keep(syntheticT distlondon) nocons nor2 ///
label dec(4) coefastr ctitle("M common") ///
addtext(% correct, `pcp1', Mean of Dep Var, `mdepvar', Controls, YES) ///
tex replace
qui probit mheir syntheticT $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui summ mheir if e(sample)==1
local mdepvar = round(r(mean),0.01)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table2_`panel'", keep(syntheticT distlondon) nocons nor2 ///
label dec(4) coefastr ctitle("M heir") ///
addtext(% correct, `pcp1', Mean of Dep Var, `mdepvar', Controls, YES) ///
tex
qui reg fmissmatch syntheticT $controls `x', cluster(byear)
qui summ fmissmatch if e(sample)==1
local mdepvar = round(r(mean),1)
outreg2 using "${dirresu}\Table2_`panel'", keep(syntheticT distlondon) nocons nor2 ///
label dec(3) coefastr ctitle("Diff abs") ///
addtext(% correct, . , Mean of Dep Var, `mdepvar', Controls, YES) ///
tex
qui reg fmissmatch2 syntheticT $controls `x', cluster(byear)
qui summ fmissmatch2 if e(sample)==1
local mdepvar = round(r(mean),1)
outreg2 using "${dirresu}\Table2_`panel'", keep(syntheticT distlondon) nocons nor2 ///
label dec(3) coefastr ctitle("Diff h-w") ///
addtext(% correct, . , Mean of Dep Var, `mdepvar', Controls, YES) ///
tex
qui probit fdown syntheticT $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui summ fdown if e(sample)==1
local mdepvar = round(r(mean),0.01)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table2_`panel'", keep(syntheticT distlondon) nocons nor2 ///
label dec(4) coefastr ctitle("M down") ///
addtext(% correct, `pcp1', Mean of Dep Var, `mdepvar', Controls, YES) ///
tex
local j = `j'+1
}
* Table 2, column [6]
use "${dirdata}\final-data.dta", clear
keep if marital_rates_sample==1
label var distlondon "Distance"
/* Note: To examine marital rates, I consider women aged 15-35 in 1861,
whether they married or not. To avoid counting women who died at an early
age as celibate, I exclude those dying before age 35 (see p.20). */
local j = 1
foreach x in "" distlondon{
local panel "A"
if `j'==2{
local panel "B"
}
qui probit celibacy syntheticT $controls `x', cluster(byear)
qui estat classification
local pcp1 = round(r(P_corr),1)
qui summ celibacy if e(sample)==1
local mdepvar = round(r(mean),0.01)
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table2_`panel'", keep(syntheticT distlondon) nocons nor2 ///
label dec(4) coefastr ctitle("never M") ///
addtext(% correct, `pcp1', Mean of Dep Var, `mdepvar', Controls, YES) ///
tex
local j = `j'+1
}
* Small cluster correction (bootstrap-t procedure of Cameron, Gelbach and Miller, 2008)
/* Note: Results stored in log file in the following location:
3 Replication package\results\Table2_smallcluster.log. */
clear all
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}"
log using Table2_smallcluster.log, replace
keep if wom==1 & nthismarriage<2 & age1861>=15 & age1861<=35 // sample
preserve
keep if nthismarr==1 & myear!=. // married
foreach x in "" distlondon{
qui probit cOut syntheticT $controls `x', cluster(byear)
keep if e(sample)==1 // keep sample of non-missing variables
clustse probit cOut syntheticT $controls `x', cluster(byear) method(pairs) reps(5000)
clustse probit mheir syntheticT $controls `x', cluster(byear) method(pairs) reps(5000)
cgmwildboot fmissmatch syntheticT $controls `x', cluster(byear) bootcluster(byear) reps(5000)
cgmwildboot fmissmatch2 syntheticT $controls `x', cluster(byear) bootcluster(byear) reps(5000)
clustse probit fdown syntheticT $controls `x', cluster(byear) method(pairs) reps(5000)
}
restore
keep if dage>35 & dyear>byear+1 & prank>1 // married and unmarried women
foreach x in "" distlondon{
qui probit celibacy syntheticT $controls `x', cluster(byear)
keep if e(sample)==1 // keep sample of non-missing variables
clustse probit celibacy syntheticT $controls `x', cluster(byear) method(pairs) reps(5000)
}
log close
* ==============================================================================
* FIGURE 6:
* Placebo tests. Dependent variable: married a commoner.
* ------------------------------------------------------------------------------
* Import placebo dataset
use "${dirdata}\placebo-data.dta", clear
graph drop _all
* Estimates for baseline sample
global controls pr4 biorder hengpee
qui probit cOut syntheticT $controls distlondon if nthismarr==1 & base_sample==1, cluster(byearclst)
qui margins, dydx(*) post
estimates store baseEST
* Placebo estimates
sort refno sample01 // drop observations in both baseline and placebo samples
by refno: gen dup=_N
by refno: drop if sample0[1]!= sample0[2] & dup>1
keep if nthismarr==1
drop syntheticT // just for label purposes
gen syntheticT = synthetic_plac
forvalues i=1811/1851{
qui probit cOut syntheticT $controls distlondon if sample01==`i' & sample0=="placebo", cluster(byearclst)
qui margins, dydx(*) post
estimates store pE`i'
}
* Figure
graph drop _all
coefplot (baseEST, msymbol(circle)) (baseEST, mcolor(white) noci) (pE1851, pstyle(p2)) (pE1850, pstyle(p2)) ///
(pE1849, pstyle(p2)) (pE1848, pstyle(p2)) (pE1847, pstyle(p2)) ///
(pE1846, pstyle(p2)) (pE1845, pstyle(p2)) (pE1844, pstyle(p2)) ///
(pE1843, pstyle(p2)) (pE1842, pstyle(p2)) (pE1841, pstyle(p2)) ///
(pE1840, pstyle(p2)) (pE1839, pstyle(p2)) (pE1838, pstyle(p2)) ///
(pE1837, pstyle(p2)) (pE1836, pstyle(p2)) (pE1835, pstyle(p2)) ///
(pE1834, pstyle(p2)) (pE1833, pstyle(p2)) (pE1832, pstyle(p2)) ///
(pE1831, pstyle(p2)) (pE1830, pstyle(p2)) (pE1829, pstyle(p2)) ///
(pE1828, pstyle(p2)) (pE1827, pstyle(p2)) (pE1826, pstyle(p2)) ///
(pE1825, pstyle(p2)) (pE1824, pstyle(p2)) (pE1823, pstyle(p2)) ///
(pE1822, pstyle(p2)) (pE1821, pstyle(p2)) (pE1820, pstyle(p2)) ///
(pE1819, pstyle(p2)) (pE1818, pstyle(p2)) (pE1817, pstyle(p2)) ///
(pE1816, pstyle(p2)) (pE1815, pstyle(p2)) (pE1814, pstyle(p2)) ///
(pE1813, pstyle(p2)) (pE1812, pstyle(p2)) (pE1811, pstyle(p2)), ///
drop(biorder hengpee pr4 pr_3 ̇_cons distlondon) ///
msymbol(d) mfcolor(white) levels(95 90) graphregion(color(white)) ///
title("Dep. Var: Married a commoner", color(black)) ytitle("mg. effect of T (synthetic prob.)") xtitle("years of placebo interruption") ///
ylabel(-0.015(0.005)0.015, nogrid) legend(on order(3 "Baseline (1861-63)" 9 "Placebo") rows(2) ring(0) position(5) bmargin(large)) ///
xlabel(0.568 "1851-53" 0.6816 "1846-48" 0.7955 "1841-43" 0.909 "1836-38" 1.023 "1831-33" 1.1366 "1826-28" 1.24939 "1821-23" 1.36392 "1816-18" 1.4775 "1811-13", angle(45)) ///
yline(0) lpattern(dash) baselevels vert ///
xline(0.568, lpattern(dot) lcolor(gray)) ///
xline(0.6816, lpattern(dot) lcolor(gray)) ///
xline(0.7955, lpattern(dot) lcolor(gray)) ///
xline(0.909, lpattern(dot) lcolor(gray)) ///
xline(1.023, lpattern(dot) lcolor(gray)) ///
xline(1.1366, lpattern(dot) lcolor(gray)) ///
xline(1.24939, lpattern(dot) lcolor(gray)) ///
xline(1.36392, lpattern(dot) lcolor(gray)) ///
xline(1.4775, lpattern(dot) lcolor(gray)) ///
text(0.00539529 0.57037 ".037", size(vsmall) orientation(vertical)) ///
text(0.00539529 0.59037 ".06", size(vsmall) orientation(vertical)) ///
text(0.00250475 0.613635 ".01", size(vsmall) orientation(vertical)) ///
text(0.00276912 0.636005 ".01", size(vsmall) orientation(vertical)) ///
text(0.0012222 0.658375 ".00", size(vsmall) orientation(vertical)) ///
text(0.00079789 0.6816 ".00", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.00076157 0.70397 ".00", size(vsmall) orientation(vertical)) ///
text(0.00128846 0.727235 ".00", size(vsmall) orientation(vertical)) ///
text(0.00131826 0.749605 ".00", size(vsmall) orientation(vertical)) ///
text(0.00379514 0.771975 ".02", size(vsmall) orientation(vertical)) ///
text(0.00402036 0.7955 ".02", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.00382393 0.81761 ".01", size(vsmall) orientation(vertical)) ///
text(0.00283652 0.840875 ".01", size(vsmall) orientation(vertical)) ///
text(0.00210621 0.86414 ".00", size(vsmall) orientation(vertical)) ///
text(0.00346859 0.88651 ".01", size(vsmall) orientation(vertical)) ///
text(0.00176394 0.909 ".00", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.0017114 0.932145 ".00", size(vsmall) orientation(vertical)) ///
text(0.00205746 0.954515 ".01", size(vsmall) orientation(vertical)) ///
text(0.00234219 0.97778 ".01", size(vsmall) orientation(vertical)) ///
text(0.00537774 1.00015 ".12", size(vsmall) orientation(vertical)) ///
text(0.0057546 1.023 ".14", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.00787327 1.04579 ".41", size(vsmall) orientation(vertical)) ///
text(0.00696466 1.06816 ".23", size(vsmall) orientation(vertical)) ///
text(0.00774832 1.09142 ".25", size(vsmall) orientation(vertical)) ///
text(0.00742015 1.11379 ".20", size(vsmall) orientation(vertical)) ///
text(0.008436 1.1366 ".31", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.00712933 1.15943 ".22", size(vsmall) orientation(vertical)) ///
text(0.00506362 1.1818 ".08", size(vsmall) orientation(vertical)) ///
text(0.00545438 1.20506 ".10", size(vsmall) orientation(vertical)) ///
text(0.00608598 1.22743 ".12", size(vsmall) orientation(vertical)) ///
text(0.00530855 1.24939 ".13", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.00537472 1.27307 ".17", size(vsmall) orientation(vertical)) ///
text(0.00442368 1.29454 ".08", size(vsmall) orientation(vertical)) ///
text(0.0031978 1.3187 ".03", size(vsmall) orientation(vertical)) ///
text(0.00497223 1.34107 ".09", size(vsmall) orientation(vertical)) ///
text(0.00586515 1.36392 ".16", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white)) ///
text(0.00497563 1.38671 ".11", size(vsmall) orientation(vertical)) ///
text(0.00487153 1.40908 ".10", size(vsmall) orientation(vertical)) ///
text(0.00658739 1.43145 ".21", size(vsmall) orientation(vertical)) ///
text(0.00614651 1.45471 ".17", size(vsmall) orientation(vertical)) ///
text(0.00672715 1.4775 ".19", size(vsmall) orientation(vertical) box bcolor(white) fcolor(white) lcolor(white))
graph export "${dirresu}\Figure6.pdf", replace
* Tests equality of coefficients
/* Note: the following tests for the equality of coefficients report all the
p-values displayed in the Figure above */
use "${dirdata}\placebo-data.dta", clear
keep if nthismarr==1
gen syntheticT0 = syntheticT // just for label purposes
replace syntheticT0 = synthetic_plac if synthetic_plac!=.
qui probit cOut syntheticT0 $controls distlondon if sample0=="baseline"
estimates store myreg0
sort refno sample01 // drop observations in both baseline and placebo samples
by refno: gen dup=_N
by refno: drop if sample0[1]!= sample0[2] & dup>1
*local count10 = 0
*local count5 = 0
*local count1 = 0
forvalues i=1851(-1)1811{
qui probit cOut syntheticT0 $controls distlondon if sample01==`i' & sample0=="placebo"
estimates store myreg`i'
* tests
qui suest myreg0 myreg`i', cluster(byearclst)
display "---- `i' ----"
test [myreg0_cOut]syntheticT0=[myreg`i'_cOut]syntheticT0
local pval = round(`r(p)',0.001)
di `pval'
}
* ==============================================================================
* FIGURE 7:
* Peer-commoner intermarriage, Probit estimation with age dummies.
* ------------------------------------------------------------------------------
* Import placebo dataset
use "${dirdata}\placebo-data.dta", clear
global controls pr4 biorder hengpee
* Panel A: baseline sample
fvset base 16 age1861
qui probit cOut i.age1861 $controls if base_sample==1, cluster(byearclst)
qui margins, dydx(*) predict(pr) post
estimates store BASE
fvset base 0 age1861
qui reg syntheticT i.age1861 if base_sample==1, nocons
estimates store BASEsynth
coefplot (BASE, label("cohort-level effect (left axis)") ylabel(-0.2(0.1)0.5)) ///
(BASEsynth, label("synthetic prob. of marrying at interruption (right axis)") axis(2) ylabel(5(5)25, axis(2)) offset(-0.05) recast(line) symbol(none)), ///
drop(biorder hengpee pr4 ̇_cons 0.age1861 15.age1861 27.age1861 28.age1861 29.age1861 30.age1861 31.age1861 32.age1861 33.age1861 34.age1861 35.age1861) ///
msymbol(d) mfcolor(white) levels(99 95 90) graphregion(color(white)) ///
title("Baseline", color(black)) xtitle(age in 1861) ytitle("prob. of marrying a commoner") ///
legend(on rows(2) region(lstyle(none))) ///
xlabel(1 "16" 2 "17" 3 "18" 4 "19" 5 "20" 6 "21" 7 "22" 8 "23" 9 "24" 10 "25" 11 "26") ///
yline(0) ylabel(-0.2(0.1)0.6) lpattern(dash) name(base, replace) vert baselevels
* Panel B: placebo
sort refno sample01 // drop observations in both baseline and placebo samples
by refno: gen dup=_N
by refno: drop if sample0[1]!= sample0[2] & dup>1
keep if nthismarr==1
drop age1861 // just for labelling purposes
rename age1861_plac age1861
fvset base 16 age1861
qui probit cOut i.age1861 $controls i.sample01 if sample0=="placebo", cluster(byearclst)
qui margins, dydx(*) predict(pr) post
estimates store PLAC
coefplot (PLAC, label("cohort-level effect (left axis)") ylabel(-0.2(0.1)0.5)) ///
(BASEsynth, label("synthetic prob. of marrying at interruption (right axis)") axis(2) ylabel(5(5)25, axis(2)) offset(-0.05) recast(line) symbol(none)), ///
keep(16.age1861 17.age1861 18.age1861 19.age1861 20.age1861 21.age1861 22.age1861 23.age1861 24.age1861 25.age1861 26.age1861) ///
msymbol(d) mfcolor(white) levels(99 95 90) graphregion(color(white)) ///
title("Placebo", color(black)) xtitle(age in 1861-X) graphregion(color(white)) ///
legend(on rows(2) region(lstyle(none))) ///
xlabel(1 "16" 2 "17" 3 "18" 4 "19" 5 "20" 6 "21" 7 "22" 8 "23" 9 "24" 10 "25" 11 "26") ///
yline(0) ylabel(-0.2(0.1)0.6) lpattern(dash) name(placebo, replace) vert baselevels
* Figure:
grc1leg base placebo, legendfrom(base) graphregion(color(white))
graph export "${dirresu}\Figure7.pdf", replace
graph drop _all
erase "${dirdata}\placebo-data.dta"
* ==============================================================================
* TABLE 3:
* The Season's interruption and marriage outcomes, IV estimation.
* ------------------------------------------------------------------------------
* Import data and keep baseline sample
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
* IV regressions
global controls pr4 biorder hengpee
* [1]
qui summ cOut
local mdepvar = round(`r(mean)',0.01)
qui reg mourn syntheticT $controls, cluster(byear)
local F1 = round(e(F),0.1)
qui cmp (cOut = mourn $controls) (mourn = syntheticT $controls), ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table3_A", keep(mourn) nocons label dec(2) coefastr ctitle("M common") addtext(Weak iv p-val, ., Mean DV, `mdepvar') tex replace
qui cmp (cOut = mourn $controls) (mourn = syntheticT $controls), ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Table3_B", keep(syntheticT) nocons label dec(2) coefastr ctitle("M interr") addtext(F-stat, `F1') tex replace
* [2]
qui summ cOut if distlondon!=.
local mdepvar = round(`r(mean)',0.01)
qui reg mourn syntheticT $controls distlondon, cluster(byear)
local F1 = round(e(F),0.1)
qui cmp (cOut = mourn $controls distlondon) (mourn = syntheticT $controls distlondon), ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table3_A", keep(mourn) nocons label dec(2) coefastr ctitle("M common") addtext(Weak iv p-val, ., Mean DV, `mdepvar') tex
qui cmp (cOut = mourn $controls distlondon) (mourn = syntheticT $controls distlondon), ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Table3_B", keep(syntheticT) nocons label dec(2) coefastr ctitle("M interr") addtext(F-stat, `F1') tex
* [3] and [4]
foreach x in "" distlondon{
qui reg mheir syntheticT $controls `x', cluster(byear)
qui summ mheir if e(sample)==1
local mdepvar = round(`r(mean)',0.01)
qui reg mourn syntheticT $controls `x', cluster(byear)
local F1 = round(e(F),0.1)
qui cmp (mheir = mourn $controls `x') (mourn = syntheticT $controls `x'), ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table3_A", keep(mourn) nocons label dec(2) coefastr ctitle("M heir") addtext(Weak iv p-val, ., Mean DV, `mdepvar') tex
qui cmp (mheir = mourn $controls `x') (mourn = syntheticT $controls `x'), ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Table3_B", keep(syntheticT) nocons label dec(2) coefastr ctitle("M interr") addtext(F-stat, `F1') tex
}
* [5] and [6]
foreach x in "" distlondon{
qui reg fmissmatch syntheticT $controls `x', cluster(byear)
qui summ fmissmatch if e(sample)==1
local mdepvar = round(`r(mean)',1)
qui ivregress liml fmissmatch (mourn = syntheticT) $controls `x', cluster(byear) first
qui weakiv, null(0) small
local wivpv = round(e(ar_p), 0.01)
qui reg mourn syntheticT $controls `x' if fmissmatch!=., cluster(byear)
local F1 = round(e(F),0.1)
qui ivregress liml fmissmatch (mourn = syntheticT) $controls `x', cluster(byear)
outreg2 using "${dirresu}\Table3_A", keep(mourn) nocons label dec(1) coefastr ctitle("Diff abs") addtext(Weak iv p-val, `wivpv', Mean DV, `mdepvar') tex
qui cmp (fmissmatch = mourn $controls `x') (mourn = syntheticT $controls `x') if fmissmatch!=., ind(1 1) cluster(byear) quietly
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Table3_B", keep(syntheticT) nocons label dec(2) coefastr ctitle("M interr") addtext(F-stat, `F1') tex
}
* [7] and [8]
foreach x in "" distlondon{
qui reg fmissmatch2 syntheticT $controls `x', cluster(byear)
qui summ fmissmatch2 if e(sample)==1
local mdepvar = round(`r(mean)',1)
qui ivregress liml fmissmatch2 (mourn = syntheticT) $controls `x', cluster(byear) first
qui weakiv, null(0) small
local wivpv = round(e(ar_p), 0.01)
qui reg mourn syntheticT $controls `x' if fmissmatch2!=., cluster(byear)
local F1 = round(e(F),0.1)
qui ivregress liml fmissmatch2 (mourn = syntheticT) $controls `x', cluster(byear)
qui margins, dydx(*) post
outreg2 using "${dirresu}\Table3_A", keep(mourn) nocons label dec(1) coefastr ctitle("Diff h-w") addtext(Weak iv p-val, `wivpv', Mean DV, `mdepvar') tex
qui cmp (fmissmatch2 = mourn $controls `x') (mourn = syntheticT $controls `x') if fmissmatch2!=., ind(1 1) cluster(byear) quietly
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Table3_B", keep(syntheticT) nocons label dec(2) coefastr ctitle("M interr") addtext(F-stat, `F1') tex
}
* [9] and [10]
foreach x in "" distlondon{
qui reg fdown syntheticT $controls `x', cluster(byear)
qui summ fdown if e(sample)==1
local mdepvar = round(`r(mean)',0.01)
qui ivregress liml fdown (mourn = syntheticT) $controls `x', cluster(byear)
qui weakiv, null(0) small
local wivpv = round(e(ar_p), 0.01)
qui reg mourn syntheticT $controls `x' if fdown!=., cluster(byear)
local F1 = round(e(F),0.1)
qui cmp (fdown = mourn $controls `x') (mourn = syntheticT $controls `x') if fdown!=., ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(pr) post
outreg2 using "${dirresu}\Table3_A", keep(mourn) nocons label dec(2) coefastr ctitle("M down") addtext(Weak iv p-val, `wivpv', Mean DV, `mdepvar') tex
qui cmp (fdown = mourn $controls `x') (mourn = syntheticT $controls `x') if fdown!=., ind(4 1) cluster(byear) quietly
qui margins, dydx(*) predict(eq(#2)) post
outreg2 using "${dirresu}\Table3_B", keep(syntheticT) nocons label dec(2) coefastr ctitle("M interr") addtext(F-stat, `F1') tex
}
* ==============================================================================
* TABLE 4:
* Contingency tables.
* ------------------------------------------------------------------------------
* Import data and keep baseline sample
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
* Treatment indicator for non-parametric analysis
_pctile syntheticT, nq(100)
gen treatment = (syntheticT > r(r80))
/* Note: For the non-parametric estimation, I compare a high- vs. a low-treatment
cohort. The high-treatment cohort are women with a synthetic probability to
marry during the interruption above 20 percent. This corresponds to the top
quintile; the 20% of women with the highest synthetic probability to marry in
1861-63. Conversely, the low-treatment cohort are women below the top quintile. */
* Contingency tables
label var grank "Wife"
capture log close // save results in log file
cd "${dirresu}"
log using Table4.log, replace
* Panel A: Low-Treatment cohorts (T < 80th percentile)
tab grank grankspo if treatment==0, expected matcell(ofreqa)
* Panel B: High-Treatment cohorts (T >= 80th percentile)
tab grank grankspo if treatment==1, expected matcell(ofreqa)
log close
* ==============================================================================
* TABLE 5:
* The interruption and sorting by title, non-parametric estimates.
* ------------------------------------------------------------------------------
* Import data and keep baseline sample
use "${dirdata}\final-data.dta", clear
keep if base_sample==1
* Treatment indicator for non-parametric analysis
_pctile syntheticT, nq(100)
gen treatment = (syntheticT > r(r80))
* Non-parametric stats
forvalues i=0/1{
tab grank grankspo if treatment==`i', chi2 lrchi2
local chi2_`i' = r(chi2)
local chi2LR_`i' = r(chi2_lr)
local chi2p_`i' = r(p)
local chi2LRp_`i' = r(p_lr)
local df_`i' = e(df_r)
local N`i' = r(N)
ktau grank grankspo if treatment==`i', stats(taub)
local taub_`i' = r(tau_b)
local taubp_`i' = r(p)
}
* Add results to matrix tab5
mat define tab5 = (.,.,.\.,.,.\.,.,.\.,.,.\.,.,.\.,.,.\.,.,.)
forvalues j =0/1{
local col = `j'+1
mat tab5[1,`col'] = `chi2_`j''
mat tab5[2,`col'] = `chi2p_`j''
mat tab5[3,`col'] = `chi2LR_`j''
mat tab5[4,`col'] = `chi2LRp_`j''
mat tab5[5,`col'] = `taub_`j''
mat tab5[6,`col'] = `taubp_`j''
mat tab5[7,`col'] = `N`j''
}
mat tab5[5,3] = `taub_0'-`taub_1'
mat tab5[7,3] = `N0'+`N1'
* Difference chi-squared and LR-chi2
local chi2_diff = `chi2_0'-`chi2_1'
local chi2_diffp = chi2tail(3,`chi2_diff')
local chi2LR_diff = `chi2LR_0'-`chi2LR_1'
local chi2LR_diffp = chi2tail(3,`chi2LR_diff')
mat tab5[1,3] = `chi2_diff'
mat tab5[2,3] = `chi2_diffp'
mat tab5[3,3] = `chi2LR_diff'
mat tab5[4,3] = `chi2LR_diffp'
* Difference tau-b
qui ktau grank grankspo if treatment==0
local convtaub_0 = sin(3.141592654*`r(tau_b)'*0.5)
qui ktau grank grankspo if treatment==1
local convtaub_1 = sin(3.141592654*`r(tau_b)'*0.5)
cortesti `convtaub_0' `N0' `convtaub_1' `N1'
mat tab5[6,3] = 0.1637
* Table 5
mat colnames tab5 = "Low-treat" "High-treat" "Diff"
mat rownames tab5 = "Pearson chi-squared" " p-val" "Likelihood ratio" " p-val" "Kendall rank correlation" " p-val" "N"
mat list tab5
esttab matrix(tab5, fmt(3 3)) using "${dirresu}\Table5.tex", mlabel("Table 5") replace
* ==============================================================================
* FIGURE 8:
* Sorting by landholdings, non-parametric estimation.
* ------------------------------------------------------------------------------
* 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))
count if fmissmatch!=. // sample (N=324)
count if fmissmatch!=. & treatment==0 // Low-treatment (N=256)
count if fmissmatch!=. & treatment==0 & age1861<22 // Low-treatment, younger (N=64)
count if fmissmatch!=. & treatment==0 & age1861>22 // Low-treatment, older (N=192)
* Kolmogorov-Smirnov tests
gen treat2`x' = treatment // for High vs. younger comparison
replace treat2 = . if treatment==0 & age1861>22
gen treat3 = treatment // for High vs. older comparison
replace treat3 = . if treatment==0 & age1861<22
foreach x in "" 2{
sort fmissmatch`x' age1861
qui ksmirnov fmissmatch`x', by(treatment) exact
local KS1`x' = round(`r(D)',0.01)
local pval1`x' = round(`r(p_exact)',0.001)
ksmirnov fmissmatch`x', by(treat2) exact
local KS2`x' = round(`r(D)',0.01)
local pval2`x' = round(`r(p_exact)',0.001)
ksmirnov fmissmatch`x', by(treat3) exact
local KS3`x' = round(`r(D)',0.01)
local pval3`x' = round(`r(p_exact)',0.001)
}
* Figure, Panel A
preserve
sort fmissmatch age1861
cumul fmissmatch if treatment>0, gen(cumTREAT)
cumul fmissmatch if treatment==0, gen(cumCONTROL)
cumul fmissmatch if treatment==0 & age1861>22, gen(cumCONTROLold)
cumul fmissmatch if treatment==0 & age1861<22, gen(cumCONTROLyoung)
twoway (line cumTREAT fmissmatch, c(J) lwidth(thick) lcolor(red)) ///
(line cumCONTROL fmissmatch, c(J) lwidth(thin) lcolor(blue)) ///
(line cumCONTROLold fmissmatch, c(J) lwidth(thin) lpattern(dash) lcolor(emidblue)) ///
(line cumCONTROLyoung fmissmatch, c(J) lwidth(thin) lpattern(solid) lcolor(emidblue)), ///
xtitle("|wife's percentile rank of acres - husband's percentile rank of acres|", size(med)) ///
ytitle("cumulative", size(med)) ///
legend(on order(1 "High-Treatment (T{&ge}80p)" 2 "Low-Treatment (T<80p)" 3 "older" 4 "younger") rows(4) ring(0) position(5) bmargin(large) region(lstyle(none))) ///
graphregion(color(white)) ///
note("High vs low: KS = `KS1' (`pval1'); High vs younger: KS = `KS2' (`pval2'); High vs older: KS = `KS3' (`pval3')")
graph export "${dirresu}\Figure8_A.pdf", as(pdf) replace
restore
* Figure, Panel B
preserve
sort fmissmatch2 age1861
cumul fmissmatch2 if treatment>0, gen(cumTREAT)
cumul fmissmatch2 if treatment==0, gen(cumCONTROL)
cumul fmissmatch2 if treatment==0 & age1861>22, gen(cumCONTROLold)
cumul fmissmatch2 if treatment==0 & age1861<22, gen(cumCONTROLyoung)
twoway (line cumTREAT fmissmatch2, c(J) lwidth(thick) lcolor(red)) ///
(line cumCONTROL fmissmatch2, c(J) lwidth(thin) lcolor(blue)) ///
(line cumCONTROLold fmissmatch2, c(J) lwidth(thin) lpattern(dash) lcolor(emidblue)) ///
(line cumCONTROLyoung fmissmatch2, c(J) lwidth(thin) lpattern(solid) lcolor(emidblue)) ///
(pcarrowi 0.95 -40 0.95 -60 "Marrying down", color(black)), ///
xtitle("difference in acres' precentile rank (husband-wife)", size(med)) ///
xline(0, lcolor(black) lwidth(thick)) ytitle("cumulative", size(med)) ///
legend(on order(1 "High-Treat (T{&ge}80p.)" 2 "Low-Treat (T<80p.)" 3 "older" 4 "younger") rows(4) ring(0) position(5) bmargin(large) region(lstyle(none))) ///
graphregion(color(white)) ///
note("High vs low: KS = `KS12' (`pval12'); High vs younger: KS = `KS22' (`pval22'); High vs older: KS = `KS32' (`pval32')")
graph export "${dirresu}\Figure8_B.pdf", as(pdf) replace
restore
* ==============================================================================
* TABLE 6:
* Women's marriages to commoners and her family's political power, IV estimation.
* ------------------------------------------------------------------------------
* Import data
use "${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 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"
* 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)
* CLR p-values clustered by family
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)
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}\Table6_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("MP (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', 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}\Table6_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)
* CLR p-values clustered by family
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)
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}\Table6_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("Years (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', 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}\Table6_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)
* CLR p-values clustered by family
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)
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}\Table6_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("local MP (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', 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}\Table6_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)
* CLR p-values clustered by family
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)
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}\Table6_A", keep(cOut) nocons label nose dec(3) noas noobs ///
ctitle("lyears (b)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', 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}\Table6_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* 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)
* CLR p-values clustered by family
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)
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}\Table6_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("MP (h)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', 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}\Table6_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)
* CLR p-values clustered by family
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)
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}\Table6_A", keep(cOut) nocons label nose dec(2) noas noobs ///
ctitle("Years (h)") addtext(CLR-pval, `CLR1', CLR-pval 2, `CLR2', 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}\Table6_B", keep(syntheticT) nocons label noas nor2 dec(3) coefastr ctitle("M common") addtext(F-stat, `F1') tex
* ==============================================================================
* TABLE 7:
* Determinants of investments in state education, IV estimation.
* ------------------------------------------------------------------------------
* 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"
/* Note: Panel B results for p-values (parenthesis) and CLR p-values
adjusted for weak instruments [brackets] are in one column in the paper,
in two columns in the table produced by this code. */
* Col [1]: Reduced-form effect
* CLR p-values
qui ivregress liml rateperpou (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 rateperpou (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 rateperpou (cOut = syntheticT biorder) $hechter $controlsrf , first
outreg2 using "${dirresu}\Table7_A", keep(cOut) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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)
qui ivreg2 rateperpound (cOut = syntheticT biorder) $hechter $controlsrf , first savefirst
est restore _ivreg2_cOut
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) replace tex ctitle("M common") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpound (cOut = syntheticT biorder) $hechter $controlsrf , first cluster(nfam) savefirst
est restore _ivreg2_cOut
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("M common") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)
* Col [2]: Any brother is MP after marriage
global polit d_MPb_aft
global polbefore d_MPf_bef
* CLR p-values
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Table7_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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 $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_d_MPb_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("MP (b)") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_d_MPb_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("MP (b)") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)
* Col [3]: Years brother is MP after marriage
global polit MPb_years_aft
global polbefore MPf_years_bef
* CLR p-values
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Table7_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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 $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_MPb_years_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("Years MP (b)") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_MPb_years_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("Years MP (b)") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)
* Col [4]: Any brother is local MP after marriage
global polit d_MPb_locaft
global polbefore d_MPf_local
* CLR p-values
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Table7_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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 $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_d_MPb_locaft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("local MP (b)") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_d_MPb_locaft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("local MP (b)") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)
* 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 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first
outreg2 using "${dirresu}\Table7_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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 $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first savefirst
est restore _ivreg2_MPb_years_locaft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("local years (b)") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsb $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_MPb_years_locaft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("local years (b)") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)
* 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 rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
outreg2 using "${dirresu}\Table7_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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 $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first savefirst
est restore _ivreg2_d_MPh_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("MP (h)") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_d_MPh_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("MP (h)") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)
* 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 rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
qui weakiv, null(0) small
local CLR1 = round(e(clr_p), 0.001)
* CLR p-values clustered by family
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first vce(cluster nfam)
local Nbobs = `e(N)' // just so that N appears after CLR p-values in table
qui weakiv, null(0) small
local CLR2 = round(e(clr_p), 0.001)
* Second-stage coefficients
qui ivregress liml rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first
outreg2 using "${dirresu}\Table7_A", keep($polit) stats(coef) nocons label noobs dec(2) noas ///
ctitle("tax rate") 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 $polbefore
local F1 = round(e(F),0.01)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first savefirst
est restore _ivreg2_MPh_years_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("Years MP (h)") label dec(3) noaster nocons nor2 stats(coef pval) addtext(F-stat, `F1', cluster, .)
qui ivreg2 rateperpou ($polit = syntheticT biorder) $hechter $controlsh $polbefore , first cluster(nfam) savefirst
est restore _ivreg2_MPh_years_aft
outreg2 using "${dirresu}\Table7_B", keep(syntheticT) tex ctitle("Years MP (h)") label dec(3) noaster nocons nor2 stats(coef pval) bracket addtext(F-stat, `F1', cluster, family)