clear clear matrix set mem 300m set more off cd "C:\otherresearch\SriLanka\SLMS\data\" use sciencerepdata.dta, clear *** Table 1: Impact of Grants on Business Closures/Attrition ** Males * Closed business reg max_closed_business mamount if mmale==1 & wave==1, robust outreg2 mamount using table1d.out, replace sum max_closed if mmale==1 & mamount<1 & e(sample) * Reporting profits reg max_reports_profits mamount if mmale==1 & wave==1, robust outreg2 mamount using table1d.out, append sum max_reports_profits if mmale==1 & mamount<1 & e(sample) ** Females * Closed business reg max_closed_business mamount if mfemale==1 & wave==1, robust outreg2 mamount using table1d.out, append sum max_closed if mfemale==1 & mamount<1 & e(sample) * Reporting profits reg max_reports_profits mamount if mfemale==1 & wave==1, robust outreg2 mamount using table1d.out, append sum max_reports_profits if mfemale==1 & mamount<1 & e(sample) *** Table 2: Do the grants affect who survives? * Baseline characteristics foreach var of varlist agebus realprof realsales K1_noland trade ownhours ownerage ednyears dspan assetindex_mg riskav { ttest `var' if wave==1 & max_reports_profits==1 & mmale==1, by(treat2) ttest `var' if wave==1 & max_reports_profits==1 & mfemale==1, by(treat2) } * F-test of joint orthogonality reg treat2 agebus realprof realsales K1_noland trade ownhours ownerage ednyears dspan assetindex_mg riskav if wave==1 & max_reports_profits==1 & mmale==1 reg treat2 agebus realprof realsales K1_noland trade ownhours ownerage ednyears dspan assetindex_mg riskav if wave==1 & max_reports_profits==1 & mfemale==1 **** Table 3: Impacts on Real Profits Conditional on Survival * Truncate profits at top 1% separately by gender sum realprof if mmale==1, de gen rprof_trunc=realprof replace rprof_trunc=28203.23 if mmale==1 & realprof>28203.23 & realprof~=. sum realprof if mfemale==1, de replace rprof_trunc=19334.37 if mfemale==1 & realprof>19334.37 & realprof~=. gen earlytreat=1 if wave==2 & amount>0 & amount~=. egen etreat=max(earlytreat), by(sheno) replace etreat=0 if etreat==. gen ltreat=etreat==0 & treat2==1 gen timesincetreat=0 replace timesincetreat=1 if wave==2 & etreat==1 replace timesincetreat=4 if wave==3 & etreat==1 for num 4/9: replace timesincetreat=3*(X-2)+1 if wave==X & etreat==1 replace timesincetreat=28 if wave==10 & etreat==1 replace timesincetreat=34 if wave==11 & etreat==1 replace timesincetreat=60 if wave==12 & etreat==1 replace timesincetreat=66 if wave==13 & etreat==1 replace timesincetreat=1 if wave==4 & ltreat==1 for num 5/9: replace timesincetreat=3*(X-4)+1 if wave==X & ltreat==1 replace timesincetreat=22 if wave==10 & ltreat==1 replace timesincetreat=28 if wave==11 & ltreat==1 replace timesincetreat=54 if wave==12 & ltreat==1 replace timesincetreat=60 if wave==13 & ltreat==1 gen year1since=amount if timesincetreat>0 & timesincetreat<=12 gen year2since=amount if timesincetreat>12 & timesincetreat<=24 gen year3since=amount if timesincetreat>24 & timesincetreat<=36 gen year5since=amount if timesincetreat>=54 & timesincetreat<=66 for var year1since-year5since: replace X=0 if X==. * Counting profits as zero if business is closed gen profclosed=rprof_trunc replace profclosed=0 if profclosed==. & closed==1 gen lprof=log(realprof) replace workerwage=0 if workerwage==. gen profwork=profclosed+(4*workerwage) *** Table 3 columns * Males xtreg realprof year1since year2since year3since year5since wave2-wave13 if mmale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, replace test year1since==year2since==year3since==year5since xtreg rprof_trunc year1since year2since year3since year5since wave2-wave13 if mmale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg lprof year1since year2since year3since year5since wave2-wave13 if mmale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg profwork year1since year2since year3since year5since wave2-wave13 if mmale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since *Females xtreg realprof year1since year2since year3since year5since wave2-wave13 if mfemale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg rprof_trunc year1since year2since year3since year5since wave2-wave13 if mfemale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg lprof year1since year2since year3since year5since wave2-wave13 if mfemale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg profwork year1since year2since year3since year5since wave2-wave13 if mfemale==1, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since ** Data for Figure 1: mean real profits by wave for treatment and control * Males table treat2 wave if (mmale==1), c(mean realprof) * Females table treat2 wave if (mfemale==1), c(mean realprof) **** Supporting Online Material Tables and Figures ***** SOM FIGURES * Figure S1: CDFs of baseline profits for survivors among male owners gen control=mamount<1 cumul realprof if (wave==1) & control==1 & mmale==1 & max_closed==0, gen(cum1ma) cumul realprof if (wave==1) & control==0 & mmale==1 & max_closed==0, gen(cum3ma) label var cum1ma "Control Survivors" label var cum3ma "Treatment Survivors" sort realprof twoway line cum1ma cum3ma realprof if (wave==1) & mmale==1 & realprof<15000 * Figure S2: CDFs of baseline profits for survivors among female owners cumul realprof if (wave==1) & control==1 & mfemale==1 & max_closed==0, gen(cum1fa) cumul realprof if (wave==1) & control==0 & mfemale==1 & max_closed==0, gen(cum3fa) label var cum1fa "Control Survivors" label var cum3fa "Treatment Survivors" sort realprof twoway line cum1fa cum3fa realprof if (wave==1) & mfemale==1 & realprof<15000 * Figure S3: Data on Standard Deviations used to construct bysort wave: sum realprof if mmale==1 & treat2==0 bysort wave: sum realprof if mmale==1 & treat2==1 **** SOM TABLES *** Table S1: Pooled Sample Results * Closed business reg max_closed_business mamount if wave==1, robust outreg2 mamount using table1d.out, replace sum max_closed if mamount<1 & e(sample) * Reporting profits reg max_reports_profits mamount if wave==1, robust outreg2 mamount using table1d.out, append sum max_reports_profits if mamount<1 & e(sample) * Impact on Profit and Income Levels xtreg realprof year1since year2since year3since year5since wave2-wave13, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, replace test year1since==year2since==year3since==year5since xtreg rprof_trunc year1since year2since year3since year5since wave2-wave13, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg lprof year1since year2since year3since year5since wave2-wave13, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since xtreg profwork year1since year2since year3since year5since wave2-wave13, cluster(sheno) fe i(sheno) outreg2 year1since year2since year3since year5since using table2.out, append test year1since==year2since==year3since==year5since sum realprof rprof_trunc lprof profwork if mamount<1 *** Table S2: Robust to removing can't locate or refusals gen cantlocateorrefuse=sheno==217|sheno==407|sheno==518|sheno==164|sheno==67|sheno==149|sheno==569|sheno==81|sheno==102 * Males * Table 1 specification Closed business reg max_closed_business mamount if mmale==1 & wave==1, robust outreg2 mamount using table1robust.out, replace sum max_closed if control==1 & e(sample) * excluding those who weren't in June/Dec surveys reg max_closed_business mamount if mmale==1 & wave==1 & cantlocateorrefuse==0, robust outreg2 mamount using table1robust.out, append sum max_closed if control==1 & e(sample) ** Females * Closed business reg max_closed_business mamount if mfemale==1 & wave==1, robust outreg2 mamount using table1robust.out, append sum max_closed if control==1 & e(sample) reg max_closed_business mamount if mfemale==1 & wave==1 & cantlocateorrefuse==0, robust outreg2 mamount using table1robust.out, append sum max_closed if control==1 & e(sample) *** Table S5 *** Is there a differential effect for poor males? gen mamount_assets=assetindex_mg*mamount reg max_closed_business mamount mamount_assets assetindex_mg if mmale==1 & wave==1, robust outreg2 mamount mamount_assets assetindex_mg using robusttab1.out, replace *** Section in text: robustness of profits to selective attrition * Bounding exercise for profits of survivors who can't trace bysort wave: egen profmean=mean(realprof) if mmale==1 gen rprofhigh1=realprof replace rprofhigh1=profmean if realprof==. & cantlocate==1 & mmale==1 & mamount<1 xtreg rprofhigh1 year1since year2since year3since year5since wave2-wave13 if mmale==1, cluster(sheno) fe i(sheno) *** Tables S3 and S4 are pieced together from round by round reports on firms which closed