**** Analysis of Development Impact of RSE in Vanuatu ** Set directory clear matrix clear cd C:\seasonalworker\ReStatReplicationFiles\ *** Open replication data use TongaReplicationData.dta, clear ** Per capita income and consumption gen pcy1=r1pcy1 gen pcy2=r1pcy2 replace pcy1=pcy if wave>=2 replace pcy2=pcy if wave>=2 replace pccons=r1pccons if wave==1 ** make sure migrants currently in New Zealand are not included in household size gen hhsizeNM=hhsize replace hhsizeNM=hhsize-1 if inNZnow==1 replace pcy2=pcy2*(hhsize/hhsizeNM) replace pccons=pccons*(hhsize/hhsizeNM) gen logpcy1=log(pcy1) gen logpcy2=log(pcy2) gen logpccons=log(pccons) *** Poor in terms of income per head below $1 and $2 per day gen poor1=pcy2<350.4 gen poor2=pcy2<700.8 *** Generate Wave dummies and Indicator for RSE household in time period since worker left. gen wave2=wave==2 gen wave3=wave==3 gen wave4=wave==4 gen RSEpost=0 replace RSEpost=1 if r2RSE==1 & wave>=2 replace RSEpost=1 if (r3repeat==1|r3newRSE==1|r3returnee==1) & wave>=3 replace RSEpost=1 if (r4repeat==1|r4newRSE==1|r4returnee==1|r4RSE==1) & wave==4 tab RSEpost wave * note RSEpost is the same as the variable everRSE *** Propensity-Score matching to create Propensity-Scores * Model 1: add last two years of lags of wage income and having a wage worker gen wage2007inc2=wage2007inc^2 gen wage2006inc2=wage2006inc^2 #delimit ; local squares "hhsize nmales1850 nadults nschoolage englitshare_male1850 postf4share_male1850 healthverygoodshare_male1850 drinkalcoholshare_male1850 meanhardlab shareadultsprevNZ numrelativesNZ assetindex pigs chickens cattle tongandwell dwellingwood tongatapu r1pcy1 r1pccons "; #delimit cr foreach x of local squares { gen z`x' = `x'*`x' } #delimit ; pscore RSEworker z* hhsize nmales1850 nadults nschoolage englitshare_male1850 postf4share_male1850 healthverygoodshare_male1850 drinkalcoholshare_male1850 meanhardlab shareadultsprevNZ numrelativesNZ assetindex pigs chickens cattle tongandwell dwellingwood tongatapu r1pcy1 r1pccons wage2007inc wage2006inc wage2007inc2 wage2006inc2 maleworker2007 maleworker2006 if wave==1, pscore(ps3); #delimit cr * Model 2: restrict to sample of applicants gen Applicant=RSEworker==1|RSEapplicant==1 #delimit ; pscore RSEworker z* hhsize nmales1850 nadults nschoolage englitshare_male1850 postf4share_male1850 healthverygoodshare_male1850 drinkalcoholshare_male1850 meanhardlab shareadultsprevNZ numrelativesNZ assetindex pigs chickens cattle tongandwell dwellingwood tongatapu r1pcy1 r1pccons wage2007inc wage2006inc wage2007inc2 wage2006inc2 maleworker2007 maleworker2006 if wave==1 & Applicant==1, pscore(ps4); #delimit cr for num 3/4: egen mpsX=max(psX), by(seq_no) gen wagegrowth=wage2007inc-wage2006inc ***** Table 1: Comparison of Baseline Means for Different Groups #delimit cr * First Compare RSE households to All others matrix R=[1,1,1,1,1,1] matrix colnames R="mean nonRSE" "mean RSE" "N nonRSE" "N RSE" "N tot" "p value" #delimit ; quietly foreach var in hhsize nmales1850 englitshare_male1850 postf4share_male1850 healthverygoodshare_male1850 drinkalcoholshare_male1850 meanhardlab shareadultsprevNZ numrelativesNZ assetindex pigs chickens cattle tongandwell tongatapu r1pcy2 r1pccons poor1 poor2 maleworker2007 wagegrowth {; ttest `var' if wave==1,by(RSEworker); matrix A=[r(mu_1),r(mu_2),r(N_1) ,r(N_2),r(N_1)+r(N_2) ,r(p)]; matrix rownames A=`var'; matrix list A; matrix R=R\A; matrix list R; }; set linesize 200; matrix list R; #delimit cr * Then compare to households matched against for PS-1 households matrix R=[1,1,1,1,1,1] matrix colnames R="mean nonRSE" "mean RSE" "N nonRSE" "N RSE" "N tot" "p value" #delimit ; quietly foreach var in hhsize nmales1850 englitshare_male1850 postf4share_male1850 healthverygoodshare_male1850 drinkalcoholshare_male1850 meanhardlab shareadultsprevNZ numrelativesNZ assetindex pigs chickens cattle tongandwell tongatapu r1pcy2 r1pccons poor1 poor2 maleworker2007 wagegrowth{; ttest `var' if wave==1 & ps3>=0.1 & ps3<=0.9,by(RSEworker); matrix A=[r(mu_1),r(mu_2),r(N_1) ,r(N_2),r(N_1)+r(N_2) ,r(p)]; matrix rownames A=`var'; matrix list A; matrix R=R\A; matrix list R; }; set linesize 200; matrix list R; #delimit cr * Matched on applicants too - PS-2 matrix R=[1,1,1,1,1,1] matrix colnames R="mean nonRSE" "mean RSE" "N nonRSE" "N RSE" "N tot" "p value" #delimit ; quietly foreach var in hhsize nmales1850 englitshare_male1850 postf4share_male1850 healthverygoodshare_male1850 drinkalcoholshare_male1850 meanhardlab shareadultsprevNZ numrelativesNZ assetindex pigs chickens cattle tongandwell tongatapu r1pcy2 r1pccons poor1 poor2 maleworker2007 wagegrowth{; ttest `var' if wave==1 & ps4>=0.1 & ps4<=0.9,by(RSEworker); matrix A=[r(mu_1),r(mu_2),r(N_1) ,r(N_2),r(N_1)+r(N_2) ,r(p)]; matrix rownames A=`var'; matrix list A; matrix R=R\A; matrix list R; }; set linesize 200; matrix list R; #delimit cr ** Median wage growth bysort RSEworker: centile wagegrowth if wave==1 bysort RSEworker: centile wagegrowth if wave==1 & ps3>=0.1 & ps3<=0.9 bysort RSEworker: centile wagegrowth if wave==1 & ps4>=0.1 & ps4<=0.9 *** Re-scale components of income so comparing on 6 month scale for all replace netremittancesall=netremittancesall*0.5 if wave==1|wave==4 replace grossremittancesall=grossremittancesall*0.5 if wave==1|wave==4 replace grossremittancesRSE=grossremittancesall*0.5 if wave==1|wave==4 replace ag_income_hh=ag_income_hh*0.5 if wave==1|wave==4 replace wageinc=wageinc*26 replace grow_value_hh=grow_value_hh*26 replace grow_valuelesspork=grow_valuelesspork*26 replace communityspend=communityspend*6 replace domtransfersin=domtransfersin*0.5 if wave==4 replace domtransfersout=domtransfersout*0.5 if wave==4 gen savingspc=pcy2-pccons *** Trimpoints for var pcy2 logpcy2 pccons logpccons savingspc:egen X_99=pctile(X), p(99) **** TABLE 3: IMPACTS ON INCOME AND CONSUMPTION MEASURES ** 8 Columns are for DD, DD-Matching on All, DD-Matching on Applicants, 3 trimmed, and FE, FE-Matching, FE-Matching on All, and 8 trimmed ** Then a Row will be for each outcome #delimit ; foreach lhs of varlist pcy2 logpcy2 pccons savingspc logpccons hhsizeNM {; * Ever in RSE as measure of exposure; reg `lhs' RSEworker RSEpost wave2-wave4, cluster(seq_no); outreg2 RSEpost using Table2row_`lhs'.out, replace slow(400); reg `lhs' RSEworker RSEpost wave2-wave4 if mps3>=0.1 & mps3<=0.9, cluster(seq_no); outreg2 RSEpost using Table2row_`lhs'.out, append slow(400); reg `lhs' RSEworker RSEpost wave2-wave4 if mps4>=0.1 & mps4<=0.9, cluster(seq_no); outreg2 RSEpost using Table2row_`lhs'.out, append slow(400); reg `lhs' RSEworker RSEpost wave2-wave4 if mps4>=0.1 & mps4<=0.9 & pcy2=0.1 & mps3<=0.9, fe cluster(seq_no) i(seq_no); outreg2 RSEpost using Table2row_`lhs'.out, append slow(400); xtreg `lhs' RSEpost wave2-wave4 if mps4>=0.1 & mps4<=0.9, fe cluster(seq_no) i(seq_no); outreg2 RSEpost using Table2row_`lhs'.out, append slow(400); xtreg `lhs' RSEpost wave2-wave4 if mps4>=0.1 & mps4<=0.9 & pcy20 & changestockRSE~=. bysort wave: tab repeatRSE #delimit cr **** TABLE 4: HOME-IMPROVEMENTS, ASSET OWNERSHIP, ETC. *** Make dwelling improvement at all egen anydwellingimprove=max(dwellingimprove), by(seq_no) egen anyassetpurchase=max(newassetpurchase), by(seq_no) reg laddertoday RSEpost if wave==4 & mps3>=0.1 & mps3<=0.9, robust outreg2 RSEpost using table4.out, replace reg laddertoday RSEpost if wave==4 & mps4>=0.1 & mps4<=0.9, robust outreg2 RSEpost using table4.out, append * robustness - check if results holding controlling for recall of ladder before reg laddertoday RSEpost ladderbefore if wave==4, robust reg laddertoday RSEpost ladderbefore if wave==4 & mps3>=0.1 & mps3<=0.9, robust reg laddertoday RSEpost ladderbefore if wave==4 & mps4>=0.1 & mps4<=0.9, robust reg anydwellingimprove RSEpost if wave==4 & mps3>=0.1 & mps3<=0.9, robust outreg2 RSEpost using table4.out, append reg anydwellingimprove RSEpost if wave==4 & mps4>=0.1 & mps4<=0.9, robust outreg2 RSEpost using table4.out, append reg anyassetpurchase RSEpost if wave==4 & mps3>=0.1 & mps3<=0.9, robust outreg2 RSEpost using table4.out, append reg anyassetpurchase RSEpost if wave==4 & mps4>=0.1 & mps4<=0.9, robust outreg2 RSEpost using table4.out, append for var hhbankaccount pigs chickens cattle s6q1_400-s6q1_423: gen base_X=X if wave==1 for var hhbankaccount pigs chickens cattle s6q1_400-s6q1_423: egen mbase_X=max(base_X), by(seq_no) gen boat=s6q1_419==1|s6q1_418==1 gen base_boat=boat if wave==1 egen mbase_boat=max(base_boat), by(rse_id) #delimit ; foreach lhs of varlist hhbankaccount pigs cattle chickens s6q1_400 s6q1_402 s6q1_403 s6q1_405 s6q1_406 s6q1_409 s6q1_410 boat s6q1_421 s6q1_423 {; reg `lhs' RSEpost mbase_`lhs' if wave==4 & mps3>=0.1 & mps3<=0.9, robust; outreg2 RSEpost using table4.out, append slow(400); reg `lhs' RSEpost mbase_`lhs' if wave==4 & mps4>=0.1 & mps4<=0.9, robust; outreg2 RSEpost using table4.out, append slow(400); }; #delimit cr *** Table 5 - child's education use IndividualDataTonga.dta, clear * New Table 5 reg attend RSEworker mbaseattend baseage if baseage>=6 & baseage<=14 & wave==4, robust outreg2 RSEworker using table5b.out, replace bdec(3) reg attend RSEworker mbaseattend baseage if baseage>=6 & baseage<=14 & wave==4 & mps3>=0.1 & mps3<=0.9, robust outreg2 RSEworker using table5b.out, append bdec(3) reg attend RSEworker mbaseattend baseage if baseage>=6 & baseage<=14 & wave==4 & mps4>=0.1 & mps4<=0.9, robust outreg2 RSEworker using table5b.out, append bdec(3) reg attend RSEworker mbaseattend baseage if baseage>=15 & baseage<=18 & wave==4, robust outreg2 RSEworker using table5b.out, append bdec(3) reg attend RSEworker mbaseattend baseage if baseage>=15 & baseage<=18 & wave==4 & mps3>=0.1 & mps3<=0.9, robust outreg2 RSEworker using table5b.out, append bdec(3) reg attend RSEworker mbaseattend baseage if baseage>=15 & baseage<=18 & wave==4 & mps4>=0.1 & mps4<=0.9, robust outreg2 RSEworker using table5b.out, append bdec(3) sum attend if RSEworker==0 & baseage>=6 & baseage<=14 sum attend if RSEworker==0 & baseage>=6 & baseage<=14 & mps3>=0.1 & mps3<=0.9 sum attend if RSEworker==0 & baseage>=6 & baseage<=14 & mps4>=0.1 & mps4<=0.9 sum attend if RSEworker==0 & baseage>=15 & baseage<=18 sum attend if RSEworker==0 & baseage>=15 & baseage<=18 & mps3>=0.1 & mps3<=0.9 sum attend if RSEworker==0 & baseage>=15 & baseage<=18 & mps4>=0.1 & mps4<=0.9