* Michler et al., 2018 * THIS FILE PRODUCES THE REGRESSION RESULTS IN THE PAPER AND APPENDICES use mm.dta, clear /*========================================================================= PREPPING DATA ===========================================================================*/ ** Generate log transformed variables foreach var of varlist landowned mainrf { bysort qnno Year: gen log`var' = asinh(`var') } lab var loglandowned "Ln land owned" lab var logmainrf "Ln main season rainfall" ** Generate dependent variables gen lny = asinh(cpprodctn/cpland) lab var lny "Ln output (kg/ha)" gen lncv = asinh(cpproductionvalue/cpland) lab var lncv "Ln value of home chickpea consumption (USD/ha)" gen lnc = asinh(totalcost_sum/cultarea) lab var lnc "Ln total cost (USD/ha)" gen lnr = asinh(salesinc_sum/cultarea) lab var lnr "Ln revenue from crop sales (USD/ha)" gen lnp = asinh((salesinc_sum- totalcost_sum)/cultarea) lab var lnp "Ln crop sales profit (USD/ha)" ** Generate independent production variables foreach var of varlist cpseed cpchemfertqt cptotchemcost cplabour cptotallabourcost cpsalescost { bysort qnno Year: gen lny_`var' = asinh(`var'/cpland) } lab var lny_cpseed "Ln seed per ha" lab var lny_cpchemfertqt "Ln fertilizer per ha" lab var lny_cptotchemcost "Ln chemical cost per ha" lab var lny_cpsalescost "Ln transport cost per ha" lab var lny_cptotallabourcost "Ln hired labor cost per ha" lab var lny_cplabour "Ln family labor days per ha" ** Generate independent cost/revenue variables foreach var of varlist seedcost totfertcost totchemcost totallabour_sum totallabourcost_sum salescost { bysort qnno Year: gen lnc_`var' = asinh(`var'/cultarea) } lab var lnc_seedcost "Ln seed per ha" lab var lnc_totfertcost "Ln fertilizer per ha" lab var lnc_totchemcost "Ln chemical cost per ha" lab var lnc_salescost "Ln transport cost per ha" lab var lnc_totallabourcost_sum "Ln hired labor cost per ha" lab var lnc_totallabour_sum "Ln family labor days per ha" /*========================================================================= GENERATE DATA SETS ===========================================================================*/ save "mm_long.dta", replace reshape wide Head_gender- lnc_salescost, i(qnno) j(Year) save "mm_wide.dta", replace /*========================================================================= OLS & FE ===========================================================================*/ use mm_long.dta, clear ******************** *** 1 Production *** ******************** *** OLS, Pooled, no covariates, district FE *** xi: reg lny icp i.Year i.District *** OLS, Pooled, covariates, district FE *** local b1 lny_cpseed lny_cpchemfertqt lny_cptotchemcost lny_cplabour lny_cptotallabourcost lny_cpsalescost local b2 Head_gender dependencyperc offincsource logmainrf shock loglandowned xi: reg lny icp `b1' `b2' i.Year i.District xtset qnno Year *** FE, no covariates *** xi: xtreg lny icp i.Year, fe *** FE, covariates *** local b1 lny_cpseed lny_cpchemfertqt lny_cptotchemcost lny_cplabour lny_cptotallabourcost lny_cpsalescost local b2 Head_gender dependencyperc offincsource logmainrf shock loglandowned xi: xtreg lny icp `b1' `b2' i.Year, fe ************** *** 2 Cost *** ************** *drop if lny == . *** OLS, Pooled, no covariates, district FE *** xi: reg lnc icp i.Year i.District *** OLS, Pooled, covariates, district FE *** local b1 lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost local b2 Head_gender dependencyperc offincsource logmainrf shock loglandowned xi: reg lnc icp `b1' `b2' i.Year i.District xtset qnno Year *** FE, no covariates *** xi: xtreg lnc icp i.Year, fe *** FE, covariates *** local b1 lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost local b2 Head_gender dependencyperc offincsource logmainrf shock loglandowned xi: xtreg lnc icp `b1' `b2' i.Year, fe **************** *** 4 Profit *** **************** *** OLS, Pooled, no covariates, district FE *** xi: reg lnp icp i.Year i.District *** OLS, Pooled, covariates, district FE *** local b1 lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost local b2 Head_gender dependencyperc offincsource logmainrf shock loglandowned xi: reg lnp icp `b1' `b2' `b3' i.Year i.District *** FE, no covariates *** xi: xtreg lnp icp i.Year, fe *** FE, covariates *** local b1 lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost local b2 Head_gender dependencyperc offincsource logmainrf shock loglandowned xi: xtreg lnp icp `b1' `b2' i.Year, fe /*========================================================================= CRE ===========================================================================*/ use mm_wide.dta, clear ******************** *** 1 Production *** ******************** ***CRE, without covariates randcoef lny2008 lny2010 lny2014, choice(icp2008 icp2010 icp2014) method(CRE) showreg ***CRE, with covariates local maincovars1 lny_cpseed2008- lny_cpsalescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lny_cpseed2010- lny_cpsalescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lny_cpseed2014- lny_cpsalescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lny2008 lny2010 lny2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRE) showreg ************** *** 2 Cost *** ************** ***CRE, without covariates randcoef lnc2008 lnc2010 lnc2014, choice(icp2008 icp2010 icp2014) method(CRE) showreg ***CRE, with covariates local maincovars1 lnc_seedcost2008- lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010- lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014- lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnc2008 lnc2010 lnc2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRE) showreg **************** *** 4 Profit *** **************** ***CRE, without covariates randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) method(CRE) showreg ***CRE, with covariates local maincovars1 lnc_seedcost2008- lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010- lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014- lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRE) showreg /*========================================================================= THREE YEAR CRC ===========================================================================*/ ******************** *** 1 Production *** ******************** use mm_wide.dta, clear ***CRC, without covariates randcoef lny2008 lny2010 lny2014, choice(icp2008 icp2010 icp2014) meth(CRC) ***CRC, with covariates local maincovars1 lny_cpseed2008- lny_cpsalescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lny_cpseed2010- lny_cpsalescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lny_cpseed2014- lny_cpsalescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lny2008 lny2010 lny2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') meth(CRC) keep showreg *Recover theta hat egen h1_bar = mean(icp2008) egen h2_bar = mean(icp2010) egen h3_bar = mean(icp2014) egen h12_bar = mean(int_4) egen h13_bar = mean(int_5) egen h23_bar = mean(int_6) egen h123_bar = mean(int_7) gen l0 = -_b[l1]*h1_bar - _b[l2]*h2_bar - _b[l3]*h3_bar - _b[l4]*h12_bar - _b[l5]*h13_bar - _b[l6]*h23_bar - _b[l7]*h123_bar gen theta = l0 + _b[l1]*icp2008 + _b[l2]*icp2010 + _b[l3]*icp2014 + _b[l4]*int_4 + _b[l5]*int_5 + _b[l6]*int_6 + _b[l7]*int_7 lab var theta "comparative advantage" gen theta1 = l0 + _b[l1]*0 + _b[l2]*0 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta1 "never adopt" gen theta2 = l0 + _b[l1]*1 + _b[l2]*1 + _b[l3]*1 + _b[l4]*1 + _b[l5]*1 + _b[l6]*1 + _b[l7]*1 lab var theta2 "always adopt" gen theta3 = l0 + _b[l1]*0 + _b[l2]*1 + _b[l3]*1 + _b[l4]*0 + _b[l5]*0 + _b[l6]*1 + _b[l7]*0 lab var theta3 "early adopters" gen theta4 = l0 + _b[l1]*0 + _b[l2]*0 + _b[l3]*1 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta4 "late adopters" gen theta5 = l0 + _b[l1]*1 + _b[l2]*0 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta5 "early dis-adopters" gen theta6 = l0 + _b[l1]*1 + _b[l2]*1 + _b[l3]*0 + _b[l4]*1 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta6 "late dis-adopters" gen theta7 = l0 + _b[l1]*1 + _b[l2]*0 + _b[l3]*1 + _b[l4]*0 + _b[l5]*1 + _b[l6]*0 + _b[l7]*0 lab var theta7 "mixed adopters" gen theta8 = l0 + _b[l1]*0 + _b[l2]*1 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta8 "mixed dis-adopters" *graph bar (mean) theta1 (mean) theta2 (mean) theta3 (mean) theta4 (mean) theta5 /// * (mean) theta6 (mean) theta7 (mean) theta8 gen r1 = _b[b] + _b[phi]*theta1 lab var r1 "returns for never adopters" gen r2 = _b[b] + _b[phi]*theta2 lab var r2 "returns for always adopters" gen r3 = _b[b] + _b[phi]*theta3 lab var r3 "returns for early adopters" gen r4 = _b[b] + _b[phi]*theta4 lab var r4 "returns for late adopters" gen r5 = _b[b] + _b[phi]*theta5 lab var r5 "returns for early dis-adopters" gen r6 = _b[b] + _b[phi]*theta6 lab var r6 "returns for late dis-adopters" gen r7 = _b[b] + _b[phi]*theta7 lab var r7 "returns for mixed adopters" gen r8 = _b[b] + _b[phi]*theta8 lab var r8 "returns for mixed dis-adopters" graph bar (mean) r2 (mean) r3 (mean) r4 (mean) r7 (mean) r8 (mean) r6 (mean) r5 (mean) r1, /// bar(1, color(navy)) bar(2, color(navy*0.75)) bar(3, color(navy*0.5)) bar(4, color(navy*0.25)) /// bar(5, color(maroon*.25)) bar(6, color(maroon*0.5)) bar(7, color(maroon*0.75)) bar(8, color(maroon)) /// legend(on order(1 3 5 7 2 4 6 8) col(4) lab(1 "Always adopter") lab(2 "Early adopter") lab(3 "Late adopter") /// lab(4 "Mixed adopter") lab(5 "Mixed dis-adopter") lab(6 "Late dis-adopter") lab(7 "Early dis-adopter") /// lab(8 "Never adopter") pos(6)) ************** *** 2 Cost *** ************** use mm_wide.dta, clear ***CRC, without covariates randcoef lnc2008 lnc2010 lnc2014, choice(icp2008 icp2010 icp2014) method(CRC) ***CRC, with covariates local maincovars1 lnc_seedcost2008- lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010- lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014- lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnc2008 lnc2010 lnc2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRC) keep showreg *Recover theta hat egen h1_bar = mean(icp2008) egen h2_bar = mean(icp2010) egen h3_bar = mean(icp2014) egen h12_bar = mean(int_4) egen h13_bar = mean(int_5) egen h23_bar = mean(int_6) egen h123_bar = mean(int_7) gen l0 = -_b[l1]*h1_bar - _b[l2]*h2_bar - _b[l3]*h3_bar - _b[l4]*h12_bar - _b[l5]*h13_bar - _b[l6]*h23_bar - _b[l7]*h123_bar gen theta = l0 + _b[l1]*icp2008 + _b[l2]*icp2010 + _b[l3]*icp2014 + _b[l4]*int_4 + _b[l5]*int_5 + _b[l6]*int_6 + _b[l7]*int_7 lab var theta "comparative advantage" gen theta1 = l0 + _b[l1]*0 + _b[l2]*0 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta1 "never adopt" gen theta2 = l0 + _b[l1]*1 + _b[l2]*1 + _b[l3]*1 + _b[l4]*1 + _b[l5]*1 + _b[l6]*1 + _b[l7]*1 lab var theta2 "always adopt" gen theta3 = l0 + _b[l1]*0 + _b[l2]*1 + _b[l3]*1 + _b[l4]*0 + _b[l5]*0 + _b[l6]*1 + _b[l7]*0 lab var theta3 "early adopters" gen theta4 = l0 + _b[l1]*0 + _b[l2]*0 + _b[l3]*1 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta4 "late adopters" gen theta5 = l0 + _b[l1]*1 + _b[l2]*0 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta5 "early dis-adopters" gen theta6 = l0 + _b[l1]*1 + _b[l2]*1 + _b[l3]*0 + _b[l4]*1 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta6 "late dis-adopters" gen theta7 = l0 + _b[l1]*1 + _b[l2]*0 + _b[l3]*1 + _b[l4]*0 + _b[l5]*1 + _b[l6]*0 + _b[l7]*0 lab var theta7 "mixed adopters" gen theta8 = l0 + _b[l1]*0 + _b[l2]*1 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta8 "mixed dis-adopters" *graph bar (mean) theta1 (mean) theta2 (mean) theta3 (mean) theta4 (mean) theta5 /// * (mean) theta6 (mean) theta7 (mean) theta8 gen r1 = _b[b] + _b[phi]*theta1 lab var r1 "returns for never adopters" gen r2 = _b[b] + _b[phi]*theta2 lab var r2 "returns for always adopters" gen r3 = _b[b] + _b[phi]*theta3 lab var r3 "returns for early adopters" gen r4 = _b[b] + _b[phi]*theta4 lab var r4 "returns for late adopters" gen r5 = _b[b] + _b[phi]*theta5 lab var r5 "returns for early dis-adopters" gen r6 = _b[b] + _b[phi]*theta6 lab var r6 "returns for late dis-adopters" gen r7 = _b[b] + _b[phi]*theta7 lab var r7 "returns for mixed adopters" gen r8 = _b[b] + _b[phi]*theta8 lab var r8 "returns for mixed dis-adopters" graph bar (mean) r2 (mean) r3 (mean) r4 (mean) r7 (mean) r8 (mean) r6 (mean) r5 (mean) r1, /// bar(1, color(navy)) bar(2, color(navy*0.75)) bar(3, color(navy*0.5)) bar(4, color(navy*0.25)) /// bar(5, color(maroon*.25)) bar(6, color(maroon*0.5)) bar(7, color(maroon*0.75)) bar(8, color(maroon)) /// legend(on order(1 3 5 7 2 4 6 8) col(4) lab(1 "Always adopter") lab(2 "Early adopter") lab(3 "Late adopter") /// lab(4 "Mixed adopter") lab(5 "Mixed dis-adopter") lab(6 "Late dis-adopter") lab(7 "Early dis-adopter") /// lab(8 "Never adopter") pos(6)) **************** *** 4 Profit *** **************** use mm_wide.dta, clear ***CRC, without covariates randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) method(CRC) ***CRC, with covariates local maincovars1 lnc_seedcost2008- lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010- lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014- lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRC) keep showreg *Recover theta hat egen h1_bar = mean(icp2008) egen h2_bar = mean(icp2010) egen h3_bar = mean(icp2014) egen h12_bar = mean(int_4) egen h13_bar = mean(int_5) egen h23_bar = mean(int_6) egen h123_bar = mean(int_7) gen l0 = -_b[l1]*h1_bar - _b[l2]*h2_bar - _b[l3]*h3_bar - _b[l4]*h12_bar - _b[l5]*h13_bar - _b[l6]*h23_bar - _b[l7]*h123_bar gen theta = l0 + _b[l1]*icp2008 + _b[l2]*icp2010 + _b[l3]*icp2014 + _b[l4]*int_4 + _b[l5]*int_5 + _b[l6]*int_6 + _b[l7]*int_7 lab var theta "comparative advantage" gen theta1 = l0 + _b[l1]*0 + _b[l2]*0 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta1 "never adopt" gen theta2 = l0 + _b[l1]*1 + _b[l2]*1 + _b[l3]*1 + _b[l4]*1 + _b[l5]*1 + _b[l6]*1 + _b[l7]*1 lab var theta2 "always adopt" gen theta3 = l0 + _b[l1]*0 + _b[l2]*1 + _b[l3]*1 + _b[l4]*0 + _b[l5]*0 + _b[l6]*1 + _b[l7]*0 lab var theta3 "early adopters" gen theta4 = l0 + _b[l1]*0 + _b[l2]*0 + _b[l3]*1 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta4 "late adopters" gen theta5 = l0 + _b[l1]*1 + _b[l2]*0 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta5 "early dis-adopters" gen theta6 = l0 + _b[l1]*1 + _b[l2]*1 + _b[l3]*0 + _b[l4]*1 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta6 "late dis-adopters" gen theta7 = l0 + _b[l1]*1 + _b[l2]*0 + _b[l3]*1 + _b[l4]*0 + _b[l5]*1 + _b[l6]*0 + _b[l7]*0 lab var theta7 "mixed adopters" gen theta8 = l0 + _b[l1]*0 + _b[l2]*1 + _b[l3]*0 + _b[l4]*0 + _b[l5]*0 + _b[l6]*0 + _b[l7]*0 lab var theta8 "mixed dis-adopters" *graph bar (mean) theta1 (mean) theta2 (mean) theta3 (mean) theta4 (mean) theta5 /// * (mean) theta6 (mean) theta7 (mean) theta8 gen r1 = _b[b] + _b[phi]*theta1 lab var r1 "returns for never adopters" gen r2 = _b[b] + _b[phi]*theta2 lab var r2 "returns for always adopters" gen r3 = _b[b] + _b[phi]*theta3 lab var r3 "returns for early adopters" gen r4 = _b[b] + _b[phi]*theta4 lab var r4 "returns for late adopters" gen r5 = _b[b] + _b[phi]*theta5 lab var r5 "returns for early dis-adopters" gen r6 = _b[b] + _b[phi]*theta6 lab var r6 "returns for late dis-adopters" gen r7 = _b[b] + _b[phi]*theta7 lab var r7 "returns for mixed adopters" gen r8 = _b[b] + _b[phi]*theta8 lab var r8 "returns for mixed dis-adopters" graph bar (mean) r2 (mean) r3 (mean) r4 (mean) r7 (mean) r8 (mean) r6 (mean) r5 (mean) r1, /// bar(1, color(navy)) bar(2, color(navy*0.75)) bar(3, color(navy*0.5)) bar(4, color(navy*0.25)) /// bar(5, color(maroon*.25)) bar(6, color(maroon*0.5)) bar(7, color(maroon*0.75)) bar(8, color(maroon)) /// legend(on order(1 3 5 7 2 4 6 8) col(4) lab(1 "Always adopter") lab(2 "Early adopter") lab(3 "Late adopter") /// lab(4 "Mixed adopter") lab(5 "Mixed dis-adopter") lab(6 "Late dis-adopter") lab(7 "Early dis-adopter") /// lab(8 "Never adopter") pos(6)) /*========================================================================= APPENDIX B: POTENTIAL ENDOGENOUS COVARIATES ===========================================================================*/ use mm_long.dta, clear foreach var of varlist lny_cpseed lny_cpchemfertqt lny_cptotchemcost lny_cplabour lny_cptotallabourcost lny_cpsalescost /// lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost { gen i`var' = icp*`var' } ******************************** **Heterogeneity by observables** ******************************** *** 1 Production *** ***FE, with covariates*** local lcpcovars lny_cpseed lny_cpchemfertqt lny_cptotchemcost lny_cplabour lny_cptotallabourcost lny_cpsalescost local icpcovars icp ilny_cpseed ilny_cpchemfertqt ilny_cptotchemcost ilny_cplabour ilny_cptotallabourcost ilny_cpsalescost local controls Head_gender dependencyperc offincsource logmainrf shock loglandowned xtreg lny `lcpcovars' `icpcovars' `controls' i.Year if icp != ., fe i(qnno) test lny_cpseed = ilny_cpseed test lny_cpchemfertqt = ilny_cpchemfertqt test lny_cptotchemcost = ilny_cptotchemcost test lny_cplabour = ilny_cplabour test lny_cptotallabourcost = ilny_cptotallabourcost test lny_cpsalescost = ilny_cpsalescost *** 2 Cost *** *** FE, covariates*** local lcpcovars lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost local icpcovars icp ilnc_seedcost ilnc_totfertcost ilnc_totchemcost ilnc_totallabour_sum ilnc_totallabourcost_sum ilnc_salescost local controls Head_gender dependencyperc offincsource logmainrf shock loglandowned xtreg lnc `lcpcovars' `icpcovars' `controls' i.Year if icp != ., fe i(qnno) test lnc_seedcost = ilnc_seedcost test lnc_totfertcost = ilnc_totfertcost test lnc_totchemcost = ilnc_totchemcost test lnc_totallabour_sum = ilnc_totallabour_sum test lnc_totallabourcost_sum = ilnc_totallabourcost_sum test lnc_salescost = ilnc_salescost *** 4 Profit *** ***FE, with covariates*** local lcpcovars lnc_seedcost lnc_totfertcost lnc_totchemcost lnc_totallabour_sum lnc_totallabourcost_sum lnc_salescost local icpcovars icp ilnc_seedcost ilnc_totfertcost ilnc_totchemcost ilnc_totallabour_sum ilnc_totallabourcost_sum ilnc_salescost local controls Head_gender dependencyperc offincsource logmainrf shock loglandowned xtreg lnp `lcpcovars' `icpcovars' `controls' i.Year if icp != ., fe i(qnno) test lnc_seedcost = ilnc_seedcost test lnc_totfertcost = ilnc_totfertcost test lnc_totchemcost = ilnc_totchemcost test lnc_totallabour_sum = ilnc_totallabour_sum test lnc_totallabourcost_sum = ilnc_totallabourcost_sum test lnc_salescost = ilnc_salescost *************************** ***Endogeneity of inputs*** *************************** use mm_wide.dta, clear ***CRC, with covariates, chemical use endogenous local maincovars1 lnc_seedcost2008 lnc_totfertcost2008 lnc_totallabour_sum2008 lnc_totallabourcost_sum2008 lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totfertcost2010 lnc_totallabour_sum2010 lnc_totallabourcost_sum2010 lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totfertcost2014 lnc_totallabour_sum2014 lnc_totallabourcost_sum2014 lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') /// method(CRC) endo(lnc_totchemcost2008 lnc_totchemcost2010 lnc_totchemcost2014) keep use mm_wide.dta, clear ***CRC, with covariates, family labor endogenous - takes around 2300 iterations local maincovars1 lnc_seedcost2008 lnc_totfertcost2008 lnc_totchemcost2008 lnc_totallabourcost_sum2008 lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totfertcost2010 lnc_totchemcost2010 lnc_totallabourcost_sum2010 lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totfertcost2014 lnc_totchemcost2014 lnc_totallabourcost_sum2014 lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') /// method(CRC) endo(lnc_totallabour_sum2008 lnc_totallabour_sum2010 lnc_totallabour_sum2014) keep use mm_wide.dta, clear ***CRC, with covariates, fertilizer endogenous local maincovars1 lnc_seedcost2008 lnc_totchemcost2008 lnc_totallabour_sum2008 lnc_totallabourcost_sum2008 lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totchemcost2010 lnc_totallabour_sum2010 lnc_totallabourcost_sum2010 lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totchemcost2014 lnc_totallabour_sum2014 lnc_totallabourcost_sum2014 lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp2008 lnp2010 lnp2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') /// method(CRC) endo(lnc_totfertcost2008 lnc_totfertcost2010 lnc_totfertcost2014) keep /*========================================================================= APPENDIX C: SEPARABILITY OF LABOR ===========================================================================*/ use mm_long.dta, clear *Real Wages from Bachewe et al., 2016 gen wage_b = 1.51 if Year == 2008 replace wage_b = 1.49 if Year == 2010 replace wage_b = 2.11 if Year == 2014 lab var wage_b "wage from Bachewe et al. (2016)" gen famlabourvalue_b = totallabour_sum*wage_b lab var famlabourvalue_b "family labour (USD)" gen labour_Bachewe = famlabourvalue_b+totallabourcost_sum lab var labour_Bachewe "total labour (USD) Bachewe et al. (2016)" gen cpfamlabour_b = cplabour*wage_b lab var cpfamlabour_b "chickpea family labour (USD)" gen cplabour_Bachewe = cpfamlabour_b+cptotallabourcost lab var cplabour_Bachewe "chickpea total labour (USD) Bachewe et al. (2016)" *Shadow Wages from Saketta and Gerber, 2017 gen wage_s = 7.134 if Year == 2010 replace wage_s = 24.86 if Year == 2014 replace wage_s = 7.134*(CPI2009/100) if Year == 2008 lab var wage_s "wage from Saketta \& Gerber (2017)" gen famlabourvalue_s = totallabour_sum*wage_s lab var famlabourvalue_s "family labour (USD)" gen labour_Saketta = famlabourvalue_s+totallabourcost_sum lab var labour_Saketta "total labour (USD) Saketta \& Gerber (2017)" gen cpfamlabour_s = cplabour*wage_s lab var cpfamlabour_s "chickpea family labour (USD)" gen cplabour_Saketta = cpfamlabour_s+cptotallabourcost lab var cplabour_Saketta "chickpea total labour (USD) Saketta \& Gerber (2017)" ** Generate new dependent variables gen lnc_b = asinh((totalcost_sum + famlabourvalue_b)/cultarea) lab var lnc_b "Ln total cost with Bachewe et al. (2016) wage (USD/ha)" gen lnc_s = asinh((totalcost_sum + famlabourvalue_s)/cultarea) lab var lnc_s "Ln total cost with Saketta \& Gerber (2017) wage (USD/ha)" gen lnp_b = asinh((salesinc_sum - totalcost_sum - famlabourvalue_b)/cultarea) lab var lnp "Ln crop sales profit (USD/ha)" gen lnp_s = asinh((salesinc_sum - totalcost_sum - famlabourvalue_s)/cultarea) lab var lnp "Ln crop sales profit (USD/ha)" ** Generate chickpea labor values foreach var of varlist cplabour_Bachewe cplabour_Saketta { bysort qnno Year: gen lny_`var' = asinh(`var'/cpland) } lab var lny_cplabour_Bachewe "Ln labour cost per har" lab var lny_cplabour_Saketta "Ln labour cost per har" ** Generate farm cost/revenue variables foreach var of varlist labour_Bachewe labour_Saketta { bysort qnno Year: gen lnc_`var' = asinh(`var'/cultarea) } lab var lnc_labour_Bachewe "Ln labour cost per har" lab var lnc_labour_Saketta "Ln labour cost per har" reshape wide District- lnc_labour_Saketta, i(qnno) j(Year) save "lab_wide.dta", replace *** 1 Production *** ***CRC, with Bachewe labour local maincovars1 lny_cpseed2008 lny_cpchemfertqt2008 lny_cptotchemcost2008 lny_cpsalescost2008 lny_cplabour_Bachewe2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lny_cpseed2010 lny_cpchemfertqt2010 lny_cptotchemcost2008 lny_cpsalescost2010 lny_cplabour_Bachewe2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lny_cpseed2014 lny_cpchemfertqt2014 lny_cptotchemcost2014 lny_cpsalescost2008 lny_cplabour_Bachewe2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lny2008 lny2010 lny2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') meth(CRC) ***CRC, with Saketta labour local maincovars1 lny_cpseed2008 lny_cpchemfertqt2008 lny_cptotchemcost2008 lny_cpsalescost2008 lny_cplabour_Saketta2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lny_cpseed2010 lny_cpchemfertqt2010 lny_cptotchemcost2008 lny_cpsalescost2010 lny_cplabour_Saketta2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lny_cpseed2014 lny_cpchemfertqt2014 lny_cptotchemcost2014 lny_cpsalescost2008 lny_cplabour_Saketta2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lny2008 lny2010 lny2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') meth(CRC) *** 2 Cost *** ***CRC, with Bachewe labour local maincovars1 lnc_seedcost2008 lnc_totfertcost2008 lnc_totchemcost2008 lnc_salescost2008 lnc_labour_Bachewe2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totfertcost2010 lnc_totchemcost2010 lnc_salescost2010 lnc_labour_Bachewe2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totfertcost2014 lnc_totchemcost2014 lnc_salescost2014 lnc_labour_Bachewe2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnc_b2008 lnc_b2010 lnc_b2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRC) ***CRC, with Saketta labour local maincovars1 lnc_seedcost2008 lnc_totfertcost2008 lnc_totchemcost2008 lnc_salescost2008 lnc_labour_Saketta2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totfertcost2010 lnc_totchemcost2010 lnc_salescost2010 lnc_labour_Saketta2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totfertcost2014 lnc_totchemcost2014 lnc_salescost2014 lnc_labour_Saketta2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnc_s2008 lnc_s2010 lnc_s2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRC) *** 4 Profit *** ***CRC, with Bachewe labour local maincovars1 lnc_seedcost2008 lnc_totfertcost2008 lnc_totchemcost2008 lnc_salescost2008 lnc_labour_Bachewe2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totfertcost2010 lnc_totchemcost2010 lnc_salescost2010 lnc_labour_Bachewe2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totfertcost2014 lnc_totchemcost2014 lnc_salescost2014 lnc_labour_Bachewe2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp_b2008 lnp_b2010 lnp_b2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRC) ***CRC, with Saketta labour local maincovars1 lnc_seedcost2008 lnc_totfertcost2008 lnc_totchemcost2008 lnc_salescost2008 lnc_labour_Saketta2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010 lnc_totfertcost2010 lnc_totchemcost2010 lnc_salescost2010 lnc_labour_Saketta2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014 lnc_totfertcost2014 lnc_totchemcost2014 lnc_salescost2014 lnc_labour_Saketta2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnp_s2008 lnp_s2010 lnp_s2014, choice(icp2008 icp2010 icp2014) controls(`maincovars1' `controls1' `maincovars2' `controls2' `maincovars3' `controls3') method(CRC) /*========================================================================= APPENDIX D: TWO YEAR CRC ===========================================================================*/ use mm_wide.dta, clear ***2008-10 CRC, with covariates local maincovars1 lnc_seedcost2008- lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars2 lnc_seedcost2010- lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 randcoef lnr2008 lnr2010 , choice(icp2008 icp2010 ) controls(`maincovars1' `controls1' `maincovars2' `controls2') meth(CRC) showreg ***2010-14 CRC, with covariates local maincovars2 lnc_seedcost2010- lnc_salescost2010 local controls2 Head_gender2010 dependencyperc2010 offincsource2010 loglandowned2010 logmainrf2010 shock2010 local maincovars3 lnc_seedcost2014- lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnr2010 lnr2014, choice(icp2010 icp2014) controls(`maincovars2' `controls2' `maincovars3' `controls3') meth(CRC) ***2008-14 CRC, with covariates local maincovars1 lnc_seedcost2008- lnc_salescost2008 local controls1 Head_gender2008 dependencyperc2008 offincsource2008 loglandowned2008 logmainrf2008 shock2008 local maincovars3 lnc_seedcost2014- lnc_salescost2014 local controls3 Head_gender2014 dependencyperc2014 offincsource2014 loglandowned2014 logmainrf2014 shock2014 randcoef lnr2008 lnr2014, choice(icp2008 icp2014) controls(`maincovars1' `controls1' `maincovars3' `controls3') meth(CRC) /*========================================================================= APPENDIX E: REDUCED FORM COEFFICIENTS ===========================================================================*/ *Reduced form coefficients for the table in Appendix E are taken from the regressions for 3 year CRC above. *END