do public works programs increase women’s economic
TRANSCRIPT
cfgafgfdsgerg
DoPublicWorksProgramsIncreaseWomen’sEconomicEmpowerment?
EvidencefromRuralIndia
SonaldeDesai
ProfessorofSociologyUniversityofMaryland
andSeniorFellow
NationalCouncilofAppliedEconomicResearch
October2018
Acknowledgement:This work was carried out with the aid of a grant from the InternationalDevelopmentResearchCentre,Ottawa,Canada.TheviewsexpressedhereindonotnecessarilyrepresentthoseofIDRCoritsBoardofGovernorsoranyotherfundingagency.The results are based on the IndiaHumanDevelopment Survey (IHDS). TheIHDSfieldwork,dataentryandanalyseshavebeenfundedthroughavarietyofsources, including the US National Institutes of Health (grant numbersR01HD041455 and R01HD061048), UK Department of InternationalDevelopment,TheFordFoundation,andTheWorldBank.Collaboration fromReeve Vanneman andAmareshDubey in IHDS project isgratefully acknowledged.Thedata filewith administrativeMGNREGAvillageleveldatawascarefullyconstructedbyMr.O.P.Sharma.Withouthishelpandsupport,thisresearchwouldnothavebeenpossible.
- 1 -
DoPublicWorksProgramsIncreaseWomen’sEconomicEmpowerment?
EvidencefromRuralIndia
Sonalde Desai
Abstract:Althoughitisoftenarguedthatlabormarketdiscrimination,resultinginblockedemploymentopportunitiesandwagediscrimination,reduceswomen’seconomicempowerment,ithasbeendifficulttoexaminethisconstraintrigorously.TheenactmentoftheMahatmaGandhiNationalRuralEmploymentGuaranteeAct(MGNREGA)of2005offersusauniqueopportunitytoexaminetheroleofexpandingopportunitiesonwomen’seconomicempowerment.Usingsurveydatacollectedbeforeandaftertheenactmentofthisprogram,weexaminechangesinwomen’sparticipationinpaidworkandtotalearningsbyusingadifference-in-differenceapproach.TheresultssuggestthatwomenlivinginvillageswithahigheravailabilityofMGNREGAworkaremorelikelytoparticipateinwagelaborandhavehighertotalwageincomes.Theseeffectsareabsentformen,possiblybecausetheyhavegreateralternativeopportunities.
- 2 -
DoPublicWorksProgramsIncreaseWomen’sEconomicEmpowerment?EvidencefromRuralIndia
SonaldeDesai
Introduction: TheprevalenceoflowemploymentratesamongwomeninSouthAsiaandtheMiddleEasthasbeenthesubjectofconsiderableattentionintheliteraturefornearlyfourdecades(Boserup1970,Dixon1982).However,afurtherdeclinefromeventheselowrateshastakenanalystsbysurprise(KlasenandPieters2012,Dasetal.2015,Kapsos,Bourmpoula,andSilbereman2014,Desai2013).TheNationalSampleSurveydata(Kapsos,Bourmpoula,andSilbereman2014)showthatin1994,42.7percentofthewomenaged15yearsandabovewereparticipatinginthelaborforcewhilethisproportionhadfallento31.2percent,withthedeclineforruralwomenbeinggreater(from49percentto35.8percent)thanthatforurbanwomen(from23.8percentto20.5percent). Howdoweexplainthisdecline?Doeseconomicgrowthcreateconditionsthatallowpoorwomentostepoutofthelaborforcetomeettheircareresponsibilities?Orarethepatternsofeconomicgrowthsuchthatruralwomenareunabletofindappropriatework?ThisisthetopicthatthepresentpaperexploresbyfocusingonauniquelabormarketexperiencecalledtheMahatmaGandhiNationalRuralEmploymentGuaranteeAct(MGNREGA),whichcameintoeffectin2006.ThisActissupposedtooffer100daysofmanuallabortoanyruralhouseholdthatdemandswork,withthesamewagesbeingofferedtomenandwomen. UsingpaneldatafromtheIndiaHumanDevelopmentSurvey(IHDS),conductedtwicein2004-05beforetheimplementationofMGNREGA,andin2011-12afterMGNREGAhadbeeninexistenceforseveralyears,thispaperexaminesthedifferencesinwomen’sworkparticipationandincomesbeforeandaftertheimplementationoftheAct.Thisexogenouschangeallowsustotesttheimpactoflabormarketinterventionsonwomen’seconomicempowerment.
ExplanationsforDecliningFemaleLaborForceParticipation:
OneofthebestknownexplanationsforaU-shapedrelationshipbetweenwomen’semploymentandeconomicdevelopmenthasbeenofferedbyClaudiaGoldin,whoarguesthatwomen’slaborforceparticipationisaffectedbytwoparametersoftheSlutskyequation—ownwage(compensated)elasticityandtheincomeelasticity.Where
- 3 -
ahighdegreeofstigmaisassociatedwiththeparticipationofwivesinwagework,risingincomesleadtoadeclineinfemalelaborforceparticipation.Thisisparticularlytruewherewomenfacepoorjobprospects.Highlevelsofeconomicdevelopmenthelpimprovewomen’sprospectswhilealsoreducingideologicalbarrierstowomen’sparticipationinpaidwork—atleastbasedontheWesternexperience—whichcombinetostrikinglyimprovewomen’slaborforceparticipationrates(Goldin1995,2006).
However,beforeweassumethattheeconomicgrowthexplainsthedeclineinwomen’slaborforceparticipationinIndia,itisimportanttonotethatempiricalstudiestryingtoanalyzetheextentofsupportfortheU-shapedrelationshipbetweenwomen’semploymentandeconomicgrowthhavefoundonlylimitedsupport(GaddisandKlasen2014). Asecondargumentforthedeclineinwomen’slaborforceparticipationisassociatedwithrisingeducation.Asgirlsandyoungwomenspendmoretimeinschoolandcollege,itmayreducetheiravailabilitytoparticipateinthelaborforce.Moreover,evenacquiringalowlevelofeducationcauseswomentonurturepreferencesforjobsthatarenoteasilyavailabletothem.Formostruralwomen,agriculturalworkisusuallytheonlyoptionavailable,beitworkonthefamilyfarmorforwages.Formenwhohaveacquired4-9yearsofeducation,itmaybeeasiertofindjobsasdrivers,postmenorconstructionsupervisors,butthesejobsarenoteasilyavailabletowomen.Consequently,womenhaverelativelyfewoptionsuntiltheycompletesecondaryeducationandcanworkatmoreskilledjobsasteachers,nursemidwives,orpre-schoolprogramworkers.Thus,aseducationgrows,initialeducationalexpansionmaybeassociatedwithwomen’slaborforcewithdrawal(Neff,Sen,andKling2012). Athirdpotentialexplanationrelatestochangesinemploymentopportunitiesinthedemandforlabor.AnoverwhelmingmajorityofruralIndianwomenworkinagriculture,eitherascultivatorsonfamilyfarmsorasagriculturallaborersinnearbyfarms.TheNationalSampleSurvey(NSS)documentsthatwhile59percentoftheruralmaleworkersareinagriculture,thecorrespondingfigureforfemaleworkersis75percent(NationalSampleSurveyOrganisation2013).However,boththeproportionofGDPcomingfromtheagriculturalsectoraswellasfarmsizeshavesteadilydropped,resultinginextremecrowdingintheagriculturalsector(Papola2012,AgricultureCensusDivision2014).Sincewomenaredisproportionatelylocatedinthissector,theymaybeadverselyaffectedbythesedevelopments. AfourthexplanationmaybethatthedatafromtheNSS,onwhichmostresearchersrelytoestimateworkparticipationrates,maybestructuredinawaythatfailstocapturethefullrangeofwomen’sactivitiesandthatstructuralchangesinthenatureoflabormarketshaveintensifiedthisomissioninrecentyears.
- 4 -
Fromapublicpolicyperspective,botheventualities,thatis,whetherwomenarecrowdedoutofthelabormarketorhavechosentowithdrawfromitvoluntarily,havesignificantimplications,bothforwomen’seconomicempowermentandfordevelopmentplanning.Whiledecliningfemaleworkparticipationremainssubjecttomuchspeculation(Neff,Sen,andKling2012,GaddisandKlasen2014,KlasenandPieters2012),ithasbeendifficulttoempiricallyevaluatethesecompetingexplanations.Fortunately,theenactmentoftheMGNREGAin2005helpsustotesttheimpactoftheexpansionofemploymentopportunitiesonwomen’sworkparticipation.MahatmaGandhiNationalRuralEmploymentGuaranteeAct:ANaturalExperiment
In2005,Indiapassedanextraordinarylegislation,theNationalRuralEmploymentGuaranteeAct(NREGA),laterrenamedastheMahatmaGandhiNationalRuralEmploymentGuaranteeAct(MGNREGA).MGNREGAguarantees100daysofmanuallabortoanyruralhouseholdthatdemandswork.ThewageratesarefixedbyeachstatebutmustbeaminimumofRs.100perday,andequalwagesaretobeofferedtomenandwomen.Householdsmaysplittheirentitlementof100daysinanywaytheychoosebetweendifferenthouseholdmembers(MinistryofRuralDevelopment2013).On-sitechildcareistobeprovided,thoughstudiesshowthatthisprovisionispoorlyimplemented(KheraandNayak2009).TheActrequiresthatatleastone-thirdoftheworkbegiventowomen.Sincefewotherjobsprovideequalwagesformenandwomen,itisnotsurprisingthatwomenhaveflockedinlargenumberstoMGNREGAwork,andconsequently,overtheyears,MGNREGAhascometobedominatedbywomenworkers(Desai,Vashishtha,andJoshi2015).
MGNREGAhasseveralcharacteristicsthatareparticularlynoteworthy:(1)Itissupposedtobeavailabletoanyruralhouseholdthatdemandsworkwithoutanytargeting.(2)Itoffersequalwagestomenandwomen.(3)Itissupposedtobeavailableondemandwithvillagesrequiredtoholdatleasttwomeetingsayearwherehouseholdscanregistertheirworkdemand. EvaluationsofMGNREGAimplementation,however,showthatthepromiseof100daysofworkisrarelyimplemented.Sincelocalandstategovernmentsaresupposedtotakealeadinprogramimplementation,considerableheterogeneityinimplementationisobservedacrossthecountry.StateslikeChhattisgarh,RajasthanandAndhraPradeshhaveprovidedsubstantialruralemploymentthroughMGNREGAwhileGujaratandOdishahavenothadastrongprogram(MinistryofRuralDevelopment2015).Moreover,evenwithinthesamestateanddistrict,somevillageleadershavefiguredouthowtoformulateandimplementprojectsthatuseMGNREGAfundswhileotherlocalleadershavebeenmorelackadaisicalintheirapproach(Desai,Vashishtha,
- 5 -
andJoshi2015).Thisfailureifoftenduetoalackofunderstandingabouttheprogramstructureandabilitytonavigatethesystem. MGNREGAprovidesanextraordinaryopportunitytoexaminetheroleofemploymentopportunitiesinshapingwomen’seconomicempowerment.Inspiteofconsiderableadvocacyamongfeministscholarsandactivistsforincreasingemploymentopportunitiesforwomenandforeliminationofthegendergapinpayscales,werarelycomeacrosssituationsthatlendthemselvestoevaluatingwhethertheexpansionofemploymentopportunitieswillactuallyresultingreateremploymentforwomenorwhethertheirdomesticandcareresponsibilitieswilldominatewithlittlechangeinthetrendsforwomen’semployment. ResearchStrategy:
Thispaperreliesonchangesinparticipationinpaidworkandtotalwageincome
forruralIndianmenandwomenaged15-64yearsbetween2004-05and2011-12.Twoaspectsoftheprogrammakeitpossibleforthispapertoexplorethewayinwhichwomen’slaborforcebehaviorrespondstoexpandingopportunities.First,bycomparingwomen’sparticipationinpaidworkbeforeandaftertheimplementationoftheprograminthesamehouseholdsandvillages,weareabletotracethechangesinculturalandsocialconditionsthatlimitwomen’semployment.Second,bycomparingchangesovertimebetweenvillagesthatexhibitedastrongimplementationofMGNREGAwiththosethathadweakimplementationofMGNREGA,weareabletotracetheprogrameffectnetofsecularchangesthataffectedthenationasawholeafter2005whentheActwaspassed.
IndiaHumanDevelopmentSurvey
TheaboveanalysisisfacilitatedbydatafromtheIndiaHumanDevelopmentSurvey(IHDS),WavesIandII.WaveIoftheIHDSwasconductedin2004-05,justbeforeMGNREGAwasimplemented.1WaveIIoftheIHDSwasconductedin2011-12,whentheActwasfullyimplemented.IHDS-Iinterviewed41,554householdsspreadacross1503villagesand971urbanblocksinallthestatesandUnionTerritoriesofIndia,withtheexceptionoftheislandsofAndamanandNicobar,andLakshadweep.IHDS-IIsetouttointervieweachoftheIHDS-Ihouseholdsandanysplithouseholdsthatlivedinthesamelocality.Itwasabletore-interview72percentoftheurbanand90percentoftheruralhouseholds.Afteragapofsevenyears,anoverallre-contactrateof83percentputsthe
1 MGNREGAwaspassedin2005andimplementedinaphasedmannerbeginningwith2006.By2011-12,whenIHDS-IIwasconducted,theruralareasofallthedistrictswerecoveredbyMGNREGA.
- 6 -
IHDSamongsomeofthebestmaintainedpanelsamplesindevelopingcountries(Aldermanetal.2001).
TheIHDSisamulti-purpose,multi-topicsurveythatcontainsinformationabout:
• Basicbackgrounddata(caste,religion,placeoforigin);• Trackingoforiginalhouseholdmembersandcollectionofproxyinformationfor
migrantsaswellasremittancesfrommigrants;• Housingconditionsandassetownership;• Detailedincomeandemployment;aconsumptionexpendituremodulewith50
broadcategories;debtandfinancialsectorparticipationdata;andlandownershipincludingintra-householddifferencesinownership;
• Morbidityandhealthexpenditure,ADL;• Education,educationalexpenditure,basicreading/arithmetictestsforyouth
aged8-11and15-18years;• Socialnetworks,trustandconfidenceininstitutions,localcrime,sexual
harassment;• Majorhouseholdeventsbetweentwointerviews;and• Genderrelations,fertility,contraception,marriagehistory(collectedfrom
womenbywomeninterviewers).TheIHDSdataareconsideredtobeofhighquality,andhavebeendownloadedby
over8,000usersandhavegeneratedmorethan220papersanddissertations.ComparisonsoftheIHDSestimatesofbasicdemographiccharacteristicswiththeCensus,NationalSampleSurveysandNationalFamilyHealthSurveyssuggestthatonmostmajorvariables,theIHDSresultsaresimilartotheseothersources(Desaietal.2010).
Forthispaper,weintegratetheIHDSsurveydataonMGNREGAimplementation
fromdataprovidedbytheMinistryofRuralDevelopment(mgnrega.nic.in).ThevillagesincludedintheIHDSweremanuallymatchedwiththenumberofdaysofMGNREGAprovidedin2010-11and2011-12andwiththevillagepopulationfromCensus2011.ThisallowsustoestimatetheaveragenumbersofdaysofMGNREGAworkprovidedperhouseholdineachofthevillagesinwhichtheIHDSrespondentslive.WehavebeenabletoobtainthisinformationforallthestatesexceptTamilNadu,whichhasthusbeenexcludedintheseanalyses.2ThisprovidesuswithanestimateoftheintensityofMGNREGAeffortsinthevillage.
2 EffortsforobtainingdataforTamilNaduareunderway.
- 7 -
MeasuringWomen’sEconomicEmpowerment:Inthispaper,wefocusonthetwomainmeasuresofwomen’seconomic
empowerment:(1)Whetherwomenparticipateinpaidwork,includingagriculturalcasuallabor,non-agriculturalcasuallabor,salariedemployment,andMGNREGAwork,duringtheyearprecedingthesurvey;and(2)Totalcashearningsduringtheyearprecedingthesurvey.InordertoexaminetheroleofMGNRGAinshapingwomen’seconomicoutcomesasopposedtomen’seconomicoutcomes,weestimateandpresentmodelsformenandwomenseparatelybuttestforsignificanceofrelevantcoefficientsinapooledmodel.
Women’seconomiccontributionstothehouseholdwell-beinginvolvebothwageworkandworkinhouseholdenterprisessuchashouseholdfarmorhouseholdbusiness.Whilehouseholdsbenefitconsiderablyfromwomen’sworkonfamilyfarms,whichoftenreleasesmentoengageinwagework(DesaiandJain1994),qualitativeaswellasquantitativestudiessuggestthatwomenthemselvesattachconsiderablevaluetotheirroleaseconomicactorsasopposedtounpaidfamilylaborandcashincomeoftenincreasestheirvoice,agencyandcontroloverhouseholdresources(Kabeer1999,PresserandSen2000,Narayan2006,Agarwal1997).Hence,itisimportanttoexaminetheroleofMGNREGAinshapingwomen’saccesstopaidwork.
Inadditiontowhetherwomenparticipateinpaidworkornot,wealsoexaminewomen’stotalcashearningsduringtheyearprecedingthesurvey.Thetotalearningsareafunctionofbothparticipationinwagelaborandthewagerate.Wefocusonthetotalearnings,settingtheearningstozeroformenandwomenwhohavenocashincome. TheChallengeofMeasuringWomen’sWorkParticipation:
Indiaishometoavibrantwomen’smovementandoneofthemostimportantcontributionsofthismovementistoensurethatemploymentdatacollectionbytheNationalSampleSurveyOrganization(NSSO)andtheCensusentailsimpartingadequatetrainingtodatacollectorstoensurethattheactivitieswomenparticipatein(forexample,helpingonfamilyfarms,caringforanimals,makingpicklesorgur[jaggery]forsale)isconsideredaseconomicallyproductiveworkratherthanmerelydomesticwork(JainandBanerjee1985).However,afocusontimeallocation,combinedwithdatacollectionstrategiesusedbytheNSSO,maynotadequatelycapturewomen’seconomicactivities,particularlyinaneraoftransition.
ThefrequentlyusedNSSOdefinitionofemploymentisacombinationofUsualPrincipalStatus(PS)andUsualSubsidiaryStatus(SS).AnindividualisdefinedasbeingemployedaccordingtoPS,ifs/heengagesintheNSSdefinitionofeconomicactivityfor
- 8 -
amajorityoftheyear.AnindividualisdefinedasbeingemployedaccordingtotheSSifs/heisengagedinaneconomicactivityforatleast30days.If,inaneraoframpantunder-employment,ayoungwomanspendsfiveweekscollectingforestproduceforownconsumption,shewillbeclassifiedasbeingemployedaccordingtothesubsidiarystatus.However,ifthesuddenavailabilityofconstructionworkleadshertospend20daysworkingatawagethatallowshertopurchasefirewood,shewouldnotbeconsideredasbeingemployed,evenbysubsidiarystatus.Moreover,ifsheworksinseveraldifferentactivitiesbutnoneoftheselastsforatleast30daysatastretch,andfor183daysincombination,wouldshebeclassifiedasbeingemployedbyeitherprincipalorsubsidiarystatuscriteria?Instructionstotheintervieweraresomewhatambiguousonthisscore.Timeuseresearchshowsthatwomenarefarmorelikelytoengageinmultipleactivitiesandtheuseofworkparticipationratesbasedontimeusedataisbetterabletocapturemultipleactivities.Consequently,workparticipationestimatesbasedontheNSSmayunder-estimatetheworkparticipationofwomen(HirwayandJose2011,Kapsos,Bourmpoula,andSilbereman2014). UnliketheNSSO,theIHDScollectsdataonbothincomeandemploymentinasinglemodule.Thus,itfirstaskswhetherthehouseholdownsorcultivatesland,thenasksaboutseason-wiseproduction,andfinallyaskswhoengagedinfarmwork.Similarly,forwageandsalarywork,itlistseverysinglepaidactivitythatindividualsundertake,regardlessofthenumberofdaystheywork.Thisallowsforagreatercaptureoffragmentedandmultipleactivities.Asaresult,IHDSworkparticipationratesforwomenarehigherthantheNSSparticipationrates,butthoseformenarecomparable.StatisticalModel:
InordertoexaminetheimpactoftheavailabilityandintensityofMGNREGAwork,weestimatethreelevelrandominterceptmodelswherethehouseholdIisnestedinvillagejandvillagejisnestedinstatek.ThisallowsustoestimatethreelevelrandominterceptmodelsusingSTATAofthefollowingform:
𝑌𝑖𝑗𝑘 = 𝛼0 + 𝛼1𝑋1 + 𝛼2𝑌𝑒𝑎𝑟 + 𝛼3𝑁𝑅𝐸𝐺𝐴 + 𝛼4𝑌𝑒𝑎𝑟 ∗ 𝑁𝑅𝐸𝐺𝐴 +𝛽0𝑗𝑘 + 𝛿00𝑘 + 𝜀𝑖𝑗𝑘
whereX1referstoindividualandhouseholdcharacteristicstobecontrolled,whileYearreferstowave2oftheIHDSsurveyconductedduringtheyear2011-12,andmeasuressecularchangeinoutcomeYovertime.NREGAisacontinuousvariableindicatingthenumberofdaysofMGNREGAworkprovidedinthevillageduringtheyearprecedingthesurvey3andtheinteractiontermYear*NREGAindicatesthechangeintheimpactof
3 SincetheIHDS-2surveyspannedtheperiodOctober2011toDecember2012,wehavetakenanaverageofvillagelevelMGNREGAdaysforFYs2010-11and2011-12.
- 9 -
NREGAvariablebetween2004-05and2011-12.Thisdifference-in-differenceequationallowsustoseetheimpactofMGNREGAimplementationintensityonlaborforcebehaviorwhilecontrollingforbothvillageandhouseholdcharacteristics.
Thecontrolvariablesincludemaritalstatus,ageoftheindividual,casteandreligion,landownership,educationoftheindividual,highesteducationlevelofanyhouseholdadults,andwhetherindividualsliveinavillagethathasrelativelylowinfrastructurefacilities.DescriptiveStatistics:WorkParticipation:
Figures1and2describechangesinthelabormarketparticipationofmenandwomenaged15-64yearsbetween2004-05and2011-12.Figures1aand1bshowthatifwedonotlimitourselvestotheNSSOdefinitionofPrincipalandSubsidiaryStatusandfocusonanywork,evenifitisundertakendiscontinuously,spreadacrossdifferentactivitiesanddoneforashorttime,theproportionofpopulationthatisnotemployeddropsforbothmenandwomen,butthedeclineislargerforwomen.Theincreaseinwomen’slaborparticipationcomesmostlyfromthenumberofwomenwhoworkforlessthanamonth(increasingfrom7.5percentto11.7percent)butasmalldeclineintheproportionofwomenworkingforatleast60days.ThissuggeststhattheNSScriterionofignoringshort-termworkmaybemissingoutsomeimportantchangesinIndianlabormarkets,particularlyforwomen.
17.9
4.0 4.4
69.2
15.9
7.1 5.6
66.6
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
No work 1 Month 2 Months 60+ days
Percent
No.ofMonthsWorkedinPrecedingYear
Fig.1a:TrendsinWorkParticipationforMenages15-64between2004-5and2011-12
2004-5 2011-12
- 10 -
Theotherreasonbehindtheunder-estimationofworkintheNSSmayhavetodowithfragmentationofwork.Whenindividualsworkinmorethanoneactivityandnoactivitymeetsthethresholdof30days,itispossiblethatenumeratorsomittheseactivitiesfromtheiractivitycount.Figures2aand2bdocumenttheconsiderableincreaseintheproportionofmenandwomenwhoundertakebothagriculturalandnon-agriculturalwork.
50.5
7.5 7.3
34.8
46.0
11.78.4
33.9
0.0
10.0
20.0
30.0
40.0
50.0
60.0
No work 1 Month 2 Months 60+ days
Percent
No.ofMonthsWorkedinPrecedingYear
Figure1b:TrendsinWorkParticipationforWomenAged15-64Yearsbetween2004-05and2011-12
2004-5 2011-12
17.9
41.6
18.422.2
15.9
32.2
27.324.6
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
No Work Agriculture Only Combination Non AgricultureOnly
Percent
TypeofWorkinthePrecedingYears
Figure2a:TrendsinWorkTypeforMenAged15-64Yearsbetween2004-05and2011-12
2004-5 2011-12
- 11 -
TheseobservationsareborneoutbythedatapresentedinTable1.Thistabledocumentsthedecreaseintheproportionofmenandwomenwhoareoutoftheworkforce.Italsodocumentsanoveralldeclineinthenumberofdaysworkedbyeachparticipantwiththedeclinebeingthelargestinagriculturalwork,whetherthelatterwasworkingonthefamilyfarmorasalaboreronthefarmsownedbyotherfarmers.Buttheresultsalsoshowsomesharpdifferencesintheworkpatternsofmenandwomenbetweenthetwosurveys.Bothmenandwomenarelesslikelytoworkasagriculturallaborersin2011-12thantheywerein2004-5,andperhapstomakeupforthedeclininguseofhiredlabor,theyincreasetheirparticipationinworkonfamilyfarms.However,thisincreaseisgreaterforwomenthanformen.Infamilybusinessalso,womenhavealargerincreaseinparticipationthanmen.Incontrast,menhavesharplyincreasedtheirparticipationinnon-farmwagelabor,atrendnotvisibleforwomenifexcludeMGNREGAwork.TheimpactoftheintroductionofMGNREGAisvisibleinTable1wheremen’sparticipationinMGNREGAincreasefromnoparticipationinpreimplementationerato12percentmenand9percentwomenundertakingMGNREGAwork.
[Table1abouthere]ParticipationinMGNREGA:
AlthoughMGNREGAissupposedtooffer100daysofworktoanyhouseholdthatdemandswork,theactualavailabilityofworkisconsiderablylowerwithlessthan5percentoftheIHDShouseholdsbeingabletogetfull100daysofwork(Desai,Vashishtha,andJoshi2015).TheissueofworkrationinginMGNREGAhasbeenextensively
50.5
39.3
4.2 6.0
46.0
34.6
10.4 9.1
0.0
10.0
20.0
30.0
40.0
50.0
60.0
No Work Agriculture Only Combination Non AgricultureOnly
Percent
TypeofWorkinthePrecedingYear
Fig.2b:TrendsinWorkTypeforWomenages15-64between2004-5and2011-12
2004-5 2011-12
- 12 -
documented(Duttaetal.2012,Das2015).StatelevelpoliciesdetermineMGNREGAimplementationbutthereisconsiderablelocalvariationinit,frequentlyduetothelackofcapacityonthepartofvillageofficialsandsometimesduetothepressureexertedbylargefarmerstoensureacontinuoussupplyofagriculturalworkwithoutcompetitionfromMGNREGA.Whilethispositsanunfortunatesituationforhouseholdsseekingfunds,itallowsustoexaminetheroleofMGNREGAinincreasingwomen’seconomicempowermentatdifferentlevelsofMGNREGAimplementation.
Figure3plotsthecumulativedistributionofperhouseholdMGNREGAworkintheIHDShouseholds.Theresultsshowthat17percentofthevillageshadnoallocationofMGNREGAworkatall,whileanadditional30percentallocatedlessthanonedayofworkperhousehold.Ofcourse,forindividualswhoparticipatedinMGNREGA,theaveragenumberofdaysworkedwillbehigher.Table2providesdescriptivestatisticsforMGNREGAparticipantsandnon-participantsin2011-12.Onanaverage,MGNREGAparticipantsareslightlyolderandlesseducatedthanthenon-participants,andtendtocomefromDalitorAdivasibackgrounds.
[Table2abouthere]MultivariateResults:
ThegoalofthispaperisexaminetherolethattheavailabilityofMGNREGAworkplaysinshapingwomen’seconomicempowerment.WedonotexaminethedirectimpactofMGNREGAsincethatisendogenous.However,sinceMGNREGAisademand-drivenprogram,itishardtoruleouttheroleofindividualdemandandendogenousprogramplacementthatafflictsmanyotherareasofresearch(Angeles,Guilkey,andMroz1998).Weargue,however,thattwoaspectsofourstrategyshieldussomewhatfromthischallenge.First,wefocusonbeforeandafterprogramimplementationinthesamevillages.Thisdifference-in-differenceallowsustotakeintoaccountthepre-existingdifferencesbetweenMGNREGAandnon-MGNREGAvillages.Second,we
0
20
40
60
80
100
120
0 3 6 9 12 15 18 21 24 27 30 33 37 40 43 46 49 55 60 64 67 71 77 88 99
%Villages
DaysofMGNREGAworkDonePerHousehold
Fig3:CumulativeDistribuitonofMGNREGAWorkintheVillage:AdministrativeData2010-2012
- 13 -
comparemenandwomeninthesamevillages,allowingustotesttherelationshipbetweenMGNREGA-inducedlabormarketchangesformenandwomen,andeconomicempowerment.Ifwefindastrongerimpactforwomenthanformen,wecanarguethatMGNREGAfillsanichethatallowswomenwiththelatentdemandforpaidworktomeettheirneeds.
[Table3abouthere] Table3presentsresultsfromtherandomeffectslogisticregressionmodelestimatedwithSTATAwithparticipationinpaidworkasthedependentvariable.Theresultsshowthattheparticipationofmenandwomeninpaidworkincreasedbetween2004-05and2011-12,withtheincreasebeingslightlygreaterforwomen.VillagesthathaveagreaterallocationofMGNREGAworkalsoseemtobevillageswherepaidlaborwashigherevenbeforeMGNREGAwasimplemented,thatis,atthetimeofthe2005survey.However,overandabovethesespatialandseculareffects,participationinpaidworkincreasedatagreaterpaceinvillageswithgreaterMGNREGAimplementationthaninthosewithlowerimplementation,butthisrelationshipisstatisticallysignificantonlyforwomen.Thedifferenceinthisinteractiontermbetweenmenandwomenisstatisticallysignificantatthe0.001levelinpooledmodels(notreportedhere).ThissuggeststhattheexpansionofopportunitiesduetoMGNREGAdrawsthosewomenintopaidlaborwhomighthaveotherwisecontinuedtoworkonfamilyfarms.OtherresearchbasedonIHDSdatadocumentsthatnearly45percentoftheMGNREGAwomenworkersworkedonfamilyfarmsduringtheprecedingwaveofIHDS. WhileMGNREGAistheprimaryindependentvariableofinterestinthispaper,someoftheothereffectsonparticipationinpaidlaborarealsointeresting.Asdocumentedbytheotherstudies,educationseemstobeassociatedwithlowerparticipationinpaidworkforbothmenandwomeninruralIndia,andasPieterandKlasen(2012)note,itisonlyatthehighestlevelofeducationthatweseewomenbeingpulledintopaidwork.Landownersarefarmorelikelytoworkontheirownfarmsthaninwageandsalarywork.DalitsandAdivasisaresubstantiallymorelikelytoworkasfarmlaborersandmanualnon-agriculturallaborersandindividualsfromtheforwardcastes,andthisrelationshipisparticularlystrongforwomen. Itisimportanttorememberthatthesemodelscontainrandomerrorsforbothvillagesandstates.Theproportionoftotalvarianceexplainedbythevillageofresidenceisfargreaterthanthatexplainedbythestateofresidence,suggestingstronglocaleffectsonlaborforceparticipation.Theplaceofresidencehasafargreaterimpactontheworkparticipationratesofwomenthanonthoseofmen,aresultthatwillnotsurpriseresearchersfamiliarwithstarkdifferencesingendersystemsacrossdifferentpartsofIndia(DysonandMoore1983,JejeebhoyandSathar2001).
[Table4abouthere] Table4presentstherelationshipbetweenvillagelevelMGNREGAworkallocationandthelogofannualwageincomeforallindividualsaged15-64years.Incomesarepresentedin2011-12constantrupeesandsetto0forthosewhodidnotparticipateinwagelabor.Theresultsshowthatwageincomesforbothmenandwomenincreasedbetween2004-05and2011-12,andthatthisincreaseisstatistically
- 14 -
significant.VillageswithgreaterMGNREGAallocationseemtohavehigherwageincomesformenandwomenevenbeforeMGNREGAwasallocated,possiblyreflectinghigherpre-existinglevelsofparticipationinpaidwork.However,theinteractionbetweenthesurveyperiodandMGNREGAallocationinthevillageshowscontradictoryeffectsformenandwomen.TherelationshipbetweenthesurveyperiodandMGNREGAavailabilityisnon-significantformenandthecoefficientisnegative.Incontrast,thewagesforwomeninvillageswithhigherlevelsofMGNREGAworkavailabilitygrewovertime,andthisincreaseisstatisticallysignificantatthe0.001level.Theinteractiontermforgender*survey*MGNREGAavailabilityisstatisticallysignificantinapooledmodel. AnumberofstudiesbasedontheMaharashtraEmploymentGuaranteeScheme,thepredecessorofMGNREGA,aswellasonstudiesbasedontheearlyyearsofMGNREGAimplementation,whichrelyonthephasedroll-outofMGNREGA,suggestthatthepresenceofpublicworksemploymenttightensthelabormarketsandleadstoanincreaseinwages(DattandRavallion1994,ImbertandPapp2011).However,theIHDSfailstosupportthis.Despitesomedisagreement(Schultz1967)mostscholarsoftheIndianeconomysinceB.S.AmbedkarandV.K.R.V.RaohavearguedthatruralIndiasuffersfromdisguisedunemployment(Krishnamurty2008,BhagwatiandChakravarty1969).Ifthisisthecase,publicworksemploymentthatcoversonlypartoftheyearshouldcauseneithertighteningofthelabormarketnoranincreaseinwages.Andreducingdisguisedemploymentshouldnotaffectthemarketlaborsupply.TheaverageincreaseinthehouseholdincomeofRs.4,000fromMGNREGAworkforoneinfourruralhouseholds(Desai,Vashishtha,andJoshi2015)canhardlycreatesubstantialchangesinthewagestructureoftheruraleconomy,norisitsubstantialenoughtoputindividualsaboveathresholdwhereleisureismorevaluablethanwork.
[Table5abouthere] TheresultspresentedinTable5suggestthatahigherallocationofMGNREGAworkraisesmarketwages(excludingMGNREGA)formenbutthisrelationshipisnotstatisticallysignificantforwomen.ThissuggeststhatpresenceofMGNREGAprogramdoeslittletoreducetheavailabilityofwomenforotherwork.Thisisconsistentwiththeargumentthatthereissubstantialunderemploymentamongruralwomenandintroductionofpublicworksprogramsisnotsufficienttoeliminatethisunderemployment.StudiesusingIHDSthatexamineindividualMGNREGAparticipants’workin2004-5and2011-12findthatabout45percentofMGNREGAparticipatingwomenwerenotinpaidlaborbeforetheprogramcameintobeing,possiblybecausedemandforfemalelaborwaslowinthevillage.
[Table6abouthere] Ifthisisthecase,thenwomen’srisingwageincomeinthepresenceofMGNREGAisalmostexclusivelyduetohigherworkparticipationbywomenratherthanrisingwages.Table6suggeststheplausibilityofthisexplanation.Hereweestimatethehouseholdlevelfixedeffectsmodelsfor2004-05and2011-12,andfindthatthoughintheoverallhouseholddecisionmakingprocess,womenarefarlesslikelytobechosentoparticipateinthepaidlaborthenmen,thisnegativeeffectismoderatedinvillageswithgreaterallocationofMGNREGAdays.Villagesthathaveachievedastrong
- 15 -
implementationoftheMGNREGAprogramaredifferentinIHDSWaveI,evenbeforetheActwaspassed,butbyWaveII,therelationshipbetweenMGNREGAallocationandfemaleworkparticipationhadnearlydoubled,withhouseholdslivinginthehighprogramimplementationareabeingfarmorelikelytofavorwomen’sparticipationinwagelabor.Discussion:Inthispaper,wehaveexaminedtheparticipationinwageemploymentandincomesofruralIndianmenandwomeninthepresenceofdifferentlevelsofMGNREGAworkallocationintheirrespectivevillages.TheresultsshowthattheprimaryimpactofMGNREGAimplementationistoincreasewomen’sparticipationinpaidwork,andtherebytoincreasetheirincomes,thoughasimilarimpactisnotfoundformen.Thissuggeststhatthedemandforlabormaybeabiggerconstraintonwomen’sworkparticipationinIndiathanlaborsupply.Thisobservationisbuttressedbythefactthatnearlytwo-thirdsofthewomenwhoarenotcurrentlyemployedreportthatifsuitableworkwereavailable,theywouldbewillingtoworkandtheirfamilymemberswouldnotobjecttothisdecision. Resultspresentedinthispaperraiseabroaderissue.Wetendtothinkofmen’sandwomen’slaborforceparticipationdecisionsasbeingindependent.However,familiesbalancethetimedifferentindividualsspendinmarketandnon-marketactivitiesandinfarming,wagelaborandfamilybusinessinawaythatmaximizesoverallfamilyincome(DesaiandJain1994).Consequently,whenwagesforagriculturalandnon-agriculturallaborersarerising,itmaymakesenseforsomeofthefamilymemberstoparticipateinwageworkandforotherstoconcentrateonworkingonfamilyfarmsandinfamilybusiness.Ifmen’smarketwagesfaroutpacewomen’smarketwages,itwouldmakesensefromafamilyperspectiveforwomentoworkinfamilybusinessandonfamilyfarmsandformentoengageinpaidemployment.However,thisexclusionfromearningindependentincomereduceswomen’sbargainingpowerinthehousehold(Agarwal1997,DwyerandBruce1988)andwhilefamily’soverallaccesstoresourcesmayincrease,women’sowncontroloverresourcesmaydecline.Byensuringequalwagestomenandwomen,MGNREGAcreatedaclimateinwhichhouseholdsarelesslikelytodesignatemenaswageearnersandwomenashelpersinfamilyenterprises. Onecaveatmustbekeptinmind.MGNREGAisademand-drivenprogram.Thehigherthedemandforwork,thegreaterwouldbethelikelihoodoftheprogramtoprovidework.Hence,MGNREGAallocationisnotstrictlyexogenous.Nonetheless,thefactthattheavailabilityofworkmobilizeswomenwhowerenotinthepaidlaborforceinWaveIisquiteanachievement.Itsuggeststhatrespondingtowomen’sneedsbyexpandingworkopportunitiesislikelytomobilizemorewomentoentertheworkforce,therebyincreasingtheirwageincomes.ThisisaverydifferentstoryfromthelaborforcewithdrawalstorythatisbeingtoldonthebasisoftheNSSdata.Itsuggeststhatthedemandforlaborisabiggerbottleneckthantherestrictionoflaborsupplyimposedbyculturalforces.
- 16 -
Inadditiontoaddressingtheroleofpublicworksprograminshapingwomen’sparticipationinpaidwork,thispaperalsoshedssomelightonthemysteryofdecliningfemalelaborforceparticipationratesinIndia.Ourresultssuggestthatalthoughtherearechangesinruralwomen’semploymentinIndia,thenetemploymentisafunctionoftwodivergenttrends.First,ifwemovebeyondthesomewhatrestrictivedefinitionusedbytheNSS,morewomenareparticipatingintheworkforcein2011-12thanin2004-5.However,theirworkismorefragmentedandforshortdurationwhichmaybeeasytooverlookinNSStypedesignthatisfairlystructuredinwhatiscountedaswork.ThedeclinethatNSShascapturedisreal,however.EvenIHDSwithitsmoreexpansivedefinitionfindsthatthereisadistincttrendtowardsdeclineinnumberofdaysworkedinayearforworkingmenandwomen,particularlywomen.ItisclearthatMGNREGAisnotabletoofferthefull100daysofworktohouseholdsandasaresult,whileitbringswomenintothelaborforce,itonlyoffersafewdaysofwork. WorkinProgress: Thisextendedabstractisbasedonworkinprogress.FutureworkonthispaperincludescompletionoflinkagesbetweenadministrativeandsurveydataforTamilNaduandestimationofindividuallevelfixedeffectsmodels.Bothofthesetasksareeasilyfeasible,wehavebeenpromisedcooperationbyMinistryofRuralDevelopmenttohelpwithcompletionoftheadministrativelinkagesandestimationofindividuallevelfixedeffectsmodelsisrelativelystraightforwardandpreliminaryresultssupporttheargumentspresentedabove.
- 17 -
References: Agarwal,Bina.1997."''Bargaining''andGenderRelations:WithinandBeyondthe
Household."FeministEconomics3(1):1-51.doi:10.1080/135457097338799.AgricultureCensusDivision.2014.AgricultureCensus2010-11:AllIndiaReporton
NumberandAreaofOperationalHoldings.NewDelhi:MinistryofAgriculture,GovernmentofIndia.
Alderman,Harold,JereR.Behrman,Hans-PeterKohler,JohnA.Maluccio,andSusanCottsWatkins.2001."AttritioninLongitudinalHouseholdSurveyData."DemographicResearch5:79-124.
Angeles,Gustavo,DavidGuilkey,andThomasMroz.1998."PurposiveProgramPlacementandtheEstimationofFamilyPlanningProgramEffectinTanzania."JournalofStatisticalAssociationofAmerica93(443):884-899.
Bhagwati,JagdishN.,andSukhamoyChakravarty.1969."ContributionstoIndianEconomicAnalysis:ASurvey."TheAmericanEconomicReview59(4):1-73.doi:10.2307/1812104.
Boserup,Esther.1970.Woman’sRoleinEconomicDevelopment.London:GeorgeAllenandUnwin.
Das,Sonali,SonaliJain-Chandra,KalpanaKochhar,andNareshKumar.2015."WomenWorkersinIndia:WhySoFewamongSoMany?"InIMFWOrkingPaperWP/15/55,editedbyInternationalMonetaryFund.WashingtonDC:InternationalMonetaryFund.
Das,Upasak.2015."RationingandAccuracyofTargetinginIndia:TheCaseoftheRuralEmploymentGuaranteeAct."OxfordDevelopmentStudiesDOI:10.1080/13600818.2015.1042445.
Datt,Gaurav,andMartinRavallion.1994."TransferBenefitsfromPublic-WorksEmployment:EvidenceforRuralIndia."TheEconomicJournal104:1346-1369.
Desai,Sonalde.2013."WomeninWorkforce:BurdenofSuccess,DeclineinParticipation."Yojana57(56-59).
Desai,Sonalde,AmareshDubey,B.L.Joshi,MitaliSen,AbusalehShariff,andReeveVanneman.2010.HumanDevelopmentinIndia:ChallengesforaSocietyinTransition.NewDelhi:OxfordUniversityPress.
Desai,Sonalde,andDevakiJain.1994."MaternalEmploymentandChangesinFamilyDynamics:TheSocialContextofWomen'sWorkinRuralSouthIndia."PopulationandDevelopmentReview20(1):115-136.
Desai,Sonalde,PremVashishtha,andOmkarJoshi.2015.MahatmaGandhiNationalRuralEmploymentGuaranteeAct:ACatalystforRuralTransformation.NewDelhi:NationalCouncilofAppliedEconomicResearch.
Dixon,RuthB.1982."WomeninAgriculture:CountingtheLaborForceinDevelopingCountries."PopulationandDevelopmentReview8(3):539-566.doi:10.2307/1972379.
Dutta,P.,RinkuMurgai,MartinRavallion,andDominiqueWanDeValle.2012."DoesIndia'sEmploymentGuaranteeSchemeGuaranteeEmployment."Economic&PoliticalWeeklyXLVII(16):55-64.
- 18 -
Dyson,Tim,andMickMoore.1983."OnKinshipStructure,FemaleAutonomy,andDemographicBehaviorinIndia."PopulationandDevelopmentReview9(1):35-60.
Gaddis,Isis,andStephanKlasen.2014."EconomicDevelopment,StructuralChange,andWomen'sLaborForceParticipation:ARe-examinationoftheFeminizationUhypothesis."JournalofPopulationEconomics27(3):639-681.
Goldin,Claudia.1995."TheU-ShapedFemaleLaborForceFunctioninEconomicDevelopmentandEconomicHistory."InInvestmentinWomen’sHumanCapitalandEconomicDevelopment,editedbyT.PaulSchultz,61-90.Chicago:UniversityofChicagoPress.
---2006."TheQuietRevolutionThatTransformedWomen'sEmployment,EducationandFertility."AmericanEconomicAssociationPapersandproceedings.
Hirway,Indira,andSunnyJose.2011."UnderstandingWomen'sWorkUsingTime-UseStatistics:TheCaseofIndia."FeministEconomics17(4):67-92.doi:10.1080/13545701.2011.622289.
Imbert,Clement,andJohnPapp.2011."Title."CSAEWorkingPaperWPS/2013-03,Oxford.
Jain,Devaki,andNirmalaBanerjee,eds.1985.TyrannyoftheHousehold:InvestigativeEssaysonWomen'sWork.NewDelhi:Shakti.
Jejeebhoy,J.Shireen,andA.ZebaSathar.2001."Women'sAutonomyinIndiaandPakistan:TheInfluenceofReligionandRegion."PopulationandDevelopmentReview27(4):687.
Kabeer,Naila.1999."Resources,Agency,Achievements:ReflectionsontheMeasurementofWomen'sEmpowerment."DevelopmentandChange30(3):435-464.
Kapsos,Steven,EvangeliaBourmpoula,andAndreaSilbereman.2014."WhyIsFemaleLabourParticipationDecliningSoSharplyinIndia?"InILOResearchPaperNo.10,editedbyInternationalLabourOrganisation.Geneva,Switzerland:InternationalLabourOrganisation.
Khera,Reetika,andNandiniNayak.2009."WomenWorkersandPerceptionsoftheNationalRuralEmploymentGuaranteeActinIndia."PaperpresentedattheFAO-IFAD-ILOWorkshoponGaps,TrendsandCurrentResearchinGenderDimensionsofAgriculturalandRuralEmployment:DifferentiatedPathwaysoutofPoverty,Rome,April2,2009.
Klasen,Stephan,andJannekePieters.2012."PushorPull?DriversofFemaleLaborForceParticipationduringIndia’sEconomicBoom."InIZADiscussionPaper6395.Bonn,Germany:IZA.
Krishnamurty,J.2008."IndianAntecedentstoDisguisedUnemploymentandSurplusLabour."TheIndianJournalofLabourEconomics51(1):53-61.
MinistryofRuralDevelopment.2015.MahatmaGandhiNationalRuralEmploymentGuaranteeAct,2005–ReporttothePeople.NewDelhi:GovernmentofIndia.
MinistryofRuralDevelopment,GovernmentofIndia.2013.MahatmaGandhiNREGA,2005-OperationalGuidelines.editedbyDepartmentofRuralDevelopment.
- 19 -
Narayan,Deepa,ed.2006.MeasuringEmpowerment:Cross-DisciplinaryPerspectives.NewDelhi:OxfordUniversityPress.
NationalSampleSurveyOrganisation.2013."KeyIndicatorsofEmploymentandUnemploymentinIndia",NSS68thRound,2011-12.NewDelhi:GovernmentofIndia.
Neff,Daniel,KunalSen,andVeronikaKling.2012."ThePuzzlingDeclineinRuralWomen’sLaborForceParticipationinIndia:ARe-examination."InGIGAGermanInstituteofGlobalandAreaStudiesWorkingPaperNo.196/2012.Hamburg,Germany.
Papola,T.S.2012."StructuralChangesinIndianEconomy:EmergingPatternsandImplications."InISIDWorkingPaperSeries2012/2.NewDelhi:InstitutionforStudiesinIndustrialDevelopment.
Presser,HarrietB.,andGitaSen.2000.Women'sEmpowermentandDemographicProcesses:MovingBeyondCairo.NewYork:OxfordUniversityPress.
Schultz,TheodoreW.1967."SignificanceofIndia's1918-19LossesofAgriculturalLabour-AReply."TheEconomicJournal77(305):161-163.doi:10.2307/2229375.
20
Tables
Table1:ChangeinWorkPatternbetween2004-05and2011-12,MenandWomenAged15-64Years
Men Women2004-05
2011-12
2004-05
2011-12
PercentParticipatingNotworking 17.3 15.6 50.7 46.0Workonownfarm 48.6 50.6 34.6 37.5Workonfamilybusiness 11.5 10.0 3.1 4.0Agriculturallabor 24.8 21.6 17.4 16.6Nonagriculturaldailylabor 20.2 24.6 4.5 3.9Workonmonthlysalary 10.4 11.4 2.5 3.3WorkinMGNREGA 0.0 13.0 0.0 8.7Workedonlyinagriculture(farmerorlaborer) 42.1 32.7 39.7 35.5Workonlyforfamily(onfarmorinbusiness) 32.8 28.7 26.8 27.6AllworkexcludingMGNREGA 82.7 83.8 49.3 52.7AllworkincludingMGNREGA 82.7 84.4 49.3 54.0AverageNo.ofDaysSpentinPrecedingYearperPerson(Population)Workonownfarm 49.5 40.8 26.1 21.6Workonfamilybusiness 25.6 23.7 5.2 8.1Agriculturallabor 36.5 27.5 21.2 17.0Nonagriculturaldailylabor 35.7 44.1 5.7 5.5Workonmonthlysalary 26.9 32.1 4.5 7.4WorkinMGNREGA 0.0 3.8 0.0 2.5Workedonlyinagriculture(farmerorlaborer) 85.7 68.0 47.2 38.5Workonlyforfamily(onfarmorinbusiness) 74.3 63.8 31.2 29.6AllworkexcludingMGNREGA 171.6 165.2 62.4 59.2AllworkincludingMGNREGA 171.6 168.8 62.4 61.6AverageNo.ofDaysSpentinPrecedingYearperParticipantWorkonownfarm 101.9 80.6 75.2 57.5Workonfamilybusiness 223.2 236.7 168.1 204.8Agriculturallabor 147.3 127.5 122.0 102.6Nonagriculturaldailylabor 177.1 179.6 126.4 138.9Workonmonthlysalary 259.2 280.9 183.5 222.6WorkinMGNREGA 29.5 28.5Workedonlyinagriculture(farmerorlaborer) 139.7 111.9 107.3 84.1Workonlyforfamily(onfarmorinbusiness) 134.2 113.3 85.2 73.6AllworkexcludingMGNREGA 207.6 197.2 126.4 112.4AllworkincludingMGNREGA 207.6 200.0 126.4 114.1TotalSampleSize 38949 41053 38629 43113
21
Table2:CharacteristicsofMGNREGAMaleandFemaleParticipantsandNon-participantsAged15-64Years,IHDS-II,2011-12
Males FemalesNon-
participant Participant Non-participant Participant
AnyPaidWork 0.49 1.00 0.19 1.00TotalEarnings 21745 18729 3406 9787No.ofDaysofNREGAWorkintheVillage 10.24 19.15 10.53 22.86Age 34.16 38.61 34.16 39.51MaritalStatusMarried(Omitted) 0.64 0.84 0.73 0.84Unmarried 0.34 0.13 0.20 0.04Widowed/Divorced 0.02 0.03 0.07 0.12
EducationofRespondentNoSchooling(Omitted) 0.20 0.36 0.42 0.691-4Grades 0.08 0.13 0.07 0.085-9Grades 0.39 0.36 0.31 0.1910-11Grades 0.16 0.07 0.10 0.0312thandSomeCollege 0.11 0.05 0.07 0.01CollegeGraduate 0.07 0.02 0.03 0.00
HighestEducationbyAnyAdultintheHouseholdNoSchooling(Omitted) 0.16 0.26 0.19 0.331-4Grades 0.06 0.10 0.06 0.095-9Grades 0.34 0.40 0.34 0.3610-11Grades 0.16 0.10 0.14 0.1012thandSomeCollege 0.15 0.08 0.13 0.07CollegeGraduate 0.14 0.05 0.13 0.04
Caste/ReligionForwardCaste(Omitted) 0.19 0.11 0.19 0.07OtherBackwardClasses(OBCs) 0.37 0.30 0.36 0.37Dalit(SC) 0.21 0.34 0.21 0.35Adivasi(ST) 0.11 0.14 0.10 0.15Muslim 0.11 0.10 0.12 0.04Christian,Sikh,Jain,etc. 0.02 0.00 0.02 0.01
LandOwnershipNoLand(Omitted) 0.39 0.34 0.39 0.41Marginal(<1Hectare) 0.38 0.47 0.40 0.39Small(1-1.99Hectares) 0.12 0.12 0.11 0.12Medium/Large(2+Hectares) 0.11 0.07 0.10 0.08
HouseholdSize 5.84 5.23 5.86 4.93No.ofChildrenintheHousehold 1.50 1.61 1.69 1.52LessDevelopedVillage 0.53 0.70 0.55 0.62SampleSize 36543 4510 39626 3487
22
Table3:AvailabilityofMGNREGAWorkandParticipationinWageLaborResultsfromRandomEffectsLogitModel,MenandWomenAged15-64Years
Variable Men WomenCoeff. SE Coeff. SE
SurveyWave2 0.332 *** 0.020 0.329 *** 0.025No.ofDaysofNREGAWorkintheVillage 0.006 *** 0.001 0.007 *** 0.002Survey*NREGAdays 0.001 0.001 0.009 *** 0.001Age -0.022 *** 0.001 -0.002 * 0.001MaritalStatus(MarriedOmitted)Unmarried -1.681 *** 0.027 -0.703 *** Widowed/Divorced -0.557 *** 0.059 0.132 ***
Education(Noneomitted)1-4Grades -0.109 *** 0.038 -0.121 ** 5-9Grades -0.339 *** 0.029 -0.540 *** 0.03010-11Grades -0.669 *** 0.037 -0.897 *** 0.05212thandSomeCollege -0.669 *** 0.044 -0.538 *** 0.064CollegeGraduate 0.059 0.056 0.515 *** 0.083
HouseholdEducation(Noneomitted)1-4Grades -0.026 0.044 -0.162 *** 5-9Grades -0.142 *** 0.031 -0.438 *** 0.02910-11Grades -0.235 *** 0.038 -0.707 *** 0.04012thandSomeCollege -0.342 *** 0.041 -0.813 *** 0.044CollegeGraduate -0.621 *** 0.048 -0.967 *** 0.053
Caste/Religion(ForwardCasteOmitted)OtherBackwardClasses(OBCs) 0.182 *** 0.027 0.390 *** 0.036Dalit(SC) 0.874 *** 0.030 0.992 *** 0.038Adivasi(ST) 0.769 *** 0.042 1.032 *** 0.049Muslim 0.156 *** 0.043 -0.090 0.057Christian,Sikh,Jain,etc. -0.312 *** 0.068 0.096 0.100
LandOwnership(NoneOmitted)Marginal(<1Hectare) -0.526 *** 0.022 -0.383 *** Small(1-1.99Hectares) -1.133 *** 0.029 -0.754 *** 0.036Medium/Large(2+Hectares) -1.814 *** 0.033 -1.447 *** 0.042
HouseholdSize -0.075 *** 0.005 -0.126 *** 0.006No.ofChildrenintheHousehold 0.093 *** 0.009 0.161 *** 0.010LessDevelopedVillage 0.113 *** 0.034 0.157 *** 0.048Constant 2.143 0.095 -0.796 *** 0.215Level3(State)Variance 0.100 0.034 0.829 0.266Level2(Village|State)Variance 0.217 0.013 0.498 0.027LikelihoodRatioTestvs.Logisticmodel 2590 *** 9578 *** ICCState 0.028 0.179 Village|State 0.088 0.287 SampleSize 79784 81431
23
Table4:AvailabilityofMGNREGAWorkandLogofAnnualWageIncomeResultsfromRandomInterceptModel,MenandWomenAged15-64Years
Men Women
Coeff. SE Coeff. SESurveyWave2 0.727 *** 0.039 0.384 *** 0.030No.ofDaysofNREGAWorkintheVillage 0.010 *** 0.002 0.008 *** 0.002Survey*NREGAdays -0.003 0.002 0.013 *** 0.002Age -0.040 *** 0.002 -0.003 * 0.001MaritalStatus(MarriedOmitted)Unmarried -3.389 *** 0.049 -0.793 *** 0.039Widowed/Divorced -1.103 *** 0.115 0.381 *** 0.048
Education(Noneomitted)1-4Grades -0.154 * 0.072 -0.145 *** 0.0505-9Grades -0.522 *** 0.055 -0.638 *** 0.03610-11Grades -1.100 *** 0.070 -0.857 *** 0.05312thandSomeCollege -1.059 *** 0.083 -0.446 *** 0.066CollegeGraduate 0.478 *** 0.105 0.633 *** 0.093
HouseholdEducation(Noneomitted)1-4Grades -0.031 0.082 -0.258 *** 0.0555-9Grades -0.223 *** 0.058 -0.747 *** 0.03810-11Grades -0.383 *** 0.072 -1.069 *** 0.04812thandSomeCollege -0.601 *** 0.077 -1.139 *** 0.052CollegeGraduate -1.002 *** 0.088 -1.123 *** 0.057
Caste/Religion(ForwardCasteOmitted)OtherBackwardClasses(OBCs) 0.332 *** 0.052 0.285 *** 0.040Dalit(SC) 1.662 *** 0.057 1.108 *** 0.044Adivasi(ST) 1.379 *** 0.078 1.392 *** 0.062Muslim 0.298 *** 0.082 -0.236 *** 0.064Christian,Sikh,Jain,etc. -0.658 *** 0.126 0.428 *** 0.099
LandOwnership(NoneOmitted)Marginal(<1Hectare) -1.269 *** 0.042 -0.631 *** 0.032Small(1-1.99Hectares) -2.494 *** 0.056 -1.076 *** 0.043Medium/Large(2+Hectares) -3.613 *** 0.058 -1.730 *** 0.046
HouseholdSize -0.136 *** 0.009 -0.116 *** 0.007No.ofChildrenintheHousehold 0.181 *** 0.016 0.164 *** 0.012LessDevelopedVillage 0.139 * 0.063 0.237 *** 0.062Constant 9.140 0.171 3.271 0.272Level3(State)Variance 0.319 1.315 Level2(Village|State)Variance 0.760 0.894 ResidualVariance 19.228 11.283 LikelihoodRatioTestvs.Logisticmodel 2450 *** 10275 *** ICCState 0.016 0.097 Village|State 0.053 0.164 SampleSize 79784 81431 Incomesetto0fornon-workers
24
Table5:AvailabilityofMGNREGAWorkandLogofNon-NREGADailyWageIncomeResultsfromRandomEffectsModel,MenandWomenAged15-64Years
Men Women
Coeff SE Coeff SESurveyWave2 0.250 *** 0.006 0.303 *** 0.009No.ofDaysofNREGAworkinvillage -0.003 *** 0.000 -0.001 0.001Survey*NREGAdays 0.001 *** 0.000 -0.001 0.000Age 0.005 *** 0.000 0.002 *** 0.000MaritalStatus(MarriedOmitted) Unmarried -0.087 *** 0.008 -0.072 *** 0.013Widowed/Divorced -0.060 *** 0.018 0.003 0.011
Education(Noneomitted) 1-4Grades 0.051 *** 0.011 0.011 0.0135-9Grades 0.123 *** 0.009 0.040 *** 0.01110-11Grades 0.226 *** 0.012 0.114 *** 0.02012thandSomeCollege 0.283 *** 0.016 0.380 *** 0.025CollegeGraduate 0.484 *** 0.020 0.648 *** 0.032
HouseholdEducation(Noneomitted) 1-4Grades -0.006 0.012 0.000 0.0125-9Grades -0.003 0.009 0.012 0.00910-11Grades 0.023 0.012 0.011 0.01412thandSomeCollege 0.033 * 0.014 0.017 0.016CollegeGraduate 0.231 *** 0.017 0.171 *** 0.021
Caste/Religion(ForwardCasteOmitted) OtherBackwardClasses(OBCs) -0.062 *** 0.009 0.009 0.014Dalit(SC) -0.077 *** 0.009 0.019 0.014Adivasi(ST) -0.082 *** 0.012 0.017 0.016Muslim -0.047 *** 0.014 -0.019 0.022Christian,Sikh,Jain,etc. -0.032 0.024 0.064 0.042
LandOwnership(NoneOmitted) Marginal(<1Hectare) -0.038 *** 0.006 -0.002 0.008Small(1-1.99Hectares) 0.021 * 0.010 0.032 ** 0.012Medium/Large(2+Hectares) 0.078 *** 0.012 0.054 *** 0.016
HouseholdSize 0.004 * 0.002 0.004 0.002No.ofChildrenintheHousehold -0.004 0.003 -0.004 0.003LessDevelopedVillage -0.050 *** 0.013 -0.018 0.017Constant 4.688 0.071 4.320 0.073Level3(State)Variance 0.093 0.031 0.092 0.029Level2(Village|State)Variance 0.037 0.002 0.060 0.003ResidualVariance 0.238 0.002 0.184 0.002LikelihoodRatioTestvs.Linearmodel 10362 *** 4743 *** ICC stateid2 0.253 0.274 idpsustateid2 0.354 0.454 SampleSize 40427 18474 Samplerestrictedtoemployedindividuals.
25
Table6:HouseholdFixedEffectsLogisticRegressionsforMenandWomenAged15-64Years,2004-05and2011-12
2004-05 2011-12
Coef. Std.Err. Coef. Std.Err.Female -2.327 *** 0.034 -2.357 *** 0.030Female*NREGADays 0.011 *** 0.001 0.019 *** 0.001MaritalStatus(MarriedOmitted) Unmarried -1.919 *** 0.044 -1.638 *** 0.036Widowed/Divorced -0.443 *** 0.067 -0.107 0.055
Education(Noneomitted) 1-4Grades -0.096 0.052 0.077 5-9Grades -0.108 ** 0.040 -0.162 *** 0.03410-11Grades -0.082 0.056 -0.443 *** 0.04712thandSomeCollege 0.004 0.072 -0.295 *** 0.054CollegeGraduate 0.699 *** 0.088 0.468 *** 0.069Age -0.011 *** 0.001 -0.005 *** 0.001LikelihoodRatio 11885 *** 13730 ***
DF(10)No.ofHouseholds 12090 13240 Fixedeffectsmodelsareestimatedonlyonhouseholdsinwhichthereisvariationinpaidworkparticipation.
26
AppendixTable1:DistributionofIndependentandDependentVariablesforMenandWomenAged15-64Years
Men Women 2004-
05 2011-
12 2004-
05 2011-
12 Any Paid Work 0.50 0.56 0.23 0.26 Total Earnings (in 2011-12 constant Rs.) 15251 21352 2640 3959 No. of Days of NREGA Work in the Village 10.97 11.40 11.14 11.60 Age 33.66 34.74 33.53 34.62 Marital Status
Married (Omitted) 0.67 0.67 0.75 0.74 Unmarried 0.31 0.31 0.18 0.18 Widowed/Divorced 0.02 0.02 0.07 0.08
Education of Respondent No Schooling (Omitted) 0.27 0.22 0.54 0.44 1-4 Grades 0.11 0.09 0.08 0.07 5-9 Grades 0.37 0.38 0.26 0.30 10-11 Grades 0.13 0.15 0.07 0.10 12th and Some College 0.07 0.10 0.03 0.06 College Graduate 0.05 0.06 0.02 0.03
Highest Education by Any Adult in the Household No Schooling (Omitted) 0.22 0.17 0.24 0.20 1-4 Grades 0.09 0.07 0.09 0.07 5-9 Grades 0.34 0.35 0.34 0.34 10-11 Grades 0.14 0.15 0.13 0.14 12th and Some College 0.10 0.14 0.10 0.13 College Graduate 0.10 0.13 0.10 0.12
Caste/Religion Forward Caste (Omitted) 0.18 0.18 0.18 0.18Other Backward Classes (OBCs) 0.37 0.36 0.37 0.36Dalit (SC) 0.23 0.23 0.22 0.23Adivasi (ST) 0.10 0.11 0.09 0.10Muslim 0.11 0.11 0.11 0.11Christian, Sikh, Jain, etc. 0.02 0.02 0.02 0.02
Land Ownership No Land (Omitted) 0.38 0.38 0.38 0.40 Marginal (< 1 Hectare) 0.33 0.39 0.34 0.40 Small (1-1.99 Hectares) 0.13 0.12 0.13 0.11 Medium/Large (2+Hectares) 0.16 0.11 0.15 0.10
Household Size 6.56 5.76 6.58 5.78 No. of Children in the Household 2.02 1.52 2.17 1.67 Less Developed Village 0.54 0.55 0.54 0.55Sample Size 38,949 41,053 38,629 43,113