How does fertility affect female employment? Evidence from Albania

Authors Andrea Cinque Cecilia Poggi Juna Miluka Claire Guiraud Coordination Cecilia Poggi

Agence française de développement

Papiers de recherche

Les Papiers de Recherche de l’AFD ont pour but de diffuser rapidement les résultats de travaux en cours. Ils s’adressent principalement aux chercheurs, aux étudiants et au monde académique. Ils couvrent l’ensemble des sujets de travail de l’AFD : analyse économique, théorie économique, analyse des politiques publiques, sciences de l’ingénieur, sociologie, géographie et anthropologie. Une publication dans les Papiers de Recherche de l’AFD n’en exclut aucune autre.

Les opinions exprimées dans ce papier sont celles de son (ses) auteur(s) et ne reflètent pas nécessairement celles de l’AFD. Ce document est publié sous l’entière responsabilité de son (ses) auteur(s) ou des institutions partenaires.

Research Papers

AFD Research Papers are intended to rapidly disseminate findings of ongoing work and mainly target researchers, students and the wider academic community. They cover the full range of AFD work, including economic analysis, economic theory, policy analysis, engineering sciences, sociology, geography and anthropology. AFD Research Papers and other publications are not mutually exclusive.

The opinions expressed in this paper are those of the author(s) and do not necessarily reflect the position of AFD. It is therefore published under the sole responsibility of its author(s) or its partner institutions.

Abstract

How does fertility affect female employment?

Evidence from Albania

AUTHORS

Andrea Cinque

Leibniz University (Hannover)

Université Panthéon Sorbonne (Paris 1)

Cecilia Poggi

AFD

Juna Miluka

University of New York (UNYT) Tirana (Albanie)

Expertise France

Claire Guiraud

Expertise France

COORDINATION

Cecilia Poggi (AFD)

This article inspects the relationship between fertility and employment, providing an explanatory mixed methods analysis to gauge their nexus for rural and urban Albania over the 2000s. Instrumenting reproductive decision with having two first born daughters in a 2SLS model of employment, we find that having an additional child influences negatively employment probability for rural mothers, but it does not influence employment decision in urban areas. Rural women are particularly dependent on their fertility decision if they have low levels of education. This effect is not particularly relevant to specific types of rural employment, but it is reinforced by demographic traits of the household, such as if there are seniors in the household or if the partner is working. The analysis highlights that there is a different and relatively worse labour market trajectory experienced by rural women than their urban counterparts. The article then examines qualitatively how structural and contextual settings influence women in their decisions. It inspects the experience of rural women from three distinct rural municipalities, exploring the differential roles that policy could play to favour their insertion

Keywords

Albania, Female employment, Fertility

Acknowledgements

This article is a collaboration between the AFD Research Department (in which C Poggi is research officer and A Cinque engaged as research intern during his PhD) and Expertise France (J Miluka and C Guiraud were main authors of the Expertise France (2021) “Needs Assessement Study” (NAS) report, redacted in partnership with and financed by the AFD). The authors would like to acknowledge the team engaged for the NAS qualitative data collection as well as DHS for sharing the quantitative data. Comments from the IAFFE 2022 Conference participants and Sophie Salomon are also much appreciated. Any result, opinions or errors reflect solely the views of the authors and do not represent the official position of the institutions affiliated to the study or the report

JEL Classification J13, J16, J22, O15

Original version English Accepted October 2022

Résumé

Cet article explore la relation entre la fécondité et l’emploi, proposant une analyse à méthodes mixtes pour évaluer leur lien pour les zones rurales et urbaines en Albanie au cours des années 2000. En instrumentant dans un modèle de probabilité d’emploi (2SLS) la décision de reproduction avec le fait d’avoir deux filles aînées, nous constatons qu’avoir un enfant supplémentaire influence négativement la probabilité d’emploi pour les mères dans les zones rurales, mais ceci n’influence pas la décision d’emploi dans les zones urbaines. Les femmes rurales sont particulièrement dépendantes de leurs décisions de fécondité si elles ont un faible niveau d’éducation. L’effet de la décision de reproduction n ai pas un effet statistiquement significatif pour différents types d’emploi dans le milieu rural. Toutefois, son effet est renforcé par des caractéristiques du ménage, comme la présence des personnes âgées dans le ménage ou si le conjoint travaille. L’analyse souligne que les femmes rurales connaissent une trajectoire d’accès au marché du travail différente et relativement plus défavorable que leurs homologues urbaines. L’article considère ensuite, d’une façon qualitative comment les cadres structurels et contextuels influencent les femmes dans leurs décisions. Il examine l’expérience des femmes rurales et des employeurs de trois municipalités rurales distinctes, en explorant les rôles différentiels que les politiques publiques pourraient jouer pour favoriser leur insertion

Mots clés

Albanie, emploi féminin, fécondité

Remerciements

Cet article est le fruit d’une collaboration entre le Département de la Recherche de l’AFD (au sein duquel C. Poggi est chargée de recherche et A. Cinque a été engagé comme stagiaire pendant son doctorat) et Expertise France (J. Miluka et C. Guiraud ont été les principales autrices du rapport « Needs Assessement Study » (NAS) par Expertise France (2021), rédigé en partenariat avec et financé par l’AFD). Les auteurs tiennent à remercier l’équipe engagée pour la collecte des données qualitatives du NAS ainsi que le DHS pour le partage des données quantitatives. Les commentaires reçus par les participants à la conférence IAFFE 2022 et par Sophie Salomon sont également bien appréciés Tout résultat, opinion ou erreur reflète uniquement le point de vue des auteurs et ne représente pas la position officielle des institutions affiliées à cette étude ou au rapport NAS.

Classification JEL J13, J16, J22, O15

Version originale Anglais

Acceptée Octobre 2022

Introduction

Theliteraturehaslonginvestigatedthe centralquestiontolabourstudiesashow fertilityorreproductivedecisioninteracts withthechoicetoprovideone’sservicesinto thejobmarket.Therehavebeenadvancementsintheunderstandingofhowfertility choice perse isembeddedinapeculiar socio-normative,culturalandpoliticalenvironmentthatmayinfluenceitseffects onemploymentdecisionsovertime.The literatureshowsthat,globally,wherecontraceptivesbecomeorareavailablethere isatrendingdecreaseinfertility(Bongaarts andCasterline,2018).Atthesametime,an increaseinthecontrolofwomenoftheir fertilitydecisionhaspositivewelfareeffects forthem(Balboetal.,2013;Bongaartsand Casterline, 2018).Moreover,overthelast decadesfemaleparticipationinthelabour markethasbeenexperiencinganupward trendworldwide.Ontheoneside,this eventisjustifiedbygrowingevidenceon aclearpositivecorrelationbetweeneducationandemployment,wheregreatereducationachievedbywomendoesnotonlyaffecttheprobabilityofworkbutalsothetype ofworkexperienced(HeathandJayachandran,2016).Ontheotherside,theeconomic andlegislativeenvironmentexperiencedby womenstronglyaffectwhetherjobscan besecured,withresearchshowingthat thelevelanddepthoffamilypolicieslike childcareorfamilyplanninghaveastrong impactonlabourforceparticipation(Bailey, 2006;Miller,2010).

Theemployment-fertilityinteractionvaries acrosscountries,wheredifferencesare identifiedinboththesocialfabricandnormativeinteractionsthatdictatereproductiveoutcomes,aswellasintheoveralleconomicenvironmentexperienced(Michaud andTatsiramos, 2011; FanelliandProfeta,

2021).Therecentliteratureonthelinkbetweenfamilysizeandfemalelaboursupply highlightsthelackofacausalrelationship fordevelopingandtransitioneconomies (Aaronsonetal., 2021; AgüeroandMarks, 2008, 2011).Moreover,inmiddle-income countries,womenarefoundtojuggletheir childrearingandemploymentdecisions withtheoverarchingissueofincomeinequalityexperiencedinthelabourmarket (Finlay, 2021).Whetherthesefindingsare truefortheAlbaniancontextisamatterof empiricalinvestigation.

Contextualisingreproductiverightsandoutcomesisessentialforsupportingwomenin theirchoiceofwhereandwhentowork,as wellasthetermsandconditionsofthatwork (Gammageetal.,2020).Theemploymentfertilityrelationshipcouldprovideimportant knowledgeofthelabourmarketevolutionin Albania,relevanttoguidingpolicymakers inthecontextofgender-transformative policies,bothtowardsgreatergenderparityandwomen’seconomicempowerment. Albaniaisamongthemostconservative countriesinEuropeintermsofgender norms,rankingsecondinthepatrilocality indexamong40lowandmiddle-income countriesincludedintheDemographic andHealthSurveys(DHS)(Grogan, 2018). AfterthefalloftheCommunistregime, whichguaranteedfemaleemployment whilemaintaininggenderrolesinthehousehold(FalkinghamandGjonca, 2001),severalmarketreformslikewageandprice liberalisationsmodifiedthecharacteristics oftheworldofworkinthecountry(World Bank, 2002).Thevoidofsocialprotection andeconomicrightsreinforcedwomen’s homemakingroles(Tarifa,1994)andoccupationalsegregation(EIGE, 2020).Today, awidespreadpatriarchalvisionintheAlbaniansocietyisstillreflectedingender wagegaps(Miluka, 2013),imbalancesin

intra-householdresourceallocations(Betti etal.,2020;Mangiavacchietal.,2018)and strongpreferencesofboysatbirth(Trako, 2019;Grogan,2018;Lerch,2013).

Thisarticlefirstproposesanemployment probabilityanalysisinspectingthefertility decisionviatheeffectofhavingadditional children.Inordertoestablishacausal linkbetweenthetwovariables,weapply aninstrumentalvariableapproachthat identifiesfertilitydecisionviaaninstrument thatindicatesifthefirsttwobornchildrenof womanarefemale.Followingtheliterature (foranextensivereview,seeBhalotraand Clarke, 2022),ourframeworklooksspecificallyatemploymentdecisionofwomen whoarealreadyparentsofmultiplekids. Thus,wedonotaccountintheanalysis forthechildlesspopulation(womenhave alreadymadethechoiceofhavingachild versusnone).However,weimplicitlyassumethatforthechosensampleofwomen withmultiplechildrentheremayhadbeen aquantity/qualitytrade-offinthedecision ofhavingmultiplechildreninthefirstplace (Aaronsonetal., 2014).1 Expanding Trako (2019)’sworkonAlbaniafortheearly2000s, weaskwhethertherehasbeenanysignificantevolutionovertimeacrossrural andurbanareas,butalsohowruralfemale workersandlocalemployersperceivesuch changestoday.Theanalysisusesqualitativelydatacollectedin2020forthree selectedmunicipalities(ExpertiseFrance, 2021).ThesewerechosenfortheExpertise France(2021)reporttoexemplifythemultiplicityofruralmunicipalitiesintermsofsize andproduction.Particularattentionisgiven tostructuralgapsperceivedbywomen,on aspectslikesocialnormsandtheflexibility oflabourmarkets,whicharefoundinthe

literaturetopromotetheconciliationofwork andfamily(Doepkeetal.,2022).

Therearetwodistinguishingfeaturesto ourresearchcontributingtotheliterature ontheemployment-fertilityrelation.First, weconstructouremploymentregressions basedonthelatesttwowavesofDHSdata. Thisisthemostrobustandrecentdataset onhouseholddynamicsandwomenfertility availableinthecontextofAlbania,used toinspectitsevolutionamongruraland urbanareasovertime.Moreover,onthe methodologicalside,wecarefullyaddress potentialendogeneityrelatedtothefertility decisionandselectivityissueswithrespect totheinstrumentanditsvalidityinthe contextofAlbania.Second,weperform anexplanatorymixed-methodsanalysis (CreswellandClark, 2017),makinguseof novelqualitativedatafromfocusgroupdiscussionswithwomenandwithemployers fromthreemunicipalitiesofruralAlbania. Thisapproachprovidesawaytotriangulate thequantitativedataandcorroborateits results.Moreover,itallowstogaugeinformationonthecontextexperiencedby ruralwomenwhenexploringtheagency overtheirwork,theirreproductiveandcare choices.

Wefindthathavinganadditionalchildinfluencesnegativelyemploymentprobability formothersinruralareas,butitdoesnot influenceemploymentdecisionofthosein urbanareas.Wefindthatindividualcharacteristicsmatter,asruralwomenemploymentisdependentontheirfertilitydecision particularlyiftheyhavelowlevelsofeducation.Theliteratureshowsthatmothers mayworkmoreintheinformalsectoronce theirfamilysizeincreases(Schmieder,2021),

1Aaronsonetal.(2014)inspecttheextensiveandintensivemarginsofafertilitytransitionmodel.Asthepriceof investinginchildrenreduceswithanadditionalchild,theirmodelpredictsincreasedinvestmentinchildrenanda declineinthechancesofhavinganadditionalchild.Moreover,theliteraturesuggeststhathavingmorethanone childresultsingreaterinter-generationalsupportforseniorparents(seeOliveira2016foranempiricalapplicationto China).

butwedonotfindanystatisticallyrelevant effectforspecifictypesofemployment thatcouldleadtoeitherinformalormore unstablejobs.Nonetheless,wefindthatthe negativerelationshipismaintainedonce weexploreendogenousdemographictraits ofthehousehold,suchasifthereareseniors livinginthehouseholdorifthepartneris currentlyworking.Theanalysishighlights thatthereisadifferentandrelativelyworse labourmarkettrajectoryexperiencedby ruralwomenthantheirurbancounterparts. Thus,weexplorehowstructuralsettings affectthecontextualsurroundingswhere ruralwomenmaketheiremploymentdecisions.Usingqualitativedata,weanalyse theexperiencereportedbyruralwomenin threedistinctruralmunicipalitiesofAlbania, exploringthedifferentialrolesthatpolicy couldplaytoreducebarrierstotheirinsertion.

Asperthefindingsofthefocus-groups,the absorptionofruralwomeninthelabour marketislimitedbythetypesofavailable jobs,themajorityofwhicharestilloffered intheurbanareas.Womenandemployers reportlackoftraining,skillsmismatch,and oftenlackofinformationregardingjobsearchandemploymentservices.Rural womenmaketheirdecisionsonverydiversetrajectories vis-à-vis urbanwomen. Amongthemainfactorsidentifiedbythe focusgroupsastheinhibitingcontextual settingofwomen’semploymentintherural areastherearethelackofchildcareand inadequatecoverageofchildcareduring

workinghours.Moreover,longdistancesto jobsorlowremunerationalsoplayabig part.Womenreportasdeterminingfactors theperceptionofdeterioratingworkingconditions,thelackofpublictransportationand itshighcosts,aswellasthereinforcingof socialnormstowardstheprimaryroleof womenascaregivers.Theseallimplydecreasingthetrade-offsbetweenthechoice ofhavingchildrenandworking.These findingsarealsosubstantiatedbysimilar studiesconductedontheprofileoflongtermregisteredunemployedjob-seekers withemploymentservices(UNDP,2021)and gender-baseddiscriminationandlabourin Albania(GADC,2022),whichidentifysimilar issuestowomen’sbarrierstoemployment andlabourmarketaccessinAlbania.

Thearticleisstructuredasfollows.Section 2 belowdescribesbrieflythehistory andlegislativecontextthatsurroundsthe analysis.Section 3 presentsthedata,the variablesusedanditdescribestheestimatingsample.Section 4 proposesthe identificationstrategyanddiscussesthe choiceoftheinstrumentforfertility,addressingthethreatstointernalandexternal validityoftheempiricalstrategy.Theresults arepresentedanddiscussedinSection 5,inwhichthequantitativefindingsare followedbythequalitativeanalysisforthe threemunicipalitiesunderstudyandanindepthcomparisonoffemaleworkersand employersviewsinruralareas.Finally,the lastsectionconcludes.

DuringthefiftyyearsoftheCommunistregime,anegalitariansystemwasputinplacein Albania,guaranteeingfullemploymentandthuschangingoutcomesforwomenoutside thehousehold,whilestillkeepinginplacethegenderrolesinsidethehouseholdunaffected

(FalkinghamandGjonca,2001).Afterthefallofcommunism,theinitialstageofthetransitional periodwascharacterisedbyatimeofshockseliminatingtheexistingsocialsupportsystem, revitalisationoftraditionalvalues,liberalisationreforms,andamassiveoutflowofmigration (Carlettoetal.,2006).Inaddition,themarketreformsthatfollowedincreasedearninginequalitiesthroughwageandpriceliberalisations,andchangedthecharacteristicsofemployment andtheworldofworkinthecountry(WorldBank,2002).Thevoidofsocialprotectionand economicrightsreinforcedwomen’shomemakingroles(Tarifa,1994).Consequently,women wereburdenwithmoreunpaidworkwithinthehousehold,butlessmobilityandchances tofindjobs.Thedegreeoffeminisationinthetraditionalfieldssuchaspublichealthand educationinheritedfromthecommunistsystemcontinuedtoincreaseinthepost-communist years.Todate,occupationalsegregationandsegregationofwomeninparticularfieldsof educationsuchaseducation,healthandwelfare,humanitiesandartsremainhigh(EIGE, 2020).Aderivativeoftheincreasedfeminisationwasthesocialdevaluationofthesejobsin termsofwages(VullnetariandKing,2016),whichpersiststoday(Miluka,2013).

Intheearly2000s,roughlyadecadeafterthefalloftheCommunistregime,largegapsexisted betweenwomenandmenintermsoflabourforceparticipationandemployment.In2003, therewas23.8percentpointsgapinlabourforceparticipationrate(70.5percentofworking agemenparticipatedinthelabourforcecomparedto46.7percentofwomen)(AMLSA, 2006).Likewise,in2004,therateoflabourforceparticipationrateformenwas68.6percent, whileforwomen46.4percent(AMLSA,2006).Employmentstatisticsshowthatin2004,the employmentratewas38.3percentforwomenand60.1percentformen.Gendergapsin labourforceparticipationandemploymentratescontinuetopersistandbeconsiderable, buttheyhavebeenslowlyreduced.Thelabourforceparticipationrateduringthistimehas averagea17.3percentagepoints,withthelatestratesfor2021of61.4percentforwomenand 77.3percentformen(INSTAT,2021).Employmentgapsinthelastdecadehaveaveraged13.9 percentagepointswiththelatestratesfor2021of53.8percentforwomenand68.2percent formen(INSTAT,2021).Likewise,womenweremainlyconcentratedinthesocial-state-service sector,where80.0percentofemployeesarewomen(AMLSA,2006).In2019,thepercentage ofemployedwomenineducation,humanhealthandsocialworkactivitieswas13.8percent comparedtoonly3.9percentformen(EIGE,2020).

Overthelastdecade,therehasbeenscatteredacademicevidenceproducedonthe determinantsofemploymentdecisioninAlbania.Estimatesfortheearly2000sshowthat, forparentsofatleasttwochildren,thereisapositiveeffectoffertilityonparentallabour supplyforyoungerlesseducatedparentswhomainlyliveinextendedfamilies(Trako,2019).

2

However,thisevidencewarrantsarevaluation.Albaniahasbeendisplayingsincethelast twodecadestheinstitutionalandpoliticalwillforputtingforwardconsiderableinterventions, demonstratedbyvariousreformsandongoingpledgestoimprovefemaleparticipation totheeconomy,thusananalysisofthemostrecenttrendsiswarrantedtobringscientific evidenceattheserviceofpolicymaking.

Itslegalframeworkongenderequalityisquitecomprehensive,earningAlbaniaascoreof

2Trako(2019)putstogetherdifferentsourcesofdatatoconductananalysistill2012.Thestudyfindsthatmothers increasedtheirlaboursupplyandhadagreaterlikelihoodtoworkoff-farm.Forfathers,theirlikelihoodtowork off-farmalsoincreased,asdidtheprobabilityofhavingasecondoccupation.Tothisend,theremightbetwo mechanismsinplacedrivingtheresults,suchaschildcarebeingprovidedbynon-parentaladultsinextended families,andgreaterfinancialcostsbecauseofbearingmorechildren(Trako,2019).

90.9percentfromtheUNWOMENGlobalSustainableDevelopmentGoals(SDG)Database, whichmonitorstheSDGsworldwide(UNWOMEN,2020).Therearevariousimportantpiecesof legislationthatprotectwomenagainstdiscrimination,suchastheLawonGenderEqualityin Society(No.9970,dated24.07.2008),whichspecificallyaimstoguaranteeprotectionfrom genderdiscrimination,aswellastheLawonProtectionfromDiscrimination(No.221,dated 4.2.2010),whichfurtherextendsthescopetoincludemanygroundsofdiscrimination.The “NationalStrategyonGenderEquality2021-2030”alsopromotesgenderequalityinmany fieldsoflifeincludingemployment,andequalinclusionofwomeninnon-traditionalfieldsof employment,aswellaseducationinthosefields,alongwithpromotingequalityinunpaid labourandequalsharingofhouseholdresponsibilities,aswellasequalparticipationand representationinpoliticalandpublicdecision-making.The“NationalEmploymentandSkills Strategy2019-2020”alsoaimstofostergenderequalityinemploymentandskillformation,as wellasgreaterinclusionandterritorialcohesiontobetterincludewomenfromruralordistant areas.Thestrategyacknowledgestheneedtobetterserveruralwomenwithemployment servicesandvocationaleducationandtraining(VET).In2019,therevisedLawNo.15/2019 “OnEmploymentPromotionintheRepublicofAlbania”specificallyincludedwomenvictims oftrafficking,womenvictimsofgender-basedviolence,andwomenvictimsofdomestic violenceaseligiblebeneficiariesofemploymentpromotionprogrammesofsubsidisedwages, on-the-jobtraining,andinternshipprogrammes.The2019lawwaspartofthereformof employmentservicesfocusingonredesigningtheActiveLabourMarketProgrammes,their delivery,andmonitoringandevaluation(inordertoimprovetheirimpactandcompatibility withthelabourmarketandprogrammesavailableintheEuropeanUnion).TheAlbanian LabourCodealsoprovisionsforgenderequalitydemandingequalpayforequalvaluework, aswellasincludingspecialprovisionsfortheprotectionofpregnantwomen,andprovisions againstdiscriminatoryhiringpractices.Inanefforttopromoteequalityinunpaidcarelabour andchildcareresponsibilities,thelatestamendmenttotheLabourCodealsoprovisionedfor therightofbothparentstodemandparentalleave,butstatisticsatthetimeofwritingare unavailable.Specificinstitutionstoprotectagainstdiscriminationarealsoinplace,suchas theCommissionairefortheProtectionAgainstDiscriminationandtheOmbudsman.

Lastly,itisimportanttohighlightthatfamilypolicyinitiatives,likechildcaresupport,mayplay anessentialroletodeterminefutureevolutionforfemaleactivelabourmarketparticipation. Thereexiststodatebothformsofpublicandprivatechildcareprovisioninthecountry.The publicchildcareprovisionincludescrèchesfortheage0-3andkindergartensatage3-6. Bothcrèchesandkindergartensaresubsidisedbythegovernment,andparentshavetopay afeeforthemealprovisions.Thisfeeisofabout1.1USDperdayforthecrècheand1.36USDfor thekindergarten(MoT,2022).Particularlyforpreschool,thereisalowleveloffinancingfrom thepartofthestate(UNESCO,2017),andtothebestofourknowledgenorecentreformhas yettakenplacetointroduceanyschoolmealprogrammeortorevisetheoveralllocationof institutesandavailability.Giventhepresentlegislativeframework,itisimportanttoenquire howwomenperceivetheirfertility-employmentrelation,tofurtherguidepolicymaking.

Data

Weadoptanexplanatorymixed-methodsapproach(CreswellandClark,2017),inspecting cross-sectionaldatafromtheDemographicandHealthSurvey(DHS)forAlbania,aswellas qualitativedatacollectedasfocusgroupdiscussionswithwomenandemployersfromrural Albania(ExpertiseFrance,2021).

Asprimaryquantitativedata,weusetwowavesoftheAlbanianDHScollectedrespectively in2008/09andin2017/18.DHSisanationallyrepresentativesurveyofwomen,menand householdsondemographic,healthandothersocio-economiccharacteristics.Thesurvey collectedbetween2008and2009interviewsof7,584womenandof3,013meninthe15-49 agegroup.Between2017and2018,DHSinterviewed10,860womenand6,142men.Wefollow theliterature(AngristandEvans,1998)inrestrictingthesampleofanalysistothepopulation ofwomenagedatleast20yearsofageandupto35yearsold,andreportinghavingall childrenbelowtheageof18(2,997observationsovertwowaves).

Wemakeuseofindividuallevelinformation,suchasnumberofchildren,employmentstatus inthelastsevendaysandinthelasttwelvemonths,typeofoccupationandstabilityofjobs. Wealsoinspectalargesetofcontrols,suchasanormalisedhouseholdwealthindexbased onhouseholdassets,age,education,religion,ethnicgroup,urban/ruralstatus,numberof adulthouseholdmembersandothercharacteristicsrelatedtothefertilityexperience,suchas ageatlastbirthandsexandageofthechildren.Table1displaysthesummarystatisticsfor oursample.Halfofthesampleisresidinginruralareas,has10yearsofcompletededucation (morethanprimaryeducation)andamajorityisMuslimofreligion.Almostallthesampleis currentlymarriedandbyconstruction,thenumberofchildrenisabove2.Onaverage,halfof thesamplehadamaleastheirfirstchild,suggestingnoreasonstobelievethereisgender selectionatfirstbirthinthissample.Theaverageageis31yearsoldandtheageatfirstbirth is22.Regardingwork,the32percentofwomeninthesampleworkedoverthelastweek(or similarly38%havebeenworkinginthelast12months).Moreover,ineachhouseholdthere aremorethantwoadultsactivelyengagedinanoccupation.

Moreover,thearticlemakesuseofanovelqualitativesurveycollectedin2020aspartofa reportbyExpertiseFrance(2021)assecondarydataforperformingacomparativeanalysisto inspectrurallabourmarketdifferencesinthreeareasofruralAlbania.3 Weusethequalitative dataforthreemunicipalitiesofLushnjë,ElbasanandKorcë.SixFocusGroup(FG)discussions wereset-uptoinspecttheconditionsofruralwomenparticipatinginthelabourmarket andtheperceptionofemployersfromthesameareas(participantswereidentifiedwith thesupportoftheNationalAgencyforEmploymentandSkillsorNAESlocalofficesandthe municipalities).ThethreeruralwomenFGdiscussionswereconductedinSeptember2020 gathering12womeninElbasan,15womeninKorcëand12womeninLushnjë.Womeninvitedto jointheinterviewsacrossthethreeselectedmunicipalitiesdepictabalancedrepresentation intermsoflivingareas,ageandlevelsofeducation,andsituationtowardsincome-generating activities.Moreover,threeemployerFGswereconductedinSeptember2020with9employers

3TheNASreportwasconductedbetweenFebruary2020andFebruary2021,byateamledbyExpertiseFrancein closerelationwiththeNAES,itslocalofficesinthreeselectedmunicipalitiesandthethreelocalmunicipalities’teams. TheNASreportalsobenefitsofadeskreviewandofinformantinterviewsatmunicipallevelperformedbetween 2020and2021.

inElbasan,9employersinKorcëand12employersinLushnjë.4 IneachFGdiscussion,a semi-structuredinterviewplanwasusedtoguidetheexchanges.

Thestructureoftheinterviewwasmainlyfocusedonidentifyingaccesstopublicemployment services,accesstoservicessupportingincome-generatingactivitiesforwomen,accessto decentworkandnon-discriminatoryworkplace,accesstochildcareandsupportingsocial services,accesstomobilityandtransportation,andaccesstoagender-equalitysupporting environmentfreefromadversesocialnorms,stereotypingandviolence.Themunicipalities accountforpossiblestructuraldifferencesinthepoliticaleconomyofruralareasinAlbania. Eachmunicipalitywaschosenasasignificantshareofitsterritoryisrural,buthasadistinct economicprofile(encompassingsectorsofeconomicactivitysuchasagriculture,tourism, etc.)andshowsatthepoliticallevelastrongwilltoaddresswomeneconomicempowerment issues.

SituatedinWest-centralAlbania(seeFigure 1),Lushnjëisthesmallestinsizeofthethree municipalities,lessdevelopedadministrativelyandintermsofinfrastructures(government servicesorpublictransports)thantheothermunicipalitiesduetoitssmallsize,agriculture accountsforapproximately60percentofemployment.Korcëisintheeast;itsagricultural employmentisat50percentandhasahigherlevelofsocialservicesofferedandnumberof womengroups.ThemunicipalityofElbasanislocatedinthecentre-northofthecountryand ithasmorethan60percentofemploymentinagriculture.Nonetheless,itisslightlybetter intermsofpolicyengagementandpublicinfrastructuresthantheothertwomunicipalities, influencedbyitseconomicandpoliticalclosenesstothecapital,Tirana.

4. Identificationstrategy

Thebaselineestimationtomeasuretheimpactofoffspringsizeontheemploymentofmothers ofmultiplechildrenisasfollows:

Where

tobeemployed.

fertilitydecision(AgüeroandMarks,2008,2011;Heath,2017)and β1 isthecausalparameter ofinterestrepresentingthelabourmarketimpactofamarginalbirth.Threevectorsfor individualcharacteristics(λilt),location(µl)andtime(τt)controltheheterogeneityofthis relationship.Themodelinequation1couldpossiblysufferfromendogeneityduetoomitted variablesbiasandreversecausality.Preferencescouldbemitigatedbyexpectationsonlife expectancy,familysizeand/orlabourmarket/careerprofiles.Theycouldalsobemediated bysocial,culturalandeconomicconstraints(RosenzweigandSchultz,1985;Oliveira,2016; Gammageetal.,2020).5 Amongvariousinstrumentsusedintheliteraturetoovercomethe

4TheemployeridentificationwasconductedwiththesupportoftheNAESlocalofficesandmunicipalitiesseeking forabalancedrepresentationofcompanysizeandsector(agricultureandagricultural-processing,manufacturing, servicesindustry,foodproduction,hospitalityandtourism.)

5Asadescriptiveexample,ambitiouswomencouldhavehigheropportunitycostsofbearingchildrenandthus

identificationissue,weinspectindetailbelowtheuseofsame-sexinstrumentfortheAlbanian context.6

4.1. Instrumentidentificationandassumptions

Wefollow AngristandEvans (1998), Trako (2019), AgüeroandMarks (2011)and Ebenstein (2009)inusinganInstrumentalVariable(IVorTwoStagesLeastSquares,2SLS)estimatorfor fertilitydecisionwiththegenderofthefirsttwochildrenasasourceofexogenousvariation toexplaindecisionoverchildbearingwithoutdirectlyaffectinglabourmarketparticipation. AngristandEvans(1998)whofirstintroducedthisIVshowthatwomenwithtwosame-sex childrenwillbemorelikelytohaveathirdchild(orhigherorderbirth),andthevalidityof theinstrumentisensuredasthegenderofafuturechildisrandomlyassigned.Toapply thisidentificationtooursetting,wenotethetraditionalvaluestowardsoffspringrearingin Albania.SimilartoKoreaandotherAsiancountries(Schultz,2001,2007),thereisastrong preferenceforboys.Thisimpliesthatseveralhouseholdscoulddecidetohaveanadditional thirdchildincasethefirsttwochildrenaregirls7

Toinvestigatethishypothesis,weinspectthesignificanceandmagnitudeofthefirststage regressioninTable2,questioningwhethertheinstrumentrepresentedassamesexofchildren orseparateinstrumentsbyspecificgenderoffirstbirthshaveadirectinfluenceonthenumber ofchildren.Wealsoassesstherelevanceofaddingabinaryvariableformalefirstchild (orsecondchild).Thisbinaryvariableisusedtocontrolforthosehavingamalefirstborn, whichisknowntopossiblyaffectsubsequentfertilitybehaviourbyreducingthelikelihoodof additionalchildbearing(DahlandMoretti,2008).Thiscontrolvariableisalsorelevanttoan employmentequation,asitwouldcaptureanydirectinfluenceoftraditionalvaluestoward thesexofoffspringthatmayleadwomentodedicatemoretimetothetaskofrearingason (Schultz,2007).Thetablerevealsthat,asreportedintheliterature,havingsamesexchildren increasesthenumberofchildrenceterisparibus,evenaftercontrollingforthegenderoffirst (andsecond)orderchildtobeamale(column1-3).Introducingseparatelyabinaryvariable forthefirstsame-sexbirthsbygender(column4)aseithertwomalechildrenortwofemale children,thecoefficientsshowthatfertilitypreferencealtersthenumberofchildrensolelyin theeventoftwofemalefirstbirths.Moreover,theassociationbetweenhavingafirstmale childandlowernumberofchildrendisappearsassoonasthegender-specificindicators voluntarilychosetohavefewerofthem.Inthiscase,ambitionispositivelycorrelatedwiththeprobabilitytobe employedandnegativelywiththenumberofchildren, β1 wouldbebiasedupwards,sothattheOLScoefficientis overestimated.Ontheotherhand,somewomenmayhavestrongpreferencesforhavingchildrenbutmightface liquidityconstraintstosustainthem,thelatterbeingalleviatedwhenemployed.Inthiscase,havingajobreduces thecostsofhavingchildren,leadingtoadownwardsbiastotheestimation.Althoughitisdifficulttosaywhich effectsprevailapriori,wecouldexpectthat,foranemergingeconomylikeAlbania,womeninurbanareasmaybe morelikelytobeconstructingtheirpreferencesinthefirstscenario,wherelabourmarketismorediverse,wages arehigherandtherepossiblyisgreateravailabilityofcontraceptives.Onthecontrary,womenfromruralvillages mightbemorelikelytoformtheirpreferencesetinenvironmentscharacterisedbystrongergendernormsandlower income,thereforeinthiscaseadownwardbiascouldprevail.

6WefurtherassesswhetherothercommonIVsfromtheliteraturewouldbeausefulrobustness,butweacknowledge thattheobservationsaretoofewintheDHSsampleatourdisposaltomakemeaningfulcomparisonsforself-reported infertility(AgüeroandMarks,2011)ormiscarriage/stillbirth.

7TheLATEconditionsofthe2SLSestimatorimpliesthatfertilitydecisionistobeinterpretedonlyattheintensive margin(howmanychildrentohave,giventhecostofinvestinginthem).Thismeansthattheexternalvalidityofthe analysisappliesonlyforthesampleofwomenwhoalreadyhavemadethedecisiontohaveachild(Angristand Evans,1998).Thissubsetofthepopulationhowevercorrespondsto80.71%ofwomenaged20+inAlbania.

areintroducedintheregression(column5).ThetablethusconfirmsthatforAlbaniafertility ismoreresponsivetounmetgenderpreferencesofwomenwithtwogirls.Moreover,the indicatorforamaleoffspringdoesnotseemtosignificantlyalterthenumberofchildrenhad. Inthesample,26.7percentofwomenwithmultipleoff-springshavehadtwodaughters.

Weusethetwofemalechildrenindicatorvariableasmainspecification,keepingmalefirst childbinaryasadditionalcontrol(column6,binaryvariablewithmean0.47).Eitheradding separatelyorjointlyasecondordermalebirthindicator(withmean0.49)doesnotimprove themodelF-statisticofthefirstsstageregression(column7-8).Amajorissuewithsuchan instrumentwouldbeifthegenderofthefirsttwochildreniscorrelatedwithmaritalpreferences. Itcouldhappenforexampleincasewomenhaveinterruptedpregnancies(directalteration ofbirth)orgotdivorcesduetounmatchedgenderpreferences(indirect).Wedonotfind anystatisticalevidenceinself-reportedmeasuresacrossthesamplethatthegenderofthe firstchildatbirthtendstobepredominantlymale(seeTable1)noranydisproportionalrate ofdivorcedwomenwithoutkidsorwithafemaleasfirst-born(andonlychildattimeofthe survey-notshown,butavailableonrequest).

WecompareinTable3thedifferenceinmeansforthosewithoutsame-femalegenderof twokidsorwith(column1and2respectively).Wearereassuredthatnostatisticaldifference existsindemographicsorothercontrols,exceptforhavinggreaternumberofchildren(the instrumentedvariable)andhavingamalefirst-bornorlastbornchild(whichbyconstruction willbelowerforhouseholdswithtwofemalechildren).WereportintheAppendix(TableA1, p.37)areducedformequationshowingthereisnodirectinfluenceofourinstrumentonthe likelihoodofemploymentoverthesample.

Empiricalstrategy

characteristicsthatcouldaffectsimultaneouslychildbearingandemploymentdecisions.

andhousehold-levelcontrols λicmt suchasrural/urbanresidence,ahouseholdwealthindex, yearsofeducation,numberofadultsinthehousehold,marriagestatus,age,age2 ,whether thefirst-bornisamaleandageatfirstbirth.Asfertilityisrelatedtosocialandculturalnorms, wealsointroducereligionandethnicgroupfixedeffects(sixandsevengroupsrespectively), astheycouldaccountforculturalheterogeneityandtraditionalvaluesthatarecorrelated withpreferencesforoffspring’sgender.Allregressionsabsorblocalandtimeheterogeneity withmunicipalityandsurveyfixedeffects(61municipalitiesintwoperiods).Finally,standard errorsareclusteredatDHSclusterlevel.

Theanalysisisperformedforasub-sampleofthefemalepopulationagedbetween20and 35withmultiplechildrenbelow18yearsofage(AngristandEvans,1998).Buildinguponthe literature(AngristandEvans,1998;AgüeroandMarks,2011;Ebenstein,2009)wealsoperform theanalysisseparatelybyurbanandruralareasandbyeducationlevels,sotoaccountfor underlyingstructuraldifferencesthatcoulddrivetherelationshipbetweenfertilitydecision andemploymentprobability(AngristandEvans,1998;AgüeroandMarks,2011).9

5. Results

Table 4 displaysresultsofthecorrelationbetweenfertilityandemploymentprobability inAlbania,reportingestimationforadependentvariabledefinedasbeingemployedin thelast7days(Columns1-3)andhavingbeenemployedinthelastyear(Columns4-6).

Column1(4)showsanegativecorrelationbetweenthenumberofchildrenandthelinear probabilitytobeemployedlastweek(lastyear)fortheentiresampleofwomenwithatleast twochildrenbelow18.Intermsofmagnitude,havingoneadditionalchilddecreasesthe probabilitytobeemployedby7%,orareductionof2.5percentagepointsfromameanof 32%ofwomenemployedlastweek(Column1).Column2(5)andColumn3(6)showthe regressionresultssplitbyrural/urbanresidenceofhouseholds.Inlinewiththeliteratureon fertilityandemploymentdecision(Aaronsonetal.,2021),wefindlargernegativeandmore significantcoefficientsforwomenresidinginwealthierurbanareas.Thisimpliesthatwomen incitieshaveaccesstobetter-remuneratedjobs,sothetrade-offbetweenthechoiceof havingchildrenandworkingshouldbelarger.

AlthoughthecoefficientsinTable4areindicativeofasubstitutionbetweenhavingchildren andbeingemployed,theeffectfromsuchalinearmodelcanbebiased.Forthisreason,we applya2SLSstrategy,withthefirststageresultspresentedinTable5.Acrossallcolumns, havinghadtwogirlsincreasesthemeannumberofchildrencomparedtomixed-gender births.AsinTable4wealsoincludereligionfixedeffectsandethnicityfixedeffectsinorderto ruleoutculturalheterogeneityintheoptimalchildrennumberandespeciallyinthedecision beusedforcomparisonwithmulti-countryDHSstudies(seeAgüeroandMarks(2008,2011)).

9AngristandEvans(1998)andAgüeroandMarks(2011)showthatthebiasfromoverestimationisreducedwithan increaseineducationlevelsandforlow-incomecountries.Forextremelylow-incomecountriesandlow-educated women,theyfindthatOLScouldevenunderestimatethechildbearingpressureonlaboursupply.Furthermore, Ebenstein(2009)comparestheeffectofchildbearingcostsonemploymentbetweenTaiwanesewomenandUS women,usingpreferenceformalechildrenasaninstrument.Hefindsthat,contrarytopreviousliterature,OLS underestimatetherealcoefficient.Heexplainsthisresultsbythemagnitudeofthefirststagecoefficient:thelowest theeducationandstrongerpreferenceforboys,thelargestthenumberofchildrenandthelargesttheeffectof childrenonemployment.

tohaveafurtherchildafterhavinghadtwogirls.Furthermore,largercoefficientsandastrong F-statisticsforthesampleofruralhouseholdssuggeststhatthepreferenceforboymay belargerintheseareasthaninthecities.Acrossallspecifications,theKleibergen-Paap F-statisticiswellabove10,indicatingthattheinstrumentisrelativelystrong.

Table6displaysthesecondstage’scoefficientsofthe2SLSmodel.Thedependentvariablein Columns1to3isemploymentlastweekandColumns4-5usesemploymentinthelast12 monthsasanoutcome.TheresultsinColumn1and4,usingtheentiresample,showthatthe negativerelationshippersists.Interestingly,oncewesplitthesamplebyrural/urbanareas, theresultsareconcentratedinruralareas,whereasforurbanareasthisisnotstatistically significant.TheIVcorrectsforthesimultaneousandendogenousdecisiontoenterthe labourmarketandhavechildren.Asexpected,thisupwardsbiasiscorrectlyreducedfor urbanareas,sothatthecoefficientsfortheurbansampleinTable6isnolongerstatistically differentthanzero.Thisimpliesthatfertilitydecisionhasnosignificantimpactonlabour marketparticipationforAlbanianwomenfromurbanareas.Ruralareashavestrongnegative coefficientsofhavinganadditionalchildonemploymentprobability,andthiscouldbedue toadifferenttypeofbias:aliquidityconstrainteffectcouldprevailandadownwardbias influencestheOLSresults.Forthisreason,theIVcoefficientsintheruralsamplearelarger inmagnitudethanOLS.Despitecorrectingopposingdirectionofbias,theeffectsoffertility onemploymentprobabilityinruralareasislargerandstilldrivestheresultoftheoverall sampleinColumn1and4.10 Wethusexploreindetailsinnextsectionpotentialmechanisms thatexplaindivergenceinresultsbetweenruralandurbanareas,lookingatheterogeneityin women’scharacteristicsandarounddifferenttrendsovertime.

5.1. Mechanismsandheterogeneityanalysis

WeinspectinTable7whetherthereisanyheterogeneityintheanalysisaccordingtolevel ofeducation.Therationalebehindthisisthattheremightbeadifferentattachmentto thetrade-offbetweenemploymentandotheractivitiesaccordingtotheemployment prospectsavailabletoaperspectiveworker.Wetheninspectthepopulationwithmore thanprimaryeducation(Col.1,forasamplecombiningwomenwithachievedsecondaryor tertiary,respectively26%and11%ofthesample).Thereisnodirectimpactofthefertilityon thedecisionoftheprobabilityofemploymentandinsteadthegreaterthenumberofyears ofage,thehighertheprobabilityonbeingemployed.Whereas,lookingatthe63%ofthe samplewithprimaryorlower(Col.2),theprobabilityofemploymentisstronglynegatively affectedbythefertilitydecisionaswellbythepresenceofafirst-bornchild.Moreover,the Tablesuggeststhatthisresultissolelydrivenbytheruralsample(Col.4).

Thenweinspecttheheterogeneityacrosswaves.Therecouldhavebeenseveralstructural modificationinboththeeconomicenvironmentaswellthesocio-normativeinteractions forwomenso,albeitacknowledgingitmightbesensitivetoareducedstatisticalpower,we

10Werunabatteryofrobustnessanalysestocorroboratethemainresultsfound.First,wemodifythemaindefinition ofemploymentquestionelicitedintheDHS,showinginTableA3thattheheadlineresultsareunalteredifweinspect theemploymentlinearprobabilityoverthelast12months(Columns1-3).Wefurtheralsoreportthemainequation replicatedwithalogformfortheinstrumentedvariable,surelyreportingthesameresultsbutthistimeexpressedas anelasticity(Columns4-6).

inspectseparatelyeachcross-sectiontoverifytheestimationpersistsacrosstime(Table A4).Foryear2008/09theestimationrevealthattheruralsampledoesindeedcapturethe negativeeffectthatfertilityhasonemployment,asopposedtonoeffectsinurbancontexts (whereasrestrictingfurtherthesampletoonlyruralwomenwithprimaryeducationalters theoverallstatisticalpowertowardszero,Col4).Thenegativecorrelationbetweengreater fertilityandreducedemploymentisstillpresentinthewave2017/18(Col.5).Interestingly, reducingthecross-sectionstotheruralsamplewithprimaryeducation(Col.8)revealsthat thefertilitydecisionstillisastrongdecidingfactorforruralwomenwithloweducationintheir employmentdecision.Thisresultsuggeststhatthereisrelevancetobetterunderstandhow ruralwomenexperiencebarrierstoemployment,inspectingboththestructuralaswellas contextualfactorsthatmayimpingeruralwomenfromattendingajob.

Aslastheterogeneityanalysis,weexplorethetypeofemploymentthatruralandurban womenengagewithinthedataset.Femalesresidinginruralareastendtoexperiencea loweremploymentparticipationthantheirurbancounterparts,butstructuraldifferences existalsointermsoftypeofjob.TableA5showsthatwomeninruralareasaremuchmore likelytoworkintheagriculturalsector(+53%),asself-employed(+11 2%)orforafamily member(+29.2%),oftenwithoutremuneration(+38%).Atthesametime,theyaremore engagedinseasonaloccupations(+32.8%).Giventhecharacteristicsofthejobsavailable tothem,womeninruralareascanreactintwowaysafterthebirthofachild.First,they leaveoutthelabourmarketastheopportunitycostofworkingisrelativelylowandwomen facestrongersocialpressuretocareofchildren.Second,theycontinueworkingasinformal andagriculturejobsallowstomoreflexiblycareofchildren(Aaronsonetal., 2021).The statisticsandmainemploymentresultsfoundpointtowardsthefirstoption,howeverwhen weperformananalysisofthespecificemploymentprobabilitiesbytypeofoccupationwe findinsignificantimpactsoffertilitydecision(Table A6,wherethedependentvariableis employmentinaspecificoccupationoverthelastyear).Wenoticeneverthelessthatfor theruralwomensampleunderanalysis,inwave2017/18thereisahigherprobabilityfor beingself-employedaswellasworkingforsomeoneelsethanin2007/08,butnosignificant differenceineitheremploymentinanunstablejoboragriculture.11 Whereasforurbanwomen thereisareducedprobabilityin2017/18ofbeingemployedinanunstablejob(occasionalor seasonal),possiblyconfirmingthattheytendtohavegreateraccesstoqualityoccupations overtime.12

Wethusinspectsomepossiblechannelsdrivingtheresultsfound:endogenoushousehold characteristicsthatcouldaffecttheliquidityconstraintsofwomen,butalsochangesinsocial normswithinthehousehold.

11Table A8 comparesemploymentmeansbytypeovertime,showingbetween2008and2018adropof41.5 percentagepointsofwomenworkingforfamilymembersinruralareas.Thisdecreaseisonly9.7p.cinurbanareas. Likewise,morewomenworkstablyinruralareas(+17.1pc,against+6.3inurbanareas)andtheyaremorelikelyto getpaid(+32.3pcinruralareas,+0.3pcinurbanareas).

12Figure A1 confirmsthedescriptivestatisticsofTable A7 and A8 byregressingemploymentoutcomeswitha timeandruraldummy.Thefigureshowsthatwomeninruralareashaveoverallworseemploymentconditions thaninurbanareas,butthegapislargelyreducedin2018.Theconvergenceinthejobmarketbetweencitiesand countrysidecouldexplainthenullresultsforruralareasin2018ofTable9.Theopportunitycostsofchildbearing increasesasthejobmarketimproves,thusruralwomenarelesslikelytobeaffectedintheirlabourmarketdecision in2018than2008withanadditionalchild.

Endogenoushouseholdcharacteristics

Weexplorehowdomesticcharacteristicsinfluencefertilityinitseffectsontheemployment decision.InTable8weinvestigateasampleofwomenthatareeitherlivinginahousehold whereanyseniormemberaged60+ispresentorthosewithout(PanelA,columns1-2for thefullsample,3-4forruralareas).Theinterestbehindsuchspecificationisthatwomen withchildrentendtobetheonesengagedinprimarycareofdependentmembers(children andold-ageindividuals).Thepresence(absence)ofsomebodyinretirementagecould provokedifferenteffectsontheemploymentdecisionofaworking-agewoman.Onthe oneside,thepresenceofanelderlymembercouldalleviatetheburdenofchildcareon thewoman,thushavinganadditionalchildcouldbefreeingsomeofhertimeallocatedto household/offspringcare,whichcouldbedivertedtowardsgreateremployment.Onthe otherside,thepresenceoftheelderlymembercouldbeinitselfanadditiontothecareneeds ofthehouseholds,whichwouldnegativelycorrelateherfertilitydecisionwithheremployment decision.Table8revealsthatthelattertraitseemstoprevailforthesample:overallrestricting thesampletohouseholdseitherwithorwithoutseniorslivinginthehouseholdskeepsthe negativefertility-employmentrelationship.However,thesituationdifferswhenwelookat ruralareas(whereismorecommonforhouseholdstolivewithmultiplegenerationsundera sameroof).Ifruralwomenliveinanuclearhousehold(seeCol.4),theiremploymentdecision isnotstatisticallysignificantlyinfluencedbythefertilitydecision,suggestingthisgroupmight bedrivenbyothertraits,liketheirlevelofeducationdirectlyinfluencingjobopportunities availabletothem.

Moreover,inPanelBofTable8weinvestigatewhetherforyear2018thereareanyemployment probabilitydifferentialsamongwomenwhosepartneriscurrentlyemployedandthosewitha partnernotemployed(PanelB,columns5-6forthefullsample,7-8forruralareas).Theresults againprovethatthehouseholdcompositionmattersinguidingtheeffectofreproductive decisionstowardsemployment.Ifthepartnerisemployed,thisshouldimplyanadditional sourceofrevenuerelievingthehouseholdliquidityconstraints,thuswomencouldaffordto trade-offworkinthepresenceofanadditionalchild(Col.5and7).However,intheabsence ofanadditionalrevenuefromthehusband’semployment,thereisapositivebutstatistically insignificantprobabilitytoengageinanemploymentactivityduetoanadditionalchild.

5.1.2. Inspectingproxiesofhouseholdgendernorms

Apossiblechannelinfluencingdifferencesbetweenruralandurbanfemalebehaviourinthe labourmarketmightberelatedtogendernormsandaspirations.Intheabsenceofquestions relatingtotheroleofwomeninchildbearingandemployment,weattemptanexercisefor proxyinghouseholdgendernormsintwoways:ifthemalepartneristhemaindecision-maker withinthehouseholdandtheidealnumberofchildren.Figure2ashowsthatmalepartnersin ruralareasdonotnecessarilyhavelargerfinancialdecisionsthaninurbanareas,butthey haverelativelylargersayinginwomenhealthcareandingivingthepermissiontovisitstheir relatives.However,thegapbetweentheareasislargelyreducedin2018.Onceweanalyse directlyhowhouseholdchangetheirfertilitypreferencesovertime,2bshowsthathouseholds inruralareasin2018reducewhattheyindicatetheidealnumberofchildren,especiallyboys,

controllingfortheiractualnumberofchildren.Thesecoefficientsarealsoindicativeofthe instrumentpowersimilarlytoEbenstein(2009):largerpreferencesfortheidealnumberof boysaremorelikelytoincreasetheprobabilitytohaveanadditionalchildifthefirsttwo bornonesarefemale.However,theseresultsdonotexplaintheoveralldifferenceinthefirst stagecoefficientsbetweenruralanurbanininTable5asthecoefficientsinFigure2barenot statisticallydifferentbetweenruralandurban.Ontheotherhand,theymightbeapossible driveroftheoveralldecreasefoundinthefirststagecoefficientsovertimeinTable9.

5.2. Learningfromruralwomen

5.2.1. Overallfindingsfromfocusgroups

Thefocusgroupdiscussionsofwomeninthethreeselectedmunicipalitiessharedvarious concernsregardingemploymentservices,childcareservices,training,infrastructure,social norms,perceiveddiscrimination,knowledgeoftheirrightsintheworkplace,etc.whichaffect theirlivelihoods,andlabourmarketstatus.Womeninthethreemunicipalitiessharedthe difficultiesofemploymentservicesreachingruralwomen,andbelievethatmoreeffortsshould beundertakentoprovideservicestoruralwomenlocatedindistantareas.Theycitedthat employmentserviceswerelocatedfarfromtheiradministrativeunits.Theinabilitytoreach employmentservicesalsolimitstheinformationavailabletothemregardingemployment services.Inaddition,limiteddigitalskills,lackofinternetaccess,lackofcomputers,and oftensharingacellularphonewiththewholefamily,furtherimpedesthemfromfindingout informationregardingemploymentservices.Jobsearchingisthereforemainlyconducted throughfamily,friends,andsocialadministratorsatthemunicipality,ratherthanemployment services.Long-termunemployedwomeninthemunicipalityofKorcëalsonotedadistrustin theabilityofemploymentservicesinfindingthemjobs.InthemunicipalityofLushnjë,women notetheworkdonebysocialadministratorsintermsofinformationsharingandbelieve thatcollaborationbetweensocialadministratorsandemploymentofficeswouldbemore beneficialintermofinformationsharingandjobsearches.

Accesstodecentworkingconditionsisalsoamajorconcernforwomen,whichoftenledto rejectionofjob-offersorlabourmarketinactivity.Aconsensusexiststhatahighproportion ofjob-offersprovidelowwagesandinadequateworkingconditions.Womendeclaredthat low-skilledjobsoftenofferedwagesbelowthelegalminimumwage,aswellasunpaid extrahoursofworkreachingtwelvehoursperdaywithnodaysoff,aswellasunavailability oftransportandchildcare.Deterioratingworkingconditionsareespeciallyreportedfor fishprocessingjobs,duetoinadequateworkingconditions(includingcoldtemperaturein theworkingenvironment,highhumidity,etc.).Whenaskedregardingtheirlabourrights stemmingfromthelegalprovisions(includingworkingconditions,payforovertime,abilityto takevacations,maternityleave,reportingoffenses,etc.),womenintheFGsreportthatthey haveverylittleknowledgeoftheirrightsandthewaytoreportoffenseswhentheirrightsare violated.Consequently,womenreportthattheyhaveneverfiledanycomplaintsorreported anyviolationsandtheyareunawareoftheproceduresforreportingcomplaintsorviolations. Theyalsoreporttohavenoknowledgeoftheirrightsduringpregnancy.

Thereisalsoperceiveddiscriminationintermsofgender,age,ethnicity,anddisability.Women reportedthattheyperceivethatemployersprefertorecruitandpaymenbetter.Inwomen’s viewemployersconsidermenmorestableandproductiveinemploymentbecausethey arenotresponsibleforparentalduties,donottakematernityleaveorhaveotherfamily obligations.Somewomenreportdifficultiesinfindingajobaftertheageof40years,andthey believethatemployerspreferyoungerwomen,becausetheyareviewedasmoreproductive. Somewomenreportdifficultiesinfindingajobafterage40.WomenfromtheRomaand Egyptiancommunitiesalsoreportperceiveddiscriminationduetotheirethnicitymakingit hardforthemtofindemployment.Likewise,womenwithdisabilitiesfeelthatpublicpolicies donotsupportthemandexperiencealackofjobopportunitiesthatareappropriatefortheir situation.

Giventheperceivedlowwagesandhardworkingconditionsinformaloccupations,women oftenresorttoeitherinformaljobs,ortoreceivingeconomicaidthattheywouldloseifthey enterformalemployment.Somewomenreportedthattheyhavebeenemployedininformal jobsinagricultureandtextileindustrywhilebeingregisteredasunemployedjobseekersso thattheywouldnotlosetheireconomicassistance.Womenandtheirhouseholdsperceive economicaidasstableandguaranteedincomeovertime,whileemploymentisperceived onlyinashorttomedium-termandisnotseenasasatisfactorysourceofincome.Assuch, womenalsoreportfamilypressureinkeepingeconomicaid.

Theissueofchildcareisastrongcommonalityamongallwomeninthefocusgroupsacross municipalities.Lackofchildcareisidentifiedasastrongreasontorejectjoboffersorcontinue tostayinunstableemploymentconditions.Childcareservicesofferedbythestatearelimited incapacityandtheyareincompatiblewithworkinghours.Therefore,familymembersget involvedtohelp,mainlythroughgrandmothers.However,thistypeofchildcareismainly short-lived,sincegrandmothersoftengiveuptheirrolesascaregiversespeciallyconsidering theirageandhealthproblems.Intheverylimitedinstanceswhenacompanyoffersnursery places,theworkinghoursdonotallowfortendingtochildren’sneeds.Inthissense,women andtheirchildrenneedtobereadyby5amtobeontheworkingsiteat6amtocontinue workuntil6pmtoarrivehomeat7pm.

Thelackoftransportationisyetanotherbarrierfacedbywomenandespeciallyruralwomen livingindistantareas.Whereasmostjobopportunitiesareoutsideoftheruralareas,public transportofferisverylimitedforruralareaswithnopublictransportincertainareasor limitedhoursoftransportationendingat5pm.Thisconstraintisalsoidentifiedbythe government,whichhasprovisionedfortransportationvoucherstotheworkplacethroughits 2019employmentpromotionlaw(LawNo.15/2019.“OnEmploymentPromotion”).However, thereisnoavailabledataonwhetherthisprovisionhasbeenroll-outyetnorontheprocedures forenforcingit.Additionally,transportationcostsarehighrelativetoincomes.Lastly,women inthefocusgroupshighlightedthepersistenceofstrongpatriarchalnormswherewomen’s employmentisstillconsideredasan‘obstacle’tothewell-beingofthefamily.Husband’s migrationalsoleaveswomenasthesolecaregiverinthefamilywithoutsufficientincome forthehousehold.Eveninthehouseholdswherewomen’semploymentisconceivable,itis neverthelesscommonforwomentogoforajobinterviewwiththeirhusbandormother-inlaw,whooftenasktheemployeraboutworkingconditionsorschedules,andsometimesmake thefinaldecisionwhethertoacceptthejob.Certainjobsthatexposewomentomenortothe

publicaresometimesprohibitedbythehousehold(suchascleaner,baker,hairdresser,etc.).

Thesefindingsarealsosubstantiatedbyaninvestigationontheprofileoflong-termregistered unemployedjob-seekerswiththeAlbanianemploymentservices(UNDP,2021).

Ontheemployer’sside,acrossthethreemunicipalities,theyconfirmeddifficultiesinrecruiting womenfromruralordistantareas.Assuch,theypointedoutthattherearevariousunfilled vacanciesandrejectionsfromthejobseekers.Morespecifically,employersinallthree municipalitiesfacedifficultiesinrecruitinginthesectorsoftextileorfoodprocessing, particularlyinskilledpositions.Thisisawidespreadsentimentthattheinterviewedfirmsreport existingacrossvarioussectors,asoftenrecruitmentagencieshavedifficultyinproviding suitableprofilesthatwouldmatchthewageproposedwiththework-dayrequirements (intermsoftransportandtime).Employersalsostatedintheirfocusgroupsthatthey perceivewomentopreferotheroptionsthanindustryjobs,suchaseconomicassistanceand involvementinsmallagriculturalactivities.Consequently,employersreporthavingsometime resortedtorecruitdifferentprofiles,likeworkersfromNorthAfricaortheMiddleEast.

Thus,thereseemstobeamismatchedperceptionoveremploymentopportunitiesthat createsafrictionbetweenwomen’sofferinruralordistantareasandemployers’demand. Womenreportsometimeschoosingeconomicaidoveremploymentbecauseofthelow wages,thelackoftransportationorchildcare,andduetoperceiveddeterioratingworking conditions.Employersfromthefocusgroupsperceivethisaswomen’sfavouringeconomicaid overemployment,withoutnecessarilyunderstandingthereasonsbehindit,andconsequently substitutewomen’semploymentforotherprofiles.Thisinturnaddsanadditionalbarrierto women’semployment.

Althoughemployersadmittofavourrecruitmentofworkersfromurbanareas,someemployers believethatruralwomenperformbetterbecausetheyadaptmoreeasilyandaremorelikely toworkindifficultworkingconditionsthanwomeninurbanareasare.Employerstoorecognize thedifficultiesintransportationandhightransportationcost.Theyalsoconsiderasapriority toinformwomenoftheirrightsanddutieslinkedtoemploymentcontract.However,onthe employers’side,thereareoftenrepeatedviolationsofthecontractualobligationsonthe sideofthewomenworkersthatcausesfinancialburdentotheemployer.Employersalso recognizetheneedforchildcareandtransportationforwomen.However,theyalsobelieve thattheyareinsufficienttoovercomethesocialbarrierstowomen’semployment.

Theemployer’sperspectivedifferedfromthatofwomenintermsofperceiveddiscrimination. Someemployersreportthatwomenoutperformmen.Theyalsostatethatthereisno discriminationbasedonage,onthecontrary,theyconsidereasiertorecruitwomenolderthan 40yearsofagesincetheydonottakematernityleaveandhavefewerfamilybarriers.

Regardingemploymentservices,employersfromtheFGsidentifytheheavygeneralbureaucracyasabarrieroftheirparticipationinemploymentpromotionprogrammes,whichprovide subsidizedemploymentandon-the-jobtrainingforvulnerableregisteredunemployed job-seekers,includingwomen.Inaddition,employersconsiderthattheindividualskills assessmentoftheregisteredunemployedjobseekersandjobintegrationpreparationbythe NAEScouldbestrengthened.Employersreportedthatwhenwomenareinformedabouttheir jobpositionandtheirtasks,theyadaptbetteranddemonstrateagreaterdeterminationto effectivelyandprofessionallyperformtheirtasks.Employersalsonotedthatthecurriculaof

vocationaltrainingareoutdatedintermsoftechnologiesandtools.Theypointedoutthat, duetothemismatchintheVETcurriculawiththeneededpracticalskills,theyneedtoprovide athree-monthtrainingforthenewrecruits.

5.2.2. Comparisonbetweenquantitativeandqualitativeresults

Theanalysisnowdigsintothespecifictraitsofrurallabourmarketsandparticularfocus onethreecharacteristicruralareasthatwesurveyedqualitativelyin2020(ExpertiseFrance, 2021).Wefirstshowthatthereisasignificantquantitativedifferencebetweenurbanandrural areasacrossimportantdimensionsindividuatedinqualitativeinterviews.Table10showsthat therearedifferencesregardingwomen’semploymentstatus,distancetowork,household decision-makingroles,andenrolmentofchildreninpre-schoolinanefforttocaptureneed forchildcare.Onaverage,women’semploymentishigherintheurbanareascomparedto ruralareasasindicatedbythedifferencesonthestatusofemploymentofwomeninthe lastweek.Furthermore,walkingtoworkismorefrequentforwomeninurbanareasasare shorterdistancestoworkcomparedtowomenlivinginruralareas.Thisalignswithlonger distancestoworkforwomenlivinginruralareas,andtheneedfortransportationtoreachtheir workplace.Consistentwithmorepervasivepatriarchalsocialnormsinruralareas,decision makingbyhusbandsonearningsandlargepurchasesislargerinruralareascompared tourbanareas.Lastly,thereisahigherpercentageofchildreninpreschoolinurbanareas comparedtoruralareas.

WeinspectinTable11threeaspectsthatmightbeessentialtofurtheradvancetheinclusivenessofthefemalelabourforceinruralAlbania:thestructurallimitationsofbeingworkingin ruralareas,thesocio-culturalnormsthatwomenliveandpracticeaswellastheservices attheirdisposaltoeffectivelymanagework-lifebalancewhenchildrenareinvolved.For eachofthesedimensions,wecomparequalitativelytheresultsfoundinthefocusgroup discussionstoDHSdata,bothatthenationalandtocountiesinwhichtheNeedAssessment Studyhadbeenconducted.DatafromtheDHSshowthatoverallwomen’semploymentislow inallthreemunicipalities.Elbasanhasthelowestreportedrateof20.6percent,compared toLushnjë(39.9percent)andKorcë(36.8percent).Womenreporttopredominantlywalk toworkwhichvariesbetween72.7percentforElbasan,78.2percentforLushnjë,and81.5 percentforKorcë.Onaveragewomenreporttowalktoworkforabout17minutesinLushnjë, 18minutesinElbasanand21minutesinKorcë.Genderrolesappeartobemorepervasivein Lushnjë,whichhasthehighestreportedratesofhusbanddecidingonearning(22.9percent comparedto12.0percentforElbasanand8.3percentforKorcë)andhusbanddecidingon largepurchases(19.4percentcomparedto6.6percentforElbasanand6.5percentforKorcë. Thelackofchildcareservicesmaybesymptomaticofthelowreportedratesbywomenof childrenattendingpreschool.Lushnjëhasthehighestreportedratesbywomenofchildren attendingpreschoolat58.6percentversus40.6percentforElbasanand45.4percentfor Korcë.

TheDHSfindingsinTable11supporttheissuesreportedqualitativelybywomen’sfocusgroups inthethreemunicipalities.Thebarriersidentifiedbythewomeninthefocusgroupsoflong distancestoworkforwomenintheruralareasandthelackorlimitedschedulesofpublic transportationisalsosustainedintheDHSdata.Itshowsthatwomenusewalkingasthe

mainmeanoftransportationtoworkforlongerdistancesintheruralareas.Furthermore, theidentificationbywomeninthefocusgroupsofthelackofchild-careservicesisalso reflectedintheDHSdata.Aspresentedabove,therearelowlevelsofreportedpreschool attendancebymothersinthethreemunicipalities,whichdoesnotgobeyond60percent,and thatmaypointtowardsunavailabilityofchild-careespeciallyinruralordistantareas.Lastly, thereportedsocialnormsbywomeninthefocusgroupswheretheemploymentdecisionis ofteninfluencedortakenbythehusbandsandmotherinlaws.Thesehouseholdmembers arealsoreportedattendingthejobinterviewstoassessthequalityoftheworkenvironment,a notioninlinewiththereportedlevelsofdecisionmakingamongmembers,inwhichhusbands decidepredominantlyonearningsandlargepurchasesinthethreemunicipalities.Such barriersidentifiedthroughtheDHSsamplereinforcetheevidencefoundforAlbania(GADC, 2022).

Conclusions

Thisarticleexploresthenexusbetweenfertilitydecisionandemploymentprobability,using latestdataavailableforruralandurbanAlbania.Ouranalysisshowsthatemployment probabilityinruralareasisdivergingthanurbanareas,andthatthefertilitydecisionisnot thesameinitsimpactonemploymentprobabilities.

Applyinga2SLSmodelwhereweinstrumentthenumberofchildrenhad,wefindthathaving anadditionalchildinfluencesnegativelyemploymentprobabilityformothersfromrural areas,butitdoesnotinfluenceemploymentdecisionofthoseinurbanareas.Exploringthe traitsofthetwolabourmarkets,weascertainthatjobshavegrownfasterinurbanareasthan ruralinabsorbingwomenlabourforce.Evenifbetween2007/08and2017/18therehasbeen asignificantimprovementinthetypesofoccupationsthatruralwomencouldaccess,they stillmaketheirdecisionsonverydiversetrajectories.

Moreover,wefindthatpersonaltraitsmatterinexplainingthisnexusforAlbania,asrural womenemploymentisdependentontheirfertilitydecisionparticularlyiftheyhavelowlevels ofeducation.Aspervariationacrosstypesofemployment,wedonotfindanystatistically significanteffectoffertilitydecisionforspecifictypesofemployment.Wehoweverfindthat contextualorendogenouslydefinedcharacterslikedemographictraitsofthehousehold domatterforthisdecision.Thepresenceofseniorslivinginthehouseholdorthecurrent employmentstatusofthepartnerreinforcethenegativeassociationbetweenfertilityand employment.

Wethenexplorehowstructuralsettingsaffectthecontextualsurroundingswhereruralwomen maketheirdecisions.Usingqualitativedata,weanalysetheexperiencereportedbyrural womenandemployersinthreedistinctruralmunicipalitiesofAlbania,exploringthedifferential rolesthatpolicycouldplaytoreducebarrierstofemaleinsertion.Consistentwiththe2SLS findings,focusgroupsintheruralareaspointoutthatpresenceofgrandmothersinthe householdisanunreliableandoftenshort-termsolutiontochildcare,substantiatingwomen’s roleascaregiverswithinthehousehold.Socialnormsarequitepervasiveinruralareas, affectingintra-householddecision-makingregardingwomen’semployment.Lackorcostly transportationandlongerdistancesforruralwomenindistantareasalsopresentabarrierto women’semployment.Thetrade-offbetweenchildcareandreceivinglowwagesinunstable jobswithdeterioratingworkingconditionsoftenworksagainstwomenacceptingemployment. Thereseemstobeaviciouscircleparticularlyforruralwomenlivingindistantareas,inwhich theyreporthavinglimitedaccesstoinformationandemploymentservices,aswellaslackof childcareandtransportation.Moreover,womenreportfacingsocialnormsthatmainlysee theirroleascareproviderswithinthehousehold.Beingconfrontedwithjobofferswithlow wages,indistantareas,andwithunsatisfactoryworkingconditionsoftenleadwomentorefuse suchjoboffers.Localemployersperceivethisasapreferenceofwomenforeconomicaid ratherthanworking,thereforetheyreportseekingforothertypesofprofiles,suchasmigrant work.Thefindingonmismatchedperceptionsoveremploymentopportunitieshighlights thedifferentialrolesthatpolicycouldplaytofavourruralwomen’sinsertioninthelabour market.

Inlightofthesefindings,thereshouldbeincreasedeffortsbypolicy-makerstoprovide

integratedemploymentandsocialservicestowomen,andespeciallywomeninruralareas thatareeitherunawareoftheseservicesorunabletoreachthem.Theseparationandlack ofcoordinationbetweenemploymentservicesandsocialservicescreatesanadditional burdenonwomen,increasingtheirtransactioncoststoseekemployment.Coveragewith employmentservicesinruralordistantareasshouldalsoexpandinordertoincrease outreachofthese(vulnerable)womenwhooftengoundetected.Furthermore,therole ofthemunicipalitiesandinvolvedstakeholdersshouldbestrengthened.Furthersupport shouldbeprovidedtosocialadministrators,whoconductthegroundworkandhaveextensive knowledgeoftheircommunitiesandvulnerablegroups,inordertoprovidewideroutreach services.Expansionofpublictransportationwithadequateschedulingandaffordablecosts shouldalsobeconsidered,especiallyforruralareas,whichcontinuetobemoreisolated. Thesamestandsforchildcareservices,theavailabilityofwhichislackinginruralordistant areas,andwheretheoldbeliefthatseniormembersofthehouseholdintheseareasmake upforthelackofservicesnolongerstands.Althoughfurtherevidenceisneededtoassess thelatterissue,thereseemstobetheneedformorechildcareservicestobeclosertothe workplace,forexamplewithinitiativesstrengtheningpartnershipsbetweenemployersand municipalities.

References

Aaronson,D.,Dehejia,R., Jordan,A.,Pop-Eleches, C.,Samii,C.,andSchulze, K. (2021).TheEffectof FertilityonMothers’Labor SupplyovertheLastTwo Centuries. TheEconomic Journal,131(633):1–32.

Aaronson,D.,Lange, F.,andMazumder,B. (2014).Fertilitytransitions alongtheextensive andintensivemargins. AmericanEconomic Review,104(11):3701–24.

Agüero,J.M.andMarks, M.S. (2008).Motherhood andfemalelaborforceparticipation:Evidencefrom infertilityshocks.InAmericanEconomicReview,volume98,pages500–504.

Agüero,J.M.andMarks, M.S. (2011).Motherhood andfemalelaborsupply inthedevelopingworld: Evidencefrominfertility shocks. JournalofHuman Resources,46(4):800–826. AMLSA (2006).National StrategyforGenderEqualityandViolenceintheFamily2007-2010.In Albanian MinistryofLabour,Social Affairs,andEqualOpportunity

Angrist,J.D.andEvans, W.N.(1998).Childrenand TheirParents’LaborSupply:EvidencefromExogenousVariationinFamily

Size. AmericanEconomic Review,88(3):450–477.

Bailey,M.J.(2006).More PowertothePill:TheImpactofContraceptiveFreedomonWomen’sLifeCycle LaborSupply*. TheQuarterlyJournalofEconomics, 121(1):289–320.

Balbo,N.,Billari,F.C.,and Mills,M. (2013).Fertility inadvancedsocieties:A reviewofresearch. EuropeanJournalofPopulation, 29:1–38.

Betti,G.,Mangiavacchi, L.,andPiccoli,L. (2020). Womenandpoverty:insightsfromindividualconsumptioninAlbania. ReviewofEconomicsofthe Household,18(1):69–91.

Bhalotra,S.andClarke, D. (2022).Analysisof twins,version18august 2022. forthcomingin “HandbookofLabor, HumanResourcesand PopulationEconomics”

Bongaarts,J.andCasterline,J.B.(2018).Fromfertilitypreferencestoreproductiveoutcomesinthe developingworld. PopulationandDevelopmentReview,44(4):793–809.

Carletto,C.,Davis, B.,Stampini,M.,and Zezza,A. (2006).A countryonthemove: Internationalmigrationin post-communistalbania.

InternationalMigration Review,40(4):767–785.

Creswell,J.W.andClark, V.L.P. (2017). Designingandconductingmixed methodsresearch.Sage publications.

Dahl,G.B.andMoretti, E. (2008).Thedemand forsons. TheReview ofEconomicStudies, 75(4):1085–1120.

Doepke,M.,Hannusch,A., Kindermann,F.,andTertilt,M. (2022).TheEconomicsofFertility:ANew Era.NBERWorkingPapers 29948,NationalBureauof EconomicResearch.

Ebenstein,A.(2009).When istheLocalAverageTreatmentClosetotheAverage? JournalofHuman Resources,44(4):955–975.

EIGE(2020).Genderequalityindexreportforalbania.

ExpertiseFrance (2021). NeedsAssessmentStudy -AccesstoEconomic OpportunitiesforWomen LivinginUncoveredRural Areas.Technicalreport, ExpertiseFranceand AgenceFrançaisede Développement,(May 2021)Tirana,Albania.

Falkingham,J.and Gjonca,A. (2001).

Fertilitytransitionin communistalbania,195090. PopulationStudies, 55(3):309–318.

Fanelli,E.andProfeta,P. (2021).Fathers’InvolvementintheFamily,Fertility, andMaternalEmployment: EvidenceFromCentraland EasternEurope. Demography,58(5):1931–1954.

Finlay,J.E. (2021). Women’sreproductive healthandeconomic activity:Anarrativereview. WorldDevelopment, 139:105313.

GADC (2022).GenderbasedDiscriminationand LabourinAlbania,Second Edition.Technicalreport, GenderAlliancefordevelopmentCentre(GADC), Tirana,Albania.

Gammage,S.,Joshi, S.,andvander MeulenRodgers,Y. (2020).Theintersections ofwomen’seconomic andreproductive empowerment. Feminist Economics,26(1):1–22.

Grogan,L. (2018).StrategicFertilityBehaviour,Early ChildhoodHumanCapital InvestmentsandGender RolesinAlbania.IZADiscussionPaper,No.11937.

Heath,R. (2017).Fertilityatwork:Childrenand women’slabormarketoutcomesinurbanghana. JournalofDevelopment Economics,126:190–214.

Heath,R.andJayachandran,S.(2016).Thecauses

andconsequencesofincreasedfemaleeducation andlaborforceparticipationindevelopingcountries.WorkingPaper22766, NationalBureauofEconomicResearch.

INSTAT (2021).EmploymentandUnemployment Statistics.

Lerch,M. (2013).PatriarchyandfertilityinAlbania. DemographicResearch,29:133–166.

Mangiavacchi,L., Perali,F.,andPiccoli, L.(2018).INTRAHOUSEHOLD DISTRIBUTIONinMIGRANTSENDINGFAMILIES. JournalofDemographic Economics,84(1):107–148.

Michaud,P.-C.andTatsiramos,K. (2011).Fertilityandfemaleemploymentdynamicsineurope: theeffectofusingalternativeeconometricmodelingassumptions. Journal ofAppliedEconometrics, 26(4):641–668.

Miller,G. (2010). Contraceptionas development?new evidencefromfamily planningincolombia. TheEconomicJournal, 120(545):709–736.

Miluka,J.(2013).NotJust Education:TheGender WageGapintheAlbanian LaborMarketsThrough OccupationalSegregation,

WorkExperience,andChild Care.InRuggeriLaderchi, C.andSavastano,S., editors, Povertyand ExclusionintheWestern Balkans:NewDirectionsin MeasurementandPolicy, pages155–176.Springer NewYork,NewYork,NY.

MoT (2022).Feesfor crèchesandkindergartens. InMunicipalityofTirana

Oliveira,J. (2016).The valueofchildren:Intergenerationalsupport,fertility,andhumancapital. JournalofDevelopment Economics,120(C):1–16.

Rosenzweig,M.and Schultz,T. (1985).The demandforandsupply ofbirths:Fertilityandits lifecycleconsequences. AmericanEconomic Review,75(5):992–1015.

Schmieder,J.(2021).Fertilityasadriverofmaternal employment. LabourEconomics,72:102048.

Schultz,P.T. (2001). Chapter8women’srolesin theagriculturalhousehold: Bargainingandhuman capitalinvestments.In AgriculturalProduction, volume1of Handbookof AgriculturalEconomics, pages383–456.Elsevier.

Schultz,P.T.(2007).Chapter52populationpolicies, fertility,women’shuman capital,andchildquality.

volume4of Handbookof DevelopmentEconomics, pages3249–3303.Elsevier.

Tarifa,F. (1994).DisappearingfromPolitics:SocialChangeandWomenin Albania.InRueschemeyer, M.,editor, Womeninthe PoliticsofPostcommunistEasternEurope,pages 133–152.W.Armonk,NY:M.E. Sharpe.

Trako,I.(2019).Fertilityand ParentalLabor-ForceParticipation:NewEvidence fromaDevelopingCountryintheBalkans. World

BankPolicyResearchWorkingPapers,8931.

UNDP (2021).ADetailed Analysis:Registered Long-termUnemployed JobseekersinAlbania. Technicalreport, ”SupportedContinuous UnemployedLearning” (SCUL)project(September 2021),NationalAgency forEmploymentandSkills andUNDPAlbania,Tirana, Albania.

UNESCO (2017).Albania: educationpolicyreview:issuesandrecommenda-

tions.

UNWOMEN (2020).AlbaniaCountryGenderEqualityBrief2020.

Vullnetari,J.andKing,R. (2016).‘washingmen’s feet’:gender,careand migrationinalbaniaduringandaftercommunism. Gender,Place&Culture, 23(2):198–215.

WorldBank (2002).Albania-countryfinancial accountabilityassessment (english).Technicalreport, TheWorldBank.

MeanSDMinMaxN

Currentlyworking0.320.470.001.002997 Employedinthelastyear0.380.480.001.002997 Rural0.510.500.001.002997 Wealth0.630.190.000.992997

Yearsofeducation10.504.460.0024.002997 Adultsinthehousehold2.891.241.009.002997 Married0.980.150.001.002997 Muslim0.830.370.001.002997 Age30.633.4420.0035.002997

Ageat1stbirth21.773.0013.0035.002997

Numberofchildren2.360.622.006.002997

Twofemalechildren0.260.440.001.002997

Firstchild:male0.490.500.001.002997

Notes. Pooledcross-sectionsDHS2007&2017-18.Sampleselection:femalerespondentsaged20to35yearsofage, withatleasttwochildrenandwhoseageisbelowtheageof18.Sampleweightsareapplied.

Table(2)Analysisofsamegenderinstruments:Firststageanalysisofnumberofchildren.

Dependent:Numberofchildren (1)(2)(3)(4)(5)(6)(7)(8)

Samesex0.161***0.156***0.160*** (0.024)(0.024)(0.024)

Twomalechildren0.002-0.007 (0.028)(0.031)

Twofemalechildren0.311***0.320***0.320***0.301***0.313*** (0.034)(0.037)(0.037)(0.037)(0.047)

Firstchild:male-0.145***-0.147***0.0180.0150.011 (0.025)(0.025)(0.027)(0.025)(0.031)

Secondchild:male-0.163***-0.013-0.007 (0.025)(0.025)(0.031)

First-stageF-stat4644464238766744

0.230.270.250.270.270.270.270.27

Table(3)Balancetable:differenceinmeansbetweentheIVsampleandnonIVsample (1)(2)(3)

VariableNoIVIV:TwofemalechildrenDifference Rural0.5060.5150.009 (0.500)(0.500)(0.021) Wealth0.6340.627-0.007 (0.185)(0.188)(0.008)

Yearsofeducation10.48510.5450.060 (4.453)(4.483)(0.186)

Adultsinthehousehold2.9042.852-0.052 (1.247)(1.215)(0.052)

Married0.9790.970-0.008 (0.145)(0.170)(0.006) Muslim0.8280.8480.020 (0.378)(0.359)(0.016)

Age30.63230.6450.013 (3.471)(3.364)(0.144)

Ageat1stbirth21.80621.650-0.157 (3.026)(2.903)(0.125)

Numberofchildren2.2802.6060.326*** (0.540)(0.741)(0.025)

Firstchild:male0.6620.000-0.662*** (0.473)(0.000)(0.017)

Lastchild:male0.6240.333-0.292*** (0.484)(0.472)(0.020)

Observations2,1958022,997

Notes. Thetablereportsthemeansforwomenwithatleasttwochildrenofmixedgender(column1),womenwithat leasttwofemalefirstbornchildren(column2)andtheirstatisticaldifference(column3).Pooledcross-sections DHS2007&2017-18.Sampleselection:femalerespondentsaged20to35yearsofagewithatleasttwochildren, reportinghavinganychildbelowtheageof18.

Table(4)Averageeffectofnumberofchildrenonemploymentprobability-OLS

Dependent:EmployedlastweekEmployedlastyear (1)(2)(3)(4)(5)(6) AllRuralUrbanAllRuralUrban Numberofchildren-0.071***-0.051**-0.094***-0.091***-0.090***-0.093*** (0.020)(0.022)(0.036)(0.020)(0.023)(0.036)

Rural-0.061**-0.071** (0.029)(0.030)

Wealth0.1350.284**0.292*-0.0600.1250.231 (0.093)(0.115)(0.176)(0.101)(0.122)(0.179) Yearsofeducation0.017***0.018***0.014***0.017***0.016***0.013** (0.004)(0.005)(0.005)(0.004)(0.005)(0.005)

Adultsinthehousehold0.019**0.019*0.025*0.022***0.0170.035*** (0.008)(0.010)(0.014)(0.008)(0.011)(0.013)

Married-0.042-0.014-0.066-0.0190.007-0.037 (0.077)(0.110)(0.097)(0.077)(0.113)(0.095) Age0.046-0.0180.140**0.038-0.0120.117* (0.041)(0.049)(0.065)(0.043)(0.053)(0.066)

Age2 -0.0000.001-0.002*-0.0000.001-0.001 (0.001)(0.001)(0.001)(0.001)(0.001)(0.001)

Ageat1stbirth-0.013***-0.013**-0.013**-0.012***-0.016***-0.007 (0.004)(0.005)(0.006)(0.004)(0.005)(0.006)

Firstchild:male-0.012-0.009-0.005-0.012-0.014-0.008 (0.023)(0.026)(0.036)(0.024)(0.028)(0.036)

MunicipalityFEYesYesYesYesYesYes YearFEYesYesYesYesYesYes

0.150.140.150.130.170.14 Observations299716621328299716621328

Notes. ***p-value<0.01,**p-value<0.05,*p-value<0.10.Pooledcross-sectionsDHS2007&2017-18.Thedependent variablecapturesthelinearprobabilitytohaveatleastworkedonceinthelastyear.Sampleselection:female respondentsaged20to35yearsofage,withatleasttwochildrenandwhoseageisbelowtheageof18.Sample weightsareapplied.GroupsFEstandsforincludingbothethnicgroupfixedeffectsandreligiousgroupfixedeffect. StandarderrorsareclusteredattheDHS-clusterlevel.

Table(5)Firststage:Averageeffectofhavingtwogirlsasfirstbornonnumberofchildren

Dependent:Numberofchildren (1)(2)(3) AllRuralUrban

Twofemalechildren0.320***0.432***0.201*** (0.037)(0.044)(0.058)

Rural-0.060* (0.032)

Wealth-0.558***-0.372***-0.849*** (0.098)(0.139)(0.201)

Yearsofeducation-0.011***-0.016***-0.004 (0.003)(0.004)(0.006)

Adultsinthehousehold-0.018**-0.005-0.039*** (0.009)(0.012)(0.014)

Married0.0980.188**0.034 (0.083)(0.093)(0.117) Age0.082*0.140**0.009 (0.049)(0.060)(0.081) Age2 -0.000-0.0010.001 (0.001)(0.001)(0.001)

Ageat1stbirth-0.060***-0.064***-0.057*** (0.004)(0.007)(0.006)

Firstchild:male0.0150.028-0.014 (0.025)(0.032)(0.039)

MunicipalityFEYesYesYes

Table(6)Secondstage:Averageeffectofnumberofchildrenonemploymentprobability

Dependent:EmployedlastweekEmployedlastyear (1)(2)(3)(4)(5)(6) AllRuralUrbanAllRuralUrban

Numberofchildren-0.128*-0.121**-0.119-0.110-0.122*-0.059 (0.068)(0.060)(0.162)(0.070)(0.062)(0.160) Rural-0.064**-0.071** (0.030)(0.031)

Wealth0.1040.253**0.288-0.0710.1100.270 (0.099)(0.117)(0.201)(0.105)(0.123)(0.206)

Yearsofeducation0.017***0.017***0.014***0.017***0.016***0.013*** (0.004)(0.005)(0.005)(0.004)(0.005)(0.005)

Adultsinthehousehold0.018**0.019*0.0240.022***0.0170.036** (0.008)(0.010)(0.015)(0.008)(0.011)(0.015)

Married-0.041-0.001-0.074-0.0190.011-0.043 (0.075)(0.106)(0.095)(0.077)(0.112)(0.095) Age0.054-0.0060.147**0.042-0.0060.121* (0.042)(0.050)(0.066)(0.043)(0.054)(0.067)

Age2 -0.0000.001-0.002*-0.0000.001-0.002 (0.001)(0.001)(0.001)(0.001)(0.001)(0.001)

Ageat1stbirth-0.016***-0.018***-0.014-0.013**-0.019***-0.005 (0.006)(0.006)(0.010)(0.006)(0.006)(0.010)

Lastchild:male-0.038*-0.021-0.065*-0.0130.002-0.038 (0.022)(0.028)(0.035)(0.023)(0.028)(0.037)

Notes.

35withatleasttwochildrenwhoseageisbelow18.Sampleweightsareapplied.Groupfixedeffectsincludereligion groupsandethnicgroups.StandarderrorsareclusteredattheDHS-clusterlevel.

Table(7)Averageeffectofnumberofchildrenonemploymentprobabilitybylevelof educationachieved.

SecondstageDependent:Employedlastweek

SamplebyeducationAboveprimaryPrimaryorlower educationAllUrbanRural (1)(2)(3)(4)

Numberofchildren-0.345-0.216**-0.403-0.152* (0.290)(0.089)(0.302)(0.083) Rural-0.070-0.078** (0.054)(0.035)

Wealth0.0320.012-0.1520.131 (0.258)(0.110)(0.315)(0.135)

Yearsofeducation0.028***-0.017-0.0250.002 (0.008)(0.012)(0.028)(0.011)

Adultsinthehousehold0.0130.022**0.0340.018* (0.017)(0.010)(0.022)(0.011) Married0.035-0.044-0.098-0.020 (0.098)(0.091)(0.155)(0.125)

Age0.1520.0360.143*-0.013 (0.104)(0.047)(0.086)(0.055)

Age2 -0.002-0.000-0.0020.001 (0.002)(0.001)(0.002)(0.001)

Ageat1stbirth-0.016-0.033***-0.057**-0.021** (0.015)(0.009)(0.025)(0.009)

Firstchild:male-0.000-0.071**-0.054-0.077** (0.052)(0.032)(0.065)(0.035)

FirststageDependent:Numberofchildren (1)(2)(3)(4)

Twofemalechildren0.190***0.393***0.248***0.476*** (0.059)(0.043)(0.075)(0.051)

Notes.

(showingthemaincovariatesforthesecondstageandinstrumentfromthefirststage).Thesampleiscomposedof womenagedbetween20and35withatleasttwochildrenwhoseageisbelow18.Column1restrictsthedatato womenhavingachievedhigherthanprimaryeducation,Column2-4withbeloworequaltoprimary.Thespecification isalsoinspectedforurban(Col.3)orruralareasonly(Col.4).Sampleweightsareapplied.Groupfixedeffects includereligiongroupsandethnicgroups.StandarderrorsareclusteredattheDHS-clusterlevel.

PanelA.ElderlyinthehouseholdPanelB.Partneremploymentstatus

AllRuralAllRural

SeniorsNoseniorsSeniorsNoseniorsPartnerPartnernotPartnerPartnernot inhhinhhinhhinhhemployedemployedemployedemployed (1)(2)(3)(4)(5)(6)(7)(8)

Numberofchildren-0.233*-0.245*-0.211*-0.153 -0.524**0.024-0.489*0.026 (0.120)(0.131)(0.115)(0.117) (0.264)(0.165)(0.258)(0.162)

Rural-0.074-0.054 -0.164***0.005 (0.045)(0.037) (0.045)(0.051)

Wealth0.111-0.0330.2300.227 0.0250.1560.3170.148 (0.154)(0.136)(0.188)(0.153) (0.233)(0.202)(0.220)(0.194)

Yearsofeducation0.0090.020***0.0090.018*** 0.005-0.0090.006-0.008 (0.006)(0.005)(0.007)(0.007) (0.005)(0.008)(0.005)(0.008)

Adultsinthehousehold0.037***0.0030.055***-0.003 0.024*0.0180.0190.018 (0.014)(0.014)(0.016)(0.019) (0.014)(0.018)(0.014)(0.018)

Age-0.0060.0850.016-0.035 0.141*0.0470.150*0.047 (0.064)(0.066)(0.069)(0.088) (0.083)(0.070)(0.084)(0.070)

Age2 0.001-0.0010.0000.001 -0.002-0.001-0.002-0.001 (0.001)(0.001)(0.001)(0.001) (0.001)(0.001)(0.001)(0.001)

Ageat1stbirth-0.022**-0.024**-0.022*-0.019* -0.033**-0.010-0.031**-0.010 (0.009)(0.010)(0.012)(0.010) (0.016)(0.011)(0.015)(0.011)

Firstchild:male-0.007-0.060-0.030-0.020 -0.039-0.086*-0.032-0.086* (0.038)(0.037)(0.043)(0.048) (0.051)(0.049)(0.050)(0.049)

Married0.030-0.0660.051-0.192 (0.107)(0.103)(0.116)(0.202)

Dependent:Employedlastweek Year:2008/09Year:2017/08 (1)(2)(3)(4)(5)(6)(7)(8) AllUrbanRuralRPrimaryAllUrbanRuralRPrimary N.ofchildren-0.160*0.161-0.215**-0.090-0.360**-0.714-0.155-0.210* (0.093)(0.242)(0.096)(0.097)(0.177)(0.525)(0.136)(0.123)

Rural0.092**-0.133*** (0.047)(0.035)

Wealth0.1920.3880.2760.2900.119-0.2650.1630.004 (0.137)(0.306)(0.171)(0.194)(0.166)(0.499)(0.152)(0.171) Education0.047***0.056***0.034***0.0110.0060.0050.009-0.003 (0.007)(0.011)(0.012)(0.029)(0.004)(0.007)(0.006)(0.013)

Adults0.0060.0270.004-0.0000.023**0.0170.026*0.030** (0.011)(0.019)(0.017)(0.016)(0.011)(0.021)(0.013)(0.013)

Married-0.090-0.156-0.094-0.350-0.025-0.1270.0950.102 (0.130)(0.149)(0.222)(0.233)(0.090)(0.142)(0.064)(0.069)

Age-0.0230.086-0.036-0.0730.113*0.1950.0440.034 (0.059)(0.104)(0.074)(0.084)(0.060)(0.126)(0.066)(0.074)

Age2 0.001-0.0010.0010.002-0.001-0.002-0.0000.000 (0.001)(0.002)(0.001)(0.001)(0.001)(0.002)(0.001)(0.001)

Ageat1stbirth-0.025***-0.011-0.030**-0.018-0.026**-0.046-0.016*-0.029*** (0.009)(0.016)(0.012)(0.012)(0.011)(0.032)(0.010)(0.011)

Firstchild:male-0.0400.054-0.054-0.041-0.036-0.042-0.013-0.099** (0.040)(0.066)(0.054)(0.057)(0.041)(0.075)(0.039)(0.044)

Year:2008/09Year:2017/08 (1)(2)(3)(4)(5)(6)(7)(8) AllUrbanRuralRPrimaryAllUrbanRuralRPrimary Twogirls0.434***0.293***0.562***0.577***0.242***0.160**0.354***0.418*** (0.064)(0.107)(0.079)(0.089)(0.042)(0.064)(0.055)(0.067)

MunicipalityFEYesYesYesYesYesYesYesYes GroupFEYesYesYesYesYesYesYesYes

First-stageF-stat47751423364239

Notes. ***p-value<0.01,**p-value<0.05,*p-value<0.10.2SLSmodelwithasdependentvariablethelinearprobability tohaveatleastworkedoncelastweek.Thesampleisrestrictedtowomenagedbetween20and45withatleasttwo childrenwhoseageisbelow18inYear2008/09(Col1-4)or2017/18(Col5-8).Dataisfurtherrestrictedtobeeither forurban(Col2and6)orruralareas(Col3and7),orruralsamplewithprimaryeducation(Col4and8).Sample weightsareapplied.Groupfixedeffectscaptureseparatelyreligiongroupsandethnicgroups.Standarderrorsare clusteredattheDHS-clusterlevel.

Table(10)Rural/Urbancomparison (1)(2)(3)

VariableRuralUrbanDifference

Workedlastweek0.2760.434-0.157*** (0.447)(0.496)(0.009)

Walktowork0.7420.4770.265*** (0.438)(0.500)(0.016)

Walkingdistancetowork20.05018.5651.485** (19.078)(16.842)(0.714)

Husbanddecidesonearnings0.1510.0890.062*** (0.358)(0.285)(0.007)

Husbanddecidesonlargehouseholdpurchases0.1130.0780.035*** (0.317)(0.268)(0.007)

Childinpreschool0.4630.500-0.037* (0.499)(0.500)(0.019)

Observations5,9004,96010,860

Notes: CrosssectionsdatafromDHS2017-18.Thetablereportsmeanlevelsaswellasthep-valueforateststatistics (t-testforlevel)ofdifferencesacrosstheruralandurbansamples.

PanelA:Unemployment

NAS:keypointsfromfocusgroupsDHS ElbasanLushnjeKorceElbasanLushnjeKorce

EmploymentFelthighoverallFelthighoverallFelthigh,butalso0.2060.3990.368 forwomenwith(0.404)(0.490)(0.483) acollegedegree

Observations563541573

PanelB:Distancefromworkplace

WalkingasmainmeantocommuteHightransportHightransportHightransport0.7270.7820.815 costscostscosts(0.447)(0.413)(0.389)

Walkingdistancetowork(minutes)18.05917.10620.904 (16.762)(14.936)(18.232)

Observations563541573

PanelC:Gendernorms

HusbanddecidesonearningsMenadministerHusbandsdecidesHusbandshaveasay0.1200.2290.083 earningsaboutfemaleonthetype(0.325)(0.420)(0.276

Husbanddecidesonlargepurchasesemploymentofjobsthewife0.0660.1940.065 canbeemployedin(0.249)(0.396)(0.248)

AtleastonechildisFeellackofFeellackofFeellackof0.4060.5860.454

Figures

Lushnje Korce

Latitude

Longitude Notes. Source:NeedAssessmentStudy2022

Elbasan

Figure(2)Gendernormsdifferencesovertime (a)Whathusbandmainlydecideson

(b)Idealnumberofchildren

Notes:Eachsymbolrepresentsadifferentregression,withdependentvariablesindicatedinthelegends.Confidence intervalsat90%.Allregressionsincludenumberofchildren,wealth,yearsofeducation,marriagestatus,agesquared andageatbirthascontrolsandmunicipalitiesfixedeffects.Thesampleisrestrictedtowomenaged20-35with childrenbelow18.

Table(A1)Reducedformequation

DependentvariableEmployedlastweekEmployedlastyear (1)(2)(3)(4)(5)(6) AllRuralUrbanAllRuralUrban

Numberofchildren-0.047***-0.035**-0.063**-0.063*** -0.070***-0.066*** (0.014)(0.016)(0.025)(0.015) (0.019)(0.025)

1stand2ndbornarefemale-0.032-0.029-0.032-0.010 -0.016-0.002 (0.022)(0.028)(0.037)(0.023) (0.029)(0.037)

Rural-0.051**-0.041* (0.022)(0.024)

Wealth0.0510.0800.358***-0.159** -0.0340.298** (0.066)(0.084)(0.131)(0.073) (0.093)(0.135)

Yearsofeducation0.020***0.017***0.019***0.022*** 0.017***0.019*** (0.002)(0.003)(0.004)(0.002) (0.004)(0.004)

Adultsinthehousehold0.011*0.0090.0160.015** 0.0120.020* (0.006)(0.007)(0.011)(0.007) (0.008)(0.011)

Married-0.027-0.007-0.049-0.008 0.024-0.025 (0.053)(0.071)(0.075)(0.055) (0.080)(0.074)

Age0.068**0.0320.135**0.053 0.0200.109* (0.032)(0.040)(0.053)(0.035) (0.045)(0.056)

Age2 -0.001-0.000-0.002**-0.000 0.000-0.001 (0.001)(0.001)(0.001)(0.001) (0.001)(0.001)

Ageat1stbirth-0.011***-0.013***-0.010*-0.013*** -0.019***-0.008 (0.003)(0.004)(0.005)(0.003) (0.004)(0.005)

Firstchild:male-0.026-0.022-0.020-0.023 -0.028-0.014 (0.019)(0.024)(0.031)(0.020) (0.025)(0.031)

MunicipalityFEYesYesYesYes YesYes

YearFEYesYesYesYes YesYes

GroupFEYesYesYesYes YesYes

Adj-R2 0.150.140.130.14 0.170.13 Observations2997166213282997 16621328

Notes. ***p-value<0.01,**p-value<0.05,*p-value<0.10.Thedependentvariableisabinaryvariabletakingthe value1ifthesexofthefirsttwochildrenisfemale,0otherwise.Thesampleisrestrictedtowomenagedbetween20 and35withatleasttwochildrenwhoseageisbelow18.Sampleweightsareapplied.Groupfixedeffectsinclude religiongroupsandethnicgroups.StandarderrorsareclusteredattheDHS-clusterlevel.

Table(A2)Analysisofinstruments(RURALONLY)

PanelA:Firststage

Dependent:Numberofchildren (1)(2)(3)(4)(5)(6)(7) Samesex0.219***0.208***0.219*** (0.033)(0.030)(0.032) Boy10.188***0.189***-0.036-0.020-0.028 (0.031)(0.031)(0.038)(0.038)(0.032) Boy20.224***0.016 (0.031)(0.042)

Twoboys0.003-0.016 (0.035)(0.042)

Twogirls0.415***0.432***0.416***0.432*** (0.042)(0.044)(0.060)(0.044)

PanelB:Secondstage

Dependent:currentlyemployed (1)(2)(3)(4)(5)(6)(7) Numberofchildren-0.212*-0.214*-0.212*-0.134**-0.187**-0.214*-0.188** (0.121)(0.126)(0.120)(0.063)(0.083)(0.126)(0.084) Boy10.0400.0400.0350.0400.035 (0.036)(0.036)(0.033)(0.036)(0.033) Boy20.0110.011 (0.038)(0.038)

Table(A3)Robustness:Dependentvariableforemploymentinlast12months&Instrumented variableinlogform.

RobustnessDependent:Employedlast12monthsIV:Lnnumberofchildren (1)(2)(3)(4)(5)(6) AllRuralUrbanAllRuralUrban Numberofchildren-0.174*-0.165**-0.175 (0.090)(0.082)(0.233)

Lnnumberofchildren-0.656***-0.511**-0.875 (0.245)(0.227)(0.651)

Rural-0.076**-0.071** (0.030)(0.030)

Wealth-0.1080.0940.1620.0400.234**0.090 (0.111)(0.121)(0.268)(0.106)(0.118)(0.274) Yearsofeducation0.016***0.015***0.012**0.015***0.016***0.013** (0.004)(0.005)(0.005)(0.004)(0.005)(0.005)

Adultsinthehousehold0.021**0.0170.031**0.016*0.019*0.016 (0.008)(0.010)(0.015)(0.008)(0.010)(0.015)

Married-0.0120.024-0.036-0.0270.013-0.060 (0.075)(0.107)(0.093)(0.072)(0.101)(0.092) Age0.045-0.0000.118*0.0600.0040.139* (0.044)(0.055)(0.068)(0.043)(0.051)(0.072)

Age2 -0.0000.001-0.001-0.0000.001-0.002 (0.001)(0.001)(0.001)(0.001)(0.001)(0.001)

Ageat1stbirth-0.017**-0.021***-0.012-0.024***-0.022***-0.026* (0.007)(0.007)(0.014)(0.007)(0.008)(0.015)

Firstchild:male-0.024-0.028-0.018-0.037-0.034-0.034 (0.029)(0.034)(0.050)(0.029)(0.033)(0.052)

(0.037)(0.044)(0.058)(0.013)(0.016)(0.021)

Table(A4)Robustness:Averageeffectofnumberofchildrenonemployment-IV

Dependent:Employedlastyear Year:2008/09Year:2017/08 (1)(2)(3)(4)(5)(6) AllRuralUrbanAllRuralUrban

Numberofchildren-0.171*-0.265***0.152-0.213-0.054-0.487 (0.095)(0.099)(0.242)(0.166)(0.128)(0.473) Rural0.097*-0.133*** (0.053)(0.035)

Wealth-0.0170.1080.3370.1390.135-0.163 (0.139)(0.166)(0.325)(0.164)(0.168)(0.458)

Yearsofeducation0.044***0.0200.057***0.0060.011*0.004 (0.008)(0.014)(0.011)(0.004)(0.006)(0.006) Adultsinthehousehold0.0110.0090.0320.029***0.0230.034*

(0.011)(0.016)(0.020)(0.011)(0.014)(0.020)

Married-0.106-0.136-0.1440.0180.127*-0.079 (0.126)(0.195)(0.146)(0.088)(0.067)(0.130) Age0.003-0.0060.1210.0870.0540.135 (0.064)(0.083)(0.107)(0.057)(0.068)(0.104)

Age2 0.0010.001-0.002-0.001-0.000-0.002 (0.001)(0.001)(0.002)(0.001)(0.001)(0.002)

Ageat1stbirth-0.028***-0.041***-0.009-0.014-0.010-0.025 (0.009)(0.011)(0.015)(0.011)(0.009)(0.029)

Firstchild:male-0.038-0.0420.057-0.0210.003-0.033 (0.041)(0.057)(0.066)(0.039)(0.039)(0.069)

Firststage Dependent:Numberofchildren

Year:2008/09Year:2017/08

Table(A5)Differenceinmeansbetweenurbanandruralareasbytypeofemployment (1)(2)(3) VariableUrbanRuralDifference

PanelA:Typeofoccupation

professional/technical/managerial0.3200.097-0.223*** (0.467)(0.296)(0.018) clerical0.0440.010-0.034*** (0.206)(0.100)(0.008) sales0.1140.049-0.065*** (0.318)(0.216)(0.013) agricultural-employee0.0200.5570.537*** (0.141)(0.497)(0.015) services0.0900.067-0.023* (0.286)(0.250)(0.012) skilledmanual0.2730.123-0.149*** (0.445)(0.329)(0.018) unskilledmanual0.1390.097-0.042*** (0.346)(0.296)(0.015)

PanelB:Forwhomworking

forfamilymember0.1940.4860.292*** (0.395)(0.500)(0.020) forsomeoneelse0.6430.238-0.404*** (0.479)(0.426)(0.021) self-employed0.1640.2760.112*** (0.370)(0.447)(0.018)

allyear0.9060.531-0.376*** (0.292)(0.499)(0.018) seasonal0.0510.3790.328*** (0.220)(0.485)(0.016) occasional0.0430.0900.047*** (0.202)(0.286)(0.011)

notpaid0.0430.4230.380*** (0.203)(0.494)(0.016) cashonly0.9340.388-0.546*** (0.248)(0.488)(0.017) cashandin-kind0.0120.0700.058*** (0.111)(0.256)(0.008) in-kindonly0.0100.1190.109*** (0.101)(0.324)(0.010) Observations1,0999162,015

Table(A6)Averageeffectofnumberofchildren,ruralandtimebinariesonemployment

Secondstage

A.Dependent:Self-employed

AllRuralUrban (1)(2)(3)

Numberofchildren0.1340.1840.386 (0.179)(0.213)(0.441)

Rural=10.132*** (0.047)

Year=2018-0.0100.132*-0.029 (0.036)(0.069)(0.050)

B.Dependent:Workingforsomeoneelse

AllRuralUrban (1)(2)(3)

Numberofchildren-0.135-0.057-0.702 (0.214)(0.209)(0.708)

Rural=1-0.184*** (0.059)

Year=20180.168***0.157**0.114 (0.047)(0.062)(0.085)

C.Dependent:Workinginunstablejob

AllRuralUrban (1)(2)(3)

Numberofchildren0.146-0.0610.423 (0.161)(0.233)(0.338)

Rural=10.113** (0.045)

Year=2018-0.0070.088-0.072** (0.034)(0.072)(0.033)

D.Dependent:Workinagriculture

AllRuralUrban (1)(2)(3)

Numberofchildren0.1090.1000.116 (0.132)(0.189)(0.108)

Rural=10.216*** (0.039)

Year=2018-0.0100.0280.002 (0.028)(0.066)(0.013)

RuralUrban 2008201820082018 (1)(2)(3)(4)

PanelA:Shareofemployedwomen

Employedlastweek0.2750.2930.4060.491 Employedlastyear0.4110.3350.4380.518 Observations1080169811681322

PanelB:Occupation

professional/technical/managerial0.0780.1200.3660.298 clerical0.0030.0190.0350.049 sales0.0760.0160.2110.067 agricultural-employee0.6550.4360.0270.017 services0.0780.0530.1220.074 skilledmanual0.0850.1700.2190.298 unskilledmanual0.0250.1860.0190.197

Observations459457514585

Notes. Thesampleisrestrictedtowomenagedatleast20withchildrenbelow18.

RuralUrban 2008201820082018 (1)(2)(3)(4)

PanelA:Forwhomworking

forfamilymember0.6710.2560.2590.162 forsomeoneelse0.1140.3920.5400.693 self-employed0.2140.3530.2010.146

PanelB:Jobstability allyear0.4540.6250.8640.927 seasonal0.4320.3140.0550.049 occasional0.1130.0620.0810.024

notpaid0.5670.2440.0760.027 cashonly0.2480.5620.9160.943 cashandin-kind0.0470.0990.0020.017 in-kindonly0.1380.0950.0060.013 Observations459457514585

Notes. Thesampleisrestrictedtowomenagedatleast20withchildrenbelow18.

Table(A9)Differenceinmeansbetweenurbanandruralareas (1)(2)(3)

VariableUrbanRuralDifference Numberofchildren2.3212.5110.190*** (0.581)(0.740)(0.018)

respondentcurrentlyworking0.4600.284-0.176*** (0.498)(0.451)(0.013)

Workedinthelastyear0.4890.373-0.116*** (0.500)(0.484)(0.014) Wealth0.7520.514-0.238*** (0.114)(0.180)(0.004)

Yearsofeducation12.3139.481-2.832*** (4.789)(3.393)(0.115) Numberofadultsinthehh4.2864.6130.327*** (1.339)(1.443)(0.038) Married0.9660.9820.015*** (0.180)(0.134)(0.004) Muslim0.7790.8420.063*** (0.415)(0.365)(0.011) Age34.80333.969-0.833*** (5.480)(5.333)(0.149)

Ageat1stbirth23.81622.908-0.907*** (3.995)(3.336)(0.102)

Firstandsecondbornchildrenarefemale0.2540.2630.008 (0.436)(0.440)(0.012)

Firstandsecondbornchildrenaremale0.2510.235-0.016 (0.433)(0.424)(0.012) Observations2,4902,7785,268

Notes. Balancetablereportingmeansforurban,ruralandtheirstatisticaldifference.Pooledcross-sectionsDHS2007 &2017-18.Sampleselection:femalerespondentsaged20to49yearsofage,reportinghavinganychildbelowthe ageof18.Thedataisreportedforthefullsample(column3),forurbanorruralsamples(2,490and2,778observations respectively).

Figure(A1)Employmenttypesprobabilitydifferencesacrossregressionscoefficientsfor ruralareasandovertime

Notes. Thefigurereportsthecoefficientsforruralandyearbinaryvariablesandtheirinteractionsforfouremployment linearprobabilityregressions(OLS).Employmentisdefinedashavingworkedatleastoncelastyeareitheras employeeforsomeone(β coefficientsrepresentedasacircle),asemployeeforfamilymembers(square),orhaving astableoccupation(triangle),orasunpaidworker(cross).Allregressionsincludenumberofchildren,wealth,years ofeducation,marriagestatus,agesquaredandageatbirthascontrolsandmunicipalitiesfixedeffects.Thesample isrestrictedtowomenaged20-35withchildrenbelow18.Confidenceintervalsat95%.

What is AFD?

Éditions Agence française de développement publishes analysis and research on sustainable development issues. Conducted with numerous partners in the Global North and South, these publications contribute to a better understanding of the challenges faced by our planet and to the implementation of concerted actions within the framework of the Sustainable Development Goals.

With a catalogue of more than 1,000 titles and an average of 80 new publications published every year, Éditions Agence française de développement promotes the dissemination of knowledge and expertise, both in AFD’s own publications and through key partnerships. Discover all our publications in open access at editions. afd.fr.

Towards a world in common.

ISSN 2492 - 2846

https://creativecommons.org/licenses/by-nc-nd/4.0/

Juliegilles,

Layout Denise Perrin,

https://www.afd.fr/en/ressources-accueil