Policy Research Working Paper 7917

Building an Ex Ante Simulation Model for Estimating the Capacity Impact, Benefit

Incidence, and Cost Effectiveness of Child Care Subsidies

An Application Using Provider-Level Data from Turkey

Meltem A. Aran Ana Maria Munoz-Boudet

Nazlı Aktakke

Poverty and Equity Global Practice GroupDecember 2016

WPS7917P

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

edP

ublic

Dis

clos

ure

Aut

horiz

ed

Produced by the Research Support Team

Abstract

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Policy Research Working Paper 7917

This paper is a product of the Poverty and Equity Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at amunozboudet@worldbank.org.

Public financing of child care can allow for more equitable access to these services in places where public provision and capacity are low. However, the mechanisms of subsidy deliv-ery will affect who benefits, and the overall cost effectiveness of such subsidies. This paper sets out an ex ante simulation model for estimating the benefit incidence of expanded capacity and enrollments resulting from different child care subsidy mechanisms. It uses a supply-side provider level and a demand-side household model, and combines the two models. The paper considers investment grants to pro-viders, operational monthly grants to child care providers, combinations of the investment and operational grants, and demand-side vouchers to households. The model is applied to empirical data from child care centers and households

in Turkey. The results reveal that the choice of the subsidy delivery model has a strong bearing on the benefit inci-dence and cost effectiveness of the subsidy. In the case of Turkey, where significant supply-side constraints exist in the market, a demand-side voucher system is shown to be the least cost-effective subsidy delivery model. A targeted demand-side voucher does not necessarily deliver the most

“pro-poor results,” and combinations show different benefits and costs. The proposed simulation model can be applied in other country contexts, with the only data require-ments being micro data on the costs and pricing structure of child care providers, as well as household data with variables on household welfare and child care utilization.

Bu i ld ing an Ex Ante S imu la t ion Mode l f or Es t imat ing the Capac i t y Impact , Bene f i t Inc idence , and Cost E f fe c t iveness o f Ch i ld Care Subs id ies :

An App l i ca t ion Us ing Prov ider ‐Leve l Da ta f romTurkey1

MeltemA.Aran DevelopmentAnalytics AnaMariaMunoz‐Boudet WorldBank NazlıAktakke DevelopmentAnalytics

Keywords:daycarecenters,subsidiesandevidence‐basedpolicy,vouchers,educationfinance,povertyJELClassificationNumbers:H22,H23,H32,I22,I24,I38.

1 The study was supported by the Swedish International Development Cooperation Agency SIDA under theUmbrella Facility for Gender Equality Trust Fund of theWorld Bank. The authorswould like to thank RıdvanKurtipek, Yıldız Yapar, Pınar Özel Ucuzal and Dilek Okkalı Şanalmış TurkeyMinistry of Development ; BurcuErtuğrul andFatihKaynar TurkeyMinistry ofEducation ;GülerÖzdoğan,AkınYumuşandFatmaKüçükbaşolTurkey Ministry of Family and Social Policies ; Facundo Cuevas TheWorld Bank ; Öykü Uluçay and TomasBergenholtz SwedishConsulate forcommentsandguidancereceivedduringmethodologyreviewmeetingsheldinAnkarainFebruary2016.

1. Motivation

Publicfinancingandpartialsubsidizationofchildcarecanallowformoreequitableaccesstochildcareinplaceswherepublicprovisionandcapacityarelow.Countrieshaveusedanumberofsupply‐anddemand‐sidesubsidiesinordertoexpandaccesstochildcareservices.Variousmechanismshavebeenexperimentedwithinvariouscountries,includingandnotlimitedto: i perchildoperationalgrantstosupply‐sideserviceproviders; ii tax subsidies to working parents and iii vouchers or direct cash transfers provided tohouseholds.

InMexico,in2007asupply‐sidegrantprogramtoexpandchildcareserviceswascreatedwiththegoalofincreasing laborforceparticipationofpoormothers MateoDiazandChamussy,2013 .Childcarecenterswereprovidedwithone‐timeinitialinvestmentsubsidiestoopenupnewcentersortoimprovetheirexistingfacilities.Additionally,monthlygrantswereprovided today‐carecentersonapercapita perchild basisCalderon,2014 .Theoperationalcostsofthesecentersarecoveredbythepublicsubsidyplusafeecollectedfrom the parents Staab and Gerhard, 2010 . Similarly, in the Republic of Korea, in 2009 the municipalgovernmentofSeoulstartedprovidingfinancialsupporttoprivatechildcarecentersinmeetingtheirlaborcosts.Inreturn,regulationsofpublicchildcarecentersareappliedinthesecenters,makingthemsemi‐public.Nearlyhalfofthecity’schildcarecentersarenowpartofthisscheme OECD,2012 .

Demand‐sidegrantsandtaxsubsidiesarealsocommonlyappliedinanumberofcountries.TheNetherlands,theUnitedStates,andAustraliaareexamplestocountriesprovidingvoucherstofamiliesforincreasingaccesstochildcareservices WarnerandGrauds,2011 .IntheNetherlandsandAustralia,thegovernment’saimhasbeen to promote female labor force participationwhile in the United States voucherswere intended forhelpinglow‐incomemothersleavewelfareandstartworking.IntheUnitedKingdom,thechildcarevoucherschemetargetsworkingparents–andallowsataxbreakonincomepartlysubsidizedbytheemployer,andpartlysubsidizedbythestate.Inallfourofthesedevelopedcountries,vouchersortaxsubsidiesforchildcareare provided to the household orworking parents throughworkplace benefits. In the United States andAustralia, the vouchers are primarily targeted to poor working mothers, while in the United Kingdom,NetherlandsandAustria,thevouchersareprovideduniversallywithnon‐poorworkingparentsalsoreceivingasubsidyalbeitsmaller.

While countries have experimented with different models and mechanisms for expanding access andutilization,littleresearchhasbeencarriedoutlookingattheimpactofthechoiceofsupply‐ordemand‐sidemechanisms in targeting these transfers.Where countries have supply‐side constraints in the child caremarket,andrationingtakesplaceforlimitedchildcarespaces–thechoiceofasupply‐ordemand‐sidetransfermightmattersignificantlyindeterminingwhobenefitsfrompublicsubsidies.Thispapersetsoutapracticalmethodologyformodelingthebehaviorofserviceproviders throughasupply‐sidemodel andthebehaviorof households through a demand‐sidemodel to consider the impact of supply‐side investment and/oroperational grants given to child care providers and demand‐side vouchers provided to households assubsidies for child care. Thepaper considers thedistributional impact of these transfers on increases incapacityandutilizationofservices,whilealsoconsideringcost‐effectivenessperU.S.dollarspent tobringaboutnewenrollments.

Thepaper isorganizedas follows:Section2providesanoverviewandliteraturereviewofexitingexantesimulationmodelsofchildcare.Insection3,wepresentabackgroundonthechildcaresysteminTurkey.InSection4,wepresentthemodelforsupply‐anddemand‐sidesubsidiesforthechildcaremarketlookingatvariouspolicyscenariosthroughinvestmentand/oroperationalgrantstochildcareprovidersandthroughtargetedhouseholdvoucherschemes.InSection5wepresentthedatausedforthesimulationsandwediscussthestep‐by‐stepmethodologyforthesupply‐anddemand‐sidesimulationmodels.InSection6,wepresentthemainresultswhen themethodology isapplied to theTurkishdataand inSection7weconcludewithpossiblepracticalimplications.

2. LiteratureReview

Theliteratureonex‐anteevaluationsofchildcarepoliciesusesmicrosimulationsmainlytoassesstheimpactof policy changes on maternal labor force participation and child care utilization. Some other outcome

variables investigated include demand for child care, hours spent in household production and thedistributionaleffectofthepolicies.Thepoliciesthattheauthorsinvestigaterangefromcapacityexpansionstofeereductionsandtheprovisionofsubsidies.Mostofthesepapersfocusonusinghouseholdleveldata,sometimeswithcommunitylevelvariablesonsupply‐sideavailability.

DelBocaandVuri 2007 mergingdatafromtwomicro‐levelsurveys,simulatetheimpactofapartialorafullsubsidyforchildcarecostsonfemaleemploymentinItaly.Theyuseadiscretechoicemodelwherechildcareand labor force decisions of mothers are jointly made. The authors find that in places where capacityconstraintsandrationingofchildcareservicesaremoreprevalenttheimpactofsubsidiesissmaller.SimilartoDelBocaandVuri 2007 ,Apps et al 2016 useadiscrete choicemodel to investigate the changes inmothers’laborsupplytopolicieseffectingchildcareprices,taxesandsubsidies.Intheirmodel,theyusethreechoicevariables;mother’slaborsupply,hertimeallocatedtohouseholdproductionandhoursofchildcarebought.Theyfindthatmothers’decisiontoworkishighlysensitivetoearningsandchildcarecosts.Bothofthesepapersusehouseholdleveldataandmodelthelaborsupplybehaviorofmotherswithchangesinchildcarepricesandavailabilityfacedbythehousehold.

Vanleenhove 2013 investigatesinanex‐antesimulationmodelimpactofacapacityexpansionintheFlemishchildcaremarketontheparents’choiceofmaternallaborsupplyandchildcareusageusingdatafromFlemishFamiliesandCareSurveyof2005.Accordingtoanewdecreebythegovernmentissuedin2012,thechildcarecoverageratewillrise to50percentby2016andby2020allchildcaredemand isplannedtobemet.Bychangingtheopportunitysetof thehouseholdsthroughincreases in formalcare, theauthorsimulatestheresponseofhouseholdstothecapacityexpansion.Onceagain,themodelisentirelybuiltusingthedemand‐sidehouseholdsurveydataandprovidingashocktotheavailabilityofchildcaretothehousehold.

Lookingatthedistributionalimpactofvariouschildcarepolicies,KornstadandToresen 2006 considerthecaseofchildcareinNorwayandlookattheimpactoffourdifferentreformsonchildcarechoicesoffamilies,maternal labor supply aswell as the policies’ distributional impact and their costs to public budget. Thereformsthatareinvestigatedare i increasingchildcarecapacity, ii reductionsinfees, iii rescindinghomecareallowance theallowanceforfamilieswhodonotusecenter‐basedcare andlastly iv introducingaworkingparentstaxcreditschemethatprovidesacertainamountofcredittothemotherworkingmorethan17hoursperweekandahigheramountifsheworksmorethan24hoursaweek.Authorsuseadiscretechoicejoint laborsupplyandchildcarechoicemodelandestimatethe impactof thesepoliciesusingHomeCareAllowanceSurveyof1998ofNorway.Accordingtotheresults,allfourreformsincreasethedemandforcenter‐basedcareandlaborsupplyofmotherswhile feereductionreform is themostsuccessfulone in termsofincreaseddemandandthenewtaxcreditschemeisthemostsuccessfulintermsofincreasedmaternallaborsupply.Yet,thefeereductionreformhasaslightlynegativeeffectonthedistributionofincome.Inotherwordsitbenefitsthehouseholdsthatarewell‐offslightlymorethanpoorerhouseholds‐sincethewell‐offarealreadyusingchildcareandreducingchildcarefeeswillbenefitthemdirectly‐whilethetaxcreditpolicyhasastrongpositiveandpro‐poorredistributioneffect.Inthepaper,rescindinghomecareallowancehasaslightpositivedistributionaleffectandthecapacityincreasescenariodoesnothaveanyeffectondistribution.

In Turkey, İlkkaracan et al 2015 evaluated ex‐ante a policy intervention thatwould increase child carecapacity.Theysimulatetheeffectsofanincreaseinpublicinvestmentinchildcareservicesonthejobscreatedandthedistributionofthesejobsbygenderandincomestatusofhouseholds.Theyinvestigatetheimpactofa20.7 billion TLworth of public investment ‐estimated annual expenditure for unenrolled preschool agedchildreninTurkeytocatchupwiththeaverageOECDpreschoolenrollmentrate‐onpreschoolsandchildcaremarketversusthesamelevelofinvestmentintheconstructionsector.Theyfindthataninvestmentofthisamountwillresultin719,000newjobscreatedintotalwhenitisspentonthechildcaresectorwhileitwillonlycreate290,000newjobswhenitisspentontheconstructionsector.

Ourpapercontributestotheex‐antesimulationsliteratureonchildcareintwodifferentways: i firstlyitusessupply‐sideproviderleveldatatocapturethebehavioralinvestmentmodelofserviceprovidersinthefaceofinvestmentandoperationalgrantsprovidedtothesector–hencetheex‐antemodelnotonlylooksatthe demand side with household data but combines the investment behavior of the service provider incalculating the capacity expansion; and ii secondly, itusesdataon the childcare sectorof adeveloping

country–Turkey‐withsignificantcapacityconstraintsinthechildcaresectorandhencecontributestotheliteraturewhichhas so far focused in termsof ex‐ante evaluations in the child care sectors in developedcountriessuchasNorway,ItalyandBelgium.Similartosomeoftheaforementionedpapers,thispaperlooksatthecapacityincreasesandchangesinutilization–payingparticularattentiontothedistributionalimpactofdifferenttypesandcombinationsofpoliciesincludinganinvestmentgranttoserviceproviders,anongoingpercapitabasedmonthlyoperationalgranttoserviceproviders,andahouseholdtargetedvoucherscheme.Hencesamelevelsofgrantsaredistributedtohouseholdsorcentersandtheimpactofeachintermsofnumberof children enrolled and number of children enrolled from the lower income households is investigated,consideringthecost‐effectivenessofeachscenario intermsofpernewchildenrolledinthesystem aswellas thebenefit incidenceof “new”childcareservicecapacitycreated.Thisapproach is insimilarveinwithKornstadandToresen 2006 wheredifferenttypesofreformsareinvestigatedatthesametimeinordertounderstandwhichoneismorebeneficialtothepublicandcosteffectiveintermsofthebudget.Ourfindingshoweverwillsignificantlydifferfromtheirfindings–giventheinitialconditionsandcapacityconstraintsinthechildcaremarketweanalyzeinTurkey.

3. BackgroundontheChildCareSysteminTurkey

The gross enrollment ratio in pre‐primary education remains low in Turkey, with significant capacityconstraintsinthechildcaresector,particularlyfortheyoungeragegroups.Thepre‐primarygrossenrollmentratewas38.6percent inTurkey in2015asopposed to82.5percent inOECDand93.9percent in theEUcountriesin2013.iCurrentlyabout2.7millionchildrenaged3‐5arenotbenefittingfromcenterbasedchildcareservicesasopposed toonly1.0millionchildrenenrolled.ii Hence,a largenumberof childrendonotbenefitfromcenter‐basedchildcareandpreschoolservicesinTurkey.Inordertoservethesechildrenwhoare currently not enrolled in center based child care or preschool services, a recent World Bank studyestimatesthatatotalof58,380facilitiesneedtobebuilt.iiiFortheyoungeragegroup0‐3,childcarecapacityisclosetozero,withnopublicprovisionofchildcareforthisagegroup.

Childrenfromdisadvantagedbackgroundsareevenlesslikelytobenefitfromcenterbasedchildcareservices.Only16.8percentofthechildrenaged3‐5livinginhouseholdsinthepoorestpercapitaincomedecilebenefitfromcenter‐basedchildcareservicesasopposedto60.0percentofthechildreninthesameagegrouplivinginhouseholdsintherichestpercapitaincomedecile.ivArecentpaperlookingatregulationsintheprivatechildcaremarketinTurkeyfindsthattheinfrastructurerelatedstandardsmaybetoobindingmakingthemodelexpensiveandunaffordableforpoorerhouseholdsinurbanareas Aranet.al, forthcoming .

Privateprovisionof center‐basedchild care services is very low.Asof2011,11.6percentof the childrenenrolledinpre‐primaryeducationinTurkeywereenrolledinprivateinstitutions,whilethisratewas30.5percentfortheworld,37.0percentfortheOECDand29.9percentfortheEUmembercountries.vComparedtotheworldortheEUcountriesthepercentageofchildrenenrolledinprivateinstitutionsremainslowinTurkey, pointing to the fact that an alternative way to increase capacity in Turkey could be throughinterventionsincreasingprivateprovisionoftheservices.

Limitedaccesstoaffordablechildcareisasignificantimpedimenttofemalelaborforceparticipationinthecountry.InTurkeyfemalelaborforceparticipationisalreadylowwith30.3percentin2014comparedtotheOECDandEUaverageswhichare51.4percentand51.5percentrespectively.viItiscommonforwomentodropoutofthelaborforceduetochildcareresponsibilities.Accordingtoahouseholdsurveycollectedin5provincesofTurkey,46percentofthecurrentlynon‐employedwomeninthesamplepreviouslyworkedbutdecidedtoleavethelaborforceandthemajorityofthesewomen 71percent reportedleavingthelaborforce“duetochildcareresponsibilities” WorldBank 2015 .

In2015,theTurkishgovernment,andinparticulartheMinistryofFamilyandSocialPolicieswasinterestedindevelopingpolicies forsupportingtheexpansionofchildcareinTurkey. TheMinistrywasconsideringoptionsforsubsidizingtheseservicesbythemunicipalitythroughinvestmentgrants,operationalgrantstoserviceprovidersorboth.Targeteddemand‐sidevoucherstohouseholdswerealsoconsideredasapolicyoption.Morespecifically,thepolicyscenariosconsideredincluded: i aninvestmentgrantonly Scenario1 ,ii anoperationalgrant withorwithoutapricecap Scenarios2Aand2B , iii throughavoucherscheme

Scenario3 ,and iv throughtheinteractionofaninvestmentgrantwithanoperationalgrantoravoucherscheme Scenarios4A,4Band5inTable1 .Thepurposeofthepolicymeasureswasthree‐fold: i toincreasethe capacity of these services, ii to increase accessibility and affordability of services for children fromdisadvantagedbackgroundsand iii toincreasefemalelaborforceparticipationbyincreasingemploymentratesofmotherswhowillutilizeservicesaswellascreatingnewemploymentatthesechildcarecentersforcare‐takerwomen.ThismodelwasinitiallypreparedinresponsetosuchdemandfromtheTurkishMinistryofFamilyandSocialPolicies.

4. TheModel:Supply‐andDemand‐SideSubsidiesforChildCare

Webuildamodelthatallowsforthreebasictypesofsubsidiesforchildcarethrough i investmentgrantstothe service provider, ii operational grants to the service provider and iii vouchers provided to thehousehold.Investmentgrantsareone‐timegrantsthatareprovidedtonewlyopeningchildcarecenters,inordertosupporttheirinvestmentandincreaseoverallcapacity.Operationalgrantsareprovidedonamonthlyandperchildbasisandtransferredtotheserviceproviders.Theoperationalgranthelpssubsidizetheservicesprovidedbyprivateprovidersvii.Theycoverpartofthecostsofprovision–andhelpreducingpartlythepriceschargedbytheproviders.Thevouchersarealsoprovidedonaperchildpermonthbasis,howeverthemaindifferencebetweenthevoucherandtheoperationalgrantisthatthevouchersareprovidedtohouseholdsand operational grants are given directly to the service provider. Because the voucher is given to thehousehold, this allows for additional "targeting” of the voucher to eligible households. In otherwords, itbecomes possible to “select” poorer households and/or households where women work etc. as voucherbeneficiaries,whilewiththeoperationalgrantthegrantbenefitswhoeverenrollsinthechildcarecenter‐withoutanypurposivetargeting.Figure1providesadescriptiveoverviewofthemodel.

The operational grants and vouchers in themodel have the same exact monthly per child values in thescenarios–andtheyonlydifferbythemechanismofdeliveryandhencethetargetingofthebenefits.Inthevoucherscheme,thehouseholdthatreceivesthevoucherbringsittotheserviceproviderandgetsservicesequivalenttotheamountofthevoucherforfree,andonlypaysfortherestofthefees.Inotherwords,thevouchersenablehouseholdstogetdiscountedservicesbythevalueofthevoucher.Theserviceprovidersthenclaimthevouchersandreceivemoneyfromthegovernmentequivalenttothetotalvalueofthevouchers.Bothoperationalgrantsandvouchershavemonthlyperchildvaluesthoughthepaymentsfromthegovernmenttotheserviceproviderscanbemadeonamonthly,semi‐annualorannualbasisdependingonhowtheprogramisdesigned.

Figure1:Astylizedmodelforsupplyanddemandsidechildcaresubsidies

Thesubsidieshaveanimpactonutilizationofservicesintwoways: i throughtheirimpactonexpandingcapacityandavailabilityofservices,and ii throughtheirimpactonaffordabilityofservices.Inthemodel,investmentgrantshaveanimpactonenrollmentsonlythroughtheirimpactoncapacity.The“benefit”‐ intermsoffinancialgain‐oftheinvestmentgrantisfullyrealizedbytheserviceprovider.Thevouchers‐ontheotherhand‐increaseaffordabilityoftheservicesforhouseholdsandtheassumptionisthattheirbenefitisfullyrealizedbythehousehold.Theimpactofthevouchersonenrollmentismainlythroughtheirimpactonaffordabilityofexistingservicesforhouseholds–henceweassumeintheshorttermtherewillbenocapacityimpactofthevouchersinthesystem.Theoperationalgrantsfallsomewhereinbetween–intermsofitsimpactonenrollments.Whiletheoperationalgrantisreceivedonaperchildbasisbytheserviceprovider,themodelassumesthatthereisa“pass‐through”oftheoperationalgrantstothehouseholdsintheformofareductioninprices.Inotherwords,inourmodeloperationalgrantsleadtodecreasesintheprices–butthedecreaseinprice isaportion in this caseassumedat50% of theoperationalgrantperchild.Thebenefits fromtheoperationalgrantareassumedtobe“shared”betweenthehouseholdsandtheserviceproviders,hencetheoperational grant has an impact on enrollments both through its impact on enhanced capacity throughbenefitsrealizedbytheserviceprovider andthroughitsimpactonaffordability throughbenefitsrealizedbythehousehold .

Weconsiderseveralscenarioswherethesegrantsareprovidedbythemselvesorincombinationwitheachother.Thelevelofthebenefitscanbevariedoncethemodelissetup,weconsiderhereonlyafixedlevelofbenefit 50,000TLone‐timeinvestmentgrantand300TLperchildpermonthoperationalgrantorvouchervalue ,thoughthisparametercanbeadjustedtotakeondifferentlevelsundereachscenario.Theoperationalgrantinthescenarioscanbeprovidedwithapricecapoptionandwithoutapricecapoption inScenario2Aand2B . The scenariosconsider the investmentgrantandoperationalgrantsprovided separatelyand incombinationwitheachothertolookatcosteffectivenesspernewenrollmentineachscenario.Thefulllistof7scenariosisprovidedinTable1.

Table1.Listofpolicyscenarios

Scenario Name Description

SimpleScenarios

Scenario1 SupplySideInvestmentGrantonly:One‐timeinvestmentgranttochildcarecenters

InvestmentGrant:50,000TLgiventotheserviceprovideronetimeforinitialinvestment.

Scenario2A SupplySideOperationalGrantonly:Monthlyperstudentoperationalgrantnopricecap

OperationalGrant:300TLperchildpermonthgiventotheserviceproviderfor5years.

Scenario2B SupplySideOperationalGrantonlywithPriceCap:Monthlyperstudentoperationalgrantwithpricecap

OperationalGrant:300TLperchildpermonthgiventotheserviceproviderfor5years.Pricecap:750TL theserviceprovideragreestolimitthemaximumpriceto750TLperchildpermonthunderthiscontractforthedurationofthesubsidy

Scenario3 DemandSideGrantonly:Voucherstohouseholdsperchildenrolled

VouchertoHousehold:300TLperchildpermonth,targetinghouseholdsinbottom40%ofincomepercentile.

CombinationScenarios

Scenario4A SupplySideInvestmentandOperationalGrantcombined:One‐timeinvestmentgrantandmonthlyperchildoperationalgrantcombined nopricecap

InvestmentGrant:50,000TLgiventotheserviceprovideronetimeforinitialinvestment.

OperationalGrant:300TLperchildpermonthgiventotheserviceproviderfor5years.

Scenario4B SupplySideInvestmentandOperationalGrantcombinedwithPriceCap:One‐timeinvestmentgrantandmonthlyperchildoperationalgrantcombined withpricecap

InvestmentGrant:50,000TLgiventotheserviceprovideronetimeforinitialinvestment.

OperationalGrant:300TLperchildpermonthgiventotheserviceproviderfor5years.Pricecap:750TL theserviceprovideragreestolimitthemaximumpriceto750TLperchildpermonthunderthiscontractforthedurationofthesubsidy

Scenario5 SupplySideInvestmentGrantandDemandSideGrantcombined:One‐timeinvestmentgrantandmonthlyvouchertohouseholdscombined

InvestmentGrant:50,000TLgiventotheserviceprovideronetimeforinitialinvestment.

VouchertoHousehold:300TLperchildpermonth,targetinghouseholdsinbottom40%ofincomepercentile.

5. DataandMethodology

The supply‐ and demand‐side simulation model is built using data sets from households and child careprovidersinTurkey.Auniquedatasetcollectedfromchildcareprovidersin2014in5provincesofTurkey,allowsustohaveinsightsintothemicroeconomicsofrunningachildcarecenterinTurkey,andhenceallowsustolookatthecostsandpricesinthemarketforchildcare.Turkeyhassignificantsupply‐sideconstraintsinthechildcaresector,andthesereflectontheresultsofthesimulationmodel.Theinitialconditionsofthechildcaremarketmattersignificantlyindeterminingtheoutcomesandbenefitincidenceofgrantsdistributedforexpandingchildcareutilization.Thissectionfirstdescribesthedatasetsthatwereusedforthesimulationsandthemethodologyindetail.

5.1DataUsedfortheSimulationInordertounderstandtheimpactofdifferentkindsofsubsidieswehavetwomainsimulationmodelsthatinteractwithoneanotherinproducingtheresults.Thesemodelsare i thesupply‐sidesimulationmodelandii thedemand‐sidesimulationmodel.Usingthesemodelsandchangingtheparametersforthesubsidytypeandthelevelofthesubsidyweestimatepossiblebenefitsofeachscenario.Below,wedescribethedatausedforthesimulationsandthedetailsofeachmodel.

5.1.1Supply‐SideSimulationDataForthesupply‐sidemodelweusethedatathatwerecollectedbyDevelopmentAnalyticsin2014fortheWorldBankstudy“SupplyandDemandforChildCareServicesinTurkey” WorldBank,2015 from603center‐basedchildcareprovidersin5provinces,namelyIstanbul,Gaziantep,Denizli,EskişehirandSamsun.Thedatasetincludesawiderangeofinformationonchildcarecenterssuchastheirset‐upandoperationalcostsandthepricestheychargeperchild.Forthepurposesofthisanalysis,weonlymakeuseofasamplecomposedoftheprivateprovidersinthedatasetmakingupatotalof226providers.ThisuniquelydesignedsurveyallowsustohaveinformationonthecostsandpricesfacedbyprivateprovidersinthechildcaremarketinTurkey,andallowsustomodel thebreak‐evenpointsand internalrateofreturnoninvestments in thechildcaresector.Combiningthisinformationwithamodelontheinvestmentbehaviorofprivateproviders,weareabletocomeupwithdifferent levelsof investmentsandcapacity increaseswithdifferent levelsof thesubsidywhetherthroughtheinvestmentoroperationalgrantstofacilities .

5.1.2Demand‐SideDataForthedemand‐sidemodelwemakeuseoftheTurkeyEU‐SILC SurveyofIncomeandLivingConditions 2012dataset.TheEU‐SILCdatasetincludesawiderangeofinformationonhouseholdsincludinginformationonchildren’senrollmentincenter‐basedchildcareandlaborforceparticipationofhouseholdmembers.Thisdatasetisusedinordertounderstandthebenefitincidenceofenrollmentsasaresultofthecapacityincreasethatoccursinthesupply‐sidemodel,thechangesinenrollmentasaresultofincreasedaffordability.Weonlymakeuseofthepartofthedatasetcollectedfromhouseholdslocatedinurbanareasastheproposedcapacityexpansionsarelikelytobenefitinitially‐andprimarily‐theurbanareasinthecountry.Thedemand‐sidedata are used to model the behavior of the household given increased availability of child care in theneighborhood,andincreasedaffordabilityofchildcaregiventheoperationalgrant whichleadstoreducedpricesinthemarketthroughapass‐throughassumption orthroughthevouchers.

5.2Methodology:BuildingtheSimulationModelInthesupply‐anddemand‐sidesimulations,usingexistingdataweaimtomodeltheeffectsofanoutsideshocktothesystem.Inourcase,thisshockcouldbechildcarecentersreceivinginvestmentoroperationalgrantsorbothorhouseholdsreceivingavoucher.Therearemultipleoptionswhichcouldbeadoptedinasubsidypolicyframeworkandweinvestigateinoursimulationmodelscombinationsofthesepolicieswithdifferentlevelsoftheinvestmentandoperationalgrantsandvouchers.ThescenariosareasdescribedinTable1.

5.2.1Supply‐SideSimulationWe build the supply‐side simulation model, using the supply‐side provider level survey collected in 5provinceswherewehavedetailedinformationonthesetupandoperationalcoststructureandpricesfacedby child care providers. Before a shock is introduced to the system, each provider faces certain setupinvestment and operational costs and certain prices and enrollments. All of these data come from theempiricaldataset,butaretreatedinthemodelashypotheticalcases asifinvestmentsinthesecentershavenotyetbeenmade .Aprobabilityofinvestmentiscalculatedforeachserviceproviderinthebaseline.Wecallthemhypotheticalcasesastheyactuallyexistinreality‐butinthemodeltheyaretreatedasprobabilisticinvestments–suchthataserviceproviderthatfaceslowercostsandbetterpricesinthemarkethasahigherinitiallikelihoodofinvestment.Anoutsideshockisintroducedtothemodelthatincreasesthelikelihoodforinvestment for the child care centers.With these simulations our aim is to findout the “expected” addedcapacityandhowmanymorecenterswillbeestablishedwhenthereareinvestmentgrantsoroperationalgrants ofvaryingsizes andtoseewhichonesleadtoahighercapacityincrease.

CalculatingtheNetPresentValueandtheProbabilitytoInvest

First,thenetpresentvalueforeachprivatechildcarecenterinthedataiscalculatedinordertofindouttheirinitialprobabilitytoinvest,giventheir i priceperchild, ii setupcostiftheyweretoopenupthecenternow, iii monthlyvariablecosts,and iv totalnumberofchildrenenrolled.Netpresentvalueofeachchildcarecenteriscalculatedforaperiodof10yearsbyalsotakingintoaccountthevalueofthecenterifitweretobesoldattheendofthis10‐yearperiod.

Theequationforthenetpresentvalueofachildcarecenteriscalculatedasfollows:

NetPresentValue SetupCost

∗ ∗ 12 ∗ 1 1

TerminalValueviii.

Eq1

Where:

Setupcostisthesetupcost e.g.ifthecenterweretobeopenedupnowwhatthesetupcostwouldbe ofthechildcarecenterasreportedbyeachrespondentinthesurvey.

Yearlycashflowisthedifferencebetweenyearlyrevenuescollectedfromfamiliesminusthevariablecosts Thesearepersonnelcosts,stationarycosts,utilitycostsandtravelcosts .

Thediscountraterisassumedtobeequalto40percent.ix Cumulativeandcombinedtaxratetakenat32percentinTurkeyx.

Secondly,thenetpresentvalue NPV ofaninvestmentismappedontoaprobabilityofinvestmentthroughafunctionalformasdescribedbelow SeeFigure2 .Thisnon‐linearfunctionalformisusedtohypotheticallydescribetheinvestmentbehaviorofchildcarecenters.WithincreasingvaluesofNPV,theprobabilitytoinvestincreasesandthisprobabilityrangesbetween0and1.

Figure2.Probabilitytoinvestforeachchildcarecentergiventheestimatednetpresentvalue

P invest

,,

Thisfunctionalformshowsthattheprobabilitytoinvestincreaseswithincreasingvaluesofthenetpresentvalue. Forthosewhohaveanegativenetpresentvalue,aprobabilityof0isdirectlyassignedwithoutusingthefunctionalform .

CalculatingtheEstimatedCapacityIncreaseImpactoftheGrants

Impactsofinvestmentandoperationalgrantsareestimatedbyincludingthesegrantsinthenetpresentvaluecalculations and recalculating the net present values of each child care center after the grant shock. Theoperational grant reduces the operational costs faced by the service provider, and the investment grantreducestheburdenoftheinvestment setup costsoftheserviceproviderwhichisreflectedinthecalculationofthenetpresentvalue‐andfromthere‐theprobabilitytoinvest.

Foraninvestmentgrantthenetpresentvalueequationbecomesasfollows:

NetPresentValue SetupCost ∗

∗ 12 ∗ 1

1TerminalValue

Eq 2

Foranoperationalgrantwithoutapricecapthenetpresentvalueequationbecomesasfollows:

NetPresentValue SetupCost

∑ ∗

∗ ∗ ∗

∑

∗ ∗ ∗

Eq 3

Operationalgrantsareassumedtobegivenforaperiodoffiveyears.

Foranoperationalgrantwithapricecapthenetpresentvalueequationisthesameasabove,butonlysomeoftheschoolsthathavelowerinitialpricesoptinfortheprogram.Forthosechildcareprovidersthatinitiallyare able to charge high prices, itmaynot be favorable to lock themselves into theprice cap.We assumethereforethatasub‐groupofchildcarecenterswillopt‐inforthegrant thosewithalreadyhighpriceswilloptoutofthisprogram .Thenewpricesarethencalculatedasfollowsforallcenters:

Price_new Price_oldifPrice_old‐operationalgrant*pass‐through Pricecap Thosewhoopt‐out

Eq 4

Price_new Price_old‐operationalgrant*pass‐throughifPrice_old‐operationalgrant*pass‐through Pricecap Thosewhoopt‐in

Eq 5

Itisassumedthatafter5years,thepricesgobacktowhattheywerebeforethepricecapwasintroduced.Forbothof theoperationalgrantscenariosapass‐throughrate isassumed.This lets themodeldistribute thebenefittothehouseholdsbyleadingtoadecreaseinprices Passthroughrateisassumedtobe0.5 .

Followingthenetpresentvaluerecalculation,theprobabilitytoinvestisalsorecalculatedforeachcenterandforeachgrantscenario.Oncethebaselineprobabilitytoinvestandprobabilitytoinvestundereachscenarioareallcalculated,theexpectedcapacityforeachcenteristhencalculatedasfollows:

E Capacity Baseline ProbabilitytoinvestinBaseline*Totalcapacityofthecenter

Eq 6

E Capacity ScX ProbabilitytoinvestinSc.X*Totalcapacityofthecenter

Eq 7

These expected capacities are then summed up and the percent capacity increase achievedwith Sc. X iscalculatedwiththefollowingequation:

%IncreaseintotalcapacityScX E Totalcapacity ScX‐E Totalcapacity Baseline /E Totalcapacity Baseline

Eq 8

5.2.2Demand‐SideSimulationThe next step is to tie the capacity increase coming from the supply‐side simulation to the demand‐sidehouseholddatainordertoseewhoisimpactedbythenewlycreatedcapacityandenrollments.Thecapacityincreasefoundinthesupply‐sidesimulationisusedasthe“exogenousshock”inthedemand‐sidesimulations.

Ouraimineachscenarioistofindouthowmuchcapacityisestimatedtoexpand,howmanynewchildrenareexpectedtoenroll ineachscenario,whobenefitsfromthecapacityincreasesandpricereductionsandthetotalestimatedimpactonmothers’employment.Wealsocalculatethecostimplicationsineachscenario netoftaxescollectedasaresultoftheintervention aswellascost‐effectivenessperchildenrolledandperwomanemployedintothelabormarket.

Theimpactonchildren’spreschoolenrollmentIn themodel,wecanexpectan increase inchildren’senrollment fromtwodifferentchannels: i the firstchannelworksthroughtheexpansionofcapacityandhenceenrollments anewcenterisopenedinthechild’sneighborhoodand thosemost likely toenrollbecomebeneficiariesof thisnewlycreatedcapacity; ii thesecondchannelworksthroughaffordability thechildbecomesenrolledbecausenowitismoreaffordableforthefamilytosendthechildtodaycare .Thescenariosconsideredmaycontributetoenrollmentsthroughtheirimpactoncapacityorthroughtheirimpactonaffordabilityforhouseholds,orboth.

The investment grants, where they are applied, are assumed to have an impact on enrollments throughcapacityincreases–andnotthroughincreasedaffordability.Hence,itisassumedinthemodel,thatthefullbenefitoftheinvestmentgrantwillbecapturedbytheserviceprovider.Theyhaveanimpactonenrollmentsonlythroughnewcapacitycreated.xi

Foroperationalgrants,the increaseinenrollmentsisduetothecapacityeffectaswellastheaffordabilityeffect incomeeffectatthehouseholdlevel ,asweassumethatthebenefitfromtheoperationalgrantsaresplitbetweentheserviceproviderandthehousehold.Forvouchers,theeffectonenrollmentsisthroughtheaffordabilityeffect theincomeeffectatthehouseholdlevel .Intheshortterm,weassumethatthereisnocapacityimpactasaresultofthevouchers.xii

Table2:SummaryofEnrollmentEffectsintheModelunderDifferentSubsidySchemes

ImpactonEnrollmentsthroughImpactonCapacity

ImpactonEnrollmentsthroughImpactonAffordability IncomeEffect

InvestmentGrantasinScenario1

OperationalGrantasinScenario2Aand2B

VouchersasinScenario3

Theestimatedexpectedcapacityincreaseduetotheinvestmentgrantsandoperationalgrantscomingfromthesupply‐sidesimulationisincorporatedintothemodel.Inordertoachievethis,thenumberofchildrenaged3‐5whoarecurrentlyenrolledinpreschoolordaycareintheEUSILC2012datasetismultipliedwiththeestimatedcapacityincreasecomingfromthesupply‐sidesimulationforeachscenario.ThisallowsustofindoutthenumberofspotsthatwillbeavailableforthechildrenwhoarenotcurrentlyenrolledintheSILCdataset.

Oncethenewcapacityiscalculated,thisvalueisdividedbythetotalnumberofchildrenattendingpreschoolinTurkeyinthebaseline.Lastly,thisvalueisusedasthepercentcapacityincreaseanditismultipliedbythenumberofchildrenattendingpreschoolinthedataset.ThisagainallowsustofindoutthenumberofspotsthatwillbeavailableforthechildrenwhoarenotcurrentlyenrolledintheSILCdataset.

AfterthecapacityincreaseamountsinSILCarealldetermined,thenumberofnewspotsduetotheincreasedcapacityareallocatedtothechildreninthedatasetaged3‐5,basedontheirpropensitytoattendpreschool.Thispropensityiscalculatedusingaprobitmodelwhichusesawiderangeofindependentvariablestopredictachild’spropensitytoattendpreschool.Theprobitequationisasfollows:

P attendingpreschoolordaycare φ X’ Eq 9

where the independent variables X are as follows: child’s age, mother’s level of education, mother’semployment status, father’s level of education, father’s employment status, per capita household income,householdsize,numberofadultsinthehouseholdaged65 ,numberofchildreninthehouseholdaged0‐6andregionthehouseholdislocatedin.

Usingthisequation,theprobabilitytoattendiscalculatedforallchildrenwhethertheycurrentlyattendacenter or not. The calculated capacity increase is allocated to the children starting from the childwho iscurrentlynotattendingbuthasthehighestpropensitytoattendchildcareamongthenon‐attendersandtheallocationstopswhentheincreasedcapacityisfilled.Thisexerciseallowsustoobservewhoismorelikelytobenefitfromacapacityincreaseasaresultofanoperationalgrantorinvestmentgrant.

Forthecaseofoperationalgrantwithapricecapaslightlydifferentexerciseneedstobeadopted.Inthepricecapscenario,thechildcarecentersinthesupply‐sidemodelwiththehighestpriceskeeptheirpricesanddonotopt‐inforthegrantsinceitwillnotbeprofitableforthemtodoso.Hence,weneedtoincorporatethisfactintoourdemand‐sidesimulation.Wetrytoachievethisbyassumingthatchildrenwhodonotcurrentlyattendpreschoolinhouseholdsinthehighestincomequintilewillnotbenefitfromthecapacityincreasessincetheschooltheirfamilieswouldpreferwouldnotopt‐inforoperationalgrants.xiii

Overall to calculate the numberof children enrolleddue to capacity increase, individualweights of thesechildrenwhoarenewlyenrolledaresummedup,usingtheequationbelow:

NumberofchildrenenrolledduetothecapacityincreaseScX ∑ % ∗

Eq 10

Operationalgrantsalsohaveapositiveeffectthroughincreasingaffordabilitybydecreasingprices.Toshowthis impact, the probability to attend preschool is calculated for the children i P1‐when there is anoperationalgrantandchildrenareallocatedthenewcapacityand ii P2‐whenthereisanoperationalgrantandchildrenareallocatedthenewcapacityandthereisalsothepositiveincomeshock.Theseprobabilitiesare calculated using the probit equationdescribed above. As a second step, expected number of childrenbenefiting fromaffordability increases are calculated bymultiplying theprobabilitieswithweights of thechildrenandaddingthemupforthewholepopulation.Intheoperationalgrants,thebenefitofthesubsidyissharedbetweenthehouseholdsandserviceproviders,hencethereisanincreaseinenrollmentsduetoboththecapacityincreaseandtheaffordabilityincrease theincomeeffect .

Forthevoucherscenario,themainimpactisthroughtheincomeeffectintroducedbythevoucher–hencetheincreasedenrollment isdueto increasedaffordabilityasaresultof thevoucher.Theprobability toattendpreschooliscalculatedforthechildrenusingtheprobitequationabove i inthebaselinewherethereisnovoucherand ii whenthereisavouchercreatingapositiveincomeeffect.Inthesecondstepexpectednumberofchildrenattendingpreschoolinthebaselineandwhenthereisanincomeeffectwillbecalculated.Notethatthereisnocapacityimpactintheshortterminthevoucherscenario,sincethebenefitisfullyabsorbedbythehouseholds.

Theexpectednumberofchildrenattendingpreschoolasaresultoftheaffordabilityincreasesiscalculatedusingtheequationbelow:

E NumberofchildrenenrolledduetoaffordabilityincreaseScX P2‐P1 *individualweightinthedataset

Eq 11

In summary, foreach scenario thenumberof childrenwhowillbe enrolleddue to the capacity impact issummedupwiththepredictednumberofchildrenwhowillbeenrolledinchildcareduetotheaffordabilityincreaseinordertocomeupwiththetotalnumberofnewlyenrolledchildren.

Number of newly enrolled childrenSc X Number of children enrolled due to the capacityincreaseScX E NumberofchildrenenrolledduetoaffordabilityincreaseScX

Eq 12

5.2.3FiscalImpactandCostEffectivenessDifferentcalculationsarenecessarytocalculatethecostoftheinvestment,operationalgrants,thevouchersandthescenarioswheremunicipalitiesopenupchildcarecenters.

Forinvestmentgrants,wefirstestimatetheexpectednumberofnewchildcarecentersthatwillbeopenedupasaresultofthesubsidy.Thisiscalculatedbydividingthenumberofnewenrollmentsbytheaveragesizeofachildcarecenter inourcase,thiswas50childrenpercenteronaverage .Theexpectednumberofnewchildcarecenters countrywide ,ismultipliedwiththeleveloftheinvestmentgrant.

Cost Number of children enrolleddue to the capacity increase/Average capacity of acenter *Investmentgrant

Eq 13

Fortheoperationalgrant,whenthereisnopricecap,alltheschoolsthatalreadyexistareassumedtoopt‐in,hencethecostof thegrantwillbeequalto thecountrywide levelofexpectedenrollmentmultipliedwithyearlyvalueoftheoperationalgrantandsummedupover5years.

Totalcost

∗

∗ 12

1

Eq 14

For theoperational grantwith theprice cap, first thecountry‐widenumberof childrenenrolled fromthebottom80%iscalculatedusingSILC.xivThentheestimatednumberofchildrenwhowillbenewlyenrolledisaddedtothisvaluetofindthetotalnumberofbeneficiaries.Similartothepreviousstep,5yearvalueoftheoperationalgrantiscalculatedtofindthetotalcost.

Totalcost

80%

∗

∗ 12

1

Eq 15

Thecostofthevouchersiscalculatedbyfirstaddingupnumberofchildrenalreadyenrolledfromthebottom40%withtheexpectednumberofchildrennewlyenrolledtochildcareafterthevouchershockisreceivedandthenmultiplyingthisvaluewiththevalueofthevoucher.Thisvalueisthensummedupfora5yearperiod.

Totalcost

40%

∗

∗ 12

1

Eq 16

Twokindsoftaxrevenuearecalculatedforeachscenario.Thesearecorporationtaxrevenuesandincometaxrevenuesfromnewlyemployedstaffatthecenters.

i Taxrevenuesobtainedfromthenewlyopeningcenters:Thisisintheformofincomeandcorporationtaxesmakingacumulativetotalof32%oftheprofitsxv;and

ii Taxrevenuesobtainedfromemployeecontributions:Thisiscomposedofthewomenworkinginchildcarecentersthatopenedupasaresultofthegrants.Henceasalowerboundestimatefortaxrevenue‐allthewomenemployedinnewchildcarecentersareassumedtobereceivingtheminimumwagexvi.Netpresentvalueoftaxrevenuesarecalculatedforaperiodof10years.

Netcostiscalculatedforeachscenariobytakingthedifferencebetweentotalcostsandtotaltaxrevenues.Finally,thecosteffectivenessofeachscenarioiscalculatedbydividingthenetcostsofeachscenariowiththetotalnumberofnewlygeneratedenrollmentsthroughthecapacityandincreasedaffordabilityeffectsofthescenarioanddividingthisnumberfurtherby10whichistheassumedyearsofoperationforthesecenters.

6. MainFindings

Theresultsofthesupply‐anddemand‐sidesimulationsarepresentedinthissectionwhenthemethodologyisappliedtothecaseofTurkeytakingintoconsiderationthe i impactoncapacityandestimatednumberofnewcenters ii impactonenrollmentswhicharederivedbothfromcapacityincreaseaswellasaffordabilityincrease in the household and hence may be higher than the amount of capacity increased iii benefitincidenceofnewenrollmentsandgrants %ofchildreninthebottom40%oftheincomedistributionthatbenefitfromtheexpansionandhouseholdsinthebottom40%thatbenefitfromthegrants , iv costsandrevenuesand v costeffectivenessintermsofthenetcostofthepolicyperchildenrolled.

6.1ImpactoncapacityandestimatednumberofnewcentersCapacityincreaseismaximizedinthescenariothatcombinestheinvestmentgrantandtheoperationalgrantanddoesnotimposeapricecap .Thisscenario providedunderScenario4A expandsexistingpreschoolcapacityby30percent.Theestimatednumberofnewlyopenedcentersinthisscenariois4,132creatingspacefor 206,608 newly enrolled children additionally 29,738 children are expected to be enrolled due to anaffordabilityincrease .Whenapricecapisintroducedtothesescenarios Scenario4B theimpactoncapacityisslightlylowerthoughstillhigherthanifaninvestmentgrantoroperationalgrantwereprovidedalone.

Whenan investmentgrant isprovidedalone Scenario1 , there is a small increase in capacity,while theimpactoncapacityofonlyprovidingoperationalgrantsisprojectedtobehigher Scenario2Aand2B .Withaone‐timeinvestmentgrantandwithoutoperationalgrants,thecapacityincreaseisestimatedtobeabout6percent.Theone‐timeinvestmentgrantdoesnotcreateenoughincentiveinthemodeltoexpandcapacityforchildcarecenters,rathertheoperationalgrantsseemtohaveamoreconvincingimpact:whenoperationalgrants–withorwithoutapricecap‐areprovidedperchildpermonthfor5years,theimpactoncapacityisestimatedtobe17percentwithpricecapand24percentwithoutthepricecap anestimated2,394and3,306newcentersrespectively .

Whereanoperationalgrantisintroducedwithapricecap Scenario2B ,thecapacityincreaseis–asexpected‐smallercomparedtonopricecapscenarios.Inthesupplysidesimulationmodel,whenanoperationalgrantisintroducedwithapricecap,childcarecentersthatchargemorethanthecapdonotfinditprofitabletooptin.Asaresult,capacity increaseremainssmallercomparedtothescenariowithnopricecapwhereeveryserviceproviderwouldhavetheincentivetoopt‐in.Henceforinstanceinthescenariowhenanoperationalgrantof300TLisprovidedwithoutapricecapacapacityincreaseof24percentisexpectedasopposedtoacapacityincreaseof17percentwhenthesamegrantisprovidedbutwithapricecapof750TLpermonthperchild.

Thevouchers‐whenprovidedalone‐arenotexpectedtocreatenewcapacityintheshortterminthemodel.Theimpactofthevouchersonenrollmentsarebuiltintothemodelthroughtheirimpacton“affordability”ofchildcareforthehousehold,hencetheyworkthroughthedemand‐sidesimulationandareprojectedtohavenoimpactonthesupply‐sidecapacityintheshortterm.

6.2Impactonpreschoolenrollment

Theincreaseintheenrollmentfigurescomefromtheincreaseincapacityinthesupply‐sidemodel whichisassumedtobe takenup fullybynewenrollments aswellas the increase inaffordabilityof childcarebyhouseholdsinthedemand‐sidemodel.Asexplainedinthemethodologysection,theinvestmentgrantshaveanimpactonenrollmentsonlythroughtheirimpactoncapacity,whereasthevouchersimpactenrollmentsthroughtheirimpactonaffordabilityofchildcare henceareassumedtohavenoimpactoncapacity .Theoperationalgrantshaveanimpactonbothcapacityandaffordabilitybythehousehold–thebenefitderivedfromthegrantsareassumedtobesharedbythechildcarecentersandhouseholdsequallyandenrollmentsincreaseinthesescenariosboththroughthecapacityincreaseonthesupplysideandtheaffordabilityincreaseonthedemandside andhenceincreasedenrollmentinexistingcapacity .NumberofnewlyenrolledchildrenasaresultofeachscenariothroughimpactfromthesetwochannelscanbeseeninFigure3.

Thehighestlevelsofnewenrollmentsareobservedinthesetofscenariosthatincludeinvestmentgrantsandoperationalgrantstogether Scenario4Aand4B .Thisisinparallelwiththefactthatthesescenariosalso

createthehighestcapacityincreases.Theincreaseinenrollmentsishighestwhenthereisnopricecap.Currentenrollmentrateincreasesby9.6percentagepointsinthecasewheretheone‐timeinvestmentgrantis50,000TLandtheoperationalgrantis300TLpermonthperchild.Atotalof236,346childrenareexpectedtobeenrolledinpreschool,themajorityofwhichareenrolledthroughthecapacityincrease 206,608children .Whenapricecapisintroduced Scenario4B theincreaseinenrollmentsisestimatedtobe7.6percentagepointsinthisscenario.

Figure3.Numberofnewlyenrolledchildrenduetocapacityoraffordabilityincrease

Source:Authors’calculationsusingSILC2012andSupplySideDataof“SupplyandDemandforChildcareServicesinTurkey”

Providingonlydemand‐sidevoucherstofamilies Scenario3 hasaminimalimpactonenrollments.Provisionofvoucherstothebottom40%ofthepopulationincreasespreschoolenrollmentratebyonly0.9percentagepoints.Asaresultofthisscenario,atotalof22,735childrenareexpectedtobeenrolledinpreschool,allofwhomare enrolleddue to an increase in affordability See Figure 3 . This is due to our assumption thatvouchersdonothaveasupply‐sideimpact,hencetheydonothaveanyimpactoncapacity.Becausethemainconstraintinthesectorisonthesupplyside acapacityconstraint ,thedemand‐sidestimulationintroducedbythesetransferstothehouseholddonotaddressthemainconstraintinthemarketandareineffectiveintermsof increasing enrollments.Wealso see that thevoucher scenariosalsohappen tobe the least cost‐effectivewayofexpandingenrollmentsintermsofmoneyspentpernewchildenrolled SeeSection5.5formoredetails .

Overall affordability increase generates a smaller impact as opposed to the capacity increase. In everyscenario,thenumberofchildrenenrolledduetothecapacityincreaseisalmostalwayshighercomparedtothenumberofchildrenenrolledduetotheaffordabilityincrease SeeAnnexTableA1 .Forinstanceinthescenariowhereonlyanoperationalgrantisprovidedandthereisnopricecap,thenumberofchildrenenrolledduetothecapacityincreaseis165,293whilethenumberofchildrenenrolledduetotheaffordabilityincreaseisestimatedat29,289.Thisisduetothesupply‐sideconstraintsinthesector–andsignalsthatanimpactonenrollmentsinthesectorwillmainlybeachievedbyinterventionsthattargetthesupply‐sidecapacityofchildcareinthecountry.

6.3Benefitincidenceofnewenrollmentsandgrants

Thedistributionalimpactofthegrantsacrossincomegroupsisanotherimportantdimensionanalyzedbythesimulation. In each scenario, who benefits from the expanded capacity andwho benefits from increasedaffordabilitydependsverymuchonwhetherthereisapricecapattachedtothegrantsandwhetherthereisatargetedvouchersystem.

0 50,000 100,000 150,000 200,000 250,000

Scenario 1

Scenario 2A

Scenario 2B

Scenario 3

Scenario 4A

Scenario 4B

Scenario 5

# of newly enrolled children

# of newly enrolled children due to capacity increase

# of newly enrolled children due to affordability increase

Inthescenarioswheretherearenopricecaps,thenewlycreatedcapacityaccruesregressivelyandbenefitsthehigherincomequintilesdisproportionately.Inotherwords,thechildrenwhoarelikelytobeenrolledcomefromfamiliesinthetopquintilesoftheincomedistribution.Forinstance,inthescenarioswiththeoperationalandinvestmentgrantsimplementedtogether scenario4A withoutthepricecap,30.3percentofcapacitygenerated is utilized by children in the bottom40%of the distribution.When there is a price cap in thescenarioswithoperationalgrants,theutilizationofnewcapacitybecomesmoreequitable.LookingatScenario4B where the operational and investment grants are provided together, we find that with the price capimposedat750TLper child, thenewcapacitygeneratedaccruesmore to thepoor–with39.9percentofcapacitybeingutilizedbychildreninthebottom40%oftheincomedistribution.

Thevoucherscanbetargetedtobenefitthepooresthouseholdsandhencetheyhaveappeal.However,whenwelookcarefullyattheresults,becausetheimpactofthevoucherschemeistoosmallonenrollments,theoverallimpactoftheschemeevenonpoorhouseholdsisestimatedtobelimited.Inthevoucherscenario,theenrollments increase by only 1.9 percent for the bottom 40%,while overall enrollments increase by 0.9percent.Thebenefitincidenceofthegrantispro‐poor,meaningthegrantwouldbereceivedonlybythelowincomehouseholds–forinstance,inthetargetedvoucherscheme,44%ofthebenefitsareestimatedtoaccruetohouseholdsinthebottom20%oftheincomedistribution,whichisthebestscenariointermsoftargetingwhen compared to the scenarioswith theoperational grants.However, because the enrollment impact islimitedunderthisscenario sincethevouchersystemworksthroughthedemandsideandthroughimprovedaffordabilityofexistingservices–ratherthanbyincreasingcapacity ,theoverallimpactonthepoorchildrenisverylimited.Wefindthattheenrollmentsofthebottom40%ofchildrenincreasemore about6%pointsinScenario4B inthescenarioswiththeinvestmentandoperationalgrantsprovidedtogether,whileinthevoucherscenarioprovidingthesamelevelofinvestmentgrantsandprovidingthevoucherstothehouseholdtheenrollmentincreaseforthebottom40%isonly2.1percent Scenario5 .Thisisaveryinterestingandimportantresultthatsuggeststhattargetingthepooresthouseholdsthroughademandsidetransfermightnotalwaysbethemostpro‐poorpolicy.InthecaseofchildcareinTurkey,itseemsthatthesehouseholdsdonotbenefitfromtheseservices,notbecausetheyarenotaffordabletothem,butbecausetheyarenotavailabletothem–hencebeforeresolvingsupplysidecapacityissuestargetingtheunaffordabilityproblemthroughademand‐sidetransfertohouseholdsturnsouttobeanineffectivepolicyforexpandingaccesstoservices.

Figure4.Preschoolenrollmentratesofchildrenaged3‐5bypercapitaincomequintile

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 1: Supply Side Investment Grant only

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 2A: Supply Side Operational Grant only

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 2B: Supply Side Operational Grant onlywith Price Cap

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 3: Demand Side Grant only

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 4A: Supply Side Investmentand Operational Grant combined

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 4B: Supply Side Investmentand Operational Grant combined with Price Cap

Figure5.Benefitincidenceofthegrantsand/orvouchersbypercapitaincomequintile

0.2

.4.6

.81

Enro

lmen

t rat

e fo

r Chi

ldre

n ag

ed 3

-5

1 2 3 4 5

Scenario 5: Supply Side Investment Grantand Demand Side Grant combined

0.0 0.0 0.0 0.0 0.0

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 1: Supply Side Investment Grant only

157.3

252.2 265.9311.3

427.6

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 2A: Supply Side Operational Grant only

151.1

247.9 264.0309.6

0.0

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 2B: Supply Side Operational Grant onlywith Price Cap

310.7

413.7

0.0 0.0 0.0

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 3: Demand Side Grant only

Baseline Capacity increase Affordability increase

165.7

284.4 287.0318.3

434.2

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 4A: Supply Side Investmentand Operational Grant combined

163.7

291.4 284.8316.2

0.0

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 4B: Supply Side Investmentand Operational Grant combined with Price Cap

317.0

417.1

0.0 0.0 0.0

020

040

060

080

01,

000

Tota

l ben

efit

(in m

illion

TL)

1 2 3 4 5

Scenario 5: Supply Side Investment Grantand Demand Side Grant combined

6.4CostEffectivenessImplementingeachscenario results in itsowncostsandpotential tax revenues. Inorder to lookat costeffectiveness,wefirstconsidernetcostsofeachscenario,lookingatcostsandpotentialtaxrevenuestobegenerated.Whilethecostsarethetotalcostsoftheinvestmentgrants,operationalgrants,orvouchers,thetaxrevenuesarecollectedintheformof i corporationandincometaxesofthenewlyestablishedcentersand ii employeetaxesoftheemployeesworkinginthesecenters.Thecostsarecalculatedasthetotalcostoftheone‐timeinvestmentgrantandthe5yearon‐goingcostofgrantsandvouchersandareprovidedinnetpresentvalues.ThecostsandrevenuesineachscenarioareprovidedinAnnexTableA1.

Providingoperationalgrantsfor5yearsinadditiontoinvestmentgrantsturnsouttobethemostcostlyalternative.Whenanoperationalgrantof300TLisprovidedforeachchildenrolledinthecentersfor5yearswithoutapricecapinadditiontoa50,000TLinvestmentgrant,thetotalcostis5.8billionTLinnetpresentvalue.Thiscostwhichisthehighestamongtheothersisequivalenttoabout0.30percentofTurkey’s2015GDP.xvii

Whilecostsarehigheringeneralwhenoperationalgrantsareprovided,revenuesarealsohighersincemorecentersareestablishedunderthesescenarios‐ leadingtothecollectionofmorecorporationandincometaxes.Infact,thescenariothatprovidesoperationalgrantsof300TLwithoutapricecapandaninvestmentgrantof50,000TLgeneratesthehighestamountofrevenueswith1.9billionTL.Incomparison,thelowestamountofrevenuesisgeneratedwhenvouchersareprovidedalone.Infactnorevenueisgeneratedfromprovidingvouchersfor5yearssincevouchersdonotleadtotheestablishmentofnewcenters.

Oneofthepoliciespresentedhereare“netrevenuegenerating”policyoptions.Accordingtothesimulationmodel,whentheinvestmentgrantsareappliedontheirown–thenetcostoftheprogramendsupbeingnegative,meaningthatthetaxrevenuefromcorporationtaxesandincometaxesfromemployeesinthesecenters‐overthe10yearperiodforwhichthesecentersareexpectedtooperate‐,exceedtheinitialsetupgrantsprovidedtothesecenters.Forinstance,providinga50,000TLinvestmentgranttothesecentersisestimatedtocost39millionTLinitially,andgenerate353millionTLintaxrevenuesinthe10yearperiodmakingthetotalnetcost‐314millionTLforthisinvestment.Themainassumptionhereisthatthechildcarecenterswouldtruthfullyreportprofitsandthattheemployeeswouldbeformallyemployedatthesecenters.Aslongastheserequirementsaretrackedandenforced,netpositiverevenuesareexpectedtobegeneratedwiththiskindofaninvestmentsubsidy.

In comparison, scenarios with only operational grants and operational grants provided together withinvestmentgrantsenduphavingthehighestnetcost.When300TLoperationalgrantwithoutapricecapisprovidedtogetherwitha50,000TLinvestmentgrantnetcostis4.0billionTL.Butitisusefultorememberthatthesearealsothescenariosthatgeneratethehighestchangesinthecapacityandenrollmentratesforchildren.Inordertoassessthecost‐effectiveness‐andthereforevalueformoney‐ofeachscenario,wehaveplottedthepercentagepointincreaseinenrollmentsasaresultofthepolicyagainsttheperchildnetcostofimplementingeachpolicyinFigure6.

Accordingtothisanalysis,atypologyofscenarioscomeupwhenthemodelisappliedtotheTurkishcase:

Revenue generating scenario Scenario 1 – Investment Grant Scenario : The investment grantscenariosareexpectedtoberevenuegeneratingwithinthe10yearspectrum.Theyhavelittleimpactonenrollmentsandhencearenotadvisedtobeappliedontheirowniftheintentionistoexpandservicesandincreaseenrollments–thoughiftheyareprovidedontheirowntheyarelikelytohaveapositiveimpactonthebudgetinthemediumterm.

LeastCostEffectivescenario: Scenario3–Onlyvoucherscenario :Thescenariothatusesvouchersonlywithoutaninvestmentsupport,isthemostexpensivescenariointheanalysisintermsofperchildenrolled.Thisisbecauseincentivizingahouseholdtousetheseservicesusingthedemand‐sidevouchersonly–whenavailabilityofservicesislow–isdifficulthencethevouchersdonotaddressthemostbindingproblemsinthechildcaresectorinthecountry.

MostCostEffectivescenarios Scenarios2A,4Aand4B–InvestmentandOperationalGrantswithorwithoutPriceCapandonlyoperationalgrantwithoutpricecap :Themostcosteffectivescenariosintermsofbudgetspent per child enrolled as a result of these policies, are represented by Scenarios 2 and 4 where theoperationalandinvestmentgrantsareprovidedtogether.Asdiscussedinsection5.3whenapricecapisintroducedthebenefitincidenceaccruesmoretothepoorhouseholds.HenceScenario4Bwhereoperationalgrantsareprovidedtogetherwithinvestmentgrantsandwithapricecapisthemostcost‐effectiveandpro‐poor scenario in terms of benefit incidence. Overall, these are themost cost effective and hencemostrecommended scenarios in the simulation – whereby the twin goals of attaining higher capacity andenrollmentsarealsomatchedwithaconcernforlowercostsperchildandapro‐poordistribution.

Figure6.Costeffectiveness

Source:Authors’calculationsusingSILC2012andSupplySideDataof“SupplyandDemandforChildcareServicesin

Turkey”

7. Conclusions

Themechanismforthedeliveryofachildcaresubsidymattersforresults.Theinitialconditionsofthechildcaremarket,notonlydeterminethemarginalbenefitincidenceofadditionalsupply,butalsohaveanimpactonthecosteffectivenessoftargetedsubsidymeasures.ThispaperhasdemonstratedapracticalfinancingmodelforchildcareusingempiricaldatafromTurkey’schildcaresector,andapplicabletoothercountries.Combiningthesupply‐sideinvestmentincentivesoffirmsgivenasupply‐sidesubsidyandthedemand‐sideincentivesofhouseholdsinthefaceofmoreavailablesupplyandreducedprices,themodelhaspredictedthecapacityenhancement,benefit incidenceanddistributionofnewcapacityaswellas the incidenceofpublicmoneyspentonhouseholdsbywelfarequintile.Byconsideringthecostofeachmeasureandthenewenrollmentsinducedbyeachscenario,thepaperhascomeupwithamethodologyforpredictingthemostcost‐effective policymeasure for the country for expanding child care throughpublic subsidies throughsupply‐sideinvestmentoroperationalgrantsordemand‐sidevouchersthattargethouseholds.Thepaperhas shown that themechanisms for the transfer alonemake a difference in themodel in howpro‐poor

Scenario 1

Scenario 2A

Scenario 2B

Scenario 3

Scenario 4A

Scenario 4B

Scenario 5

0

2

4

6

8

10

12

-2,000 -1,000 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000

Incr

ease

in e

nrol

lmen

t rat

e (p

erce

ntag

e po

int)

Net annual cost per newly enrolled child (TL)

Revenue generating

Most cost effective

Least cost effective

delivery of services is ‐ and simply targeting “poor households”may not be themost pro‐poorway ofexpandingservicesorsubsidizingchildcare.

Annex

TableA1.Summaryoftheresultsundereachscenario

ScenarioType

Scenario1 Scenario2A Scenario2B Scenario3 Scenario4A Scenario4B Scenario5

SupplySideInvestmentGrantonly:One‐timeinvestmentgranttochildcarecenters.

SupplySideOperationalGrantonly:Monthlyperstudentoperationalgrant nopricecap

SupplySideOperationalGrantonlywithPriceCap:Monthlyperstudentoperationalgrant withpricecap

DemandSideGrantonly:Voucherstohouseholdsperchildenrolled

SupplySideInvestmentandOperationalGrantcombined:One‐timeinvestmentgrantandmonthlyperchildoperationalgrantcombined nopricecap

SupplySideInvestmentandOperationalGrantCombinedwithPriceCap:One‐timeinvestmentgrantandmonthlyperchildoperationalgrantcombinedwithpricecap

SupplySideInvestmentGrantandVouchersCombined:One‐timeinvestmentgrantandmonthlyvouchertohouseholdscombined

Grantlevels

InvestmentgrantlevelTL‐onetimegrant

50,000₺

‐₺

‐₺

‐₺

50,000₺

50,000₺

50,000₺

OperationalgrantlevelTLpermonthperchild

‐₺

300₺

300₺

‐₺

300₺

300₺

‐₺

VoucherlevelTLpermonthperchild

‐₺

‐₺

‐₺

300₺

‐₺

‐₺

300₺

PricecapTLpermonthperchild

‐ ‐ 750₺

‐ ‐ 750₺

‐

Results %capacityincrease A

6 24 17 0 30 24 6

#ofnewlyenrolledchildrenduetocapacityincrease

B 39,145 165,293 119,688 0 206,608 166,974 39,145

#ofnewcenters C B/50B/100forSc.1

783 3,306 2,394 0 4,132 3,339 783

#ofnewlyenrolledchildrenduetoaffordabilityincrease

D 0 29,289 20,371 22,735 29,738 19,430 25,442

Totalnumberofnewlyenrolledchildren

E B D

39,145 194,581 140,060 22,735 236,346 186,404 64,586

Increaseinenrollmentrate percentagepoint

F E/ Totalnumberofchildrenaged3‐5

1.6 7.9 5.7 0.9 9.6 7.6 2.6

Increaseinenrollmentrateofthebottom40% percentagepoints

G 0.0 4.1 3.6 1.9 6.0 6.2 2.1

%ofthenewlyenrolledchildrenfrombottom40%

H

0.0 25.2 30.8 100.0 30.3 39.9 39.4

TotalbenefitaccruingtohouseholdsmillionTL

I

0 1,414 973 724 1,490 1,056 734

%ofthebenefitgoingtobottom40% J

. 29 41 100 30 43 100

%ofthebenefitgoingtobottom20% K

. 11 16 43 11 16 43

Totalcost millionTL L

39 5,353 3,612 1,292 5,832 4,081 1,619

Totalnumberofnewpersonnelneeded M

2,349 9,918 7,181 0 12,396 10,018 2,349

Totaltaxrevenuefromnewlyemployedwomenincenters TL

N

20 84 61 0 105 85 20

Totaltaxrevenuefromcorporationandincometax 32%ofprofits

O

333 1,407 1,019 0 1,759 1,422 333

Totaltaxrevenue millionTL P N O

353 1,492 1,080 0 1,864 1,507 353

TotalNetCost millionTL Q L‐P

‐314 3,861 2,532 1,292 3,968 2,574 1,266

NetannualcostpernewlyenrolledchildTL

R Q/E /10

‐803 1,985 1,808 5,683 1,679 1,381 1,960

References

Aran,M.A.,Munoz‐Boudet,A.M,andAktakke,N.“CanRegulationsMakeitMoreDifficulttoServethePoor?TheCaseofChildCareServicesinIstanbul,Turkey”JournalofHumanDevelopmentandCapabilities forthcoming .

Amarante, Verónica, Rodrigo Arim, Gioia De Melo, and Andrea Vigorito. "Family allowances and child schoolattendance:anex‐anteevaluationofalternativeschemesinUruguay."InChildWelfareinDevelopingCountries,pp.211‐245.SpringerNewYork,2010.

Apps,Patricia,JanKabátek,RayRees,andArthurvanSoest."Laborsupplyheterogeneityanddemandforchildcareofmotherswithyoungchildren."EmpiricalEconomics 2016 :1‐37.

Bourguignon,François,FranciscoHGFerreira,andPhillippeG.Leite."Conditionalcashtransfers,schooling,andchildlabour:micro‐simulatingBrazil'sBolsaEscolaprogram."TheWorldBankEconomicReview17.2 2003 :229‐254.

Del Boca, Daniela, and Daniela Vuri. "The mismatch between employment and child care in Italy: the impact ofrationing."JournalofPopulationEconomics20.4 2007 :805‐832.

İlkkaracan,İpek,KijongKim,andTolgaKaya."TheImpactofPublicInvestmentinSocialCareServicesonEmployment,GenderEquality,andPoverty:TheTurkishCase."ResearchProjectReport,LevyEconomicsInstituteofBardCollege.August 2015 .

Kornstad,Tom,andThorO.Thoresen."EffectsoffamilypolicyreformsinNorway:Resultsfromajointlaboursupplyandchildcarechoicemicrosimulationanalysis."FiscalStudies27.3 2006 :339‐371.

Leite, Phillippe G., Ambar Narayan, and Emmanuel Skoufias. "How do ex ante simulations compare with ex postevaluations?evidencefromtheimpactofconditionalcashtransferprograms."EvidencefromtheImpactofConditionalCashTransferPrograms June1,2011 .WorldBankPolicyResearchWorkingPaperSeries,Vol 2011 .

Vanleenhove, Pieter. "Full childcare coverage: higher maternal labour supply and childcare usage?." EUROMODWorkingPaperNo.EM19/13.AvailableatSSRN2381527 2013 .

Calderon,Gabriela.TheEffectsofChildCareProvisioninMexico.BancodeMexicoWorkingPapers2014‐07.BancodeMexico, 2014. Available at: http://www.banxico.org.mx/publicaciones‐y‐discursos/publicaciones/documentos‐de‐investigacion/banxico/%7B0AFBA0F8‐0973‐907E‐9ED4‐5AC6E445E064%7D.pdf

MateoDíaz,Mercedes,andLourdesRodriguezChamussy.Childcareandwomen'slabourparticipation:evidenceforLatinAmericaandtheCaribbean.Inter‐AmericanDevelopmentBank,2013.

Staab, Silke, and Roberto Gerhard. "Childcare Service Expansion in Chile and Mexico." Gender and DevelopmentProgrammePaperno:10.UnitedNationsResearchInstituteforSocialDevelopment,2010.

OECD.EncouragingQuality inEarlyChildhoodEducationandCare.OECDpolicynote.OECD,2012.Available fromhttp://www.oecd.org/education/school/49143900.pdf.

Warner,Mildred E., andRaymondHJMGradus. "The consequences of implementing a child care voucher scheme:EvidencefromAustralia,theNetherlandsandtheUSA."SocialPolicy&Administration45.5 2011 :569‐592.

WorldBank.2015.SupplyanddemandforchildcareservicesinTurkey:amixedmethodsstudy.Washington,D.C.:World Bank Group. http://documents.worldbank.org/curated/en/2015/09/24997835/supply‐demand‐child‐care‐services‐turkey‐mixed‐methods‐study

Endnotes

iRateforTurkeyisfromtheMinistryofEducation’sNationalEducationStatistics2015/16,grossenrollmentratefor3‐5yearolds.RatesforOECDandtheEUarefromWorldBank,WorldDevelopmentIndicators grossenrollmentratio,pre‐primary,bothsexes % iiWorldBank2015.SupplyandDemandforChildCareServicesinTurkeyiiiWorldBank2015.SupplyandDemandforChildCareServicesinTurkeyivWorldBank2015.SupplyandDemandforChildCareServicesinTurkeyv World Bank, World Development Indicators. Percentage of enrollment in pre‐primary education in privateinstitutions % viRateforTurkeyisobtainedfromTurkishStatisticalInstitute’swebsite,itistheannualfigurefor2014forwomen15yearsoldandolder.RatesforOECDandEUareobtainedfromWorldDevelopmentIndicatorsofWorldBank,theyareforwomen15yearsoldandolderandtheyarenationalestimates.viiNotethattheseproviderscanalsobemunicipalitiesorcrecheswithinworkingplaces.Aslongastheyareregulatedandconformtocertainstandards,theycanbeeligibleforthisfunding.

viiiTheterminalvalueiscalculatedusingtheEBITDA EarningsBeforeInterest,Taxes,DepreciationandAmortization Multiple. The final year’s profits is multiplied by a constant, in our model this value is set at 10.5 followinginternational benchmarking. See source: BAIRD 2015. Education Services Report.http://content.rwbaird.com/RWB/sectors/PDF/TechnologyServices/education‐services‐report.pdfixAssumptionhereisthatabovetheriskfreeinterestratethereneedstobeabouta30%premium.xThisisassumedtobe32%ofprofitsbasedontaxratesinTurkey;calculatedbyfirsttaking20%oftheprofitascorporatetaxandthentaking15%oftheremainingprofitasincometax.Thelevelofcumulativetaxesonprofitswouldvarybycountry.xiThisalsoholdsforthescenariosforthemunicipalitiesopeningupcrèchesaswell.Thesescenarios Scenario1 alsoonlyhaveanimpactonenrollmentsthroughincreasedcapacity.

xiiInotherwords,thefullbenefitofthevoucheriscapturedbythehouseholdsthatreceivethem withoutanincreaseinprices andthefulleffectofaninvestmentgrantiscapturedbytheserviceprovidersthatreceivethem withoutareductioninprices .Thesesimplifiedassumptionsallowustorunthemodelsmoothly,althoughvariationsofthesescenarioscanbeapplied.Forthepurposesofthissimplemodel,weproposetosticktothesedefinitions.xiiiThisassumptionisreasonableduetotheinformationgatheredinthedemandsidesurveyin2014fortheWorldBankstudySupplyandDemandforChildcareServicesinTurkey.Accordingtothedata,motherslivinginhouseholdswithpercapitaincomegreaterthan821TL theminimumpercapitamonthlyincomeforthe9thdecileaccordingtoSILC2012dataset or1012.59TLin2014values,haveawillingnesstopayof780TLonaverage.Hencewithanypricecaplowerthanthisvalue,ourassumptionisreasonable. Inordertofindthe2014nominalvalueofthe2011percapitaincomevalue,weusedDecember2011TUFEindexof200.85andDecember2014TUFEvalueof247.72obtainedfromTurkishCentralBank’swebsite. xivDuetotheassumptionthattop2decileswilloptoutoftheprogramsincetheirwillingnesstopayishigher.xvCumulativetaxiscalculatedas:20percentofprofitsascorporatetaxesand15%oftheremainder 80percentofrevenue asincometaxes.xviThetaxcollectedbythegovernmentforeachwomaniscalculatedbyaddingupthetaxesontheminimumwage,makingupatotalof98.98TLaspostedbytheMinistryofLabourandSocialSecurityforthefirst6monthsof2016.Using the information related to minimum wage in The Ministry of Labor and Social Security’s website.http://www.csgb.gov.tr/Contents/Istatistikler/AsgariUcretxvii Turkey’s 2015 GDP is 1,953,561 million TL in nominal terms. Source:http://www.tuik.gov.tr/PreHaberBultenleri.do?id 21510