Directing remittances to education with soft and hard commitments: Evidence from a lab-in-the-field experiment and new product take-up among Filipino migrants in Rome

(1)

ContentslistsavailableatScienceDirect

Journal

of

Economic

Behavior

&

Organization

jo u r n al ho me p ag e :ww w . e l s e v i e r . c o m / l o c a t e / j e b o

Directing

remittances

to

education

with

soft

and

hard

commitments:

Evidence

from

a

lab-in-the-field

experiment

and

new

product

take-up

among

Filipino

migrants

in

Rome

!

Giuseppe

De

Arcangelis

a,∗

_,

_Majlinda

_Joxhe

b

_,

_David

_McKenzie

c

_,

Erwin

Tiongson

d

_,

_Dean

_Yang

e

a_Department_of_Social_Science_and_Economics,_Sapienza_University_of_Rome,_Italy

b_Department_of_Economics_and_Social_Science,_American_University_of_Rome_“John_Cabot”,_Italy c_Development_Research_Group,_World_Bank,_United_States

d_School_of_Foreign_Service,_Georgetown_University,_United_States

e_Department_of_Economics_and_Gerald_R._Ford_School_of_Public_Policy,_University_of_Michigan,_United_States

a r t i c l e i n f o

Articlehistory:

Received17August2014

Receivedinrevisedform6December2014 Accepted19December2014

Availableonline7January2015 JELclassification: F24 O15 D19 C9 Keywords: Remittances Dictatorgame Inter-householdtransfers Philippines a b s t r a c t

Thispaper testshowmigrants’willingnesstoremitchangeswhengiventheabilityto

directremittancestoeducationalpurposesusingdifferentformsofcommitment.Variants

ofadictatorgameina lab-in-the-fieldexperimentwithFilipinomigrantsinRomeare

usedtoexamineremittingbehaviorundervaryingdegreesofcommitment.Theserange

fromthesoftcommitmentofsimplylabelingremittancesasbeingforeducation,tothe

hardcommitmentofhavingfundsdirectlypaidtoaschoolandthestudent’seducational

performancemonitored.Wefindthattheintroductionofsimplelabelingforeducation

raisesremittancesbymorethan15%.Addingtheabilitytodirectlysendthisfundingtothe

schooladdsonlyafurther2.2%.Werandomlyvarytheinformationasymmetrybetween

migrantsandtheirmostcloselyconnectedhousehold,butfindnosignificantchangeinthe remittanceresponsetotheseformsofcommitmentasinformationvaries.Behaviorinthese gamesisthenshowntobepredictiveoftake-upofanewfinancialproductcalledEduPay, designedtoallowmigrantstodirectlypayremittancestoschoolsinthePhilippines.Wefind thistake-upislargelydrivenbyaresponsetotheabilitytolabelremittancesforeducation, ratherthantothehardcommitmentfeatureofdirectlypayingschools.

1. Introduction

Migrantremittancesareoneofthelargestinternationalfinancialflowstodevelopingcountries,exceedingUS$400billion in2012(WorldBank,2013).Theseremittancesrepresentthemosttangiblewayinwhichmigrantssharethetremendous

! IsabelHernando(InnovationsforPovertyAction)providedtop-notchfieldimplementationinthePhilippines.Weappreciatethekeyrolesplayedbythe followingindividualsincollaboratinginstitutions:RaulDimayuga,RoyYu,andAnneDelosReyes(BankofthePhilippineIslands);ChichoyCampos,Joseph Estrada,andMaylaSampa(PhilippineAssociationofPrivateSchools,CollegesandUniversities);CarolPorio(FundforAssistancetoPrivateEducation); EmbassyofthePhilippinesinRome,Italy;andCristinaLiamzon(OverseasFilipinosSocietyforthePromotionofEconomicSecurity).Fundingforthisstudy wasprovidedbyUSAIDDevelopmentInnovationVentures(DIV).

∗ Correspondingauthor.Tel.:+390649910489.

E-mailaddresses:[email protected](G.DeArcangelis),[email protected](M.Joxhe),[email protected](D.McKenzie),

[email protected](E.Tiongson),[email protected](D.Yang).

http://dx.doi.org/10.1016/j.jebo.2014.12.025

(2)

gainsinincomeachievedthroughmigratingabroadwithremainingfamilymembersintheirhomecountries.Onecommon useofthesefundsistofundtheschoolingofthemigrant’schildrenandotherrelativesinthehomecountry,withseveral studiesfindingincreasesineducationasaresult(e.g.EdwardsandUreta,2003;Yang,2008;Theoharides,2014).However, migrantsmaydifferfromremittancerecipientsintheirpreferencesforhowmoneysentshouldbeused(Ashrafetal.,2015), withphysicalseparationandlimitedinformationmakingitdifficultformigrantstoensuremoneyisusedthewaythey intend.

Insuchacontext,financialinstrumentswhichprovidemigrantswithgreaterabilitytomonitorandcontrolhowfunds arespentshouldhavepositivetake-upbysomemigrants,andmayleadtoanincreaseinhowmoneyisremitted.Inthis paperwetestthisideaviaalab-in-the-fieldexperimentwithFilipinomigrantsinRome,Italy.Inadictatorgame,migrants areaskedtoallocate1000D betweenthemselvesandpeopleoftheirchoiceinthePhilippines.Differentgamesthenallow themtoexplicitlysetasidesomeofthemoneysentforeducationalpurposes,varyingtheamountofcontroloffered.Three levelsofcontrolaretested:theoptiontosimplylabelsomeamountofmoneyasforeducationalpurposesonly;labelingthe fundingandhavingitsentdirectlytoaschooltopayforaparticularstudent’seducationaltuition;andcombininglabeled moneysentdirecttoaschoolwithreportsonstudentattendanceandgrades.Wealsorandomizewhetherthehousehold inthePhilippinesistoldthemigrant’schoicesinthegame,totestwhethermigrantsareabletoexhibitmorecontrolwhen theremittancereceivershavelessinformation.

Individualswhotookpartinthisexercisewerethensubsequentlyofferedtheopportunitytouseanewfinancialproduct, EduPay,developedbytheauthorsincollaborationwiththeBankofthePhilippineIslands(BPI).Thisproductallowedmigrants tousearemittancetodirectlypayfortuitionineducationalinstitutionsinthePhilippinesforstudentsoftheirchoosing, withthispaymentgoingdirectlytotheschool,andtheschoolsubsequentlyprovidingattendancerecordsandgrades.We usethistoexaminetheactualdemandforcontroloverremittancesforeducationpurposes.Wecanthencombinethiswith thelab-in-the-fieldevidencetoexaminetheextenttowhichbehaviorinthegamepredictstake-upofthenewproduct,and tohelpbetterunderstandthelikelymechanismsbehindanytake-up.

Wefindthatintroducingsimplelabelingofremittancesforeducationincreasesthetotalamountremittedtoindividuals inthePhilippinesby15%.Thisincreasecomesfromincreasesinboththenumberofdistinctindividualstheychoosetosend moneyto,andfromtheamountsenttoeachindividual.Addingtheabilitytodirectlysendthisfundingtotheschooladds onlyanadditional2.2%totheamountremitted.Theadditionalremittinglargelyoccurswithinthemostcloselyconnected householdinthePhilippines,anddoesnotvarysignificantlyaswevarytheinformationthishouseholdisprovidedabout thechoicethemigrantismaking.Wefindthatchoicesinthesegamesaresignificantpredictorsoftake-upoftheEduPay product:individualswhoallocatemoreofthe1000Dtothisproductinthegamearemorelikelytowanttouseitinpractice. Moreover,thistake-upseemstobedrivenlargelybypreferencesforeducationlabeling,withnoadditionalpredictivepower fromuseofdirectpaymentinthegame.

Thispapercontributestotwomainliterature.Thefirstisveryrecentliteraturewhichexamineshowmigrantsrespond to changes in information asymmetry and to the opportunity to exert more control over how remittances are used. Ambler (2015) finds in a lab-in-the-field experiment that migrants share more of a windfall when relatives are told the migrant received this windfall. Ashraf et al. (forthcoming) find that migrants from El Salvador save more in the homecountrywhenofferedaccountsprovidinggreaterabilitytomonitorandcontrolsavings. Themost closelyrelated work to our paper is Ambler et al. (2015), which, in a sample of Salvadoran migrants in Washington, DC, examines the demand for a new financialproduct that allowed migrants to target remittances toward the education of a spe-cific studentthey selected in ElSalvador. Money was sent via an ATM card, that the student was told wasintended to support their education, but for which there was no enforcement on how the money was used. They find zero demand forthis product when offeredby itself, butdo findpositivedemand forthe product when it isaccompanied by matchingfunds. Ourstudy contrastswith thisfinding by showingthere canbepositivedemand frommigrants for education-labeledremittances,andthroughtheuseofalab-in-the-fieldexperimenttoexaminethemechanismsdriving thisdemand.

Thispaperalsocontributestoasecond,relatedliterature,whichexaminestheroleofsoftandhardcommitmentson financialdecisions.KarlanandLinden(2014)testthedemandandimpactoftwosavingsaccountsforeducationinUganda: aschool-basedcommitmentaccountthatcanonlybeusedforeducationalpurposes,andasoftcommitmentproductthat waslabeledforeducationusebutcouldbewithdrawnforanypurpose.Theyfindmoresavingsundertheweaker com-mitmentproduct, suggestingindividuals may wishtobind theirfuturebehavior, but nottoo tightly.Benhassine etal. (2015) showthat simplylabeling acashtransfer asintended foreducationcan havesimilar positiveeffectsonschool participationasimposingthattransfersbeconditionalonschoolattendance.Consistentwiththeideathatear-marking, or labeling,money foraspecific purposecanincreasesaving, Somanand Cheema (2011) findthat low-income Indian workerssavemorewhensalariesareearmarked,especiallywhenpicturesofthehousehold’schildrenareplacedonthe ear-markedenvelope.Ourfindingsofincreasedremittanceswhenmigrantscanlabelaccounts,andlittleadditionalremittances fromdirectingthepaymentsstraighttotheschool, suggestthatsoftcommitmentsmayalsobeenoughforeducational remitting.

(3)

2. Whymightremittancesrespondtolabelingandaddedcontrol?

Remittancesoccurforawidevarietyofreasons,includingaltruism,insurance,exchange,andasrepaymentofloans. RapoportandDocquier(2006)provideanexcellentsummaryofdifferenttheoriesofremitting,andnotethatakeyfeature ofremittancesisthatdistancemakesinformationalproblemsmorepervasiveandhencestrategicbehaviormorelikelythan isthecasewithmanyothertypesoftransfers.Inparticular,thestandardunitarymodelofthehouseholdseemsparticularly unlikelytoholdinamigrationcontextduetoinformationasymmetriesandtolimitsonmonitoringwithinthehousehold (DeLaat,2014;Chen,2013;Ambler,2015;Doietal.,2014;Genicotetal.,2014).

Inparticular,Ashrafetal.(forthcoming)findmigrantspreferthatremittancesbeusedlessforimmediateconsumption andmoreforlonger-terminvestmentandsavingsthanisthecaseforremittancereceivers.Evenwhenmigrantsand remit-tancerecipientssharethesamepreferencesoverhowremittancesshouldbespent,self-controlissuesmayleadremittance receiverstosaveandinvestlessthanboththeyandthemigrantprefer.

Moneysentforeducationpurposesmaybesubjecttoanumberoftheseissues.Migrantsmaynotbeabletomonitor howmuchschoolexpensesactuallyare,whethermoneysentforschoolingisdivertedtootheruses,andwhetherstudents actuallyattendschooloncesuchfeesarepaid.Sincethereturnstoschoolingoccurinthefuture,whilecostsoccurinthe present,impatientortime-inconsistentremittancereceiversmaybetemptedtospendmoneyonitemsthatbringimmediate benefits.Knowingthis,themigrantmaysendfewerremittancesforschoolingpurposesthanwouldbethecasewithperfect informationandcompleteenforcement.

Howmightlabelingaremittanceasbeingforeducationpurposeschangethis?Theliteratureonself-controlhaslong consideredtheroleofrulestorestrictopportunities(e.g.ThalerandShefrin,1981).Onewayofdoingthisisthroughmental accounts(Thaler,1999),wherebymoneyisimplicitlyorexplicitlysetasideindifferentlabeledaccountsfordifferent pur-poses.SomanandCheema(2011)notethatoncesuchrulesareset,thereisalargepsychologicalliteraturedetailinghow breakingtheserulescanbecostly,intermsofleadingtonegativeemotionssuchasguilt,remorse,andregret.Bylabelinga remittanceasbeingforeducationpurposes,amigrantmayeffectivelybeabletocreatethismentalaccountfortheremittance receiver,withthesepsychologicalcostsmakingitcostlyforthereceivertospendthemoneyonotheritems.

Ofcourselabelsmaynotalwayswork,inwhichcasemigrantsmaywishtoexertevenmorecontroloverhowmoney isspent.Directlysendingmoneytotheschoolforfeesisamuchstrongerwayofexertingcontrol,andmaythereforebe preferredbymigrantsiftheybelievelabelsdonotexertenoughpressureonthereceivertousemoneyfortheintended purpose.

Giventhatmigrantsregularlycommunicatewiththeirfamilymembers,anaturalquestionisthenwhetherweshould expectanoutsideentity’soffertolabelmoneyforeducationtoaddanythingbeyondanylabelsmigrantscanalreadyattach? Thatis,ifmigrantswouldsendmoremoneyifthismoneywerelabeledasbeingforeducationpurposes,whydon’tthe migrantssimplylabelremittancesthis way?Thereareatleasttwopossiblereasonswhyanoutsidepartymayprovide avaluablelabel.First,thelabelmaybeseenasmorecredibleorstrongerwhencommunicatedbyathirdparty.Second, attachingalabelmaybecostlyformigrants,asitcouldbeseenbyreceiversasasignaloflackoftrust.Thissamecostmay notexistwhenthecommunicationcomesfromathirdparty,especiallyifthereceiverisunsureofthecircumstancesunder whichthemigrantismakingthisdecision.

Incontrast,it iseasiertosee howathirdparty offeringtheabilitytodirectlysendremittancestoschooltopayfor expenses,andtoprovidemonitoringreportsonthechild’sattendanceandprogresscanaddvalue,sincethereisnoeasy wayformigrantstodothisbythemselves.However,thishardformofcommitmentcouldbeviewedbytherecipientas indicatingalackoftrustfromthemigrant,whichcouldhaveoffsettingnegativecostsforthesender.Anadditionaldownside ofhardcommitmentsisthattheyreducetheflexibilityofrecipientstouseremittancesforotherpurposes,forexampleif healthorotheremergenciesarise.

3. Setting,baselinesample,anddescriptionoflab-in-the-fieldexperiment 3.1. Setting

OurexperimentwasconductedamongFilipinoworkersinRome,Italy.ThePhilippinessendsworkerstoaverydiverseset ofmigrantdestinations.Thereareestimatedtobeapproximately113,000FilipinomigrantsinItaly,remittingaboutUS$500 million,onaverage,backtothePhilippineseachyearinrecentyears.Educationisakeydesiredusefortheseremittances, withUS$233.5millionofremittancessentinexpectationthatthiswouldfinancetheeducationofarelative(IOM,2010).1 3.2. EduPay

WeworkedwiththeBankofthePhilippinesIslands(BPI)andthePhilippineAssociationofPrivateSchools,Colleges,and Universities(PAPSCU)todevelopandpilotanewremittanceproduct,calledEduPay,thatwasintendedtoprovidemigrant workerswithgreaterabilitytocontroltheuseofremittancesforeducation.Thisproductallowsmigrantstochanneltuition

(4)

Table1

Baselinesummarystatistics.

Mean SD Min Median Max Observations

Migrantisfemale 0.73 0.44 0 1 1 501

Migrant’sage 42.25 10.32 19 42 71 499

Migrantismarried 0.68 0.47 0 1 1 501

Migrant’snumberofchildren 1.95 1.47 0 2 8 501

Migrant’syearsinItaly 9.68 8.56 0 7 38 499

Migrantisemployed 0.98 0.15 0 1 1 499

Migrant’smonthlyIncome 1045.18 566.42 0 900 7000 481

Migrant’shoursworking 42.66 18.87 0 40 88 499

Migrantremitsmonthly 0.72 0.45 0 1 1 501

Monthlyremittancessent 412.54 299.17 0 380 3000 499

Annualremittancesforeducation 1383.72 1724.83 0 970 12,000 500

Averagecostofremittance 5.64 1.97 0 5 15 498

Sponsoredstudentisafemale 0.51 0.5 0 1 1 490

Ageofthesponsoredstudent 14.35 4.72 2 15 28 488

paymentsforparticularstudentsdirectlytothosestudents’educationalinstitutionsinthePhilippinesfromaBPIremittance

branchinRome.Thismechanismallowedmigrantstoavoidsendingtuitionpaymentsviafamilymembersorothersinthe

Philippineswhomightnotbecompletelytrustedtomakesuchpaymentsreliably.Themigrantwouldthenalsosubsequently

receiveattendancerecordsandschoolgradesfromthechild’sschool,enablingthemtobettermonitortheschoolingbeing

paidfor.

3.3. Baselinesample

GiventhatFilipinosareasmallminorityoftheoverallpopulationofRome,weusedintercept-pointsamplingtoobtaina

sampleforthisstudy.BetweenAugust2012,andJanuary2013,weintercepted2291Filipinomigrantsatcommonmeeting

pointsinRomeandatthemainbranchofBPIinRome.2_Migrants_were_given_a_general_introduction_to_the_research_project,

whichwasdescribedas“aboutthelivesandfinancialdecisionsofOFWsinRome,andabouttheremittancestheysendhome tothePhilippines.”3_Migrants_were_also_told,_referring_to_the_EduPay_product,_“We_will_also_be_offering_you_a_new_product relatedtoeducationandremittancesattheendofthesurvey,andyoumaybenefitfromusingthisproduct.”

Migrantsweretheninvitedtoanswersomepreliminaryscreeningquestionstodeterminetheireligibilitytoparticipate inthepilotphase.Tobeconsideredeligible,arespondenthadtomeettwocriteria:(a)theprovinceoforiginofthemigrant andofhis/her“mostlycloselyconnectedhousehold”(i.e.thehouseholdwheretheylivedbeforemigrating,orthehousehold theysendthemostremittancesto)shouldbeinoneofthreeregionsofthePhilippinesinwhichPAPSCUhadagreedto assistinrunningthepilot;4_and_(b)_had_a_relative_in_the_Philippines_aged_5–22._If_the_migrant_did_not_refuse_to_answer furtherquestions,theinterviewwouldbegin.Thecompletesurveyinstrument,includingintroductorytextreadtopotential respondents,isprovidedintheOnlineAppendix.

Thisresultedinasampleof501migrantworkers,whowerethenadministeredabaselinesurveyandalab-in-the-field experiment.Attheconclusionofthesurveyandexperiment,participantswerethentoldaboutthenewEduPayproduct, providedwithsomerelatedmarketingmaterialand,iftheywantedtousethisnewproduct,askedtosigntwoforms:(i) ageneralinformationformonthesponsorablestudent(s)and(ii)anauthorizationformrequiredbyPAPSCUtoobtainall relevantinformationonthestudent(e.g.theirschoolIDnumber)fromtheschoolssoastoimplementEduPay.

Thebaselinesurveycollectedinformationondemographicbackground,labormarketstatus,remittingbehavior,andthe qualityoftheirrelationshipwiththeirmostcloselyconnectedhouseholdinthePhilippines.Summarystatisticsareprovided inTable1.Thesampleis73%femaleandtheaverage(andmedian)ageoftherespondentsis42.Most(70%)ofthemigrants haveacollegeorauniversitydegreeandhavebeenlivinginItalyforabout7years(median).Only21%arenevermarried, butonly5%havetheirspousewiththeminItaly.Nearly68%oftherespondentsareemployedasdomesticworkers,with theothermainoccupationsbeinghousecleaners(8%)andnannies(6%).Themedianwageis900D permonth.Almost96% haveremittedregularlytothemostcloselyconnectedhouseholdinthePhilippinesinthelast12months,and72%remit monthly.Themedianamountofremittancesis380D/month.81%reporthavingsentremittancesinthepastyearforthe specificpurposeoffundingtheeducationofsomeoneinthePhilippines,withamedianof970D/yearsentforthispurpose. ItisworthwhilecomparingthesebaselinecharacteristicstothoseoftheoverallFilipinomigrantcommunityinItaly. AccordingtoItalianlaborforcesurveydata(Italialavoro,2013),58%ofFilipinoimmigrantsinItalyarefemale,80%ofthose

2 _Intercepts_were_scheduled_at_various_times_on_a_variety_of_days_of_the_week._The_intercept_points_were_five_fixed_locations:_the_Santa_Pudenziana

Filipinocommunitychurch,theBankofthePhilippineIslandsRomebranch,theEmbassyofthePhilippines,theheadquartersofanimportantFilipinoNGO (OFSPES),andthecentraltrainstationinRome(TerminiStation).

3 _“OFW”_is_a_commonly_used_acronym_among_Filipino_migrants,_and_stands_for_“overseas_Filipino_workers.”

4 _These_regions_are:₍₁₎_Region₃_[Central_Luzon];₍₂₎_Region₄_[the_provinces_of_Batangas,_Cavite,_Laguna,_Quezon,_Rizal,_Marinduque,_Occidental_Mindoro,

(5)

intheworkforcehavemonthlyincomesbelow1000D,themedianagecategoryis30–39years,and53%haveatleastan uppersecondaryeducation.Comparingthesefigureswithoursamplestatisticsrevealsthatourstudysampleismorefemale, older,andbettereducatedthanthebroaderFilipinoimmigrantpopulationinItaly.5

ThemostcloselyconnectedhouseholdtothemigrantinthePhilippineshasanaverageof4.45individualsinit.Thehead isthemigrant’sparentsin29.5%ofcases,themigrant’sspouse22.6%ofthe time,and themigrant’ssisterorbrotherin 17.2%ofcases.62.9%ofmigrantshaveoneoftheirownchildreninthishousehold,and32.1%haveanephewornieceinthis household.Communicationbetweenthemigrantandthishouseholdisfrequent,with52%communicatingdaily,and93%at leastweekly.71%ofmigrantsdeclarethemselvestobe“verywellinformed”onwhathappensinthishousehold.However, despitethis,55%claimedtheywouldliketohavemoreinfluenceonhowtheoverallbudgetofthemostcloselyconnected householdisspentand37%statedthattheywishtohavemoreinfluenceonthespendingdecisionsoftheremittances.This suggeststheremaybedemandforinterventionswhichprovidethemwithmorecontroloverremittances.

3.4. Thelab-in-the-fieldexperiment

Theexperimentalinterventionwasadministeredtoparticipantsface-to-face,immediatelyaftertheyansweredthe base-linesurvey.Therewerethreemainobjectivesofthisexperiment.Thefirstwastotesttheresponsivenessoftotalmigrant remittanceswhenmigrantshavetheabilitytocontrolsomeofthisforeducationuse.Thesecondobjectivewastoattemptto unbundlethedifferentfeaturesoftheEduPayproduct,inordertodeterminewhetheranypositiveresponsivenesstoEduPay wasdrivenbytheabilitytochannelfundsdirectlytoschools,orwhetherEduPaysimplyprovidedawayformigrantsto labelthatatransferwasintendedforeducation.Thethirdobjectivewastoexaminehowuseofthesefeaturesvarieswith thedegreeofasymmetricinformationbetweenthemigrantandtherecipient.

Migrantsweretoldthattheywereenteredintoalotterytowina1000D prize,andaskedhowtheywouldliketoallocate anywinningsbetweenthemselves,andbetweenoneormoreotherpeopleinthePhilippines.Theyweretoldthatanyamount sharedwithpeopleinthePhilippineswouldberemittedbytheresearchprojectatnocharge.AsseenfromTable1,1000D correspondstojustoveronemonth’smedianearnings,andalmostthreemonthsofmedianremittances,soisasizeable sum.Weaskthemtomakethischoiceunderfourdifferentsettings,usingawithin-participantexperimentinwhichwe randomizedtheorderinwhicheachpersonwaspresentedwiththesefourcases.

Thecaseswereasfollows:

(i) Basic:thischoicecorrespondstoasimpledictatorgame,inwhichthemigrantdecideshowmuchtokeepforheror himself,andhowmuchtosharewithothers,withanyfundsallocatedtopeopleinthePhilippinesprovidedtothem directlyintheformofcash.

(ii) Educationlabeloption:Underthischoice,migrantsweregiventheoption(butnorequirement)oflabelinganyamount sharedasbeingforeducation,sothatwhenitwasdeliveredtotherecipientitwouldbeaccompaniedbyanotesaying “thesefundsareintendedtobeusedfortheeducationofsomeoneinyourhousehold”.Thiscapturesthelabelingaspect oftheEduPayproduct.

(iii) DirectPaymentoption:Thiscaseisidenticalto(ii),inthatmigrantscouldchooseforeachamountsharedwhetheror nottoattachaneducationlabel.Inaddition,thiscaseaddstheoptionofchoosingtohavethemoneypaiddirectlyto thestudent’sschoolastuition.ThiscapturesthelabelinganddirectpaymentaspectsoftheEduPayproduct.

(iv) Directpaymentwithperformancemonitoring: Thiscaseisidenticalto(iii),exceptthatifthemigrantchosetosend moneydirectlytotheschoolfortuition,theywouldalsoreceivereportsontheattendanceandgradesofthisstudent. ThislastcasethenmimicsthethreekeyfeaturesoftheEduPayproduct:labeling,directpayment,andmonitoring. Tobeclear,cases(i)through(iv)arenested,inthateachsuccessivecaseonlyaddsoptionsforthemigrant,andnever takesanyaway.Case(ii)addstheeducationlabeloption,butallowssimplysendingcashwithoutalabel,asincase(i). Case(iii)addsdirectpaymentstoschools,butalsoallowssimplelabelingasincase(ii)orjustcashprovisionasincase(i). Respondentshavefreechoicetouseany,all,ornoneoftheoptionsavailabletotheminaparticularcase.Forexample,incase (iii),amigrantscouldsend,foraparticularrecipient,acashremittance,alabeledremittance,oradirectpaymenttoaschool, oranycombinationofthesethree.What’smore,respondentscouldmakethesetransfersforanynumberofrecipientsinthe Philippines.

Thisnestedstructureallowsustointerpretdifferencesinremittancesbetweentreatmentsasreflectingtheimpacton remittancesofaddingorsubtractingparticularsharingoptions.Forexample,thedifferenceinremittancesbetweencases (iii)and(ii)reflectsthenetimpactofaddinganoptionfordirectpaymentstoschools,overandaboveanabilitytoattachan educationlabelortosimplysharecashwithoutalabel.

5 _Because_the_sample_appears,_in_particular,_to_be_more_female_than_the_broader_Filipino_population_in_Italy,_all_empirical_results_in_the_main_tables_(in

(6)

Choiceswereincentivizedbytellingrespondentsthatonerespondentinthestudywouldactuallywinthe1000D prize, andforthatwinneroneoftheirfourcases(i,ii,iii,oriv)wouldberandomlyselectedtobeimplementedastheyhadspecified. Migrantsweretoldthattheywouldnotbeallowedtochangetheirallocationdecisioniftheylaterlearnedthattheyhad wonthelottery,sotheyshouldtaketheallocationdecisionseriously.Thelotterywasactuallyimplementedon28March 2013.

Inthisexperimentwewerealsointerestedinwhethermigrantswouldmakedifferentchoicesunderdifferent assump-tionsaboutwhattheirfamilymembersinthePhilippinesknewabouttheirchoiceenvironment.6_In_particular,_we_were interestedinthepossibilitythatmigrantsmightbelesswillingtousetheeducationlabelordirectfundstoschoolsifthe mostcloselyconnectedhouseholdsinthePhilippinesfeltthatthemigrantwasexplicitlyseekingtocontroltheirbehavior. Thismightbethecaseifbeneficiariesviewedattemptstocontrolthemasreflectingthemigrant’slackoftrustinthemor disapprovaloftheirdecision-making.

Totestthis,werandomlyassignedthemigrantsintothreetreatmentgroups,eachof167individuals7_:

• Treatment1(Privateinformation):Migrantsassignedtothistreatmentweretoldthatthemostcloselyconnectedhousehold inthePhilippineswouldnotbeinformedofanyofthedecisionsorchoicesthemigrantmade.Ifthemigrantdecidesto sharemoneywithanyonefromthis household,theywouldreceivethe moneyandbetolditcame fromthe migrant answeringasurveyinItaly,butnotwhatthedecisionprocesswasthatresultedinthisamountbeingtransferred. • Treatment2(Informationsharing):MigrantsassignedtothistreatmentweretoldthatthehouseholdinthePhilippines

wouldbeinformedofallthechoicestheymade.Thismeansthehouseholdwouldknowthatthemigranthadtoallocate 1000D,andexactlyhowtheyhaddecidedtoallocateit.

• Treatment3(Socialexcuse):Migrantsassignedtothistreatmentweretoldthat,asinTreatment2,thehouseholdinthe Philippineswouldbeinformedofallchoicesmade.However,ifthemigrantchoseanyoftheEduPayoptions(directed payment,ordirectedpayment+monitoring),thesurveyteamwouldinformthehouseholdthatasmalldonationtoa Fil-ipinocommunityorganizationinRomewasmadewhentheEduPayoptionwaschosen.Thiswastoprovidethehousehold withasocialexcuseforusingEduPay.

OurpriorwasthatmigrantswouldbeleastwillingtosharemoneywiththehouseholdinthePhilippineswhentheir choicewaspublicinformation(treatment2),andthattheywouldbemorewillingtouseeducationlabelingordirected paymentwheneitherthischoicewasprivateinformation,orwhentheyhadasocialexcusetojustifytothehouseholdwhy theyweredoingthis(treatments1and3,respectively).

4. Lab-in-the-fieldresults

Webeginbypoolingresultsacrossthedifferentinformationtreatmentsandexamininghowthemigrant’swillingnessto share(remit)moneywithothersinthePhilippinesvarieswiththedifferentchoiceswhichprovidegreaterorlesserdegrees ofcontroloverhowmoneyisusedforeducation.Wethenexaminetheextenttowhichthesechoicesvaryaccordingtothe informationprovidedtothemostconnectedhouseholdinthePhilippines.Inthenextsectionwetheninvestigatetheextent towhichbehaviorintheselab-in-the-fieldexperimentshelpspredicttake-upofthenewEduPayproduct.

4.1. Howdoestheamountremittedvarywithdifferenteducationcommitments?

Foreachindividualweobservefourdifferentallocationdecisions,correspondingtoonedecisionforeachofthe four choicecases(i)–(iv)above.Toestimatetheimpactofthesechoiceoptionsonoutcomeyformigrantjweestimatefixed effectsmodelsoftheform:

yj=˛+ˇEduLabelj+!DirectPayi+ıDirectPay&Monitorj+"j+εj

WhereEduLabelindicatesthedecisionistakenunderchoice(ii),wherethemigranthastheoptionoflabelingremittances asforeducationpurposes;DirectPayindicatesthedecisionistakenunderchoice(iii),wherethemigrantalsohastheoption ofchoosingtohavemoneydirectlypaidtotheschool,andDirectPay&Monitorindicatesthedecisionistakenunderchoice (iv),wherethemigrantcanchoosetohavemoneydirectlypaidtotheschoolandreceivemonitoringreportsinreturn."j

isanindividualfixedeffect,enablingustoidentifyimpactsfromintra-migrantdecisions.Thisfixedeffectalsocapturesthe 24randomizationstratawhichdeterminetheorderingofthefourchoices.Thestandarderrorεjisthenclusteredatthe

individuallevel.

6 _Existing_research_(such_as_Ambler,_2015;_Genicot_et_al.,₂₀₁₄₎_suggests_that_remittance_{recipients’}_knowledge_about_the_migrant’s_choice_environment,

andthemigrant’seconomicconditionsinparticular,affectsmigrantremittancebehavior.

7 _{Randomization}_occurred_by_computer,_and_was_a_simple_random_draw,_without_{stratification,}_since_no_baseline_data_were_available_to_stratify_on_at_the

(7)

Table2

ImpactsofeducationlabelingandEduPayinlab-in-the-fieldexperiment.

Proportion choosingto remit Totalamount remitted Numberof people remittedto Amount remittedper beneficiary Numberin mainhh remittedto Remitsto someone outsidemain household Amount remittedto someone outsidemain household Education labeling 0.0459*** (0.0120) 93.66*** (11.31) 0.148*** (0.0470) 30.30*** (9.585) 0.188*** (0.0390) 0.0140 (0.0153) 17.25** (7.550) Directpayment toschool 0.0579*** (0.0119) 107.4*** (11.58) 0.162*** (0.0495) 28.19*** (10.35) 0.269*** (0.0410) −0.00399(0.0160) 5.573 (7.930) Directpayment to school+monitoring 0.0519*** (0.0121) 103.0*** (11.50) 0.152*** (0.0478) 28.31*** (10.42) 0.251*** (0.0413) 0.00200 (0.0154) 7.579 (7.990) Meanfor choicewith no-labels 0.878*** (0.00847) 614.6*** (8.077) 2.032*** (0.0327) 422.7*** (7.240) 1.481*** (0.0273) 0.238*** (0.0104) 90.17*** (5.214) p-Valuesfortesting

Education label-ing=direct payment 0.0832 0.0486 0.7124 0.7041 0.0101 0.1394 0.0569 Direct payment=direct payment+monitoring 0.3182 0.4771 0.7538 0.9836 0.5563 0.6401 0.7541 Allthree education versions equal 0.1942 0.1424 0.9217 0.9253 0.0199 0.3035 0.1188 Observations 2004 2004 2004 1838 2004 2004 2003

Notes:estimationincludesindividualfixedeffects.Robuststandarderrorsinparentheses,clusteredattheindividuallevel. Column4conditionsonremittingatall.

** _Significance_at_the_5%_level. *** _Significance_at_the_1%_level.

Wearetheninterestedintestingwhetherˇ=0,thatiswhetherornotthereisanychangeinremittancebehaviorwhen theoptiontolabelremittancesasforeducationisgiven,aswellastestingforequalityofˇ,!andı,whichtellsuswhether theformofeducationalcommitmentmatters.

Weconsiderseveraloutcomes.Thefirstistheextensivemarginofwhetherornotthemigrantchoosestoremitany ofthemoneyatallwithothersinthePhilippines.87.8%ofmigrantsdochoosetoremitinthebasic,no-labeloption.The secondoutcomeisthetotalamountremittedtoanyoneinthePhilippines.Migrantscoulddividethemoneyupwithas manyindividualsinthePhilippinesastheyliked,withthemaximumbeing9individuals.Wetotalthisup,andfindthe meanamountremittedis614.6D underthebasic,no-labeloption.Thatis,migrantsaresharingmorethantheykeepfor themselves.WealsoconsiderthetotalnumberofpeopleinthePhilippinestheychoosetoremitto(anaverageof2.03under thebasicoption),andtheamountremittedtoeachbeneficiary(conditionalonmakinganyremittances).Thentoexamine theextenttowhichdifferencesarearisingfromchangingremittancebehaviorwiththemostcloselyconnectedhousehold, versuswithotherhouseholdsinthePhilippines,weexaminethenumberofindividualswithinthemostcloselyconnected householdthattheychoosetoallocatemoneyto,whetherornottheychoosetoallocatemoneytosomeoneoutsidethis mainhousehold(23.8%dounderthebasicchoiceoption),andtheamountremittedtopeopleoutsidethemainhousehold (90.2D onaverageunderthebasicoption).

Table2reportstheresults.Firstconsidertheeffectofbeingabletoincludeaneducationlabel.Thisresultsina4.6%point increaseinthelikelihoodofremittingatall,anda93.7D increaseinthetotalamountremitted,bothsignificantatthe1% level.Thisincreaseinthetotalremittedisapproximatelya15%increaseontheamountremittedunderthebasicchoice option.Weseethatthisincreasereflectsmigrantsremittingtobothmorepeople(0.15morepeople),andremittingmoreto eachpersonthattheydoremitto.Mostoftheincreaseappearstobeoccurringwithinthemostcloselyconnectedhousehold, withnoincreaseinthelikelihoodofremittingtosomeoneoutsidethismainhousehold,and82%ofthetotalincreasein remittancesgoingtoindividualsinsidethemainhousehold.

(8)

0 .2 .4 .6 .8 1 Cumulative Pr obabili ty 0 200 400 600 800 1000 Total Amount Remitted (Euros)

No labelling Education label Direct Pay Direct Pay + monitoring Fig.1. CDFsofamountremittedunderthefourdifferentchoiceoptions.

oneexceptionisinthenumberofindividualsinthemainhouseholdthatmoneyissharedwith.Thisis0.06–0.08persons higherwitheitherdirectpaymentoption,a4–5%increaseonthebasicoptionmean.

Fig.1showstheCDFsofthetotalamountremittedunderthefourchoicestructures.Weseethechangeintheamount remittedwhenmovingfromthebasic(nolabel)optiontoanyoftheotherthreeoptionsoccursacrosstheentiredistribution, whiletheotherthreechoiceshaveverysimilardistributionstooneanother.

Insum,itdoesappearthatmigrantsarewillingtoremitmorewhengiventheabilitytodirecttheseremittancestoward education,butthatthemaineffectappearstocomefromthesoftercommitmentofeducationlabeling,ratherthanthrough theabilitytoexertmorecontrolthroughdirectpayment.8

4.2. HowdothesechoicesvarywiththeinformationprovidedtothehouseholdinthePhilippines?

Nextweexaminewhethertheresponsivenessofmigrantstotheabilitytolabelremittancesforeducationortobeable todirectlypaytheschoolvarieswiththeinformationthemostcloselyconnectedhouseholdinthePhilippineshasabout thischoice.9_The_information_{randomization}_occurred_at_the_individual_migrant_level,_with_a_migrant_making_each_of_her_or hisfourchoicesunderthesameinformationcondition.Thereforetoidentifytheimpactofdifferentinformationsettings, werelyonrandomassignmentofinformationtreatmentsacrossindividuals,andestimatethefollowingregressionatthe individuallevel:

yj =˛+ˇEduLabelj+!DirectPatj+ıDirectPay&Monitorj+$1InfoSharedj+$2InfoSharedjEduLabelj

+$3InfoSharedjDirectPayj+$4InfoSharedjDirectPay&Monitorj+%1SocialExcusej

+%2SocialExcusejEduLabelj+%3SocialExcusejDirectPayj+%4SocialExcusejDirectPay&Monitorj+εj

WhereInfoShareddenotestheindividualwasassignedtoinformationtreatment2,inwhichallchoicesaresharedwith themainhouseholdinthePhilippines,andSocialExcusedenotestheindividualwasassignedtoinformationtreatment3, inwhichallchoicesaresharedwiththemainhousehold,butasocialexcuseisgivenforusingEduPay-likeoptions.Here˛ isthemeanoutcome(e.g.totalremittances)forthebasicchoiceoptionundertreatment1(privateinformation).Wethen

8 _To_test_differences_in_treatment_effects_across_male_and_female_migrants,_Online_Appendix_Table₁_presents_regression_results_analogous_to_those_of Table2butwhereinteractiontermsareaddedbetween“male”andthethreetreatmentdummies.(The“male”maineffectneednotbeincludedinthe regression,becausetheregressionincludesindividualrespondentfixedeffects,whichabsorballtime-invariantrespondentvariablessuchasgender.)The onlystatisticallysignificantdifferencethatappearsisinthe“totalamountremitted”regression,whereformalestheeffectoftheeducationlabelappears tobelarger,by46D,thantheeffectforfemales(81D).(Bycontrast,theeffectsofeducationlabelingonotheroutcomesfoundinTable2donotseemto bedifferentialbygenderatconventionalsignificancelevels.)BecauseourstudypopulationismorefemalethanthegeneralpopulationofFilipinosinItaly, thisresultsuggeststhatourestimateinTable2ofthepopulation-averageeffectofeducationlabelingontotalremittancessentisdownward-biased.

9 _A_relevant_background_statistic_is_that_80.8%₍₄₀₅_out_of₅₀₁₎_study_participants_remit_to_their_most_closely_connected_household._Given_that_87.8%

(9)

test$1=%1=0totestwhetherinformationtreatmenthasnoeffectondecisionsunderthebasic(nolabel)choiceoption;

testˇ=$2=%2fornodifferenceintheaddedeffectofeducationlabelingrelativetothebasic,no-label,caseacrossthe

informationtreatments;andlikewisetest!=$3=%3andı=$4=%4.

Table3presentstheresults.Considerfirsttheoutcomesunderthebasic,no-labelchoice.Thepointestimateson infor-mationsharingandonhavingasocialexcusesuggestthat,ifanything,migrantsareremittinglessintotalwhenthechoice decisionistobecommunicatedtothemostcloselyconnectedhousehold.ThiscontrastswiththefindingofAmbler(2015), butthis differenceisnotstatisticallysignificant,andfornone oftheoutcomescanwerejectequalityofmeansforthe basic(nolabel)choiceacrossthedifferentinformationtreatments.Considernextchoicesundereducationlabeling.Wesee migrantssendanadditional102.7D relativetotheno-labelchoicewhenthechoiceismadeprivately,anadditional85.9D relativetothebasicchoicewhenthechoiceismadeunderinformationsharingwiththemigrant,andanadditional36.4D relativetotheno-labelchoicewhenmigrantsareprovidedwithasocialexcuse.Thep-valuefortestingequalityofthese effectsis0.127,sowecannotrejectthatthereisnoimpactofinformationtreatmentonthedecisionsmade.

Weseethisisgenerallytrueacrossthedifferentoutcomesandtests:wecannotrejectequalityofdecisionsunderthe differentinformationtreatmentsfor23ofthe28testsperformedatthebottomofthetable.Threeofthefivesignificanttests arefortestingequalityofinformationtreatmenteffectsfortheoutcomeofthenumberofindividualsinthemostclosely connectedhouseholdremittedto.Theincreaseseeninthisoutcomeunderprivateinformationappearstobelowerwhena socialexcuseisprovidedforusingeducationlabelingordirectpayment.Thiscontrastswithourpriorthatasocialexcuse wouldleadmigrantstousethesefeaturesmore.Thepointestimatessuggestthatoneexplanationmightbethatmigrants shiftmoreoftheirfundingtowardindividualsoutsidethe mostcloselyconnectedhousehold(perhapsnowbecausethe migrantcanusethesocialexcuseasajustificationforwhytheyarehelpingothersoutsidethemainhousehold),butthis effectisnotstatisticallysignificant.10

5. Dochoicesinthelab-in-the-fieldexperimentpredictproducttake-up?

Followingthelab-in-the-fieldexperiment,ourfieldteamexplainedthenewEduPayproductdevelopedwithBPIwiththe helpofsomemarketingmaterial.Migrantswhowereinterestedinusingtheproductsignedarequestlettertotheschool, askingthemtoreleasethestudents’identificationnumber,makeavailableaninvoiceforpaymenttotheschool,provide detailsofthebankaccountoftheschool,andalsoreleasethegradesofthestudent.Theseformswerethensenttoourproject coordinatorinthePhilippines,whoworkedwithPAPSCUtocontacttheschoolsandarrangethelogistics.Overall27.1%of individualsofferedtheproductsignedthisletterofintent.Weusethisasourtake-upmeasure,sinceinpracticeduringthe pilotphase,manyoftheschoolsdidnotprovidetheneededinformationinatimelyfashion,sothatthemajorityofthese intendedtransactionswerenotexecuted.11

Weexaminethedeterminantsoftake-upviaprobitregressionsoftheform: Pr(UseEduPay)=Probit(ı′_GameBehavior₊_&′_X)

whereXareadditionalcontrolvariablesthatmaybelikelytodeterminethedesireofthemigranttoremitforeducation.We includeheregender,thenumberofchildreninthePhilippines,whethertheindividualhastheirspouseinthePhilippines, whethertheyhavenephewsandniecesofschool-ageinthePhilippines,theirowneducationlevel,whethertheyhavebeen inItalylongerthanthemedianof7years,whethertheyearnlessthanthemedianof800D permonth,andtheamount theyreporthavingsenttothePhilippinesforeducationalpurposesinthepast12months.Theregressionanalysiswilluse asampleof483individualsoutofouroriginalsampleof501.12

Weconsiderseveralmeasuresofbehaviorinthelab-in-the-fieldgames.Thefirstmeasureissimplytheamountofmoney theydecidetoremitexplicitlyintheformofadirectpaymenttotheschoolunderthedirectpaymentchoiceoption,case (iii).Themeanis316D withastandarddeviationof364D.Foreaseofinterpretation,wedivideby100D,sothecoefficient representsthemarginaleffectofallocating100D moretodirectpaymentinthechoiceexperiment.

ThesecondmeasureisthedifferenceinthetotalamounttheychoosetoremittothePhilippineswhengiventheoption tolabelmoneyforeducationcomparedtothebasic,no-label,choice.Themeanis93.7D,withastandarddeviationof253D. Wethenalsoconsidertheadditionaldifferenceinthetotalamounttheychoosetoremitwhengiventhefurtheroption ofdirectingpaymentsstraighttotheschoolversuseducationlabelingalone.Themeanforthisvariableis13.8D,witha standarddeviationof156D.Againwestandardizethesevariablesintermsofhundredsofeuro.

10 _Online_Appendix_Table₂_presents_regression_results_analogous_to_those_of_Table₃_but_where_interactions_are_added_between_“male”_and_the_relevant

treatmenttermsinTable3.Becausetheregressionsdonotincludeindividualrespondentfixedeffects,the“male”maineffectisincludedintheregression. Asitturnsout,thereisnoevidencethattheestimatesinTable3aredifferentacrossgenders:noneoftheinteractiontermswithmalearestatistically significantfromzeroatconventionallevels.

11 _There_were_several_reasons_for_this._First,_a_number_of_schools_did_not_have_bank_accounts_set_up,_and_were_not_prepared_to_do_so_for_only_a_small

numberoftransactions.Second,schoolstypicallytooktwotothreeweekstoprovidetheinvoiceandbankaccountinformation,andsomanymigrants, beingnervousaboutmeetingtuitiondeadlines,optedtosendmoneythroughothermeans.Ultimatelythisresultedinonly21individualsmakingan EduPaytransaction,11inthefirstpilotphaseand10inanextensionphase.

12 _Four_individuals_were_dropped_because_of_missing_data_on_control_variables,_and_an_additional₁₄_were_dropped_because_of_lack_of_information_on_the

(10)

Table3

Dotheimpactsvarywiththedegreeofinformationasymmetry?

Proportion choosingto remit Totalamount remitted Numberof people remittedto Amount remittedper beneficiary Numberin mainhh remittedto Remitsto someone outsidemain household Amount remittedto someone outsidemain household Educationlabeling 0.0299 102.7*** _0.114 _38.25** _0.269*** _−0.0359 _−2.898 (0.0198) (20.57) (0.0808) (16.14) (0.0688) (0.0267) (12.61) Directpaymentto school 0.0599*** _124.6*** _0.174* _28.88 _0.341*** _−0.0479 _−7.090 (0.0203) (20.19) (0.0887) (18.06) (0.0750) (0.0292) (13.57) Directpaymentto school+monitoring 0.0539*** _125.4*** _0.126 _40.71** _0.305*** _−0.0359 _−5.293 (0.0195) (19.71) (0.0863) (18.36) (0.0735) (0.0281) (13.18) Infosharedwith

recipienthousehold −0.0419 −25.03 −0.0539 −44.31 −0.0539

0.00599 9.599 (0.0360) (35.81) (0.184) (32.37) (0.133) (0.0467) (22.40) Education

labeling*infoshared

0.0479 85.90*** _0.216 _−10.19 _0.132 _0.0539 _39.10 (0.0293) (32.94) (0.186) (32.74) (0.135) (0.0481) (24.54) Directpayment*info

shared 0.0299 81.11** _0.108 _5.716 _0.180 _0.0240 _18.74 (0.0309) (33.56) (0.180) (33.54) (0.133) (0.0473) (23.26) Directpayment+ monitoring*info shared 0.0299 76.32** _0.150 _−8.423 _0.186 _0.0299 _16.37 (0.0309) (33.51) (0.181) (32.98) (0.136) (0.0474) (23.04) Socialexcuse −0.0180 −31.02 −0.246 −7.791 −0.228* _0.00599 _11.08 (0.0336) (35.64) (0.174) (33.99) (0.128) (0.0466) (22.23) Education labeling*social excuse 0.0000 36.35 −0.186 16.63 −0.120 0.0359 36.23 (0.0323) (34.58) (0.172) (33.13) (0.133) (0.0476) (24.07) Directpayment*social

excuse 0.0240 60.60* −0.0958 3.807 0.00599 0.0240 25.75 (0.0304) (33.98) (0.167) (32.83) (0.130) (0.0472) (23.26) Directpayment+ monitoring*social excuse 0.0120 51.32 −0.120 7.842 −0.0180 0.0240 32.93 (0.0314) (34.55) (0.169) (33.25) (0.132) (0.0472) (24.24) Meanfornolabel

choiceinprivate

0.898*** _633.2*** _2.132*** _438.3*** _1.575*** _0.234*** _83.23*** (0.0235) (24.75) (0.136) (24.06) (0.0974) (0.0329) (15.31) p-Valuesfortestingequality

nolabelchoices acrossinformation treatments 0.5074 0.6489 0.3050 0.3324 0.1603 0.9890 0.8621 labeledchoices acrossinformation treatments 0.2638 0.1273 0.0383 0.3370 0.0162 0.1175 0.1504 Directpayment choicesacross information treatments 0.2985 0.1224 0.1875 0.7020 0.0450 0.1773 0.3306 Direct payment+monitoring choicesacross information treatments 0.2999 0.0698 0.1497 0.3249 0.0488 0.2752 0.2773 Observations 2004 2004 2004 1838 2004 2004 2003

Notes:Robuststandarderrorsinparentheses,clusteredattheindividuallevel.Column4conditionsonremittingatall.

* _Significance_at_the_10%_level. ** _Significance_at_the_5%_level. *** _Significance_at_the_1%_level.

(11)

Table4

DoLabchoicespredictactualproductinterest?Dependentvariable:signingletterofauthorizationforEduPay.

(1) (2) (3) (4) (5)

Amounttaggedwithdirectpaymentindirectpaymentchoiceexperiment 0.0240*** _0.0207*** (0.00529) (0.00540)

Differenceintotalremittancesinbasicversuseducationlabel 0.0154** _0.0201** _0.0194** (0.00748) (0.00792) (0.00823) Differenceintotalremittancesindirectpaymentversuseducationlabel −0.00503 (0.0122)

Female 0.0350 0.0558 0.0570

(0.0451) (0.0439) (0.0442) NumberofchildreninthePhilippines 0.0499*** _0.0543*** _0.0548*** (0.0161) (0.0164) (0.0164) MarriedwithspouseinthePhilippines 0.0766 0.0728 0.0720

(0.0559) (0.0557) (0.0558) Numberofnephewsandniecesaged5–22inthePhilippines −0.00448 −0.00532* −0.00521*

(0.00294) (0.00301) (0.00300) AttendedCollegeorUniversity −0.0480 −0.0409 −0.0418

(0.0472) (0.0472) (0.0472) FirstcametoItalybefore2005 −0.0496 −0.0631 −0.0632 (0.0421) (0.0424) (0.0424) IncomeinItalylessthan800D/month −0.0773* −0.0915** −0.0927**

(0.0424) (0.0421) (0.0420) AmountsenttothePhilippinesforeducationinlast12months −0.000233 0.000543 0.000538

(0.00122) (0.00124) (0.00124)

Meantake-uprate 0.27 0.27 0.27 0.27 0.27

Observations 483 483 483 483 483

Notes:Coefficientsaremarginaleffectsfromprobitestimation,estimatedatthemean.Robuststandarderrorsinparentheses. Amountsexpressedintermsofhundredsofeuros.

* _Significance_at_the_10%_level. ** _Significance_at_the_5%_level. *** _Significance_at_the_1%_level.

thatmigrantswithchildreninthePhilippinesaremorelikelytosignupforthisproduct,whereasthosewithlowerincomes arelesslikelytousetheproduct.

Thethirdandfourthcolumnsshowtheassociationwiththedifferentialamounttheychoosetoremitinthegamewhen giventheeducationlabelingoption.Column3showseach100D moreremittedwhengiventhelabelingoptionisassociated witha1.5%pointincreaseinthelikelihoodoftake-up,whichisstatisticallysignificantatthe5%level.Thisrelationship strengthensslightlywhenweaddadditionalcontrolsincolumn4.Hereaonestandarddeviationincreaseinthedifferential amountremittedwhengiventheoptionforeducationlabelingisassociatedwitha5.1%pointincreaseinthelikelihoodof take-up,representinga19%increaserelativetothemeantake-uprate.

Thelastcolumn addsthe differentialamountremittedwhengiven thedirect paymentoptionversus theeducation labelingoption.Weseethatthisvariablehasasmallandnotstatisticallysignificantassociationwithtake-up,whilethe effectoflabelingversusthebasicoptionremainsstatisticallysignificant.Thissuggeststhattake-upoftheEduPayproduct islargelydrivenbyademandforbeingabletolabelremittancesforeducation,ratherthanforademandfordirectpayment toschools.Thisisconsistentwiththefindingsofthelab-in-the-fieldexperiment.13

6. Conclusion

Wefindinalab-in-the-fieldexperimentthatmigrantsarepreparedtoremitmoremoneybacktothehomecountry whengiventheoptiontolabelsomeofthismoneyasexplicitlyforeducationpurposes.Strengtheningthiscommitment

13 _Online_Appendix_Table₃_presents_regression_results_that_correspond_to_those_of_Table₄_,_but_in_which_interaction_terms_are_added_between_“male”

(12)

fromthesoftcommitmentoflabelingtothehardcommitmentofdirectlypayingtheschool(andpotentiallymonitoringthe schoolingofthesponsoredstudent)resultsinlittleadditionalchangeinremittingbehavior.Furthermore,wefindbehavior inthisgamepredictstake-upofanewfinancialproductdesignedtoenablemigrantstodirectremittancestoschoolsinthe Philippines.Migrantswhoremitmoreforeducationpurposesinthegame,andwhoremitmorewitheducationlabelingthan without,aremorelikelytowanttousethisnewproduct.Thedemandforthenewproductseemsdrivenmorebyademand fortheabilitytolabelremittancesforeducation,thanforthehardcommitmententailedbypayingtheschooldirectly.

Thesefindingsareconsistentwithrecentevidencefromotherdomainssuggestingthatsoftcommitmentsintheform oflabelingcanchangespendingandsavingbehaviors.Theysuggestthatmigrantsmaybewillingtoincreasetheamount theyremitforeducationifgiventheabilitytoexertsomesoftcontroloveritsuse.Thechallengeforfutureworkisthento explorefurtherthelogisticsofhowbesttodothis,aswellastodeterminewhenharderformsofcommitmentsuchasdirect paymenttotheschoolswillbemoreeffective.RelatedworkbyAmbleretal.(2015)pointstoachallengeindoingso,since theyfindnotake-upamongSalvadoranmigrantsforoneproductlabeledforeducation.However,severalprivatesector companiesarenowpilotingproductsthatchannelremittancesdirectlytoeducation,includingPhilSmileforthePhilippines, andtheremittanceoperatorIMEinNepal.Weviewmeasuringtheimpactofsuchproductsonoverallremittancesandon schoolingoutcomesasapromisingdirectionforfutureresearch.

AppendixA. Supplementarydata

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jebo.2014.12.025.

References

Ambler,K.,2015.Don’ttellonme:experimentalevidenceofasymmetricinformationintransnationalhouseholds.J.Dev.Econ.(forthcoming).

Ambler,K.,Diego,A.,Dean,Y.,2015.Subsidizingremittancesforeducation:afieldexperimentamongmigrantsfromElSalvador.Am.Econ.J.Appl.Econ. (forthcoming).

Ashraf,N.,Diego,A.,Claudia,M.A.,Dean,Y.,2015.Savingsintransnationalhouseholds:afieldexperimentamongmigrantsfromElSalvador.Rev.Econ. Stat.(forthcoming).

Benhassine,N.,Florencia,D.,Esther,D.,Pascaline,D.,Victor,P.,2015.Turningashoveintoanudge?A‘labeledcashtransfer’foreducation.Am.Econ.J. Econ.Policy(forthcoming).

Chen,J.,2013.Identifyingnon-cooperativebehavioramongspouses:childoutcomesinmigrant-sendinghouseholds.J.Dev.Econ.100(1),1–18.

Edwards,A.C.,Ureta,M.,2003.Internationalmigration,remittancesandschooling:evidencefromElSalvador.J.Dev.Econ.72(2),429–461.

DeLaat,J.,2014.Householdallocationsandendogenousinformation:thecaseofsplitmigrantsinKenya.J.Dev.Econ.106,108–117.

Doi,Y.,McKenzie,D.,Zia,B.,2014.Whoyoutrainmatters:identifyingcombinedeffectsoffinancialeducationonmigranthouseholds.J.Dev.Econ.109, 39–55.

Genicot,G.,deWeerdt,J.,Alice,M.,2014.AsymmetryofInformationWithinFamilyNetworks,WorkingPaper.GeorgetownUniversity.

InternationalOrganizationforMigration(IOM),2010.TheItaly–PhilippinesMigrationandRemittanceCorridor.InternationalOrganizationforMigration, Makati,Philippines.

Italialavoro,2013.“ThePhilippineCommunityinItaly”,AnnualReportonthePresenceofImmigrants.MinisterodelLavoroedellePoliticheSociali, GovernmentofItaly.

Karlan,D.,Leigh,L.,2014.LooseKnots:StrongVersusWeakCommitmentstoSaveforEducationinUganda.Mimeo,YaleUniversity.

Rapoport,H.,Docquier,F.,2006.Theeconomicsofmigrants’remittances.In:Kolm,S.C.,Ythier,J.M.(Eds.),HandbookoftheEconomicsofGiving,Altruism andReciprocity,vol.2.Elsevier,pp.1135–1198.

Soman,D.,Cheema,A.,2011.Earmarkingandpartitioning:increasingsavingbylow-incomehouseholds.J.Market.Res.48,S14–S22.

Thaler,R.,1999.Mentalaccountingmatters.J.Behav.Decis.Mak.12,183–206.

Thaler,R.,Shefrin,H.M.,1981.Aneconomictheoryofself-control.J.Polit.Econ.89(2),392–406.

Theoharides,C.,2014.ManilatoMalaysia,QuezontoQatar:InternationalMigrationandtheEffectsonOrigin-CountryHumanCapital,WorkingPaper. AmherstCollege.

WorldBank, 2013.Migration andDevelopment BriefNo.20.http://siteresources.worldbank.org/INTPROSPECTS/Resources/334934-1110315015165/ MigrationandDevelopmentBrief20.pdf