1

TheChoiceamongAidDonors:TheEffectsofMultilateralvs.BilateralAidonRecipientBehavioralSupport1

MichaelG.Findley3.108BATTSHall

DepartmentofGovernmentUniversityofTexasatAustin

Austin,TX78712mikefindley@utexas.edu

512.232.7208

HelenV.Milner431RobertsonHallDepartmentofPoliticsPrincetonUniversityPrinceton,NJ08544

hmilner@princeton.edu609.258.0181

DanielL.Nielson

745SWKTDepartmentofPoliticalScienceBrighamYoungUniversity

Provo,UT84602dan_nielson@byu.edu

801.422.3417

20February2016

1Becauseoneofthearticleauthorswasalsoaneditorofthespecialissue,forthismanuscriptthestandingeditorexclusivelyhandledallcorrespondenceanddecisions.AnonlineappendixandreplicationdataaccompanythisarticleandappearontheReviewofInternationalOrganizationswebsiteaswellasatwww.michael-findley.com.WethanktheeditorsandreviewersforvaluablefeedbackaswellasPulapreBalakrishna,KateBaldwin,ChrisBlatt-man,ThadDunning,GuyGrossman,JoshGubler,SaadGulzar,JudeHayes,DarrenHawkins,MacartanHumphreys,SusanHyde,EvanLieberman,RobertKeohane,KosukeImai,QuinnMecham,ScottMorgenstern,KevinMorrison,PaulPoast,DanPosner,JessicaPreece,PiaRaffler,JoelSelway,DustinTingley,MikeTomz,andJeremyWeinsteinfortheirveryhelpfulcomments.TorbenBehmer,PeterCarroll,ColbyClabaugh,MaddyGleave,RaymondHicks,CarloHorz,JoanRicart-Huguet,BrandonMillerdelaCuesta,andElizabethNugentalsoprovidedinvaluableresearchassis-tance.PreviousversionspresentedattheAmericanPoliticalScienceAssociation,Chicago,IL,Aug29-Sept1,2013,theMidwestPoliticalScienceAssociationMeeting,Chicago,IL,April11-14,2013,theCAPERSconferenceatColum-biaUniversity,Dec.6,2013,andthePrincetonResearchFrontiersinForeignAidConference,April26-27,2013,andtheEvidenceinGovernanceandPolitics(EGAP)meeting,Oct.23-24,RiceUniversity.

2

Abstract

Scholars studying foreign assistance differ overwhethermultilateral aid is preferable tobilateralaidforpromotingdevelopment,butnearlyallbuildtheircasesprimarilyonhighlyaggregatedcross-national time-seriesdata.We investigate this topicexperimentally fromtheperspectiveofthosewhomtheforeignaiddirectlyaffects:recipientcitizensandelites.We thus report results of a survey experimentwith behavioral outcomes onmore than3,000Ugandancitizensandover300membersofUganda’sParliament.Inspiteofalargeliteraturesuggestingdifferences,thefindingsgenerallyrevealfewsubstantivedifferencesin citizens’ and elites’ preferences and behavior toward the two types of aid. While nostrong pattern of differences emerges, limited evidence suggests that the public evincesgreater trust inmultilateral institutions,andbothmassesandelites feel thatmultilateralaid is more transparent. Overall, these null results inform an ever-expanding literature,which is increasingly articulating distinctions between multilateral and bilateral aid. Atleastinthemindsoftherecipients,however,multilateralandbilateralaidmaynotinfactbeallthatdifferent.Thisaccordswiththeliteraturenotingthestrongoverlapinaidorgan-izations and bemoaning the fact that they do not specialize more. Our results raise thequestionaboutwhyhavebothmultilateralandbilateralaiddonorsiftheyineffectdothesamething.

Keywords:Foreignaid;experiments;foreigndonors;internationalorganizations

3

Introduction

Scholarshavedebatedwhetherbilateralaidormultilateralassistancedoesmoreto

promotedevelopment.2Theoretically,multilateralaidisoftenseenaslesspoliticalsinceit

islessspecificallytiedtodonors’foreignpolicyagendas,whicharebelievedtobedrivenin

turnbytheirpoliticalinterests.AsMartensetal.(2002)write,“Multilateralaidagencies

maybesomewhatshieldedagainstdirectpoliticalpressurefromtheirmemberstates.”

Rodrik(1996)addsthatmultilateralaidagenciesprovidemoreinformationaboutrecipi-

entcountriesandallowconditionalitytobemoreeffectivelyimposedonthem,concluding

that“multilateralflowsarelessgovernedbypoliticalconsiderationsthanbilateralones.”

Somemacro-levelempiricalstudieshaveproducedresultssuggestingthatmultilateral

agenciesfunddifferentcountriesandprojectsfrombilateralones,andthatmultilateral

projectstendtogotopoorercountriesandtothosewithgreaterneedscomparedtobilat-

eralaid(MaizelsandNissanke1984;Tsoutsoplides1991;FreyandSchneider1986;

BurnsideandDollar2000;Neumayer2003;Girod2008).

Ontheotherhand,dissentingscholarshavecontendedthatmultilateralaidcanbe

highlypoliticalaswell(GartzkeandNaoi2011).EvidencesuggeststhatUNSecurityCoun-

cil(UNSC)membershipinfluencesWorldBank(WB)loans(Dreheretal.2009)andthat

WorldBankprojectsappeartoactuallyswayvotesintheUnitedNations(UN)(Dreherand

Sturm2012;Dreheretal.2009).Indeed,somehavearguedthatbecausedevelopingcoun-

2Ontheeffectivenessofbilateralversusmultilateralaid,seeAlviandSenbeta(2012);Headey(2008);Kizhakethalackaletal.(2013);MinoiuandReddy(2007,2010);Ram(2003,2004).Atthesubnationallevel,seeDreheretal.(2016);IsakssonandKotsadam(2016).

4

triesaremembersofmultilateraldevelopmentbanksandsometimesjointlyholdnearor

fullmajoritiesofvotingshares,recipientscanmorereadilyturntheinternalpoliticsofthe

multilateralstowardtheirinterests(Lyneetal.2006,2009;Christensenetal.2011).

Bothcampsofscholarsthereforecontendthatthewayinwhichaidisdelivered—

specifically,whetheritisgivenbyamultilateralorabilateraldonor—mayaffectitsim-

pact.Theyhavebuilttheircompetingcasesonhighlyaggregated,large-nstatistical,obser-

vationalevidence.Inanotherliteratureonaid,scholarshavenotedthatthereisgreatdo-

norfragmentationandoverlap(Acharyaetal.2006;FrotandSantiso2009).Recipientsare

oftengivenaidbymanydonors,multilateralandbilateral;andthesedonorsoftencontrib-

utetotheexactsamesectorsandlocations,andoftentothesameprojects(Djankovetal.

2009;KnackandRahman2007).Afailurebydonorstocoordinateandspecializeisoften

bemoanedinthescholarlyliterature,andcallshavebeenmadetoincreaseharmonization

ofaiddonors(Easterly2007;Steinwand2015;KnackandSmets2013).Thesestudiessug-

gestthataiddonorsmaybeindistinguishablefromeachother,astheyallprovideaidtothe

samesetsofcountriesforthesametypesofprojectsinthesameareas.Thesestudiesraise

questionsaboutwhytherearesomanyaidagenciesandespeciallywhymultilateraland

bilateralonesbothexistiftheyoverlapsomuch.

Inanattempttobringadifferenttypeofevidencetobearonthedebate,wepremise

thisstudyontheideathattheactualrecipientsoftheaidshouldperceiveanymeaningful

differencesbetweenthetypesofforeignassistanceandshouldreflecttheseviewsintheir

attitudesandbehaviortowardprojectsfromdifferentdonors.Recipientcitizens'abilityto

discernbetweendifferentdonorsandtheirabilitytodevelopdifferingpreferencesoveraid

fromdifferentsourcesisimportantfromanaideffectivenessstandpoint.Citizens'ability

5

todiscerntheoriginsofforeignfundingforaprojectiscriticalbecauseitwillincreaseac-

countabilityfortheprojectsandthus,presumably,improvetheirperformance.Indeed,we

identifyfivereasonsthatindividualsmightgivefortheirpreferencesaboutaidprograms:

politicization,conditionality,transparency,efficiency,andalignment,whichwedefinelater.

Doesgreatersupportformultilateraloverbilateralprojects,orviceversa,havetodowith

howpoliticizeditis,howmuchconditionalityisimposed,howtransparentandaccountable

theprojectsare,howefficientlytheprojectiscarriedoutand/orhowalignedtheprojectis

withrecipients’preferences?

Weexploretheseissueswithwhat,tothebestofourknowledge,isthefirstnation-

allyrepresentative,large-N(n=3,017)studyaboutaidperceptionsandbehaviorsinan

aid-dependentdevelopingcountry.Weaccompaniedtheexperimentwithanextensive

surveytoprobethecausalmechanismsbehindthebehavioraloutcomes.Wealsoper-

formedasubstantivelysimilarsurveyexperimenton339membersofUganda’sparliament.

Thisenablesanexperimentalanalysisofbothmassandeliteattitudesandbehaviortoward

aidfromdifferenttypesofdonors.Likemanyaidrecipients,Ugandaisapoorcountrywith

highlevelsofaidflows:on-budgettogetherwithoff-budgetaidequaledroughly43percent

ofnationaleconomicanddevelopmentbudgetexpendituresin2012,theyearofthisstudy

(Tierneyetal.2011;Kiwanuka2012).Thusitisagoodcandidateforastudyofrecipients’

reactionstoaid.Micro-leveldata,suchasthatwecollected,provideanimportanteviden-

tiarysupplementinaddressingquestionsprobingdifferencesbetweenthetwomaintypes

ofaid.Oursurveyworkthusprovidesevidenceenablingustolearnwhethereithercitizens

orelitesseemultilateralorbilateralaidasbetteratmeetingtheirinterests.Thesurveyex-

6

perimentsadditionallyenableexplorationofthereasonsrecipientcitizensandelitesgive

fortheirpreferencestowardaidfromdifferentdonors.

Fortwoactualaidprojectsinthepipelinefinancedbymultipleinternationalorgani-

zationsandgovernments,werandomlyassignedpromptsnamingthedifferentdonorsand

assessed the effectson respondents’ supportmeasuredby their attitudes andactions. In

thesurveyexperimentweincludedthemajordonorstoUganda:theWorldBankandAfri-

canDevelopmentBank formultilateral fundsandtheUnitedStatesandChinaaid for the

bilateral ones. Prior research suggests some differences, especially, between Chinese aid

and the other donors (Dreher et al. 2015, though see alsoDreher and Fuchs (2015)). In

general, however, neither citizens nor elites expressmajor differences in their attitudes

andbehaviorstowardaidfrombilateralcomparedtomultilateraldonors.Anydifferences

are small substantively and only rarely significant statistically. Given the set of possible

analyses,thefewstatisticallysignificantresultsdonotadduptoaunifiedandrobustcon-

clusioninsupportofmultilateralorbilateralaidintheviewsofrecipients.Thisseemscon-

sistentwiththeliteraturethatnotestheheavyoverlapintheoperationsofaiddonorsand

theirfailuretocoordinateandspecialize.

Among the isolated results that surface, citizenswhoare familiarwith thedonors

aresignificantlymoresupportiveoverallforprojectsfundedbymultilateralorganizations

(WorldBankandAfricanDevelopmentBank) compared tobilateraldonors (theU.S. and

China)intwoofsixconditions(andalsoinacombinedoutcomeindex).However,citizen

familiaritywith thegiversof aidvariesacross thedonors, so treatmenteffectsare likely

biasedbyselection.FortwoofsixoutcomesMPspreferbilateraldonorstomultilateraldo-

nors,oppositeofwhatthepublicexpresses,buttheresultsareisolatedanddonotemerge

7

intheoutcomeindex.Comparing individualdonors,citizenrespondentsaremorewilling

toexpresstheirsupporttolocalleadersandtosendanSMSmessageforAmericanprojects

than forChineseaid,but theyremain indifferent for theotheroutcomes. Inaddition, the

reasonsthatpublicsmaypreferthedonorstheyknowarenotonesidentifiedasmuchin

theprior literaturebut rather citizens seem to respondmore to issues related to lackof

conditionality, better transparency,more trust, and greater efficacy. Inwhat followswe

motivatethestudy,describetheresearchdesign,andanalyzeresults.

TheoryandHypotheses

Whywouldrecipientshavedifferentpreferences foraidprojectsgiventhetypeof

provider? Itseems likely thatrecipientswillpreferdonorswhogive themaidwithmore

benefitsat lowercost. Itmaybethecasethatdifferentdonors,becauseoftheirpractices

andpreferences,giveaid inwaysthathavedistinctconsequences.Somerecipientsmight

gainmuchmoreasdifferentaidpackagesareprovidedbydifferentdonors(relativetooth-

eraiddonors’packages).Donorsappeartohaveaparticulartypeofaidpackagetheypre-

fertodeliver:thismightentailthesectortheaidtargets,themeansofdeliveringaid,orthe

amount of conditions attached to aid (Bermeo 2011, 2016, 2010; Dietrich 2016). Other

scholars,suchasAutesserre(2010),havesuggestedthatforeignactorsindevelopingcoun-

triesapproachtheirprojectswithadominantnarrativethatarisesfromtheirowndomes-

ticsituationsorpast interventionsandarenotnecessarilyappropriate forthecountryat

hand. Someevidencesuggeststhatdifferentdonorsprovidedistinctlydifferentaidpack-

ages(e.g.,Dietrich2016). Itisthereforepossiblethatrecipientshavesomeknowledgeof

8

how thesedifferentdonors’practicesaffect theprojects thatarebeing implementedand

thustheirwell-being.

Consistentwiththeliterature,wesubdivideaiddonorsintotwobroadtypes:bilat-

eralandmultilateral.Bilateraldonorsarerepresentedbysinglecountryagenciesthatpro-

videaiddirectlytodevelopingcountriesorNGOs.USAIDisanexampleofabilateralagen-

cy,andsince2000ithasbeenthebiggestbilateraldonortoUganda,thecountryinwhich

the present study was executed, followed by China, the UK, Denmark, Netherlands and

Norway (Tierney et al. 2011; Strange et al. 2015). Many scholars contend that, because

domesticpoliticsaremuchmorelikelytobereflectedbybilateralagencieswhosemarch-

ingorderscomedirectlyfromdomesticpoliticians,thesepoliticalinterestsarethoughtto

distortbilateralaidaway fromtheneedsof the recipient countries—especiallypoverty

reduction—andtowardthepolicygoalsofthedonors(MaizelsandNissanke,1984;Frey

andSchneider,1986;Tsoutsoplides,1991).

Alternatively,multilateraldonorsexistwheremorethantwobilateraldonorspool

theiraid flowsand, through the internationalorganization’sowndecisionprocesses that

aggregatethemembercountries’preferences,thenprovidetheaidtodevelopingcountries

orNGOs.TheWorldBankisthemostwell-knownmultilateralagency,andithasgenerally

beenthelargestmultilateraldonortoUganda,followedbytheAfricanDevelopmentBank

(AFDB), the European Union (EU), European Bank for Reconstruction and Development

(EBRD),AsianDevelopmentBank(ADB),andUNagencies(Tierneyetal.2011).

Some studies suggest thatmultilateral agencies fund different countries and pro-

jectscomparedtobilateraldonors,andmultilateralassistancetendstotargetpoorercoun-

tries with greater needs (Maizels and Nissanke 1984; Tsoutsoplides 1991; Frey and

9

Schneider 1986). Additional evidence suggests thatmultilateral aid also tends to be less

political,isassociatedwithbetteroutcomes,andappearsbetterabletoimposemoreeffec-

tive conditions (Maizels andNissanke 1984;Martens et al. 2002; Rodrik 1996). For in-

stance,Maizels andNissanke (1984) find that the recipient's need is relativelymore im-

portantforaggregatemultilateralthanforbilateralaidflows,whereaspolitical,economic,

andmilitarystrategicinterestsdominatetheallocationofbilateralaid.

Tsoutsoplides(1991)showsthatqualityoflifemeasuresexertastatisticallysignifi-

cant influenceuponaidallocationbythemultilateralEuropeanCommunityagencyinthe

1975to1980period.BurnsideandDollar(2000)findthatmultilateralaidresultsinbetter

outcomesforrecipientcountriesthandoesbilateralaid.Astheyconcludefromtheirquan-

titativeanalysisofcountry-yeardata,“aidthatismanagedmultilaterally(aboutone-third

ofthetotal)isallocatedinfavorofgoodpolicy.”AndNeumayer(2003)pointsoutthatthe

donorinterestbiasesinherentinbilateralaidarenotalwayspresentinmultilateralaidgiv-

ing.Asheconcludes,“theUNagenciestrytocounteracttosomeextentthebiasthatisap-

parentintheaidallocationofmanyotherdonors.”Milner(2006)showsthatmultilateral

aidseemstobemoreconnectedtodevelopmentgoalsinthemindofthedonorpublicthan

isbilateralaid.Focusingon thedistinctionbetweendonors,Girod(2008)claimsthatbe-

causemultilateraldonorsarenotbeholdentostrategicinterests,theycandistributeaidfor

developmental purposes and effectively target aid to countries thatpursue economic re-

forms.

However,otherscholarscontendthatmultilateralaidcanalsobehighlypoliticized

(GartzkeandNaoi2011).Whiletheearlierliteratureindeedseemedtoidentifypoliticiza-

tion among bilateral agencies, later scholarship appears to find similar patterns among

10

multilateraldevelopmentbanks.Inparticular,evidencesuggeststhatWorldBankaidtends

to flow disproportionately tomembers of the UN Security Council, indicating significant

levels of politicization and instrumentality in donation decisions (Dreher et al. 2009).

World Bank loansmay even influence UNGA votes toward the interests of the powerful

countriesintheGroupof7(DreherandSturm2012).Moreover,powerfulcountriesmaybe

abletoswaythemultilateralstowardtheirinterestsinformallybyinfluencinggeographic

targeting, disbursements, and the pace of approval (Kilby 2006; Kilby andDreher 2010;

Kilby2013).Allof this scholarship implies thatpolitics, andnotnecessarilyconcerns for

povertyalleviation,drivesmultilateraldonations.

Politicsmayalsoinfluencemultilateralassistanceinadifferentway.Asithappens,

developingcountriesjointlyholdsignificantvotingsharesatallofthedevelopmentbanks,

andindeedinsomemultilateraldevelopmentbanks(MDBs)—suchastheAfricanDevel-

opmentBank(AfDB)andtheInter-AmericanDevelopmentBank(IADB)—recipientcoun-

triesactuallyexercisevotingmajorities.Thismayenablerecipientstocoalescetocounter

thepoliticalinfluenceofOECDcountriesinwaysthatmightprove,intheend,tobeequally

political (Lyne et al. 2006, 2009). Indeed, some evidence suggests that the multilateral

banksmaybeless—ratherthanmore—sensitivetoproblemsofcorruptionthanbilateral

agenciesandthatthismaymakemultilateralassistancelesseffectiveatpromotingdevel-

opmentgoals,forexample,inbringingaboutbettereducationoutcomes(Christensenetal.

2011).

Strongobservationalstudiesthereforeexistonbothsidesofthisdebate.Inthis

studyweexplorethepossibilitythat,ifeitherbilateralormultilateralaidworksbetterto

promotetheinterestsofindividualrecipientcitizensandelites,theserecipientsmightbe

11

expectedtoperceivethedifferencesandthereforeshowgreatersupportformultilateralor

bilateralaid.Itisimportanttonotethatanotherstrandintheaidliteraturepointsoutthat

manyaiddonorsineffectgivetothesamecountriesforthesameprojectsinthesameare-

as.Thesestudiesnotetheproliferationofdonorswith,byonesomewhatdatedestimate,

27officialbilateraldonorsandroughly20officialmultilateraldonorsaroundtheworld

(Djankovetal.2009).AidDatainlate2016lists55bilateraldonorsand63multilaterals.

Multilateralagenciesandsubsidiarieshavecontinuedtogrow,andAidData’sinformation

baseshowsthatnon-DACdonorsaremoreactivethanoncethought(Tierneyetal.2010).

Manynotethatdonorshavefailedtocoordinateandspecializetheiraidgivingandpointto

anoverlapintheirprojects(Aldasoroetal.2010;AnnenandMoers2016;Bigstenand

Tengstam2015;BourguignonandPlatteau2015;FrotandSantiso2009;Easterlyand

Williamson2011;Fuchsetal.2015;KnackandRahman2007;Acharyaetal.2006).During

theearly2000s,forexample,Ugandahad14.8donorspersector(a2%increasefrom

2005)and7.7sectorsperdonor(a11%increase)in2009.MorerecentdatafromUganda

confirmsthistrend.Nunnenkampetal.(2015)findthattheduplicationofeffortsamong

aiddonorsincreasedbetween2006-2009and2010-2013forsixoutofUganda'sninema-

jordonors.Usingsubnationaldata,forinstance,theyfindthat$100millionofaidbythe

topninedonorsinUgandawereonlyspreadoverlessthanfourdifferentdistrict-sector

combinations(Nunnenkampetal.2015).Thisoverlapamongdonorsexpandedforevery

sectorbetween2006-2009and2010-2013,saveforhealthandeducation.Thisresearch

suggeststhatrecipientsmaynotperceiveanydifferencesamongdonorssincetheyare

fundingsuchsimilarprojects,anditraisesthequestionofwhycountriesusebothbilateral

andmultilateralaid.

12

Our studyprobeswhether recipientsdo finddifferences amongdonors.Recipient

citizens'abilitytodiscerntheoriginsof foreignfundingforaproject iscriticalbecause it

enables accountability for projects and thusmight improve their performance. Further-

more,weexaminetheviewsofbothmassesandelites.Politicalelitesandcitizensmayhave

distinctinterestsinrelationtoforeignassistance.Researchsuggeststhatpoliticalleaders,

especially those in the government, and citizens may thus react very differently to aid

(BuenodeMesquitaandSmith2007,2009;Findleyetal.2016).First,weexpectMPsinour

surveytohavemoreknowledgeaboutaidagenciesanddeliverythantheaveragecitizen.

Second,weexpectthatifonetypeofaidisseenasmoresubjecttopoliticalcontrolbyre-

cipient governments, then elites should favor that formof aid. For thepublic,we expect

thatpoliticalcontroloveraidasopposedtousingaidfordevelopmentandpovertyallevia-

tionwillbeopposed.Thus,ifonetypeofaidisseenasmorepoliticized,lessefficacious,less

abletomeettheirneeds,andlesstransparent,thepublicwillpreferthattypeofaidlesson

average.Priorresearchhaspresentedevidencethatthepublicprefersprojectsthatareless

likelytobepoliticizedandcapturedbypoliticalelites(Milneretal.2016).

Thus,simplehypothesescapturingbothschoolsofthoughtfollow.

H1:Citizensandpoliticalelitesinaidrecipientcountriesshouldhavedifferentprefer-

encesaboutmultilateralandbilateralaidandassociateddonors.

Ofcourse it ispossible thatneithermultilateralnorbilateraldonorsarepreferred

bycitizensandelitesinrecipientcountries.Thiscouldbeduetothefactthataidagencies

oftenoverlapheavilyintheiraidgivingorduetooffsettingeffectswhereinsomemultilat-

13

eralsandbilateralsarepoliticizedwhereasothersarenot,oritcouldsimplybethatdespite

characteristicsofaidgiving,citizensandpoliticalelitesdonotholdstrongpreferencesover

thetypesofdonorsofferingassistance.

Ifinfactrecipientsprefermultilateralorbilateralaid,anarrayofmechanismsmight

explain why. We propose five reasons that individuals might give for their preferences

aboutaidprograms.Thesefivereasonsforaidpreferencesarepoliticization,conditionali-

ty,transparency,efficiency,andalignment.First,citizensmightbeconcernedaboutthepo-

liticizationoftheaidprogram.Thatis,theymightthinkthatsomedonorswillfavorcertain

groups,regions,orprojectsoverothersduetopoliticalconsiderations.Donorsmightdirect

aidinthiswayortherecipientgovernmentmaybeabletocontrolaidinordertodistribute

itinwayspoliticallyusefultothem.Andsomeaiddonorsmaybemoreabletobecaptured

thanothers.Priorresearchsuggeststhatcertainaidprogramscanbetargetedtoassistpo-

liticallyimportantgroups,ratherthanbeingassignedonthebasisofneed(Jablonski2014).

Ourassumption is thatgovernmentsandrulingpartieswanttoremain inpower; foreign

aidisjustanotherresourcethattheycanusetodoso.Itiswellknownthatgovernments

usealltypesofprogramstodistributefavorstopoliticallyimportantgroups,andaidisjust

one more form of resource they can distribute (Morrison 2009; Pepinsky 2008; Bates

1981).Thus,aidmightbedirectedmoretowardsregionsorgroupswhoprovidemorepo-

liticalsupport for thegovernmentorrulingparty(Dreheretal.2015).Oraidmight flow

towardprojectsthatthegovernmentanditsrulingpartyfavorsforelectoralreasonssuch

asareas thatdonotsupport therulingparty inorder towinvotes.MPs,especially those

withinthegovernment,maypreferthistypeofaidsinceitmightenhancetheirchancesof

stayinginoffice.

14

Politicization,ontheonehand,implicatestherecipientgovernment,butitalsomay

haveimplicationsforthedonor.Foreigndonorsappeartogiveaidasamechanismforex-

tractingachangeinpolicyfromtherecipientgovernment(BuenodeMesquitaandSmith

2009,2007).Politicizationcanalsorefertothewaydonorstargetaidtosatisfytheirown

goals.Hence,aidmightbedirectedbytheforeigndonortowardgroups,regions,orsectors

thatthedonorseesaspoliticallyimportant.Ifdonorswantuseofamilitarybaseinsome

regionoftherecipientcountry,theymaytargetaidtowardthatregioninhopesofbuying

supportorquiescencefortheforeignmilitarypresence.Likewise,donorsmayseektobuy

supportfortheirpositionsininternationalorganizationssuchastheUnitedNationsSecuri-

tyCouncil.WhileMPsmaypreferaidthatismorepoliticallyuseful,theaveragecitizenmay

notsincethismeanshisorherneedsarelesslikelytobemet.Ifpoliticizationmatters,then

weexpect:

H2a:Amongcitizensthelesspoliticizedformofaidshouldbethemorepreferred.

H2b:AmongMPs,themorepoliticizedformofaidshouldbemorepreferred.

Second, the conditions that donors attach to aid may matter for what recipients

thinkof it. Themore costly the changes that thedonordemands, theharsher the condi-

tions.Wegenerallyexpectthatthemoreconditionsandtheharshertheyare,thelesslike-

lyrecipientsofanytypearetopreferthattypeofaid.WeanticipatethatMPsparticularly

willnotlikeconditionsonaidsincethisreducestheflexibilitywithwhichthegovernment

canuse theaid. Citizensmayormaynotdislikeconditionsdependingonperceptionsof

government.Totheextenttheydonottrusttheirowngovernment,citizensshouldprefer

15

moreconditionalaid(Milneretal.2016).Domultilateralandbilateralaidagenciesdiffer

intheirconditionality?Rodrik(1996)arguesthatmultilateraldonorsaremorecapableof

adding (more) conditions to aid and implementing them. BuenodeMesquita and Smith

(2009,2007)maketheargument,however,thatbilateraldonorsareprincipallyinterested

inaidinexchangeforpolicyconcessionsbyrecipientgovernments. Thesepolicyconces-

sionsaremorelikelytobepolitical(andevengeopolitical)thaneconomic,however.Stone

(2004,2002)pointsoutthatmultilateralagenciesareoftenoverruledbypowerfuldonor

countriesintheirattempttoimposeconditionsonstrategicallyimportantrecipientcoun-

tries, suggesting that enforcement may be weak for both multilateral and bilateral aid

agencies.Soitisunclearwhichtypeofaidagencymayimposemoreconditionalityandbe

betterabletoenforceit.

H3a:Amongcitizens,themoreconditionalformofaidshouldbethemorepreferred.

H3b:AmongMPs,thelessconditionalformofaidshouldbethemorepreferred.

Third,multilateralandbilateralaidshouldvaryaccordingtothedegreeoftranspar-

encyandmonitoringtheyallow.Rodrik(1996)againclaimsthatmultilateraldonorsmay

bemoreabletoextractinformationaboutrecipientsandhowtheyuseaid.Itseemslikely

thatbilateraldonorsarelessinterestedintheexactoutcomeofaidprojectsthantheyare

inthepolicyconcessionstheyreceiveinexchangeforaid.Hencetheirneedfortransparen-

cyandmonitoringofaidislowandtheirdesiretohavethepolicyconcessionsunmaskedto

thepublicisprobablyevenlower.However,someimportantbilateraldonors,suchasthe

UnitedStates,publicizevoluminousdocumentationandthereforeseektobehighlytrans-

16

parent.EasterlyandPfutze (2008)developan indexof transparency.Theyconclude that

multilateral aid organizations generally do better than bilateral ones, but some bilateral

agenciesperformwellonthetransparencyscale.Forthedonorsweconsider,theU.S.ranks

6thamongbilateraldonorsandChinawasnotmeasured;theWorldBankisdisaggregated

andtheInternationalDevelopmentAssociation(IDA)ranks1stamongmultilateraldonors

whereas theAfDBranks4th.3Again, thepublicandpolitical elitesmaydifferon theiras-

sessmentofthedesirabilityoftransparency.Thepubliconaverageshouldfavoritandsee

this as ameans formaking sure aid helps them.However, forMPs the situationmaybe

more complicated. Greater transparencymaywork against getting the aid to the people

and projects theymost value politically.We expect that if these factorsmatter thenwe

shouldseethefollowing:

H4a: Among the public, themore transparent form of aid should be themore pre-

ferred.

H4b:AmongMPs,however, the less transparent formofaidshouldbe themorepre-

ferred.

Fourth, theefficacy,efficiencyorsuccessoftheaidmaymattermosttorecipients.

Improvingwelfarebypromotinghealth,sanitation,employment,education,nutrition,lon-

gevity,and/ortransportationmaybe foremost inrecipients’minds. Multilateralaidmay

3PublishWhatYouFundprovidestransparencyrankingsforallthesedonors,poolingmultilateralsandbilaterals.Of46donors,theUnitedStatesisratedfairlyhigh(USMilleniumChallengeCorporationranks2ndandUSAgencyforInternationalDevelopmentranks19th),Chinaranksnearthebottomat45th,theWorldBankIDAranksquitehighat6thandtheAfDBranks10th.TheseratingsarethusbroadlyconsistentwithEasterlyandPfutze(2008)inidentifyingmultilateraldonorsasmoretransparentthanbilaterals.(Seehttp://ati.publishwhatyoufund.org/,ac-cessedDecember12,2016.)

17

bemoreeffectiveandefficientsinceitislikelytobeaimedateconomicdevelopmentmore

exclusivelyandmorelikelytobemonitoredcarefully.EasterlyandPfutze(2008)pointout

thatthreetypesofaidareusuallyviewedasleasteffective:tiedaid,foodaid,andtechnical

assistance.Multilateralaidagenciesdonotprovidemuchifanyofthesethreetypesofaid,

whilebilateralagenciesdo.Moreover,EasterlyandPfutze(2008)developanindexofse-

lectivitythatmeasureswhetheraidgoestopoor,autocratic,andcorruptcountries. They

showthatmanymultilateralaidagenciesdobetteronthesedimensionsthandobilateral

ones.Theyarguethatmultilateralsaremorelikelytogivetopoorercountries,butthisof-

tenmeanstherecipientcountriesaremorelikelytobeautocraticandcorrupt.

However,multilateraldonorsmayhavestricterpracticesforpreventingthediver-

sionofaid.Themajormultilateraldevelopmentbanks(MDBs)havealwayshadrules for

sanctioning corrupt practices, but they recently upgraded their rules and procedures to

rootoutcorruptioninaidprojects.InApril2010,thefiveleadingMDBs—theAfricanDe-

velopmentBank,theAsianDevelopmentBank,theEuropeanBankforReconstructionand

Development, the Inter-American Development Bank and the World Bank—signed an

agreementproviding formutualandreciprocalenforcementofdebarmentdecisions.The

AgreementforMutualEnforcementofDebarmentDecisionsthusincreasedtheriskfaced

bycommercialorganizationsthatdobusinessinthedevelopingworld,whileaffirmingthe

MDBs’ commitment to combating fraudulent, corrupt, and collusive practices.4 It is not

clearthatmostbilateralagencieshaveanythingclosetothissetofrulesandpowers.

4Priortothemid-1990s,MDBsreliedprimarilyontheirprocurementpoliciestocurbcorruptpractices.Asageneralrule,MDBsprovidefundingforpublicsectorprojectsontheconditionthattheborrowerselectsthecontractorsinacompetitiveprocess,carriedoutinaccordancewiththeprocurementpoliciesoftherelevantMDB.Thenin2006,thefivemainMDBs,togetherwiththeInternationalMonetaryFundandtheEuropeanInvestmentBank(EIB),establishedtheInternationalFinancialInstitutionsAnti-CorruptionTaskForceto

18

However,someevidencesuggeststhatmultilateraldonorsareequallyindifferentto

corruption as bilateral donors (Alesina andWeder 2002). Some evidence even indicates

that themultilateralbanks—becausedevelopingcountrieshavesignificantvoting influ-

enceanddislikeanti-corruptionconditionality—arelesssensitivetocorruption.Multilat-

eralsmaythusprovelesseffectiveinpromotingdevelopment,atleastintheeducationsec-

tor (Christensen et al. 2011). We expect both MPs and the public to desire that aid be

delivered in themost efficient and effectivemanner since thismeans theywill gain the

mostfromit.Againtheliteraturesuggestscompetinghypotheses.

H5a:Citizensandpoliticaleliteswhoperceivemultilateralaidasprioritizingeffective

andefficientdeliveryrelativetobilateralaidshouldprefermultilateraloverbilateral

aid.

H5b:Citizensandpoliticaleliteswhoperceivebilateralaidasprioritizingeffectiveand

efficientdeliveryrelative tomultilateralaidshouldpreferbilateralovermultilateral

aid.

Finally,wealsoconsidered theextent towhichsubjects felt thatmultilateralsand

bilateralsmatchedprojectswiththeirneeds. Prominentmultilateralandbilateraldonors

arethoughtbysometoallocateaidinquitedifferentways.Inparticular,itispossiblethat

developacatalogueofmeasuresaimedatharmonizingtheeffortsoftheparticipatinginstitutionsagainstfraudandcorruption.TheTaskForcerecommendationswerepublishedinSeptember2006inadocumenttitled‘UniformFrameworkforPreventingandCombatingFraudandCorruption’,whichwassubsequentlyendorsedbytheparticipatinginstitutionsandhencewasacrucialfirststepintheMDB’seffortstocoordinatetheireffortsagainstfraudandcorruption.TheUniformFrameworkcontainedasetofharmonizeddefinitionsforsanctionablepracticestobeusedbytheparticipatinginstitutionsinalltheiroperations.In2010fiveMDBs—AfDB,ADB,EBRD,IADB,andWB,signedtheMutualEnforcementAgreement(SeilerandMadir2012).

19

thesedistributionpatternsshapeperceptionsaboutaddressingneed. Generally, roughly

20percentofbilateralaidisdisbursedthroughNGOs,whilemultilateralagenciesdisburse

lessthan5percentthroughNGOs.Oftheaidthatcanbecategorizedintodistinctchannels,

35 percent ofmultilateral aid goes to the recipient government directly and close to 50

percentofbilateralaiddoes.Ifwecomparethetwolargestdonorscentraltothisstudy—

theWorldBankandUSAID—theproportionsareespeciallydifferent.Recipientgovern-

mentsmanagenearlyallWorldBankprojects.Contrastingly,USAIDcontractswithprivate

companiestomanageprojects,andgovernmentsrarelyseethefundsdirectly.Thechannel

ofdeliverymighthavesignificantinfluenceoveraideffectiveness(Dietrich2016).Theex-

tenttowhichrecipientpublicsandevenelitescanappreciatethesedifferencesisofcourse

opentodebate.Again,onthisdimensionweexpectMPsandthepublictoagreethatmeet-

ing community needs is important. They may define those needs differently but both

groupsshouldwantaidtoservetheircommunities.Wethushypothesize:

H6a:Citizensandpoliticaleliteswhoperceivemultilateralaidasmeetingcommunity

needsbetterthanbilateralaidshouldprefermultilateraloverbilateralaid.

H6b: Citizens and political elites who perceive bilateral aid as meeting community

needsbetterthanmultilateralaidshouldpreferbilateralovermultilateralaid.

TheResearchContext:Uganda

WhythefocusonUganda?Ugandaisapoordevelopingcountry,whichexperienced

civilwarintheearly1980sandpartialdemocratizationthereafter.Sincethemid-1980s

Ugandahasbeenmorestable,fastergrowing,andaleaderamongthedemocratizingAfri-

20

cancountries.Ithasalsobeenamagnetforforeignaid.Asonestudynotes“Uganda’seco-

nomicandpoliticalreformshaveattractedagreatdealofpraisesincePresidentYoweri

Museveniassumedpowerin1986.RegularlycitedasoneofAfrica’sfew‘donordarlings,’

Uganda’sstructuraladjustmentprogramandwide-rangingpoliticalreformshavebeen

heldresponsibleforitshigheconomicgrowthratesandstablegovernanceoverthepast

twodecades.Inparticular,theprocessbywhichpowerhasbeendeconcentratedandde-

volvedtofivelevelsoflocalgovernmenthasbeencalled‘oneofthemostfar-reachinglocal

governmentreformprogramsinthedevelopingworld’”(Green2010).Sincethe1990s,aid

hasbeenequaltoroughly80percentofUganda’sgovernmentexpendituresand15percent

ofitstotalGDP,thoughthesetotalshavedecreasedrecentlyduetothegrowingUgandan

economy,governmentbudget,andpublicexpenditures.Nevertheless,Ugandaremains

heavilyaiddependent.IfgroupswithinUgandahavelittle,ornoknowledgeofaidprojects

anddonors,thenitisunlikelythatgroupswithinotherdevelopingcountrieswillknow

muchmore.

Inaddition,itisinterestingtonotethatbilateralandmultilateralaidgotodifferent

sectorsinUganda.Bilateralaidtendstofundmorehumanitarianaidandcommodityand

generalprogramassistance.Multilateralaidtendstofocusoneconomicinfrastructureand

productionsectors.5Whydothesedifferencesexist?Aretheyareasonforthesupposed

preferenceofmultilateraloverbilateralaid?Ourresearchshouldallowustoseeifthese

differencesinsectorsmatterforcitizenandeliteperceptionsofaideffectivenessandsup-

portforprojectsdependingonthedonor.

5Thesearethefourofthefivemajor“sectors”definedbyOECDforcategorizingaid.Thefifthsector,socialinfra-structure,isprettyequallyfundedbythetwotypesofdonors.

21

Asnotedabove,ifwecomparethetwolargestdonorscentraltothisstudy—the

WorldBankandUSAID—theproportionsofaidthroughdifferentchannelsareespecially

different.RecipientgovernmentsmanagenearlyallWorldBankprojects;USAID,onthe

otherhand,contractswithprivatecompaniestomanageprojects,thusbypassinggovern-

ments.Ininterviews,multipleofficialsatUSAIDandtheWorldBankinUgandatoldusa

similarstoryaboutthedifferentaidmanagementstyles.ForUSAIDthehardesttaskis

monitoringthecontractorsandNGOstominimizeagencylosses.FortheWorldBank,the

challengeisplacingstrictauditingandprocurementrequirementsongovernments.

BasedonbroaderpatternstrackedbyAidData(Tierneyetal.2011)fortheperiod

2000-2013,UgandahasslightlyhigherthanaveragelevelsofaidchanneledthroughNGOs,

slightlylowerthanaveragelevelsoftiedaid,andrelativelysimilarpatternsofbudgetsup-

port.ForaidchanneledthroughNGOs,whereAidDatahasinformationonchannelofdeliv-

ery,36%ofaidprojectsinUgandaarechanneledthroughNGOscomparedto28%forthe

restofsub-SaharanAfrica;intermsofamountsofaid14%ofoverallaidinUgandaischan-

neledthroughNGOscomparedto11%intheremainingcountries.Countriessimilarto

UgandaintheirNGOpatternsincludeRwanda,SouthAfrica,Niger,SouthSudan,Malawi,

Mali,Eritrea,Sudan,Ethiopia,SierraLeone,Mauritania,BurkinaFaso,Liberia,CentralAfri-

canRepublic,DemocraticRepublicofCongo,andBurundi.

Ugandahaslowerlevelsoftiedaidthanaveragewith7%tiedinUgandacompared

to10%fortheremainderofsub-SaharanAfrica.CountriessimilartoUgandainclude

BurkinaFaso,Malawi,Benin,Lesotho,Gabon,Tanzania,Mali,CentralAfricanRepublic,

Madagascar,Rwanda.BudgetsupportlevelsinUgandaarequitesimilartotherestofsub-

SaharanAfricawith1%ofprojectsallocatedtobudgetsupportinbothUgandaandthere-

22

mainingcountries,thoughwith13%oftotalUgandanaiddevotedtobudgetsupportrela-

tiveto9%inothercountries.Inallcases,mostcountriesarefairlysimilarineachofthese

categories,andUgandaisrelativelyclosetotheaverageamongsub-SaharanAfricancoun-

tries.

Wethusexploretheseissueswithwhat,tothebestofourknowledge,isthefirstna-

tionallyrepresentative,large-N(n=3,017)studyofaidrecipientsinadevelopingcountry.6

Wealsosurveyed339currentandformermembersofParliamentinUganda. Weaccom-

paniedthemassexperimentwithanextensivesurveytoprobethecausalmechanismsbe-

hindthebehavioraloutcomes.TheMPsurveywassimilarbutbriefer.Asnoted,likemany

aidrecipients,Ugandaisapoorcountrywithhighlevelsofaidflows.Thusitisagoodcan-

didateforastudyofrecipients’reactionstoaid.

ResearchDesign

Thisstudydrawsontheexperimentalcontextanddesignreportedintwoother

studies(Milneretal.2016;Findleyetal.2016).7Weinvestigatetheattitudesandbehavior

ofmorethan3,000Ugandancitizens(N=3017)andover300MPs(N=339)towardforeign

aidthroughanationallyrepresentativeexperimentandsurveyofrecipients’preferences

overdifferentfunders.Theexperimentincorporatedbehavioralresponsesinwhichsub-

6Totalnforthestudywas3,582.Wedonotfocusononeconditionfromtheexperimenthereandhenceourob-servationsarereduced.Resultsforotherexperimentalconditionsreportedelsewhere((Milneretal.2016;Findleyetal.2016).7Incontrasttotheseearlierstudies,whichtakeonthequestionofpreferencesforaidvs.governmentspending,thepresentpaperfocusescentrallyonthedifferencesbetweenmultilateralandbilateraldonors.Intheearlierstudies,theauthorsnoteinpassingthattherearenodifferencesamongdonors,butonlyconsiderasimpletestthatallowsthemtopoolinanalysesofaidvs.governmentspending.Giventheattentiondevotedtodistinctionsbetweenmultilateralandbilateralaidinthebroaderaidliterature,thecurrentstudytakesonthisimportantques-tionsandprovidesathoroughexaminationofthepossibledistinctionsatboththemassandelitelevel,andinalargevarietyofsubgroupanalyses.

23

jectscouldsubstantiate(ornot)theirstatedpreferencesbyundertakingcostlypersonal

actions.Werandomlyassigneddescriptionsofactualpipelineprojectstorespondents.The

projectswereco-financedbymultiplecountriesandagencies,whichallowedustomanipu-

latethedonorpresented–namingpossiblecontributorsoneatatimeinseparatecondi-

tions–tothesubjectsaswellasthetypeofprojectwithoutusingactivedeception.

Forthemasssurvey,weusedarandomsamplingprocedureinwhichanyUgandan

adulthadroughlyanequalchanceofbeingselectedtoparticipateinthestudy.Westarted

withcensusdatatoselectthesubjectpool,matchingthenumberofparliamentaryconstit-

uenciesbyregionproportionaltothecensusdata.Fifty-fiveconstituencieswereselected,

with15intheCentralregion,15intheNorth,14intheWest,and11intheEast.Wethen

selected two sub-counties in each constituency, one parish in each sub-county, and one

pollingstationineachparishsothat,finally,eachparliamentaryconstituencyhadtwopoll-

ingstationsthatservedastheSamplingStartPoints(SSPs).Uganda’sone-partydominance

promptedustooversampleoppositionstrongholds.

Eighty-four local Ugandan enumerators administered the instrument to 3,017 re-

spondentsintheneighborhoodsandvillagesofthefourdifferentregionsofUgandaduring

themonthsof Juneand July2012.Theaverage interviewtimewas59.7minutes.The in-

strumentwastranslatedinto11locallanguagesthattheenumeratorsspoke;420(12per-

cent)oftheinterviewswereconductedinEnglish.

Wefurtherrandomizedtheadultwithinthehouseholdtowhomtheinstrumentwas

administered.Toaccomplishthis,enumeratorsobtainedalistofalladultsinthehousehold

(by gender, alternating homes) and then randomly chose one of those adults and asked

whethertheywouldcompletetheinterview.Ourprocedureworkedreasonablywell;gen-

24

der,education,age,party,religion,andregionalvariableswerenotsignificantlyrelatedto

whethersubjectsreceivedgivenexperimentalconditions.

WedrewconstituencieswithOppositionMPsinproportiontothenumberofOppo-

sitionMPsinParliament,usingdataonMPsfromthecurrent(9th)Parliament.Wedidthis

byregionaswell.Ouroversamplingofoppositionstrongholdsgaveusthebreakdownby

partyofthesampledconstituenciesseeninonlineAppendixTableA1columns2&3,which

isnotverydifferent fromthemakeupof thecurrentparliament (AppendixTableA1col-

umns4&5).

At the assigned polling stations, enumerators began at themain intersection and

each walked in a different direction, away from the other enumerators. They surveyed

housesontheleftsideofthestreet,startingwiththesecondhouseandeveryotherhouse

thereafter.Uponcompletion,theycountedonehousetoskipandsurveyedagain.Atwenty-

pagetrainingmanualspellsoutourprocessandisavailableuponrequest.

Theexperimentincorporatedbehavioralresponsesinwhichrespondentscould

substantiate(ornot)theirstatedpreferencesbyundertakingactionsimposingpersonal

costs.Werandomlyassigneddescriptionsoftwoactualforthcomingdevelopmentprojects

inthe“pipeline.”Theprojectswereco-financedbytheWorldBankandtheAfricanDevel-

opmentBankandthereforefundedbyallofthebanks’membergovernments,whichal-

lowedustorandomlyassignthenameddonorpresentedtotherespondentswithoutactive

deception.Thatis,becausemanystatesandorganizationscontributetothemultilateral

funds,wewereabletonamespecificdonorswhomightbecontributorstomultilateralef-

fortsofthissort.

25

Thetwoprojectsprovidedelectricityandeducation.Thetextoftheeducationpro-

jectwas: “The Post PrimaryEducation andTrainingAdaptable ProgramLending Project

seeks to increaseaccess to lowersecondaryeducation, improve thequalityof lowersec-

ondaryeducation, andenhanceprimaryeducationand training.Theprojectmay require

yourcommunitytoprovidingfundingformaintenance inthefuture. [Thisprojectwillbe

fundedby theRANDOMLYASSIGNEDFUNDER].Howmuchwould you support this pro-

ject?” Neither project type was significantly preferred over the other in the between-

subjectsdesign,whichlikelyreflectstheUgandans’perceptionthatbothtypesaredesper-

atelyneeded.Wethuspooledtheproject-typeconditions.See theappendix(p.1) for the

specificlanguageusedintheelectricityproject.

The fundingorganizations randomlyassigned for themasspublicwere theWorld

Bank,theAfricanDevelopmentBank,theGovernmentoftheUnitedStates,theGovernment

ofChina,agenericmultilateralinstitution(“aninternationalorganizationfundedbymany

countries”), agenericbilateral agency (“a single foreigncountry”), andNoDonor.Due to

samplesizeconstraints,fortheMPswerandomlyassignedonlytheWorldBankandUSAID

aswellas thegeneric conditions.Wereporta randomizationcheckanalysis inAppendix

TableA7,whichdemonstratesthatrandomassignmenteffectivelyproducedbalanceacross

anumberof factorsforwhichwehavedata.Results forthe“NoDonor”conditionarere-

portedelsewherebecausetheyarenotrelevantforthecomparisonsmadebetweenmulti-

lateralandbilateralaidhere(Milneretal.2016;Findleyetal.Forthcoming).Belowwefirst

reporttheresultsforthemassespoolingthebilateraldonors(U.S.Government,China,ge-

nericbilateral)andthemultilateraldonors(WorldBank,AfricanDevelopmentBank,and

26

genericmultilateral).ForMPsthenwepoolthebilateraldonors(USAIDandgenericbilat-

eral)andmultilateraldonors(WorldBankandgenericmultilateral).

Our studyemploysabetween-subjectsdesign, so subjectsarenot comparingpro-

jectsdirectly.Subjectsonlyseeonecondition,whichenablesustolookformeaningfuldif-

ferencesinlevelsofsupportbetweenidenticalprojectsthatarerandomlyassignedasorig-

inating from different foreign donors. With random assignment of the treatment, the

characteristics of individuals and their prior experiences or beliefs shouldnot affect our

results.

After the aid project prompt, enumerators inquired about several attitudinal out-

comesandinvitedtherespondentstosupporttheprojectbysigningapetitionandsending

an SMS message.8 Citizens could endorse or oppose the projects verbally. Enumerators

theninvitedrespondentstosignapaperpetitionandsendanSMStextmessageinsupport.

Onceverbalintentionswererecorded,enumeratorspresentedthemanactualpetitionand

recordedwhetherornot theysigned.Enumeratorsalsogaverespondentsaslipofpaper

withtheSMSnumberandaskedthemtosendatextlaterthatday.SMStextscostUgandans

between50and130USh,sothetextrepresentedanactualcosttothecitizensthattheydid

notexpecttorecover.9Giventheaveragesubject’slowdailyincomeof2,935USh(1.08U.S.

dollars),forthevastmajorityofsubjectsthecostlikelyappearedmeaningful.

MPswereaskedslightlydifferentquestions.Wequeriedthemontheirwillingness

tocoordinatewiththeirpeersinsupportof(orinoppositionto)theproject,tellconstitu-

8Manipulationchecksforthemassesshowthatsubjectsrecalledthetypeofprojectandthetypeofdonorinmostcases(89%forprojectand63%fordonor).Themanipulationcheckwasaskedmuchlaterthanthema-nipulationitself,whichmayexplainthedropoff.Table3reportsthetworefinements.First,weestimatedtheresultswhendroppingsubjectsthatdidnotpassthemanipulationcheck.Second,weestimatedcomplierav-eragecausaleffectsusingassignmenttotreatmentasaninstrumenttopredictcompliance(passingthema-nipulationcheck),whichinturnpredictslevelsofsupport.9Subjectsexpectedthattheywouldpaythecost.Afterwards,however,wereimbursedthem.

27

entsabouttheproject,rallylocalsinsupportof(orinoppositionto)theproject,andsigna

lettertothePresidentinsupportoforoppositiontotheprojects.BecausetheMPswere

presentedwithbothprojects,wehavetwoobservationsforeachonalloftheseoutcomes,

exceptthepetitiontothepresident.EachMPwasaskedtosignasinglepetitionthatre-

portedtheirlevelofsupportforbothprojectstothePresident,thuswehaveoneobserva-

tionforeachMPonthisoutcome.Thisdesignchoicewasmadetoreducetheburdenonthe

MPsandtolessenredundancyofsendingtwonearlyidenticalletterstothepresident.Be-

causetheMPsreceivedthesamedonoracrossthetwoprojectsthisshouldnotaffectthe

resultsbecausewearecomparingdifferencesindonorsandnotsectors(giventherewas

nomeaningfuldifferencebetweenprojecttypes).Thesevariousmeasuresofsupportpre-

senttherespondentswithvaryinglevelsofcost(attitudinalvs.behavioralresponses)and

areusedasthekeyoutcomevariablestogaugesupportforprojectsacrosstreatmentarms.

ThesummarystatisticsforeachofthedependentvariablesappearsinTable1.We

alsoutilizeanaggregated formof thedependentvariables,calledoverallsupport foraid.

Wesummed thedifferentdependentvariables intoan index,Aid support. Wewanted to

measureoveralllevelsofsupportoroppositiontoaidbydonor.Andwecraftedthesurvey

instrument so that each dependent variable represented a further step in a chain of in-

creasinglycostlyactionsshowingsupportoropposition.Sinceallofthedependentvaria-

blesaredichotomous,AidSupportsimplysumsthem.Missingvaluesaretreatedas0,but

respondentswhodidnotrespondtoanyoftheDVsaredropped.Theminimumvaluefor

thisvariableis0andthemaximumis6.

[TABLE1ABOUTHERE]

28

Toachievegreatergeneralizability,weusedtwodifferentprojecttypesandsixdif-

ferentforeigndonortypes.Thefourspecificdonorsnamed–theWorldBank,theAfrican

DevelopmentBank, theUnitedStates, andChina–are themostactive inUgandaandac-

counted for54%of totalaiddisbursements. Wealsochose theelectricityandeducation

projectsbecausetheyrepresentthetypesofprojectsthatcanbegivenselectivelytocon-

stituenciesthatsupportpoliticians.Forthemasssurvey,werandomlyassignedthedonor

and theproject type.Neitherproject type in themass surveywas significantlypreferred

overtheotherinthebetween-subjectsdesign,whichmayreflectthefactthatbothtypesof

projectsaredesperatelysoughtafter inUganda.Becausetherewerenosignificantdiffer-

encesbetweenprojecttypes,wefocusonlyonthedifferenceamongaiddonorsacrossboth

projecttypespooledtogether.

Twoadditionalfeaturesofourapproachincludedasimilarexperimentandsurvey

onmorethantwo-thirdsofmembersoftheUgandan9thparliament.Anindividual’sstatus

in society— elite versusmass public— could differentially determine preferences over

aid.Eliteswillgenerallypossessmorepoliticalknowledge,greatercontroloveraidpackag-

es,andmoreincentivestoactinafashionthatadvancestheirpoliticalinterests(i.e.,stay-

ing or getting into political office). Thesedistinctions from the general publicwill often

givethemdifferentpreferencesoveraidthantheaveragerecipientinadevelopingcountry.

Our hypotheses above address a number of themost importantways inwhich political

elitesandcitizensmaydifferintheirpreferencesoveraiddonors.Wecarriedoutanexper-

imentonaconveniencesampleof339MPs.10Weattemptedtoconductacensusofallcur-

10Wesurveyed354MPstotal.Butsomereceivedanothercondition.Andofthetotal,276wereofthe375Mem-bersofthe9thUgandanParliament(thesittinglegislature)and78wereformerMPsfromthe8thParliament.Forour339MPs,264werecurrentmembersand75wereformer.

29

rentMPsandachieveda72percentresponserate.Fortheelites,wewerenotabletoran-

domlysample.Wethuscomparedourconveniencesampletoasetofcharacteristicsforthe

9thParliamentandshowthoseresultsinTable2below.OursamplereflectstheactualPar-

liamentreasonablywell.Thesurveysandexperimentsonthemassesandelitesweresimi-

lar,butnotidentical,andwereperformedbetweenJuneandOctober2012bylocalUgan-

danenumerators.

WechosetoconductourexperimentonMPsasopposedtoothergovernmentoffi-

cialsforanumberofreasons.First,parliamentiswherethebudgetandtheacquisitionof

aid(bothbudgetsupportandprojectaid)isdiscussedanddecided.Second,afterconduct-

inginterviewswithMPsandlocalcouncilors(LC-VandLC-III,whichareroughlyequivalent

togovernorsandmayors),itbecameclearthatlocalofficialshadlittletonodirectman-

agementofproject-levelaidfunds.MPs,however,veryclearlyhadexperiencewithaidboth

inparliamentarydebatesandinmanagingaidfunds(53%ofourMPintervieweessaidthey

hadpersonallymanagedaidfunds).Moreover,MPsvaluesuchprojectsintheirdistricts;a

majorityofthemininterviewssaidthattheyreceivedpraiseandappreciationfromcitizens

forsuchprojects.Third,Uganda’sparliamentarysystemmergestheexecutiveandlegisla-

tivebranches,andthusweareabletoalsosurveycabinetministerswhoplayanimportant

roleindecisionmaking.Infact,theexperimentincludes49governmentministers(thisin-

cludesdeputyministers),22shadowcabinetmembers(theopposition’scabinet),andboth

governmentandoppositionchiefwhips.

[TABLE2ABOUTHERE]

Inaddition,weaskedrespondents—massesandelites—fortheirreasonsforsup-

portingoropposingtheseprojects.Wewereinterestedinthecausalmechanismslinking

30

theirpreferencestotheoutcomes.Wefocusedonfivedifferentcausalreasons:politiciza-

tion,conditionality,transparency,efficiency,andcommunityneeds.Toprobethis,weuse

thelanguagedescribedbelowwhichcorrespondstothemechanismsidentifiedabovein

thetheorysection.

Forpoliticization,wemeanthatthedonoror implementinggovernmentwill favor

certaingroups,regionsorprojectsoverothersforpoliticalconsiderations.Needsrelatedto

development and poverty reductionwill be less important in driving aid allocation than

calculationsaboutmaintainingpoliticalsupport.Toexplorethis,weaskedtwoquestions,

oneeachforbilateralormultilateraldonors:“whichofthefollowingstatementsisclosest

toyourview?ChooseStatement1orStatement2. Statement1: Aid from[an individual

foreign government or an international organization]most helps the neediest people in

your country. Statement 2: Aid from [an individual foreign government or international

organization]mosthelpsthefriendsoralliesofthecountry.”

Theconditionsthatdonorsorgovernmentsattachtoprojectsmaymatterforhow

recipientsreact to them.Againweaskedtwoquestionscorrespondingtodifferentdonor

types: “[Foreign governments or International organizations] often ask poor recipient

countriestochangesomeoftheirpoliciesinexchangeforforeignaid.Whichofthefollow-

ingstatements isclosest toyourview?ChooseStatement1orStatement2. Statement1:

Theconditionsorrequirementssetby[aforeigngovernmentoraninternationalorganiza-

tion]inorderforUgandatoreceiveaidareunfairandhurtUganda.Statement2:Thecondi-

tionsorrequirementssetby[aforeigngovernmentoraninternationalorganization]inor-

derforUgandatoreceiveaidhelpUgandatoreformandbecomeabettercountry.”

31

Transparency implies that recipients canmonitor and follow the progress that a

projectistakingandcanseewherethefundsarebeingapplied.Weask:“Whichofthefol-

lowingstatementsisclosesttoyourview?ChooseStatement1orStatement2.Statement

1:Itiseasiertoseewhereaidfromaninternationalorganizationisspentandtomonitor

howitisused.Statement2:Itiseasiertoseewhereaidfromanindividualforeigngovern-

mentisspentandtomonitorhowitisused.”

Fourth,theefficacy,efficiencyorsuccessoftheaidmaymattermost.Recipientsmay

caremostabouttheeffectsofaidprojects.Weask:“Whichofthefollowingstatementsis

closesttoyourview?ChooseStatement1orStatement2.Statement1:Aidfromaninter-

nationalorganizationhasthemostimpactandtheleastwasteinachievingitsgoals.State-

ment2:Aidfromanindividualforeigngovernmenthasthemostimpactandtheleastwaste

inachievingitsgoals.”

Fifth, the alignment of the preferences of donors and recipientsmay be closer in

sometypesofprojectsthanothers.Weask:“Whichofthefollowingstatementsisclosestto

yourview?ChooseStatement1orStatement2. Statement1:Projects fundedbyinterna-

tionalorganizationsmostoftenmatchtheneedsofmycommunity. Statement2:Projects

funded by individual foreign countries’ governmentsmost oftenmatch the needs ofmy

community.”

Results

First,itisimportanttonotethatdonorsdoseemtochannelaiddifferently;howev-

er,theymayprovideaidformanyofthesametypesofprojects,astheresearchondonor

fragmentationnotes.Forinstance,USAIDprimarilycontractswithU.S.-basedcompanies,

32

NGOs,andtheirpartners.11China,incontrast,mostlyprovidestiedaidinthattheysend

theirowncontractorsespeciallyintheimplementationoflargeinfrastructureprograms

(DreherandFuchs2015).Incontrast,theWorldBankandAfricanDevelopmentBankpro-

videnearlyalloftheiraidfundsdirectlytothegovernment,whichthenimplementsthe

projectswhileabidingbystrictprocurementandaccountingcriteria.12Otherbilateralaid

agenciesaresomewhereinthemiddle,givingbothdirectgovernmentaidandcontracting

todotheirownprojects.Second,in(non-experimental)surveyresponses,citizensappear

topickupondifferencesacrossdonors.Inthesurveywhenaskedwhichaidtypehadthe

mostimpactandtheleastwasteinachievingitsgoals,61percentofrespondentsbelieved

multilateralsdidbettercomparedto34percentinfavorofbilaterals.Whenaskedwhich

typemostoftenmatchedtheneedsoftheircommunity,nearlytwiceasmany(59to34

percent)saidthatthemultilateralsdidbetter.Third,recipientsaregenerallyawareofdo-

nors.Despitethefactsthattheaverageeducationlevelofoursamplewas7years,that

mostofourrespondentswereverypoor,andthatmorethan60percentwereunemployed,

manyhadheardofthemainaidagencies(accordingtonon-experimentalsurveyrespons-

es).Themostwell-knowndonor(asadonor)wastheUnitedStatesat86%,followedby

Chinaat75%,theWorldBankat68%,andtheAFDBat35%.13Finally,whengivenachoice

betweenforeignaiddonorsandtheirowngovernmenttoundertakeprojects,thepublicat

leastexpressedsignificantlymoresupportfortheforeignaiddonors,whiletheMPsper-

11ForUganda,theOECDCreditorReportingSystemshowsthatitreceivedonly$0.2min2010forbudgetsupportfromtheUSAID,whichwas0.05%oftotalU.S.ODAreceived.12In2010,accordingtotheOECDCreditorReportingSystem,theWorldBankgave$100.9minbudgetsupporttoUganda.Ugandareceived30.7%ofIDAfundsasbudgetsupport,whileotherdevelopingcountriesreceivedonly21.1%.Budgetsupportisnottheonlygroupoffundsthatgoesdirectlytothegovernment,butitistheeasiesttocount.13OursurveyalsoshowedthatUgandansknewalotaboutpolitics.Over80%correctlyidentifiedtheirMP,andalmost70%correctlyidentifiedtheirwomanMPaswell.

33

ceivedtheirowngovernmenttobepreferable((Milneretal.2016;Findleyetal.2016).

ThusthepublicandMPsseemtobeabletoformviewsaboutdifferenttypesofprojectfun-

ders.

Turning to experimental results, therewere no significant differences in levels of

supportor inbehavioralmeasuresof supportacrossexperimental conditions formasses

andveryfewfortheMPs.(SeeTable3.)Inourexperiment,thesubjectsdidnotdifferenti-

ateamongdonors.Supportfortheelectricityandeducationprojectswasveryhigh,around

80%andoftenrunningto90%.Thesehighlevelsofsupportforaidsuggestthatceilingef-

fectsaregoingtomakefindingtreatmenteffectsdifficult.MPspreferredthebilateralpro-

jectsinmostinstancesbuttheresultsweresignificant(at0.1level)inonlytwocases,and

thereforeshouldnotbegivenmuchweight.Indeed,thebroaderlessonfromthemainanal-

ysisisthatmassesandMPsdonotprefermultilateralorbilateralaidovertheother,which

doesnotsupportourfirsthypothesis.Itisworthconsideringsubgroupanalysestoconsid-

ertowhatextentthisnullresultholds.Wenoteherethatwhilesomesubgroupeffectsap-

pear,theyarefewinnumberandwithmultipletestingadjustmentswouldnotamounttoa

definitiveconclusionthatmultilateralorbilateralaidissignificantlyandrobustlypreferred

bymassesorMPs.Weturntotheseresultsnow.

[TABLE3ABOUTHERE]

Notallsubjectsweresufficientlyawareofthedonors,whichsuggestsconfiningthe

analysistothosewhoknowthedonors.Whenweconsideredonlythesubjectswhowere

familiarwithspecificbilateralormultilateraldonors,wesawsignificanttreatmenteffects

intwoofsixcases,andalsointhecombinedindex,buthastentoaddthatweshouldnot

infertoomuch,especiallysincetheresultsholdinonlytwoofsixcasesandthepossibility

34

of selection bias lingers.We also treated subject familiarity as a compliance problem in

whichtherewassomedrop-offincomplianceassubjectsdidnotunderstandthemanipula-

tionbynotbeing familiarwith thedonor.Toaccurately computeeffects in this case,we

conductedacomplieraveragecausaleffect(CACE)(GerberandGreen2012)analysisusing

Two-Stage Least Squares regression and show substantively similar results. (See discus-

sionofAppendixTableA2.)

Beyond the broadmultilateral vs. bilateral comparison, the public inUgandaper-

ceivedsignificantdifferencesbetweentheU.S.andtheChinesebilateralprograms in two

cases,butnotinthecombinedindex.AndnodifferencesbetweentheWorldBankandthe

AFDB emerged. For the U.S. and China, individuals were significantly more willing (p =

0.014)totell their local leadersoftheirsupport forU.S.projects(0.96)comparedtoChi-

neseones(0.91).AndcitizenswithcellphonesalsosenttheSMSsignificantlymoreoften(p

=0.021) intheUScondition(8.8%)thanintheChinacondition(2.9%).U.S.-Chinadiffer-

encesarereportedinTable4.Whilenotrobust,thesesubgroupeffectsareconsistentwith

otherresearchonChineseaidsuggestingitisdifferentthanUSaid,beingmuchmorefun-

gible,lessconditional,lesstransparent,anddirectedtowardthegovernment(Dreheretal.

2015;Bräutigam2009,2011).

[TABLE4ABOUTHERE]

Acrossthemanypossiblecausalmechanismsweexplored,thereisagainverylim-

itedevidenceaboutsomeofthemechanisms,asshowninTables5and6.(AlsoseeAppen-

dixTablesA3-A6.)Thisevidenceisclearerinthecaseofthemasses(Table5)whenthey

knowofthedonorthanwiththeelitesinpartbecausewehavemorepowertodetectthese

subgroupseffectswiththe largermasssample.Afterreceivingthetreatment information

35

aboutwho carriedout the aidproject, the respondentswere asked a series of questions

aboutthereasonsforpreferringonetypeofaidoveranother.Thesequestionswereasked

post-treatment,butcouldnotfeasiblybeaskedpriortothetreatmentattheriskofbiasing

responses.Becausethesequestionswereaskedpost-treatment, theappropriatecompari-

sonis, forexamplelookingatmultilateralaidtransparency,betweensubjectsassignedto

the bilateral condition who nonetheless said that multilateral aid is more transparent

against thoseassignedto themultilateralconditionwhothensaid thatmultilateralaid is

moretransparent.Thesamelogiccouldbeappliedtoeachofthecomparisons.

Lookingat theoverallaidsupport indexvariable,wefindevidencethat thepublic

seems to prefer multilateral aid donors depending on their views about conditionality,

transparency, and efficacy.14 For themasses, politicization reflecting the influence of the

foreigndonorsand their friends in therecipientcountryshows littleeffect.Whenrecipi-

entsbelievethatconditionalityhurtsthecountry,theyaremoresupportiveoftheaidpro-

jectwhentolditisbyamultilateraldonor.Wesuspectthisisbecausetheybelievethereis

lessconditionalitywithmultilateralthanbilateralaid.Amongrecipientswhobelievemulti-

lateralaidismoretransparentthanbilateralaid,theyalsoaremoresupportiveoftheaid

projectwhen it comes frommultilateral donors,which is consistentwith hypothesis 4a.

Amongthosewhobelievemultilateralaidismoreeffectivethanbilateralaid,theytooare

moresupportiveoftheprojectsinthemultilateralcondition.Andamongthosewhobelieve

multilateraldonorsaremoreefficaciousthanbilateralsources,theyalsoaremoresupport-

14Theresultsforthetransparency,conditionality,andefficacymechanismsseemtobecapturingasimilarphe-nomenongiventhatthesupportindexresultsarenearlyidenticalacrossthethree.

36

iveofaidprojectswhentoldtheprojectisbyamultilateraldonor,consistentwithhypoth-

esis5a.15

[TABLES5and6ABOUTHERE]

In sum, themain results reveal little evidence that citizensorpolitical eliteshave

strong preferences formultilateral versus bilateral assistance. Subgroup analysis reveals

isolatedevidencethatcitizensprefermultilateralaidandelitespreferbilateralaid;howev-

er,wehasten toadd that in lightof the setofpossibleanalyses, these results are few in

numberandwouldbeevenweakerifexplicitlyincludingmultipletestingadjustments.16

Discussion

The lackof treatment effects could arise fromvarious factors. First and foremost,

theabsenceoftreatmenteffectsmayrevealalackofpreferenceformultilateralrelativeto

bilateraldonorsamongbothmassesandelites. As the researchondonor fragmentation

notes,recipientsreceiveaidfrommanysourceswhooftenfundtheidenticaltypesofpro-

jectsinthesameareas.Donorsmaydeliveraidindistinctwaysbutiftheyoverlapheavily

inwhattheydothentheymayappearindistinguishable.Second,citizensandelitesmaynot

knowenoughaboutwhichdonorsaredoingwhichprojects,orhowdifferentdonorsoper-

ate.Itisnotable,however,thatinthesamesurveycitizensandMPsdidhavesignificantly

different preferences for foreign donors as a group relative to their own government,

15Intwoofthecases(transparencyandefficacy),respondentspreferredmultilateralaidtobilateralaidevenwhentheythoughtthatbilateralaidwasmoretransparentandefficacious.Thusthepreferenceformultilateralaidisstrongacrosstheboard,thoughstatisticallydifferentonlyamongthosewhoperceivemultilateralaidasmoretransparentandefficacious.16Usingaglobalsampleofelitesandadifferentsetofquestions,Parksetal.(2016)findevidenceindicatingthatelitesratemultilateraldonorshigherthanbilateraldonors.

37

whichtheyseemtohaveenoughknowledgeabouttomakesuchdistinctions((Milneretal.

2016;Findleyetal.2016).

Third,weaskedaboutpublicgoodsprojectsthatthepublicandMPsinUgandades-

peratelywantandsotheirsupportfortheseprojectsisveryhighatbaseline.Theseceiling

effectsmean thatusing the treatment tomovepeople toevenhigher levelsof support is

likely to be very difficult. The peoplewho do notwant these projects nomatterwho is

fundingthemarelikelytobeatinyminoritywhoobjectstosuchpublicgoods.Somelitera-

turesuggeststhereshouldbedifferencesbetweenmultilateralsandbilaterals,evenifthe

literatureisdividedonwhichshouldbemorepreferablebasedonlevelsofpoliticization,

conditionality,andothercharacteristics.If infactrecipientsdonotpreferoneforeigndo-

nortypeovertheother,thenthesenullresultsareinformativefortheliteraturethatcon-

tinuestodrawdistinctionsbetweenmultilateralandbilateraldonors.Moreover,theyseem

tolendsupporttotheliteratureondonorfragmentationthatnotestheheavyoverlapand

lackofcomplementarityamongforeigndonors.Thisliteraturesuggeststhatbecausethey

tendtodothesametypeofprojectsinthesameareas,recipientsshouldnotperceivedif-

ferencesamongthedonors.

[TABLES7AND8ABOUTHERE]

What limited evidence for distinctions there is indicates that if the public prefers

multilateraldonors,consistentwithhypotheses4aand5a,suchsupportrelatestomultilat-

eraldonors’ lackofconditionality,greatertransparencyandefficacyaspublicgoodspro-

viders; see tables 5 and 6. Probing recipient’s trust for the different donors also reveals

strongerpreferencesformultilateralaid.Whenaggregatingtrustacrossmultilateralsand

bilaterals,citizenshavemuchhigherlevelsoftrustformultilateralorganizations(mean=

38

3.27forMLvs.2.80forBL,p=0.00),asshownintable7.Intables7and8wecanseethat

foreachdonormultilateralonesaremoretrustedthanbilateralonesbythemasspublic.

AndthisfitswithourfindingabovesincetheWorldBankisthemosttrustedamongalldo-

norsandChinaistheleasttrusted.Multilateraldonorsarealsoseenasmoretransparent

bybothelitesandmassesintable8.Weaskedrespondentstochoosebetweentwostate-

ments: “Statement1: It is easier to seewhereaid from [an international organization] is

spentandtomonitorhowitisused.OrStatement2:Itiseasiertoseewhereaidfrom[an

individualforeigngovernment]isspentandtomonitorhowitisused.”Thefirstwassignif-

icantlymorelikelyforinternationalorganizationsthanforindividualcountrygovernments

asshownintable8,thusindicatingthatcitizensandMPsperceivemuchgreatertranspar-

encyinmultilateralsthanbilaterals.Whilenotprimaryexperimentalevidence,thesurvey

results informthehypothesizedmechanisms tosomeextent.Ugandancitizensseemulti-

lateraldonorsasmoretransparent,moreeffectiveandmoretrustworthy,butnotlesspo-

liticizednormorelikelytoimposeconditionalityforcefully.

Conclusions

Weexploreddifferencesinattitudinalandbehavioralsupportofrecipientcitizens

towarddifferenttypesofdonorsofaidprojects.Theliteraturedrawsdistinctionsbetween

multilateralsandbilateralsintermsofanumberoffactors.Theconventionalwisdomindi-

catesthatmultilateralaidismoreeffectivethanbilateralaidbecauseitislesspoliticized,

moreabletoenforceconditionality,andlesspronetocorruption,thoughtherearemany

criticsofthisviewwhocounterthatmultilateralaidisalsopoliticized.Ourexperiment

providesevidencecontrarytoargumentspositingdifferencesbetweenmultilateralandbi-

39

lateraldonors.Themainresultsofouranalysisaregenerallynull,andthesubgroupanal-

ysesthatemergearesuggestivebutlargelyisolatedtospecificconditions.Thesenullre-

sultsseemmorepersuasivesincewedofindsignificantdifferencesinpreferencesforciti-

zensandMPswhenfacedwithachoicebetweentheirowngovernmentandalltypesof

foreigndonors.Furthermore,thislackofdifferencesamongforeigndonorsareconsistent

withthelargeliteraturethatseesdonorsasoverlappingheavilyintheiraidprovision,fail-

ingtocoordinateandspecialize,andgenerallygivingaidforthesametypesofprojectsin

thesameareasinthesamecountries.Thisviewofforeigndonorsasfailingtocomplement

oneanotherwouldleadtoperceptionsbyrecipientsthatallforeigndonorsarealike.

Whatlimitedevidencedoesemergefromtheexperimentandcompanionsurvey

suggeststhatcitizensseemultilateralaiddonorsasbeingmoretransparent,moretrust-

worthy,andmoreeffective.Politicalelitesseemtohaveaslightpreferenceforbilateraldo-

nors.Theseanalysesdonot,however,constitutestrongevidencewhenconsideringtheset

ofpossiblecomparisonsandthemultipletestingadjustmentsneededtomakedefinitive

conclusions.

Inamoregeneralsense,thepresentstudysuggeststhatcitizensandelitespossess

informationthatmayproverelevantinanalysesofaid.Itisoddthatsomuchisstudiedand

writtenaboutaidbutthatsolittleofthatworkactuallyasksindividualsinrecipientcoun-

triesfortheirviews.Thisprojectseekstoamelioratesomeofthatoversightandservesto

encourageotherstudiesinthefuturethattakeseriouslytheopinionsandbehaviorofthe

peopleonthegroundmostaffectedbytheaid.Inthepresentanalyses,welearnfromrecip-

ientsthatthedistinctionsbetweenmultilateraldonorsandbilateralonesidentifiedinmac-

ro-levelanalysesmaynotbefeltorconsideredimportantwithinrecipientcountries.Ifthis

40

isbecausetheforeigndonorsareallgenerallyseenasdoingthesamething,thenthisraises

questionsaboutwhywehavebothbilateralandmultilateralaidagenciesandwhytheyare

unabletobettercoordinateandspecialize.Numerousstudiespointoutthatdonorandaid

projectproliferationarecostly.Ourresultssuggestthatpolicymakersindonorcountries

shouldworktocoordinateanddifferentiatetheireffortsmore.Multilateralagencies

shouldalsofindabetterdivisionoflaborwiththebilateralones.Suchdifferentiationcould

improveaidgiving.

41

References

Acharya,A.,DeLima,A.T.F.,&Moore,M.(2006).Proliferationandfragmentation:Transactionscostsandthevalueofaid.TheJournalofDevelopmentStudies,42(1),1-21.

Aldasoro,I.,Nunnenkamp,P.,&Thiele,R.(2010).Lessaidproliferationandmoredonorcoordination?Thewidegapbetweenwordsanddeeds.JournalofInternationalDevelopment,22(7),920-940.

Alesina,A.,&Weder,B.(2002).DoCorruptGovernmentsReceiveLessForeignAid?TheAmericanEconomicReview,92(4),1126-1137.

Alvi,E.,&Senbeta,A.(2012).DoesForeignAidReducePoverty?JournalofInternationalDevelopment,24(8),955-976,doi:10.1002/jid.1790.

Annen,K.,&Moers,L.(2016).Donorcompetitionforaidimpact,andaidfragmentation.TheWorldBankEconomicReview,lhw019.

Autesserre,S.v.(2010).ThetroublewiththeCongo:localviolenceandthefailureofinternationalpeacebuilding(Cambridgestudiesininternationalrelations,Vol.115).Cambridge;NewYork:CambridgeUniversityPress.

Bates,R.H.(1981).MarketsandStatesinTropicalAfrica.Berkeley,CA:UniversityofCaliforniaPress.

Bermeo,S.B.(2010).DevelopmentandStrategy:AidAllocationinanInterdependentWorld.Durham,NC:DukeUniversity.

Bermeo,S.B.(2011).ForeignAidandRegimeChange:ARoleforDonorIntent.WorldDevelopment,39(11),2021-2031,doi:10.1016/j.worlddev.2011.07.019.

Bermeo,S.B.(2016).AidIsNotOil:DonorUtility,HeterogeneousAid,andtheAid-DemocratizationRelationship.InternationalOrganization,70(1),1-32.

Bigsten,A.,&Tengstam,S.(2015).InternationalCoordinationandtheEffectivenessofAid.WorldDevelopment,69,75-85,doi:http://dx.doi.org/10.1016/j.worlddev.2013.12.021.

Bourguignon,F.,&Platteau,J.-P.(2015).TheHardChallengeofAidCoordination.WorldDevelopment,69,86-97,doi:http://dx.doi.org/10.1016/j.worlddev.2013.12.011.

Bräutigam,D.(2009).Thedragon'sgift:therealstoryofChinainAfrica.Oxford;NewYork:OxfordUniversityPress.

Bräutigam,D.(2011).Aid‘WithChineseCharacteristics’:ChineseForeignAidandDevelopmentFinanceMeettheOECD-DACAidRegime.JournalofInternationalDevelopment,23(5),752-764,doi:10.1002/jid.1798.

BuenodeMesquita,B.,&Smith,A.(2007).ForeignAidandPolicyConcessions.JournalofConflictResolution,51(2),251-284,doi:10.1177/0022002706297696.

BuenodeMesquita,B.,&Smith,A.(2009).APoliticalEconomyofAid.InternationalOrganization,63(2),309-340,doi:doi:10.1017/S0020818309090109.

Burnside,C.,&Dollar,D.(2000).Aid,PoliciesandGrowth.AmericanEconomicReview,90(4),847-868.

42

Christensen,Z.,Homer,D.,&Nielson,D.L.(2011).DodgingAdverseSelection:HowDonorTypeandGovernanceConditionAid’sEffectsonSchoolEnrollment.WorldDevelopment,39(11),2044-2053,doi:10.1016/j.worlddev.2011.07.018.

Dietrich,S.(2016).DonorPoliticalEconomiesandthePursuitofAidEffectiveness.InternationalOrganization,70(1),65-102.

Djankov,S.,Montalvo,J.G.,&Reynal-Querol,M.(2009).Aidwithmultiplepersonalities.JournalofComparativeEconomics,37(2),217-229.

Dreher,A.,&Fuchs,A.(2015).Rogueaid?AnempiricalanalysisofChina'saidallocation.CanadianJournalofEconomics/Revuecanadienned'économique,48(3),988-1023.

Dreher,A.,Fuchs,A.,Hodler,R.,Parks,B.,Raschky,P.A.,&Tierney,M.J.(2015).AidonDemand:AfricanLeadersandtheGeographyofChina'sForeignAssistance.CESifoWorkingPaperNo.5439.Munich:CenterforEconomicStudies,IfoInstitute.

Dreher,A.,Fuchs,A.,Hodler,R.,Parks,B.,Raschky,P.A.,&Tierney,M.J.(2016).Aidondemand:AfricanleadersandthegeographyofChina'sforeignassistance.AidDataWorkingPaper#3Revised.Williamsburg,VA:AidData.

Dreher,A.,&Sturm,J.-E.(2012).DotheIMFandtheWorldBankinfluencevotingintheUNGeneralAssembly?PublicChoice,151(1-2),363-397.

Dreher,A.,Sturm,J.-E.,&Vreeland,J.R.(2009).DevelopmentAidandInternationalPolitics:DoesMembershipontheUNSecurityCouncilInfluenceWorldBankDecisions?JournalofDevelopmentEconomics,88(1),1-18.

Easterly,W.(2007).Areaidagenciesimproving?EconomicPolicy,22(52),633-678,doi:10.1111/j.1468-0327.2007.00187.x.

Easterly,W.,&Pfutze,T.(2008).WhereDoestheMoneyGo?BestandWorstPracticesinForeignAid.JournalofEconomicPerspectives,22(2),29-52.

Easterly,W.,&Williamson,C.R.(2011).RhetoricversusReality:TheBestandWorstofAidAgencyPractices.WorldDevelopment,39(11),1930-1949,doi:10.1016/j.worlddev.2011.07.027.

Findley,M.G.,Harris,A.,Milner,H.,&Nielson,D.L.(2016).EliteandMassSupportforForeignAidVersusGovernmentPrograms:ExperimentalEvidencefromUganda.InternationalOrganization,Forthcoming.

Frey,B.S.,&Schneider,F.(1986).CompetingModelsofInternationalLendingActivity.JournalofDevelopmentEconomics,20(2),225-245,doi:10.1016/0304-3878(86)90022-2.

Frot,E.,&Santiso,J.(2009).CrushedAid:FragmentationinSectoralAid.SITEWorkingPaperNo.6,2009

Fuchs,A.,Nunnenkamp,P.,&Öhler,H.(2015).Whydonorsofforeignaiddonotcoordinate:Theroleofcompetitionforexportmarketsandpoliticalsupport.TheWorldEconomy,38(2),255-285.

Gartzke,E.,&Naoi,M.(2011).MultilateralismandDemocracy:ADissentRegardingKeohane,Macedo,andMoravcsik.InternationalOrganization,65(03),589-598.

Girod,D.M.(2008).CutfromtheSameCloth?Multilateralvs.BilateralAid.PaperpresentedattheAnnualConferenceofInternationalPoliticalEconomySociety,Philadelphia,November2008

Green,E.D.(2010).Patronage,DistrictCreation,andReforminUganda.StudiesinComparativeInternationalDevelopment(SCID),45(1),83-103,doi:10.1007/s12116-009-9058-8.

43

Headey,D.(2008).Geopoliticsandtheeffectofforeignaidoneconomicgrowth:1970–2001.JournalofInternationalDevelopment,20(2),161-180,doi:10.1002/jid.1395.

Isaksson,A.-S.,&Kotsadam,A.(2016).Chineseaidandlocalcorruption.AidDataWorkingPaper#33.Williamsburg,VA:AidData.

Jablonski,R.S.(2014).HowAidTargetsVotes:TheImpactofElectoralIncentivesonForeignAidDistribution.WorldPolitics,66(2),293-330.

Kilby,C.(2006).Donorinfluenceinmultilateraldevelopmentbanks:ThecaseoftheAsianDevelopmentBank.TheReviewofInternationalOrganizations,1(2),173-195,doi:10.1007/s11558-006-8343-9.

Kilby,C.(2013).Thepoliticaleconomyofprojectpreparation:AnempiricalanalysisofWorldBankprojects.JournalofDevelopmentEconomics,105(0),211-225,doi:http://dx.doi.org/10.1016/j.jdeveco.2013.07.011.

Kilby,C.,&Dreher,A.(2010).Theimpactofaidongrowthrevisited:Dodonormotivesmatter?EconomicsLetters,107(3),338-340,doi:10.1016/j.econlet.2010.02.015.

Kiwanuka,M.(2012).PrioritiesforRenewedEconomicGrowth&Development.BudgetAddressdeliveredbytheMinisterofFinance,Planning&EconomicDevelopmentDeliveredattheMeetingofthe2ndSessionofthe9thParliamentofUganda.

Kizhakethalackal,E.T.,Mukherjee,D.,&Alvi,E.(2013).Quantileregressionanalysisofhealth-aidandinfantmortality:anote.AppliedEconomicsLetters,20(13),1197-1201,doi:10.1080/13504851.2013.799744.

Knack,S.,&Rahman,A.(2007).Donorfragmentationandbureaucraticqualityinaidrecipients.JournalofDevelopmentEconomics,83(1),176-197.

Knack,S.,&Smets,L.(2013).AidTyingandDonorFragmentation.WorldDevelopment,44(0),63-76,doi:http://dx.doi.org/10.1016/j.worlddev.2012.09.006.

Lyne,M.M.,Nielson,D.L.,&Tierney,M.J.(2006).WhoDelegates?AlternativeModelsofPrinciplesinDevelopmentAid.InD.G.Hawkins,D.A.Lake,D.L.Nielson,&M.J.Tierney(Eds.),DelegationandAgencyinInternationalOrganizations(pp.41-76).NewYork:CambridgeUniversityPress.

Lyne,M.M.,Nielson,D.L.,&Tierney,M.J.(2009).Controllingcoalitions:Sociallendingatthemultilateraldevelopmentbanks.TheReviewofInternationalOrganizations,4(4),407-433.

Maizels,A.,&Nissanke,M.K.(1984).MotivationsforAidtoDevelopingCountries.WorldDevelopment,12(9),879-900,doi:10.1016/0305-750x(84)90046-9.

Martens,B.,Mummert,U.,Murrell,P.,Seabright,P.,&Ostrom,E.(2002).TheInstitutionalEconomicsofForeignAid:CambridgeUniversityPress.

Milner,H.V.(2006).WhyMultilateralism?Foreignaidanddomesticprincipal-agentproblems.InD.G.Hawkins,D.A.Lake,D.L.Nielson,&M.J.Tierney(Eds.),DelegationandAgencyinInternationalOrganizations(pp.107-139).NewYork:CambridgeUniversityPress.

Milner,H.V.,Nielson,D.L.,&Findley,M.G.(2016).Citizenpreferencesandpublicgoods:comparingpreferencesforforeignaidandgovernmentprogramsinUganda.ReviewofInternationalOrganizations,11,forthcoming.

Minoiu,C.,&Reddy,S.G.(2007).AidDoesMatter,AfterAll:RevisitingtheRelationshipBetweenAidandGrowth.Challenge,50(2),39-58,doi:10.2753/0577-5132500203.

44

Minoiu,C.,&Reddy,S.G.(2010).Developmentaidandeconomicgrowth:Apositivelong-runrelation.TheQuarterlyReviewofEconomicsandFinance,50(1),27-39,doi:10.1016/j.qref.2009.10.004.

Morrison,K.M.(2009).Oil,NontaxRevenue,andtheRedistributionalFoundationsofRegimeStability.InternationalOrganization,63(1),107-138,doi:doi:10.1017/S0020818309090043.

Neumayer,E.(2003).TheDeterminantsofAidAllocationbyRegionalMultilateralDevelopmentBanksandUnitedNationsAgencies.InternationalStudiesQuarterly,47(1),101-122.

Nunnenkamp,P.,Rank,M.,&Thiele,R.(2015).AidfragmentationanddonorcoordinationinUganda:Adistrict-levelanalysis.KielInstitutefortheWorldEconomy(IfW).

Pepinsky,T.B.(2008).CapitalMobilityandCoalitionalPolitics:AuthoritarianRegimesandEconomicAdjustmentinSoutheastAsia.WorldPolitics,60(3),438-474.

Ram,R.(2003).RolesofBilateralandMultilateralAidinEconomicGrowthofDevelopingCountries.Kyklos,56(1),95-110,doi:10.1111/1467-6435.00211.

Ram,R.(2004).Recipientcountry's‘policies’andtheeffectofforeignaidoneconomicgrowthindevelopingcountries:additionalevidence.JournalofInternationalDevelopment,16(2),201-211,doi:10.1002/jid.1071.

Rodrik,D.(1996).WhyIsThereMultilateralLending?InM.Bruno,&B.Pleskovic(Eds.),AnnualWorldBankConferenceonDevelopmentEconomics,1995(pp.167-193).Washington,DC:InternationalMonetaryFund.

Seiler,N.,&Madir,J.(2012).FightAgainstCorruption:SanctionsRegimesofMultilateralDevelopmentBanks.JournalofInternationalEconomicLaw,15(1),5-28,doi:10.1093/jiel/jgr037.

Steinwand,M.C.(2015).CompeteorCoordinate?AidFragmentationandLeadDonorship.InternationalOrganization,69(02),443-472,doi:doi:10.1017/S0020818314000381.

Stone,R.W.(2002).LendingCredibility:TheInternationalMonetaryFundandthePost-CommunistTransition.Princeton,NJ:PrincetonUniversityPress.

Stone,R.W.(2004).ThePoliticalEconomyofIMFLendinginAfrica.AmericanPoliticalScienceReview,98(4),577-591.

Strange,A.M.,Dreher,A.,Fuchs,A.,Parks,B.,&Tierney,M.J.(2015).TrackingUnderreportedFinancialFlows:China’sDevelopmentFinanceandtheAid–ConflictNexusRevisited.JournalofConflictResolution,doi:10.1177/0022002715604363.

Tierney,M.J.,Nielson,D.L.,Hawkins,D.G.,Roberts,J.T.,Findley,M.G.,Powers,R.M.,etal.(2011).MoreDollarsthanSense:RefiningOurKnowledgeofDevelopmentFinanceUsingAidData.WorldDevelopment,39(11),1891-1906,doi:http://dx.doi.org/10.1016/j.worlddev.2011.07.029.

Tsoutsoplides,C.(1991).TheDeterminantsoftheGeographicalAllocationofECAidtotheDevelopingCountries.AppliedEconomics,23(4A),647-658.

45

TablesandFigures

Table1:SummaryStatisticsonOutcomeVariables

OutcomeVariable Obs Mean Std.Dev. Min MaxMasses

StrongSupport 3007 0.770 0.421 0 1Tell 2967 0.940 0.237 0 1WillingtoSign 3008 0.831 0.374 0 1SignPetition 3017 0.803 0.398 0 1WillingtoSMS 3017 0.635 0.481 0 1SentSMS 1143 0.049 0.216 0 1AidSupport 3017 3.993 1.461 0 6MPs

StrongSupport 567 0.827 0.378 0 1Tell 567 0.986 0.118 0 1WillingtoSign 567 0.824 0.381 0 1SignPetition 570 0.747 0.435 0 1WillingtoSignPres. 292 0.747 0.436 0 1SignedPres. 292 0.682 0.467 0 1TellConstituents 567 0.984 0.125 0 1RallyLocalOfficials 501 0.970 0.171 0 1CoordinatewithPeers 567 0.970 0.171 0 1AidSupport 570 6.898 1.841 0 9

46

Table2:CompositionoftheSamplevs.the9thUgandanParliament.

Sample 9thParliamentGender %Male 67 65%Female 33 35Party %NRM 74.6 73.5%Independents 10.2 11.2%FDC 8.5 8.8%DP 3.1 3.4%UPC 3.1 2.6%CP 0.25 0.25%JEEMA 0.25 0.25Region %fromCentral 28 25%fromEastern 28 27%fromNorthern 18 22%fromWestern 26 26MPType %ConstituencyMPs 59 62%DistrictWomenMPs 28 29%SpecialInterestMPs 6 7%Ex-OfficioMPs 8 2

47

Table3:BasicResultsfromMPandMassSurveys.

PanelA:MPandCitizenOutcomes Strong

SupportTell Willing

tosignSigned Willing

toSMSSentSMS AidSup-

portMasses Bilateral 0.77 0.94 0.83 0.80 0.64 0.05 4.01N 1532 1512 1533 1537 1537 595 1537Multilateral 0.76 0.94 0.83 0.80 0.63 0.05 3.97N 1475 1455 1475 1480 1480 548 1480Difference -0.01 -0.00 0.00 -0.00 -0.02 -0.00 -0.04MPs Strong

SupportTell Willing

tosignSigned Willing

toSignPres.

SignedPres.

AidSup-port

Bilateral 0.85 0.99 0.83 0.76 0.75 0.67 6.92N 293 293 293 294 145 145 294Multilateral 0.80 0.98 0.81 0.74 0.74 0.69 6.87N 274 274 274 276 147 147 276Difference -0.05* -0.01 -0.02 -0.02 -0.01 0.02 -0.05PanelB:EliteOnlyOutcomes TellConstituents RallyLocalOfficials Coordinatewith

PeersMPs Bilateral 0.99 0.98 0.98N 293 251 293Multilateral 0.97 0.96 0.96N 274 250 274Difference -0.02* -0.02 -0.01Note:Anegativedifferencemeansthattheproportionofsupport forprojects inthebilateralcondition is largerthantheproportionunderthemultilateralcondition,implyingthebilateralconditionispreferredtothemultilat-eralone.Apositivedifferenceimpliesthatthemultilateralconditionispreferredtothebilateralcondition.Notethatifasubjectstateds/hedidnotwanttosignthepetition(thirdcolumn)westillpresentedthemthepossibilityofsigningthepetition(fourthcolumn).ThehigherNsforwillingnesstoSMSinthefifthcolumnarearesultofsub-ject refusals toanswer thepetitionquestions.That is, if a subject refused toanswerpetitionquestions,westillaskedaboutSMSandfewersubjectsdeclinedtoanswerSMSquestions.Also, theNsmaydecrease inthe“SentSMS” condition (relative to “Willing to SMS”) because we only calculate Sent SMS for subjects who owned aphone.Note:Two-tailedtestsofsignificance:*0.10;**0.05.

48

Table4:CitizenPreferencesforAmericanvs.ChineseAid

Allmasssubjects Strong

SupportTell Willing

tosignSigned Willing

toSMSSentSMS

AidSup-port

ChineseAid 0.80 0.91 0.83 0.81 0.68 0.03 4.05N 391 387 391 391 391 139 391U.S.Aid 0.81 0.96 0.84 0.81 0.68 0.09 4.12N 448 442 448 450 450 182 450Difference -0.01 -0.04** -0.01 -0.01 0.00 -0.06** -0.06Note:Two-tailedtestsofsignificance:*0.10;**0.05.

49

Table5:MassSupportandAidSupport:Respondentsknowingdonor

Bilateral N Multilateral N DifferencePoliticization

Aidhelpstheneediestpeople 4.07 335 4.21 222 0.13 Aidhelpsfriendsofdonor 4.14 443 4.29 289 0.15

Conditionality

ConditionsonaidhurtUganda 3.93 356 4.16 245 0.23* ConditionsonaidhelpUganda 4.28 383 4.33 253 0.04

Transparency

Bilateralaidmoretransparentthanmultilateralaid

4.24 277 4.34 191 0.10

Multilateralaidmoretransparentthanbilateralaid

4.01 444 4.22 293 0.21**

Efficacy

Bilateralaidmoreeffectivethanmultilateralaid

4.28 285 4.36 181 0.08

Multilateralaidmoreeffectivethanbilateralaid

4.02 448 4.21 308 0.18*

Alignment

Bilateralaidmatchesneedofcommunity

4.22 268 4.29 187 0.06

Multilateralaidmatchesneedofcommunity

4.02 474 4.28 302 0.26**

Note:Allvaluescapturethescoresontheoverallsupportindex,ratherthanonanyoneindividualoutcomemeas-ure.Anegativedifferencemeansthattheproportionofsupportforprojectsinthebilateralconditionislargerthanthe proportion under themultilateral condition, implying the bilateral condition is preferred to themultilateralone.Apositivedifferenceimpliesthatthemultilateralconditionispreferredtothebilateralcondition.Two-tailedtestsofsignificance:*0.10;**0.05;***0.01.

50

Table6:MPSupportandAidSupport

Bilateral N Multilateral N DifferencePoliticization

Aidhelpstheneediestpeople 6.92 61 7.01 78 0.09 Aidhelpsfriendsofdonor 6.97 184 6.92 212 -0.05

Conditionality

ConditionsonaidhurtUganda 6.77 172 6.70 155 -0.07 ConditionsonaidhelpUganda 7.28 103 7.23 111 -0.05

Transparency

Bilateralaidmoretransparentthanmultilateralaid

6.94 187 6.93 174 -0.01

Multilateralaidmoretransparentthanbilateralaid

6.90 92 6.84 90 -0.06

Efficacy

Bilateralaidmoreeffectivethanmultilateralaid

6.76 82 6.60 75 -0.16

Multilateralaidmoreeffectivethanbilateralaid

7.02 198 7.06 188 0.04

Alignment

Bilateralaidmatchesneedofcommunity

7.04 77 6.94 81 -0.10

Multilateralaidmatchesneedofcommunity

6.89 192 6.93 171 0.04

Note:Allvaluescapturethescoresontheoverallsupportindex,ratherthanonanyoneindividualoutcomemeas-ure.Anegativedifferencemeansthattheproportionofsupportforprojectsinthebilateralconditionislargerthanthe proportion under themultilateral condition, implying the bilateral condition is preferred to themultilateralone.Apositivedifferenceimpliesthatthemultilateralconditionispreferredtothebilateralcondition.Two-tailedtestsofsignificance:*0.10;**0.05;***0.01.

51

Table7:CitizenTrustLevelsforDomesticversusInternationalInstitutions. Multilateral WB UNDP ADB AnymultilateralBilateral Means 3.34 3.29 3.15 3.30U.S. 2.99 0.35 0.30 0.16 0.31 (0.03) (0.04) (0.03) (0.05)China 2.59 0.75 0.70 0.56 0.71 (0.03) (0.04) (0.04) (0.05)Anybilat-eral

2.80 0.55 0.49 0.35 0.50

(0.03) (0.03) (0.03) (0.04)Note:Meantrustinitalics.Cellsreportdifferencesinmeanswithstandarderrorsinparenthesesunderneath.Boldentriesindicatesignificantdifferences.Positivedifferencesandt-statsmeanthatthemultilateraldonorismoretrustedthanthebilateralone.

Table8:SurveyOutcomeResultsforMultilateralvs.BilateralAid.

Mean

(MultilateralAid)

Mean

(BilateralAid)Difference

Masses Trust 3.27 2.80 0.470***

Transparency 0.63 0.37 0.261***

MPs Transparency 0.65 0.35 0.306***

Note:Two-tailedtestsofsignificance:*0.10;**0.05;***0.01.