J01150

(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

Context-dependent

cheating:

Experimental

evidence

from

16

countries

David

Pascual-Ezama

a,∗

,

Toke

R.

Fosgaard

b

,

Juan

Camilo

Cardenas

c

,

Praveen

Kujal

d

,

Robert

Veszteg

e

,

Beatriz

Gil-Gómez

de

Lia ˜no

f

,

Brian

Gunia

g

,

Doris

Weichselbaumer

h,i

,

Katharina

Hilken

j

,

Armenak

Antinyan

k

,

Joyce

Delnoij

l

,

Antonios

Proestakis

m

,

Michael

D.

Tira

n

,

Yulius

Pratomo

o

,

Tarek

Jaber-López

p

,

Pablo

Bra ˜nas-Garza

d

a_Department_of_Financial_Economy_and_Accounting,_Universidad_Complutense_Madrid,_Campus_Somosaguas,_Madrid_28223,_Spain b_Department_of_Food_and_Resource_Economics,_University_of_Copenhagen,_Rolighedsvej_23,₁₉₅₈_{Frederiksberg}_C,_Denmark c_Facultad_de_Economia,_CEDE,_Universidad_de_Los_Andes,_Calle_19A_No._1-37_Este,_Bloque_W_(W-803),_Bogotá,_Colombia d_Economics_Department,_Business_School,_Middlesex_University_London,_The_Burroughs,_NW4_4BT,_London,_UK e_School_of_Political_Science_and_Economics,_Waseda_University,_1-6-1_Nishiwaseda_Shinjuku-ku,_Tokyo_169-8050,_Japan

f_Department_of_Social_Psychology_and_Methodology,_Universidad_Autónoma_de_Madrid,_Campus_de_Cantoblanco_s/n,_Madrid_28049,

Spain

g_The_Johns_Hopkins_Carey_Business_School,₁₀₀_{International}_Drive,_Baltimore,_MD_21202-1099,_USA h_Department_of_Economics,_University_of_Linz,_{Altenbergerstr.}_68,₄₀₄₀_Linz,_Austria

i_Department_of_Women’s_and_Gender_Studies,_University_of_Linz,_{Altenbergerstr.}_68,₄₀₄₀_Linz,_Austria j_Department_of_Applied_Economics,_Vrije_Universiteit_Brussel_(VUB),_Pleinlaan_2,₁₀₅₀_Brussels,_Belgium k_University_of_{Erlangen-Nuremberg,}_Chair_of_Economic_Theory,_Lange_Gasse_20,₉₀₄₀₃_Nuremberg,_Germany l_Utrecht_University_School_of_Economics,_Utrecht_University,_{Kriekenpitplein}_21-22,₃₅₈₄_EC_Utrecht,_The_Netherlands

m_Institute_for_Health_and_Consumer_Protection,_Joint_Research_Centre,_European_Commission,_Via_Enrico_Fermi_2749,₂₁₀₂₇_Ispra_(VA),

Italy

n_Department_of_General_Psychology,_University_of_Padua,_Via_Venezia_12,₃₅₁₃₁_Padova,_Italy

o_Department_of_Economics,_Faculty_of_Economics_and_Business,_Satya_Wacana_Christian_University,_Jalan_Diponegoro_52-60,_Salatiga 50711,Indonesia

p_Laboratorio_de_Economía_{Experimental,}_Department_of_Economics,_Universitat_Jaume_I,_Castellón,_Spain

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received25September2014

Receivedinrevisedform24February2015 Accepted25April2015

Availableonline21May2015

Keywords: Honesty Corruption Culturaldifferences

a

b

s

t

r

a

c

t

Policymakersuseseveralinternationalindicesthatcharacterizecountriesaccordingtothe

qualityoftheirinstitutions.However,noefforthasbeenmadetostudyhowthehonestyof

citizensvariesacrosscountries.Thispaperexploresthehonestyamongcitizensacross16

countrieswith1440participants.Weemployaverysimpletaskwhereparticipantsfacea

trade-offbetweenthejoyofeatingafinechocolateandthedisutilityofhavingathreatened

self-conceptbecauseoflying.Despitetheincentivestocheat,wefindthatindividualsare

mostlyhonest.Further,internationalindicesthatareindicativeofinstitutionalhonestyare

completelyuncorrelatedwithcitizens’honestyforoursamplecountries.

BY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

∗ Correspondingauthorat:DepartmentofFinancialEconomyandAccounting,UniversidadComplutenseMadrid,CampusSomosaguas,Madrid,28223, Spain.Tel.:+34630176315;fax:+34910102708.

E-mailaddresses:david.pascual@ccee.ucm.es(D.Pascual-Ezama),tf@ifro.ku.dk(T.R.Fosgaard),jccarden@gmail.com(J.C.Cardenas),

pkujal@gmail.com(P.Kujal),rveszteg@gmail.com(R.Veszteg),bgil.gomezdelianno@uam.es(B.Gil-GómezdeLia ˜no),brian.gunia@jhu.edu(B.Gunia), Doris.Weichselbaumer@jku.at(D.Weichselbaumer),khilken@vub.ac.be(K.Hilken),armenak.antinyan@fau.de(A.Antinyan),J.M.J.Delnoij@uu.nl (J.Delnoij),antonios.proestakis@ec.europa.eu(A.Proestakis),michael.d.tira@gmail.com(M.D.Tira),yulius.pratomo@staff.uksw.edu(Y.Pratomo), tarekjaberlopez@gmail.com(T.Jaber-López),branasgarza@gmail.com(P.Bra ˜nas-Garza).

http://dx.doi.org/10.1016/j.jebo.2015.04.020

(2)

1. Introduction

Imagineyourselfonauniversitycampus.Youseethefollowingannouncement:“Isthisyourluckyday?Flipacoinandwin aChocolate”.YouapproachthedeskandreceivetheopportunitytowinadeliciousandbeautifullywrappedLindtLindor chocolatetrufflebyflippingacoinwithablackandawhiteside.Youflipthecoininaboxatanearbytable;theboxprotects youranonymity.Youwinachocolateifyoureportthatthewhitesidecameupandnothingifyoureportblack.Ifyouactually rolledblack,theonlythingkeepingyoufromenjoyingthetruffleisyourmoralcompass.Youfaceatradeoffbetweenthe joyofeatingafinechocolateandtheindividualdisutilityofhavingathreatenedself-conceptbecauseoflying.Thereisno socialshamingorostracism.Weranthisexperiment(threetreatments)in16countriestotesthowsomeregularcitizens aroundtheworldbehaveinsuchasituation.Interestingly,wedidnotfindanysignificantdifferencesacrossanotherwise veryheterogeneoussetofcountries.

Moststudiesaboutculturaldifferencesregardingdishonestbehaviorhaveusedthesurveymethodology.Transparency InternationalreportslargedifferencesincorruptionaroundtheworldandtheWorldValueSurveydocumentscross-cultural differencesinopinionsregardinghow“justifiable”itistocheatontaxesorpublictransportationfares.However,corruption indicesmeasureperceptionsandnotactualbehaviors.Further,variationsacrosscountriesinacademicdishonesty(Rawwas etal.,2004)andtaxevasion(Almetal.,1995;Cummingsetal.,2009)reinforcetheimpressionthatcheatingisbothabundant anddiversearoundtheworld.Inthesameline,recentresearchinbehavioralandexperimentaleconomicshasshownthat alargefractionofindividualsarepronetocheating(Ariely,2012;Gneezy,2005).AnexceptiontothistrendisAbeleretal. (2014)whoreportaphone-basedincentivizedexperimentwitharepresentativesampleinGermany.Theyfindthatmost peoplereporthonestlyafterflippingacoininabsoluteprivacywitha50%chanceofgettingapayoffof15eurosincashor throughanAmazongiftcertificate.Abelerandcolleaguesalsocomplementedtheirstudywithalaboratoryexperimentand findthatthereisaslightlyhigherlevelofdishonestreportinginthissetting.

Thedominantviewintheliteratureisthatindividualsengageindishonestbehavioraslongastheycanmaintaina positiveself-imagewhileobtainingthemaximumpayoffsfromtheirdishonesty(Mazaretal.,2008).Further,researchhas shownthatcreatingajustificationofapositiveself-imagewhilebehavingdishonestlyiscontextspecific(Fosgaardetal., 2013;GinoandGalinsky,2012;Pascual-Ezamaetal.,2013).

Inthispaper,weexaminepreciselythis.Thatis,whethercheatingpersediffersacrossdifferentcountriesandwhether thecontext1_influences_this_behavior._We_conducted_an_experiment_in_which_participants_reported_the_outcome_of_a_coin tosstowinaprize.Wereplicatedtheexperimentinadiversesetof16countriesaroundtheworldwith1440subjects,90 ineachcountry(30pertreatment).WeusedasimplecheatingtaskbasedonBucciolandPiovesan(2011).Wetookgreat caretoruneachsessionundersimilarconditions.Thelocationoftheexperimentwasalwaysabusycommonareason universitycampuses(seeTable1foranoverview).Participantshadtoflipablack/whitecoin;iftheoutcomewasreported white,theyobtainedaredLindtLindorTruffle;ifreportedblack,theyobtainednothing.Asoursampleincludessubstantial culturalvariation–includingparticipantsfromAnglo-Saxon,Latin,Germanic,NordicandAsiaticcountries–onewould expectsubstantialheterogeneityincheating.Inaddition,andconsistentwithasocialconstructionistview,theeffectsof culturedependonthespecificsofthechoicecontext(Gelfandetal.,2013;KramerandMessick,1995).

Inthefirsttreatment(theSelf-ReportedTreatment,SRT)eachparticipantflippedacoininaprivateareawithoutthe presenceoftheresearchersorotherparticipants.Afterwardsthesubjectfilledareportsheet–indicatingsexandthe white/blackcoin-tossresult.Itwasclearlyindicatedthatthefilledformshouldbeleftintheboxnearby.Nointeractionwith theexperimentersoccurredinthistask.

Takingintoaccountthatthereareheterogenoustypesofdishonestpeople(Gneezyetal.,2013),thistaskisagood measureofdishonestyforseveralreasons.Firstly,asagrowingbodyofresearchsuggeststhatincompletelyanonymous settings,wheretheriskofbeingcaughtisinexistent,finding100%cheatingisnotunusual(Pascual-Ezamaetal.,2013;Shu etal.,2012),however,peoplemayalsorestricttheamountofcheating(AyalandGino,2011;Gneezy,2005).

Inourtaskthereisnopossibilitytorestricttheamountofcheating,thedecisionissimplytobehonestornot(report blackorwhite).Further,decisionmakingisimmediateandintuitive,insteadofdeliberativeandmeditativeinacognitive dualsystem(BazermanandTenbrunsel,2011;Kahneman,2011).Inourexperiment,cheatingisanautomaticresponseand theneedforjustificationmattersonlywhenpeoplehaveenoughtimetodeliberate(Shalvietal.,2012).Finally,thereward isasimplechocolatethatgivesinstantgratification.

ThetwoothertreatmentsweretheWrittenandtheVerbalReportedTreatments(WRTandVRT,respectively).InWRT, participantscompletedareportsheetinprivateandsubmittedittotheexperimenter.Theexperimentermadeanoteofthe reportedoutcome,and,ifdue,handedthechocolatetotheparticipant.InVRT,participantswerenotaskedtofillanyform. Theyverballyreportedtheoutcomeofthecoinfliptotheexperimenter.Ifreportedwhite,theexperimentalisthandeda chocolatetothem.

Ourdifferenttreatmentsallowustounderstandhowthelevelofcheatingisshapedbycontext,i.e.thedifferencesin reportingacrosstreatments.Buildingonthetheoryofself-imagemaintenance(Mazaretal.,2008),wepredictthatour threetreatmentswillhavedifferentimplicationswithregardtothemoralprocessesofreportingincorrectoutcomes.We

(3)

Table1

Summaryofcountriesstudied.

Country City University

Austria Linz JohannesKeplerUniversityLinz

Belgium Brussels VrijeUniversiteitBrussel(VUB)

Colombia Bogotá UniversidaddelosAndes

Denmark Copenhagen CopenhagenUniversity

Finland Oulu UniversityofOulu

Germany Nuremberg UniversityofErlangen-Nuremberg

Greece Rethymno UniversityofCrete

India Delhi HansrajCollege,UniversityofDelhi

Indonesia Salatiga UniversitasKristenSatyaWacana

Italy Padova UniversitàdiPadova

Japan Tokyo WasedaUniversity

Netherlands Utrecht UtrechtUniversity

Spain Madrid UniversidadAutónomaMadrid

Turkey Istanbul Bahc¸es¸ehirÜniversitesi

USA Baltimore JohnsHopkinsUniversity

UK London MiddlesexUniversity

conjecturethatmisreportingintheSRTistheleastmorallydifficultdecisionsincereportingdoesnotinvolveanykindof socialinteractionand,subsequently,nosocialpressurefromtheexperimenter.Consequently,thedecisiontobedishonestis onlyatradeoffbetweentheinternalmoralpressure(andtheconsequentguiltexperienced)againstthepleasureofenjoying aLindtchocolate.

InWRT,anadditionalmoralpressureoriginatesfromthefactthatthedecisionisdocumentedonpaperandhandedto anotherperson.Thisadditionalsocialprocessislikelytoinvolveshame(ontopofguilt)incaseofdishonesty,andhence weexpectlessdishonestyoccurringinthistreatment.Relativetoverbalreports,weconjecturethatwrittenreportsare moredistantandmakeiteasiertotakeawayinternalmoralcontrol(Shuetal.,2012).Therefore,weexpecttheVRTtobring evenstrongerself-imageerosionincaseofcheating,sincetheuntruthfuldecisionnowmustbestatedface-to-face.Inthis situationweexpectthemostmoralpressuretobepresentandhencetheleastdishonestbehavior.Thisisconsistentwith

Banduraetal.(1996)whofindthatverbalreportsarelikelytobelesspronetodishonestbehaviorthanwrittenreports. Giventhiswehypothesizethatself-reportingwithnointeraction(withtheexperimenter)increasethetemptationtobe dishonest.

2. Experimentaldesign

WereplicateasimplecointaskbyBucciolandPiovesan(2011).Asnotedbefore,participantshadtoflipablack/white coin.IfthereportedtosswaswhitetheyobtainedaredLindtLindorchocolateandnothingotherwise.

Atfirstglance,thechoiceofchocolateasanincentivemightseemasunusualcomparedtotheconventionalchoice ofmoney.Webelievethatchocolatesareagoodsolutionforstudyingdecision-makingacrosscountries,giventhatour studyisinafieldsettingandnotinthelaboratory.Infact,webelievethatitisadvantageoustotellstudentsyouhave achanceofwinningachocolate,ratherthantellingthemyouhaveachancetowinadollar(oranycorrespondinglocal currency).Consumingasmallsnackinsuchasituationissimplymuchmoreofaneverydayevent,andhencefeelsmore natural.Andamorenaturalsettingenablesgreaterexperimentercontrolandexternalvalidity.Furthermore,achocolateis achocolateanywhere,andsubjectsneverhavetocalculatethemonetaryvalueofthegood,justreceiveasmalldelicious snack.Clearly,therearemanyotherpotentialincentiveapproachesonecouldapply.Amonetaryprizecouldbescaledbased onBigMacprizes,GDP,oranypurchasingpowerindex.Althoughtheremightbecountrydifferencesinchocolatepreference (asindicatedbyMesserli,2012),thechoiceofchocolateaspaymentconstitutesjustasmanyorjustasfewconcernsasany othercross-countrypaymentvehicle,includingcountry-adjustedmonetarycompensations.Anotherpossibleconcernisthe verysmallsizeoftheincentive.Weacknowledgethesmallsize,yetwewanttoemphasizethatconsideringtheextremely shortdurationoftheexperiment(perhaps1or2min),ourpaymenttranslatesintoanhourlypaymentsimilartowhatis usedinmostlaboratorystudies.

Ourparticipantsareuniversitystudents.2_The_coin_was_flipped_only_once_and_by_one_person_at_a_time._The_task_was performedinaprivateareaatasafedistancefromtheexperimentersorotherparticipants.Thelocationoftheexperiment waswithinthecollege/universitycampus,forexample,inahallwayinthecafeteriaarea,henceensuringasteadyflowof potentialparticipants.Toensurethatwedostudycountrydifferences,weensuredthatonlynativepeopleparticipatedin thevariouslocations.Theimportanceofensuringthisissupportedbyevidenceoftheeffectofimmigrants’originalcountry

(4)

socialnormsinethicalbehavior(KountourisandRemoundou,2013).Theexperimentwascarriedoutinthelocalworking languageforboththeverbalandthewrittenpart.3

Thewholestudycomprisesthreedifferenttreatmentsaccordingtothelevelofproximitybetweentheparticipantand theexperimenter:

•Self-ReportedTreatment(SRT):Astudentapproachedtheexperimenter(attableA,seeFigureA1intheappendixfor details),he/shewashandedareportsheet,andwasguidedtotableB(farenoughawayfromtheresearcherarea)toflip theblack/whitecoinprivately(seeBinFig.A1).Thecoinflipwascarriedoutinsideasmallcardboardboxtomakeittruly private.ThestudentnotedgenderandtheresultofthecoinfliponthereportsheetattableB.Afterfillingoutthereport sheet,thestudentwalkedovertoathirdtable(tableCinanotherprivateareaatadistancefromtheexperimenter,seeC inFig.A1)todepositthereportsheetinabox,andtotakethechocolatehim/herselfincaseofawhitetoss.

•WrittenReportedTreatment(WRT):ExactlythesameasSRT,butnowwiththefilledoutreportsheetinhand,thestudent wentbacktotableA,gavethereportsheettotheexperimenterandreceivedthechocolateifawhitetosswasreported. TableCisthereforenotused(seeFig.A1,Appendix).

•VerbalReportedTreatment(VRT):ExactlythesameasWRT,butnowthestudentdidnothavetofilloutthereportsheet (onlyreportgender)andhadtoinformtheexperimenterverballyabouttheoutcomeofthecointoss.

Werantheexperimentwith90participantsineachcountry.Therewere10studentsforeachtreatmentandthiswas repeatedthreetimes:10(WRT),10(VRT),10(SRT),10(WRT),10(VRT),10(SRT),10(WRT),10(VRT),10(SRT).Westarted eachtreatmentwithtenchocolatesonaplateontableB(VRTandWRT)ortableC(SRT).Whenwechangedthetreatmentwe refilledthechocolatestakenbytheparticipantsandregisteredtheexactnumberofchocolatesreplaced4_._At_the_beginning ofeachtreatmentwestartedwithtenchocolatesintheplate.Attheendoftheexperimentwecomparedthenumbersin ourrecordswiththereportsbytheparticipants.InWRTandSRTparticipantswrotethecoloronthereportsheet.InVRT participantsonlywrotethegender.Inthiscase,whenparticipantsgaveusthereportsheetandtoldusthecolorwetore alittlecornerofthereportsheetwhentheytoldusblacksoattheendwecouldknowhowmanyblacksandwhiteswere declared.

Notethat,aswekeptnonamesorindividual-specificrecords,wehadtobeverycarefulaboutwhethersomeonereturned. Experimenterswereinstructedtoinformanyonewhotriedtoreturntoflipthecoinagainthatitwasonlypossibleto participateonce.Incase,facultymembersshowedupforparticipation,itwasemphasizedthatitwasintendedtobea studentevent.Completingtheexperimenttookaboutaminuteortwo.

3. Results

3.1. Differencesbytreatments

Toourgreatsurprise,overallonly57%oftheparticipantsreportedwhiteacrossallthreetreatmentsandcountries.Since theprobabilityofobtainingwhiteis50%,thisresultindicatesthat86%resistedthetemptationtolie[100%−2(57%–50%)]. Weconcludethatlittledishonestyisobservedacrossallcountriesandtreatments.Lookingatthedifferenttreatments,the degreeofdishonestbehavior,listedasthepercentageofwhiteoutcomes,confirmsourinitialconjectures.Weobservea greateramountoflyingintheSRT,withWRTinthemiddle,andwithVRThavingthesmallestamountoflying.

62%(SelfRT)>57%(WrittenRT) >53%(VerbalRT)

ThedifferencesbetweenSRTandWRTarenotstatisticallysignificant(2=.316;p-value=.574),althoughtheaverage

acrossthesixteensocietieswere5percentagepointshigherfortheformer.However,usingacontingencytable(Pearson

2)thedifferencebetweenSRTandVRTwaslarger(9percentagepoints)andsignificantatthe5%level(2=4.479;

p-value=0.034).Further,wedonotfindanysignificantdifferencesbetweenWRTandVRT(2=2.422;p-value=0.120).Thus,

comparingthedifferenttreatmentswefindthattheverbalreportsarelikelytobelesspronetodishonestbehaviorthan writtenreportsinlinewithpreviousliterature(Banduraetal.,1996;FestingerandCarlsmith,1959;Mazaretal.,2008)and wefurtherfindthatself-reportingtheanswers(andnothavingtodeliverittotheexperimenter)increasesthiseffect.

3 _A_few_students_asked_why_the_{coin-chocolate}_event_was_carried_out._Our_prepared_explanation_was_that_the_purpose_was_to_understand_the_students’ interestforchocolates.Thisstatementwasintentionallymadeimprecise,suchthatstudentsdidnotthinkthiswasatestbutrathersawitasachocolate promotion(weneversaiditwasaLindtpromotion).

(5)

Table2

Resultsbyconditionandcountry.

Country Self-Report(SRT)

(n=30p/t)

WrittenReported (WRT)

(n=30p/t)

VerbalReported (VRT) (n=30p/t)

All (n=90)

Austria 40% 77%*** _60% _59%*

Belgium 57% 47% 53% 52%

Colombia 57% 67%** _43% _56%

Denmark 70%** _43% _40% _51%

Finland 63%* _67%** _53% _61%**

Germany 73%*** _50% _53% _59%*

Greece 60% 60%** _37% _52%

India 47% 67% 50% 54%

Indonesia 60% 57% 40% 52%

Italy 60% 47% 57% 54%

Japan 57% 60% 57% 58%*

Netherlands 67%** _60% _50% _59%*

Spain 77%*** _67%** _63%* _69%***

Turkey 53% 67%** _57% _59%*

USA 70%** _53% _63%* _62%**

UK 63%* _33% _70%** _56%

Notes:Columnsshowthepercentagesofparticipantswhohavetakenchocolatesineachtreatment(denoted“p/t”). *_p-value_<_0.10.

**_p_-_value_<_0.05. ***_p_-_value_<_0.01.

3.2. Differencesbycountries

Asecondimportantresultisthatwefindnostatisticallysignificantdifferencesacrossthe16countriesacrossany treat-ment:SRT(2=16.953;p-value=0.322);VRT(2=15.691;p-value=0.403)andWRT(2=22.674;p-value=0.091).This

suggeststhatthereisahighlevelofhomogeneityinbehavioracrossoursamplecountries.

Additionalinterestingresultsshowupfromacloserexaminationofthedata.Sincetossingacoinfollowsabinomial distribution,welookedforthosecountrieswherethereisananomalousnumberofwhite(awarded)coinsreported.Table2

reportsdetailedinformationbycountryandtreatments.5_At_the_country_level,_the_share_of_whites_reported_in_excess_of_50% (acrossalltreatments),issignificant(ata5%significancelevel)inthreeoutofsixteencountries.Yet,theoverallcountry leveleffectscanmaskunderlyingeffectsatthetreatmentlevel.Wethereforefocusonthetreatmenteffectsatthecountry level.Obviouslyasamplesizeof30pertreatmentallowsusonlytomakeroughassessmentofthedegreeofcheating,but westillbelievethisroughmeasureprovidesausefulcomparisonacrosscountriesandtreatments.6

ResultfromtheSelf-reporttreatment(SRT)isdisplayedinFig.1.InSRTwefindthatthenullhypothesisofhonestbehavior isrejectedforanumberofcountries.ThisfigureshowsthatUKandFinlandrejectthenullat˛=10%(dottedlinelabeledwith

*).Mostimportantly,GermanyandSpainrejectat˛=1%(continuouslinelabeledwith***).Finally,Netherlands,Denmark

andtheUSrejectfor1%<˛<10%.

Fig.2displaysdatafromTable2forthewrittenandtheverbaltreatments.Blue(leftside)barsinFig.2refertoWRT (written).ThenumberofcountriesthatrejectthenullisreducedtosixandtheyarenotthesameasinSRT:Colombia, Finland,TurkeyandIndiarejectat1%<˛<10%whileAustriadoessoat˛=1%.Spainweaklyrejects(˛=10%).DatafromVRT

(verbal)areshownintheredbars(rightside)ofFigure2.Onlytwocountriesappearsignificant:Spainrejectsat1%<˛<10%

whiletheUKrejectsatthe1%level.TheUSweaklyrejects(˛=10%).

Therearesomeinterestingissuestoemphasize:Spainalsorejectedthenullinalltreatments.UnexpectedlyUKappears undertheverbalandtheself-reportedtreatmentbutnotinthewritten.Thelattercaseisexactlythecomplementaryof Austriawhichappearsinthewrittencaseonly.AlsotheUSandFinlandappearintwotreatments.Itisalsoimportantto remarkthatBelgium,Greece,Italy,IndonesiaandJapanneverappearasdishonest.7_Although_it_is_interesting_to_note_that thesedifferenceattheconditionleveldonotaimatexplainingeachcountry-specificeffects,wesimplywanttohighlight thatbesidesouroverallconclusionofnosignificantacross-countryeffects,wedoobservesomevariationinthedata.

5_We_have_also_conducted_a_Logit_analysis_to_explain_the_outcome_of_individual_coin_tosses._In_the_regression,_among_other_factors,_we_controlled_for differentincomelevelsacrosscountries(withthehelpofper-capitaGNI)andforcountry-levelper-capitacocoaconsumption.Whileincomedoesnot seemtohaveasignificantimpact,per-capitacocoaconsumptionhasanegativeeffectoncheating,i.e.thehighertheper-capitaconsumptiontheloweris cheating.

6_We_also_conduct_the_analysis_by_using_{Inglehart-Welzel}_cultural_categories_(based_on_the_World_Value_Surveys)._Again_we_do_not_find_any_robust_effect ofcultureoncheatingbehavior.

(6)

Fig.1. PercentageofchocolatestakenintheSRT.*p-value<0.10;***p-value<0.01.

Fig.2.PercentageofchocolatestakenintheWRTandVRT.*p-value<0.10;***p-value<0.01.

3.3. Genderdifferences

Onanexploratorybasis,wealsoanalyzedtheeffectofgender(seeCrosonandGneezy,2009,forareview).Although previousstudieshavefoundcorrelationsbetweengenderandcheating(e.g.,DreberandJohannesson,2008;Fosgaardetal., 2013),nostatisticallysignificantdifferencesbetweenthecountrieswerefoundforthemale(2=15.898;p-value=0.389)or

forfemale(2=10.679;p-value=0.775)participantsinlinewithveryrecentstudies(Abeleretal.,2014).Analyzingby

treat-ment,nodifferenceswerefoundintheSRT[(2=16.374;p-value=0.358)(2=17.051;p-value=0.316)],WRT[(2=19.872;

p-value=0.177)(2=19.341;p-value=0.199)]ortheVRT[(2=8.581;p-value=0.898)(2=12.618;p-value=0.632)]

(7)

Table3

Pearsoncorrelationsofthefourindexeswiththethreetreatments.

BSI2011 (11countries)

PCI2012 (16countries)

GCB2013 (14countries)

AOC2014 (16countries)

SRT 0.057 −0.334 −0.590** −0.096

p-value 0.868 0.205 0.026 0.725

WRT 0.232 0.317 0.340 0.115

p-value 0.492 0.231 0.235 0.673

VRT −0.483 −0.381 −0.339 −0.336

p-value 0.132 0.146 0.236 0.203

Notes:(.)indicatesthenumberofcountriesinthesample.SomecountriesofoursamplearemissingintheBSI(Austria,Colombia,Denmark,Finland, Greece)andintheGCB(Austria,Netherlands);p-valueis2-tailed.

**_p-value_<_0.05.

3.4. Comparisonwithotherevidenceofunethicalbehavior

Theanalysisofourresultcanbecomplementedwithacomparisonofthe(published)rankingsofcorruption.Notethatthe linkbetweencheatingandcorruptionhasbeenreportedasstrong(Magnusetal.,2002).Therefore,wehavetestedwhether ourexperimentaldatacorrelatewithanyofthefollowinginternationalcorruptionindexesfromTransparencyInternational andWJRRuleofLawIndex:i.e.theBSI2011:BriberyPayingIndex(fromTransparencyInternational);thePCI2012:Perception ofcorruptionIndex;theGCB2013:GlobalCorruptionBarometer(averageofallsectors)andAOC2014:Absenceofcorruption.8

Table3summarizesallthecorrelations(12comparisons:4indices×3treatments),forthecountriesinoursamplethatare alsorepresentedintheindices.

AsshowninTable3,onlyonecorrelationisstatisticallysignificant(GCB2013vs.SRT;p<0.05).Ontopofthat,thesignof thecorrelationsdonotfollowanypattern:weget7negativesand5positives.Wecansafelysaythatinternationalindices indicativeofinstitutionalhonestyarenotcorrelatedwithourexperimentaldata.

Asafurtherrobustnesscheckwealsoanalyzedtherelationbetweenthebehaviorintheexperimentsandanswerstothe worldvaluesurveysquestionsaboutmoralityattitude.9_We_find_that_neither_the_answer_of_individual_item_(all_p_>_0.100) northeaverage(p=0.179)ofthemiscorrelatedwithourexperimentalevidence.

4. Conclusions

Whypeoplecheatand howitvariesacrossculturesisanimportantpolicyquestion.Mostofthestudiesregarding thishavefocusedoninstitutionaldishonesty(orcorruption).However,whetherinstitutionalhonestydirectlymapsinto itscitizens(innate)honestyhasnotbeenexplored.Notethat,whileinstitutionalhonestyisafunctionofits(self)rules, individualhonestyismuchmoreprimal,andhencedifferentfromitsmorepopularcounterpart,i.e.institutionalhonesty. Interestingly,thereislittleworkinthisregard.Thispapersattemptstostudycitizen’shonestyinamulti-countrystudy.To ourknowledgethisisthefirststudyofthiskindduetothespanofcountriesandtheincentivizedmechanismused.

Moststudiesoninstitutionalhonestyrelyonnon-incentivizedself-reportedsurveys.Inourstudy,weusechocolatesas anincentiveandtemptationdevice.Therewardisannouncedandisthemainmotivatortoprospectiveparticipantsinour experiment.Theyhopetowinachocolatebytakingpartintheexperiment.Thisselfselectionalsoimpliesthattheyhave apreferencefortherewardandhavefallenforthe“temptationofwinningachocolate.”Byusingthismechanismwethus studytheinnatetendencytolieamongstcitizensof16countries.

Ourmainfindingisthattherearenostatisticallysignificantdifferencesacrossoursamplecountriesregardingtheir honestylevels.Infactweobserveahighlevelofhonestyandsomeimportantculture-specificinteractions.Recallthatin ourtaskwecomparehonestbehavioracrosscountrieswhensubjectscanreporttheoutcomeunderthreeconditions,i.e. Self,WrittenorVerbal.Wearguethatthesethreeconditionsdifferintermsofthemoralweightoflying.Thatis,underSelf reportingandnointeractionwiththeexperimenter,themoralweightoflyingistheleast,meanwhile,itincreasesaswe movefromtheVerbaltotheWrittencondition.

Overall,wedofindsupportforourinitialhypothesisinthatthereisgreaterdeceptionundertheSelfreportingcondition followedbyWrittenandVerbal.However,weobservenodifferencesacrosscountries.Ourlackofresultsinthisregardcould beduetothesmallerdatasizeatthetreatmentlevelineachcountry.Wehaveadditionallylookedwhetherweobserve significantdifferencesacrosscountriesbygroupingtheaccordingtocommoncharacteristics(i.e.European,Asian,etc.). Againwefindnosignificantdifferencesacrossoursamplecountries.Finally,thoughhandicappedbyevensmallernumber ofobservations,wedonotfindanygenderdifferencesacrossthethreeconditionsorcountries.

Asmentionedearlierwemeasuretheinnatetendencyofcitizenstocheatandwouldliketostressthatthisisverydifferent fromstudyinginstitutionalhonesty.Ourresultsshouldnotbeconfoundedwiththosestudyinginstitutionalhonestyasthe

8_Note_that,_countries_were_ordered_from_least_to_most_corrupt_for_all_indices_for_the_analysis.

(8)

factorsthatdetermineonearedifferentfromtheother.Further,thekindofcheatingwestudyisinstinctive.Thatis,subjects donotgettimetodeliberatetheiractions.Thesameappliestotheirparticipationintheexperiment.Thatis,weexplicitly statetherewardandhenceexpectthatweattractparticipantsthathaveanaffinitytowardchocolateconsumption.Given this,thehighlevelsofhonestyweobserveacrosscountriesisindeedsurprising.Further,ourresultsshowthatapparently thisistrueacrossoursamplecountries.

Welookatthesideofcitizensandnotatthesupplysideofcorruptionintheprivateandpublicsectoroftheeconomy. Surveystudiessufferfromempiricalissuessuchasconfoundingfactorsandhypotheticalbiasesthatareinherentin ques-tionnaires(FalkandHeckman,2009).Further,reliabledataonissuessuchascorruption,dishonesty,cheating,etc.arehard tocomeby.Itisinthiscontextthattheexperimentalmethodologyisveryuseful.Wetestedordinarycitizens’behavior, asopposedtosubjectiveperceptionsofnationalinstitutions.Importantly,however,ourdatasuggestthatordinarycitizens aremuchmorehonestthanimpliedbysurveysfocusedonnationalinstitutions.Perhapsitisthecorruptingnatureofthese institutionsthatisreflectedinthesesurveystudies,butnotthecorruptionoftheaveragecitizen.Thelackofconnection betweenthesupplyofcorruptionandthehonestyofindividualcitizensalsorequiresfurtherinquiry.Ourresultssuggest thatthevariationinhonestyacrosscountriesisnotcorrelatedwithcorruptionindices,openingquestionsaboutthe account-abilityofprivateandpublicinstitutionsandthepoorconnectionwiththecitizens’behaviorandpreferencesforhonestyin theirprivatedecision.Ourresultsclearlyshowtheneedforfurtherincentivizedandcontrolledexperimentstoexplorethe issueofordinarycitizens’honestyacrosscountries.

Acknowledgements

Financialsupportfromthefollowing:MinistryofSciencefromSpain:ECO2013-44879-RandJuntadeAndaluciaExcelllent GroupsP12-SEJ-1436(Spain).

AppendixA. Supplementarydata

Supplementarydata associated with this article can be found, in theonline version, at http://dx.doi.org/10.1016/ j.jebo.2015.04.020.

References

Abeler,J.,Becker,A.,Falk,A.,2014.Representativeevidenceonlyingcosts.J.PublicEcon.113,96–104. Alm,J.,Sanchez,I.,deJuan,A.,1995.Economicandnoneconomicfactorsintaxcompliance.Kyklos48(1),3–18.

Ariely,D.,2012.TheHonestTruthAboutDishonesty:HowWeLietoEveryone–EspeciallyOurselves.Harper;Honestedition,Boston,MA.

Ayal,S.,Gino,F.,2011.Honestrationalesfordishonestbehavior.In:Mikulincer,M.,Shaver,P.R.(Eds.),TheSocialPsychologyofMorality:Exploringthe CausesofGoodandEvil.AmericanPsychologicalAssociation,Washington,DC.

Bandura,A.,Barbaranelli,C.,Caprara,G.V.,Pastorelli,C.,1996.Mechanismsofmoraldisengagementintheexerciseofmoralagency.J.Personal.Soc.Psychol. 71(2),364–374.

Bazerman,M.H.,Tenbrunsel,A.E.,2011.BlindSpots:WhyWeFailtoDoWhat’sRightandWhattoDoaboutIt.PrincetonUniversityPress,UK. Bucciol,A.,Piovesan,M.,2011.Luckorcheating?Afieldexperimentonhonestywithchildren.J.Econ.Psychol.32(1),73–78.

Croson,R.,Gneezy,U.,2009.Genderdifferencesinpreferences.J.Econ.Lit.47(2),1–27.

Cummings,R.G.,Martinez-Vazquez,J.,McKee,M.,Torgler,B.,2009.Taxmoraleeffectstaxcompliance:evidencefromsurveysandanartefactualfield experiment.J.Econ.Behav.Organ.70,447–457.

Dreber,A.,Johannesson,M.,2008.Genderdifferencesindeception.Econ.Lett.99(1),197–199.

Exadaktylos,F.,Espín,A.M.,Bra ˜nas-Garza,P.,2013.Experimentalsubjectsarenotdifferent.Sci.Rep.3(1231),http://dx.doi.org/10.1038/srep01213. Falk,A.,Heckman,J.,2009.Labexperimentsareamajorsourceofknowledgeinthesocialsciences.Science326(5952),535–538.

Festinger,L.,Carlsmith,J.M.,1959.Cognitiveconsequencesofforcedcompliance.J.Abnorm.Soc.Psychol.58,203–210.

Fosgaard,T.,Hansen,L.G.,Piovesan,M.,2013.Separatingwillfromgrace:anexperimentonconformityandawarenessincheating.J.Econ.Behav.Organ. 93,279–284.

Gelfand,M.J.,etal.,2013.Towardaculture-by-contextperspectiveonnegotiation:negotiationteamsintheUnitedStatesandTaiwan.J.Appl.Psychol.98 (3),504–513.

Gino,F.,Galinsky,A.D.,2012.Vicariousdishonesty:whenpsychologicalclosenesscreatesdistancefromone’smoralcompass.Organ.Behav.Hum.Decis. Process.119(1),15–26.

Gneezy,U.,2005.Deception:theroleofconsequences.Am.Econ.Rev.95(1),384–394.

Gneezy,U.,Rockenbach,B.,Serra-García,M.,2013.Measuringlyingaversion.J.Econ.Behav.Organ.93,293–300. Kahneman,D.,2011.Thinking:FastandSlow.FarrarStrausandGiroux,NewYork.

Kountouris,Y.,Remoundou,K.,2013.Isthereaculturalcomponentintaxmoral?EvidencefromimmigrantsinEurope.J.Econ.Behav.Organ.96,104–119. Kramer,R.M.,Messick,D.,1995.NegotiationasaSocialProcess.Sage,ThousandOaks,CA.

Magnus,J.R.,Polterovich,V.M.,Danilov,D.L.,Savvateev,A.V.,2002.Tolerancetocheating:ananalysisacrosscultures.J.Econ.Educ.33,125–135. Mazar,N.,Amir,O.,Ariely,D.,2008.Thedishonestyofhonestpeople:atheoryofself-conceptmaintenance.J.Mark.Res.45(6),633–644. Messerli,F.H.,2012.Chocolateconsumption,cognitivefunction,andNobellaureate.N.Engl.J.Med.367,1562–1564.

Pascual-Ezama,D.,Prelec,D.,Dunfield,D.,2013.Motivation,money,prestigeandcheats.J.Econ.Behav.Organ.93,367–373.

Rawwas,M.Y.A.,Al-Khatib,J.A.,Vitell,S.J.,2004.Academicdishonesty:across-culturalcomparisonofU.S.andChinesemarketingstudents.J.Mark.Educ. 26,89–100.

Shalvi, S., Eldar, O.,Bereby-Meyer, Y., 2012. Honesty requires time (and lack of justifications). Psychol. Sci. 23, 1264–1270, http://dx.doi.org/ 10.1177/0956797612443835.