testing ambiguity theories with a mean-preserving design 2017 chun-lei yang资料.pdf

该【testing ambiguity theories with a mean-preserving design 2017 chun-lei yang资料 】是由【李十儿】上传分享，文档一共【20】页，该文档可以免费在线阅读，需要了解更多关于【testing ambiguity theories with a mean-preserving design 2017 chun-lei yang资料 】的内容，可以使用淘豆网的站内搜索功能，选择自己适合的文档，以下文字是截取该文章内的部分文字，如需要获得完整电子版，请下载此文档到您的设备，方便您编辑和打印。:..QuantitativeEconomics8(2017),219?2381759-7331/20170219Testingambiguitytheorieswithamean-preservingdesignChun-LeiYangInstituteforSocialandEconomicResearch,NanjingAuditUniversityandResearchCenterforHumanitiesandSocialSciences,AcademiaSinicaLanYaoSchoolofEconomics,ShanghaiUniversityofFinanceandEconomicsProminentmodelssuchasmaxminexpectedutility/alpha-multiprior(MEU/α-MP)andKlibanoff,i,andMukerji(KMM)interpretambiguityaver-sionasaversionagainstsecond--colorurns,-preservingprospects,MEU/α-MP,KMM,andSavage?ssub-jectiveexpectedutilityallpredictunequivocallythatrisk-aversedecisionmakers(DMs)willavoidthe50?50urnthatexhibitsthehighestriskconceivable,whilerisk-,,whichisdistinctfromnotionsofsecond-?ndingisrobustevenwhenthereisonlypartialambiguity,,Ellsbergparadox,expectedutility,experiment,meanpre-serving,monotonicity,partialambiguity,second-orderrisk,,?s(1961)thoughtexperiments,thatmissinginformationaboutobjectiveprobabilitiescanaffectpeople?sdecisionmak-inginawaythatisinconsistentwithSavage?s(1954)subjectiveexpectedutility(SEU)?stwo-colorproblem,onewith50redand50blackballs(theriskyurn)bina-tionofredandblackballs(theuncertainorambiguousurn),mostpeopleprefertobetontheriskyurn,-?rmEllsberg?s?nding,as,forexample,surveyedinCamererandWeber(1992).Chun-LeiYang:******@:yao.******@,YanChen,Soo-HongChew,DanHouser,MingHsu,BinMiao,JackStecher,RobertoWeber,SongfaZhong,(SUFE),MinistryofEducation,Shanghai200433,?-::..220YangandYaoQuantitativeEconomics8(2017)(KangandStulz(1997),CovalandMoskowitz(1999)).In-vestors?expectedexcessreturnsarehigherwheninformationqualityismoreuncer-tain(EpsteinandSchneider(2008)).Limitedstock--holdsiswellillustratedasarisingfromtheheterogeneityamonguncertainty-aversein-vestors(Cao,Wang,andZhang(2005)).Uncertaintyaboutfuturerewardsmaycausejob-eptlesspaymentatanearliertimethanwouldbeoptimalotherwise(CoxandOaxaca(2000),Oprea,Friedman,andAnderson(2009)).Thepresenceofambi-guityisshowntoincreasetheprovisionofpublicgoods(EichbergerandKelsey(2002)),andtoaffectequilibriumpricesinoligopoly?rms(Eichberger,Kelsey,andSchipper(2009)).,GilboaandSchmeidler(1989)developthemaxminexpectedutility(MEU)theory,wherethede-cisionmaker(DM)-alizedtotheso-calledα-MP(multiprior)modelbyGhirardato,heroni,andMari-i(2004).MEUsolvestheparadoxandhasbeenappliedtostudiesonassetpricinginDowandWerlang(1992)andEpsteinandWang(1994)-tionalformisthesmoothmodelofambiguityaversionbyKlibanoff,i,andMukerji(2005;KMM).KMMassumethattheDMisaSavage-typesubjectiveexpectedutilitymaximizer,onthespaceofsecond-,Ju,andMiao(2014),Hansen(2007),HansenandSargent(2010),JuandMiao(2012),heroni,i,andRuf?no(2013)essfullyappliedKMMtostudiesofassetpricing,toobtaininternallyconsistentcalibrationofambiguityattitudes,(CEU)bySchmeidler(1989),(2004)surveyapplicationofCEUinvariousareasofeconomicssuchasin-surancedemand,assetpricing,?elds,manynewexperimentalstudieshavebeenconductedtotestthesemodelsandcharacterizesubjects?-ever,,onecan-notdistinguishwhethertheobservedambiguityaversionre?ectswillingnesstopayanambiguitypremiumforthesecond-orderriskassociatedwiththeuncertainact,whichα-MP(MEU)andKMMpredict,orfortheissueofambiguityperse,whichseemstobe1Forfurthertheoreticalmodelsofmultiplepriors,second-ordersophistication,andrank-dependentutility,seeSegal(1987,1990),Casadesus-Masanell,Klibanoff,andOzdenoren(2000),Nau(2006),ChewandSagi(2008),ErginandGul(2009),andSeo(2009)(2008),EichbergerandKelsey(2009),andEtner,Jeleva,andTallon(2012)offerexcellentsurveys.:..QuantitativeEconomics8(2017)Testingambiguitytheories221behindtheideasofsource-dependencestudiesinitiatedbyHeathandTversky(1991)andFoxandTversky(1995).Wepresentalotterydesign,whichisasimplemodi?cationofEllsberg?stwo-,theDMdrawstwicewithreplacementfromanurnwith2Nballscol-?hwhiteballs,thewinningchanceforthepayoff0,andsymmet-ricallythesamefor2x,ish(2N?h)/4N2,andthatforxis1?h(2N?h)/(N?h)/N2,,werestrictourselvestoamean--,DMswillfacebothanambiguityanda(uniformdistri-pound-,accordingtothewellknowntheoriesofSEU,α-MP(MEU),andKMM,arisk-averseDMpreferstheambiguousurntothe50?50riskyurn,whilearisk-seekingDM?,irrespectiveoftheriskattitudetest,thesetheorieswithsomeproperextensionsalsopredictthattheDMistoconsis-pound-,wefurtherdesignaso-calledpartialambiguity(PA)?,inaconservativeestimationbytreatingallrisk-neutralDMsasnon-violationcases,27?52%subjectsindifferenttreatmentsviolatedthepredictionsofSEU,α-MP(MEU),,23?33%subjectsshowedincon-sistentbehavioracrossthetwomaindecisiontasks,inviolationofthetheoreticalpre-(FA).These?ndingspresentconsiderablechallengestothedominantapproachofinter-pretingambiguityaversionasaversionagainstassociatedsecond-;thatis,somepsychologicalorneuralfactorsmayhelpshapedecisionsbeyondsecond-,wediscusstherelevantpreferencemodels,ourexperimentalde-sign,?,C,,,?rst:..222YangandYaoQuantitativeEconomics8(2017)drawisredandthesecondiswhite,aparticipantgets100yuan;ifthetwodrawsareofthesamecolor,hegets50yuan;butifthetwocolorsareintheorderofwhite?rstandredsecond,?+,subjectshavetochooseeitherthesurepayoffortheriskyone,choiceB,-price-list(MPL)procedurethatcanalsobeviewedasamodi?edversionoftheBecker,DeGroot,andMarschak(BDM),,(FA),(PA),,urwherethenumberdifferencebetweenthetwocolorsisatleast6,whichadmitsonly12outofthe17possi-(#M=11inFAvs.#M=12inPA).WekeepthefeasiblesetofpriorssimilarbetweenPAandFA,buttheambiguityrangeassociatedwithPAisnotsosmallastomaketheam-,however,thatthefeatureofadifferentcolorwinningeachroundensuresthatthemeanofthelotteryisalways50yuan,-,alllotteriescanberankedbyvariance,(100,1/4;50,1/2;0,1/4)here,itwouldlosethestructuralcongruencetochoicesCandD,,Zingales,andMaestripieri(2009),DeGroot,andMarschak(1964)(2002)formulti-price-list(MPL)?m(2008)andTrautmann,Vieider,andWakker(2011).DetaileddiscussioncanbefoundintheAppendix,availableinasupplementary?leonthejournalwebsite,/460/,aPAdesignwithN=6and#M=10wouldmakeittoosimilartoFA,whileonewithN=500and#M=,?rst,togetawayfromindifferenceinthetheorypredictionthatisnotfeasibleunderFA,and,second,,seeChew,Miao,andZhong(2013).:..QuantitativeEconomics8(2017)Testingambiguitytheories223pletelistoffeasiblelotteries,N=**********White109876543210p(0)00090160210240250240210160090p(50)10820680580520500520580680821p(100)00090160210240250240210160090Mean5050505050505050505050Variance04508001050120012501200**********Note:Table1summarizesallpossible?rst-orderlotteriesgiventhispayoffrule,withπhcodingforthelotterywithhredballsand10?,itsmean,,theurnwith4redand6whiteballs,π4,givesustheprobabilitiesof024,052,and024toearntheprizeof0,50,and100yuan,respectively,,ourmodi?edEllsbergriskyprospect,π5,hasthehighestvarianceof1250,={rwrrwwwr}esintheexperi-:S→XwithX={050100}beingthespaceofpay-∈Δ(S),h=012N,denotetheprobabilitydis-htributiononthestatespacewithhredand2N?hwhiteballs,whileπN∈Δ(