驱动全球股票收益的因素

WhatFactorsDriveGlobalStockReturns?ÕKeweiHoua,G.AndrewKarolyib,BongChanKhocAbstractThisstudyseekstoidentifywhichfactorsareimportantforexplainingthetime-seriesandcross-sectionalvariationinglobalstockreturns.Weevaluatefirmcharacteristics,suchassize,earnings/price,cashflow/price,dividend/price,book-to-marketequity,leverage,momentum,thathavebeensuggestedintheempiricalassetpricingliteraturetobecross-sectionallycorrelatedwithaveragereturnsintheUnitedStatesandindevelopedandemergingmarketsaroundtheworld.Formonthlyreturnsof29,000individualstocksfrom49countriesoverthe1981to2003period,weperformcross-sectionalregressiontestsofaveragereturnsattheindividualfirmlevelandweconstructfactor-mimickingportfoliosbasedonthesefirm-levelcharacteristicstoassesstheirabilitytoexplaintime-seriesreturnvariationincountry,industryandcharacteristics-sortedportfolios.Wefindthatthemomentumandcashflow/pricefactor-mimickingportfolios,togetherwithaglobalmarketfactor,capturesubstantialcommonvariationinglobalstockreturns.Inaddition,thethreefactorsexplaintheaveragereturnsofcountryandindustryportfolios,andawidevarietyofsingle-anddouble-sortedcharacteristics-basedportfolios.JELclassification:F30,G14,G15.Keywords:Internationalfinance;assetpricingmodels;commonfactors.CurrentVersion:December27,2006ÕWethanktheDiceCenterforResearchonFinancialEconomics,BSIGAMMAFoundation,andINQUIRE-UKforfundingsupport.HelpfulcommentswerereceivedfromMichaelAdler,MichaelBrandt,FrancescaCarrieri,MagnusDahlquist,KenFrench,GabrielHawawini,SteveHeston,DonKeim,MarkLang,Kuan-HuiLee,RogerLoh,DavidNg,MikeRoberts,DavidRobinson,AnaPaulaSerra,RobStambaugh,RenéStulzandAlvaroTaboadaaswellasfromseminarparticipantsatISCTE(Portugal),UniversidadedoPorto,OhioStateUniversity,UniversityofPennsylvaniaWhartonSchool,BaruchCUNY,YorkUniversity,UniversityofNorthCarolina,DukeUniversity,VanderbiltUniversity,PurdueUniversity,CRSPForum2006,andtheFirstInternationalConferenceonAsia-PacificFinancialMarkets.aFisherCollegeofBusiness,OhioStateUniversity.Email:[e-mail],OhioStateUniversity.Email:[e-mail]stration,SeoulNationalUniversity.Email:[e-mail].1WhatFactorsDriveGlobalStockReturns?Therehasbeenconsiderableevidencethatthecross-sectionofaveragereturnsarerelatedtofirm-levelcharacteristicssuchassize,earnings/price,cashflow/price,dividend/price,book-to-marketequity,leverage,momentumbothintheUnitedStatesandindevelopedandemergingmarketsaroundtheworld.Measuredoverlongsampleperiods,smallstocksearnhigheraveragereturnsthanlargestocks(Banz,1981;Reinganum,1981;Keim,1983;KatoandSchallheim,1985;HawawiniandKeim,1999;Heston,RouwenhorstandWessels,1995).FamaandFrench(1992,1996,1998),Capaul,RowleyandSharpe(1993),Lakonishok,ShleiferandVishny(1994),ChuiandWei(1998),Achour,Harvey,HopkinsandLang(1999a,1999b),EstradaandSerra(2005)andGriffin(2002)showthatvaluestockswithhighbook-to-market(B/M),earnings-to-price(E/P),orcash-flow-to-price(C/P)ratiosoutperformgrowthstockswithlowB/M,E/PorC/Pratios.Moreover,stockswithhighreturnoverthepast3-to12-monthscontinuetooutperformstockswithpoorpriorperformance(JegadeeshandTitman,1993,2001;Carhart,1997;Rouwenhorst,1998;Chan,HameedandTong,2000;Chui,TitmanandWei,2003;Griffin,JiandMartin,2003;Hou,PengandXiong,2006a,2006b).Theinterpretationoftheevidenceis,ofcourse,stronglydebated.Somebelievethatthepremiumsassociatedwiththesecharacteristicsarecompensationforpervasiveextra-marketriskfactors,othersattributethemtoinefficienciesinthewaymarketsincorporateinformationintoprices.Yetothersproposethatthepremiumsarejustamanifestationofsurvivorshipordata-snoopingbiases(Kothari,ShankenandSloan,1995;MacKinlay,1995).Manyofthestudieslistedabovethatfocusoninternationalmarketsmotivatetheireffortsasaresponsetothislattercriticism.Thatis,totheextentthatdevelopedoremergingmarketsmoveindependentlyfromU.S.markets,theyprovideindependentverificationofthesize,valueandmomentumpremiums.Wemotivateourstudyinthissamespirit,butwebroadentheinvestigationtoover29,000stocksfrom49countriesusingmonthlyreturnsoverthe1981to2003periodtore-examinethesize,value/growthandmomentumeffects.Tothisend,wetakeadvantageofthebreadthandcoverageofThomsonFinancial’sDatastreamInternationalandWorldscopedatabases.Weassessavarietyoffirmattributes(includingmarketcapitalization,B/M,E/P,C/P,momentum,dividendyield,andfinancialleverage)forthecross-sectionofexpectedstockreturnsattheindividualfirmlevel.Perhapsmoreimportantly,weseektoidentifywhichfactorsareimportantforexplainingthecommonvariationinglobalstockreturns.Foreachofthefirmattributesdiscussedabove,weconstructazero-2investmentfactor-mimickingportfolio(inthespiritofHuberman,KandelandStambaugh,1987,usingthemethodologyofFamaandFrench,1993,andChan,KarceskiandLakonishok,1998)bygoinglonginstocksthathavehighvaluesofanattribute(suchasB/M)andshortinstockswithlowvaluesoftheattribute.Examiningthereturnsbehaviorofthedifferentmimickingportfolioscanhelpusevaluateandinterprettheunderlyingfactors(CharoenrookandConrad,2005).Finally,weassesstheperformanceofdif