Fama-French-1992-3-factor-model三因素模型

AmericanFinanceAssociationTheCross-SectionofExpectedStockReturnsAuthor(s):EugeneF.FamaandKennethR.FrenchSource:TheJournalofFinance,Vol.47,No.2(Jun.,1992),pp.427-465Publishedby:BlackwellPublishingfortheAmericanFinanceAssociationStableURL::25/08/201008:16YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://=black.EachcopyofanypartofaJSTORtransmissionmustcontainthesamecopyrightnoticethatappearsonthescreenorprintedpageofsuchtransmission.JSTORisanot-for-profitservicethathelpsscholars,researchers,andstudentsdiscover,use,andbuilduponawiderangeofcontentinatrusteddigitalarchive.Weuseinformationtechnologyandtoolstoincreaseproductivityandfacilitatenewformsofscholarship.FormoreinformationaboutJSTOR,[e-mail]ericanFinanceAssociationarecollaboratingwithJSTORtodigitize,preserveandextendaccesstoTheJournalofFinance.*VOL.XLVII,NO.2*JUNE1992TheCross-SectionofExpectedStockReturnsEUGENEF.FAMAandKENNETHR.FRENCH*ABSTRACTTwoeasilymeasuredvariables,sizeandbook-to-marketequity,combinetocapturethecross-sectionalvariationinaveragestockreturnsassociatedwithmarket3,size,leverage,book-to-marketequity,andearnings-priceratios.Moreover,whenthetestsallowforvariationin3thatisunrelatedtosize,therelationbetweenmarket/3andaveragereturnisflat,evenwhen3istheonlyexplanatoryvariable.THEASSET-PRICINGMODELOFSharpe(1964),Lintner(1965),andBlack(1972)haslongshapedthewayacademicsandpractitionersthinkaboutaveragereturnsandrisk.Thecentralpredictionofthemodelisthatthemarketportfolioofinvestedwealthismean-varianceefficientinthesenseofMarkowitz(1959).Theefficiencyofthemarketportfolioimpliesthat(a)expectedreturnsonsecuritiesareapositivelinearfunctionoftheirmarketO3s(theslopeintheregressionofasecurity'sreturnonthemarket'sreturn),and(b)marketO3ssufficetodescribethecross-sectionofexpectedreturns.ThereareseveralempiricalcontradictionsoftheSharpe-Lintner-Black(SLB)model.ThemostprominentisthesizeeffectofBanz(1981).Hefindsthatmarketequity,ME(astock'spricetimessharesoutstanding),addstotheexplanationofthecross-sectionofaveragereturnsprovidedbymarketOs.Averagereturnsonsmall(lowME)stocksaretoohighgiventheirfestimates,andaveragereturnsonlargestocksaretoolow.AnothercontradictionoftheSLBmodelisthepositiverelationbetweenleverageandaveragereturndocumentedbyBhandari(1988).Itisplausiblethatleverageisassociatedwithriskandexpectedreturn,butintheSLBmodel,leverageriskshouldbecapturedbymarketS.Bhandarifinds,how-ever,thatleveragehelpsexplainthecross-sectionofaveragestockreturnsinteststhatincludesize(ME)aswellasA.Stattman(1980)andRosenberg,Reid,andLanstein(1985)findthataver-agereturnsonU.S.stocksarepositivelyrelatedtotheratioofafirm'sbookvalueofcommonequity,BE,toitsmarketvalue,ME.Chan,Hamao,andLakonishok(1991)findthatbook-to-marketequity,BE/ME,alsohasastrongroleinexplainingthecross-sectionofaveragereturnsonJapanesestocks.*GraduateSchoolofBusiness,UniversityofChicago,1101East58thStreet,Chicago,IL60637.WeacknowledgethehelpfulcommentsofDavidBooth,Nai-fuChen,GeorgeConstan-tinides,WayneFerson,EdwardGeorge,CampbellHarvey,JosefLakonishok,RexSinquefield,ReneStulz,MarkZmijeweski,andananonymousreferee.ThisresearchissupportedbytheNationalScienceFoundation(Fama)andtheCenterforResearchinSecurityPrices(French).427428TheJournalofFinanceFinally,Basu(1983)showsthatearnings-priceratios(E/P)helpexplainthecross-sectionofaveragereturnsonU.S.stocksinteststhatalsoincludesizeandmarketF.Ball(1978)arguesthatE/Pisacatch-allproxyforunnamedfactorsinexpectedreturns;E/Pislikelytobehigher(pricesarelowerrelativetoearnings)forstockswithhigherrisksandexpectedreturns,whatevertheunnamedsourcesofrisk.Ball'sproxyargumentforE/Pmightalsoapplytosize(ME),leverage,andbook-to-marketequity.Allthesevariablescanberegardedasdifferentwaystoscalestockprices,toextracttheinformationinpricesaboutriskandexpectedreturns(Keim(1988)).Moreover,sinceE/P,ME,leverage,andBE/MEareallscaledversionsofprice,itisreasonabletoexpectthatsomeofthemareredundantfordescribingaveragereturns.OurgoalistoevaluatethejointrolesofmarketA,size,E/P,leverage,andbook-to-marketequityinthecross-sectionofaveragereturnsonNYSE,AMEX,andNASDAQstocks.Black,Jensen,andScholes(1972)andFamaandMacBeth(1973)findthat,aspredictedbytheSLBmodel,thereisapositivesimplerelationbetweenaveragestockreturnsand,3duringthepre-1969period.LikeReinganum(1981)andLakonishokandShapiro(1986),wefindthattherelationbetweenA3andaveragereturndisappearsduringthemorerecent1963-1990period,evenwhenAisusedalonetoexplain