Chp4-3－参考－理性资产定价

RationalAssetPricesGEORGEM.CONSTANTINIDES*ABSTRACTThemean,covariability,andpredictabilityofthereturnofdifferentclassesoffinancialassetschallengetherationaleconomicmodelforanexplanation.Theunconditionalmeanaggregateequitypremiumisalmostsevenpercentperyearandremainshighafteradjustingdownwardsthesamplemeanpremiumbyintro-ducingpriorbeliefsaboutthestationarityoftheprice–dividendratioandthe~non!-forecastabilityofthelong-termdividendgrowthandprice–dividendratio.Recognitionthatidiosyncraticincomeshocksareuninsurableandconcentratedinrecessionscontributestowardanexplanation.Alsoborrowingconstraintsovertheinvestors’lifecyclethatshiftthestockmarketrisktothesavingmiddle-agedconsumerscontributetowardanexplanation.Acentralthemeinfinanceandeconomicsisthepursuitofaunifiedtheoryoftherateofreturnacrossdifferentclassesoffinancialassets.Inparticular,weareinterestedinthemean,covariability,andpredictabilityofthereturnoffinancialassets.Atthemacrolevel,westudytheshort-termrisk-freerate,thetermpremiumoflong-termbondsovertherisk-freerate,andtheaggregateequitypremiumofthestockmarketovertherisk-freerate.Atthemicrolevel,westudythepremiumofindividualstockreturnsandofclassesofstocks,suchasthesmall-capitalizationversuslarge-capitalizationstocks,the“value”versus“growth”stocks,andthepastlosingversuswin-ningstocks.Theneoclassicalrationaleconomicmodelisaunifiedmodelthatviewsthesepremiaastherewardtorisk-averseinvestorsthatprocessinformationrationallyandhaveunambiguouslydefinedpreferencesoverconsumptionthattypically~butnotnecessarily!belongtothevonNeumann–Morgensternclass.Naturally,thetheoryallowsformarketincompleteness,marketimperfec-tions,informationalasymmetries,andlearning.Thetheoryalsoallowsfordifferencesamongassetsforliquidity,transactioncosts,taxstatus,andotherinstitutionalfactors.Thecauseofmuchanxietyoverthelastquarterofacenturyisevidenceinterpretedasfailureoftherationaleconomicparadigmtoexplainthepricelevelandtherateofreturnoffinancialassetsbothatthemacroandmicro*UniversityofChicagoandNBER.IthankJohnCampbell,GeneFama,ChrisGeczy,LarsHansen,JohnHeaton,RajnishMehra,L’ubosˇPástor,DickThaler,andparticularlyAlonBravandJohnCochrane,fortheirinsightfulcommentsandconstructivecriticism.Finally,IthankLiorMenzlyforhisexcellentresearchassistanceandinsightfulcommentsthroughoutthisproject.Naturally,Iremainresponsibleforerrors.THEJOURNALOFFINANCEVOL.LVII,NO.4AUGUST20021567levels.Acelebratedexampleofsuchevidence,althoughbynomeanstheonlyone,isthefailureoftherepresentative-agentrationaleconomicparadigmtoaccountforthelargeaveragepremiumoftheaggregatereturnofstocksovershort-termbondsandthesmallaveragereturnofshort-termbondsfromthelastquarterofthe19thcenturytothepresent.Dubbedthe“EquityPre-miumPuzzle”byMehraandPrescott~1985!,ithasgeneratedacottagein-dustryofrationalandbehavioralexplanationsofthelevelofassetpricesandtheirrateofreturn.Anotherexampleisthelargeincreaseinstockpricesintheearlyandmiddle1990s,whichFederalReserveChairmanAlanGreenspandecriedas“IrrationalExuberance”evenbeforetheunprecedentedfurtherincreaseinstockpricesandprice–dividendratiosinthelate1990s.Myobjectiveistorevisitsomeofthisevidenceandexploretheextenttowhichtherationaleconomicparadigmexplainsthepricelevelandtherateofreturnoffinancialassetsoverthepast100
years,bothatthemacroandmicrolevels.InSectionI,Ireexaminethestatisticalevidenceonthesizeoftheuncon-ditionalmeanoftheaggregateequityreturnandpremium.First,Idrawasharpdistinctionbetweenconditional,short-termforecastsofthemeanequityreturnandpremiumandestimatesoftheunconditionalmean.Iarguethatthecurrentlylowconditionalshort-termforecastsofthereturnandpre-miumdonotlessentheburdenoneconomictheorytoexplainthelargeun-conditionalmeanequityreturnandpremium,asmeasuredbytheirsampleaverageoverthepast130years.Second,Iarguethateventhoughonemayintroduceone’sownstrongpriorbeliefsandadjustdownwardsthesample-averageestimateofthepremium,theunconditionalmeanequitypremiumisatleast6percentperyearandtheannualSharperatioisatleast32per-cent.Thesenumbersarelargeandcallforaneconomicexplanation.InSectionII,Idiscusslimitationsofthecurrenttheorytoexplainempir-icalregularities.Iarguethatpercapitaconsumptiongrowthcovariestoolittlewiththereturnofmostclassesoffinancialassetsandthisimpliesthattheobservedaggregateequityreturn,thelong-termbondreturn,andtheobservedreturnsofvarioussubclassesoffinancialassetsaretoolarge,toovariable,andtoopredictable.Intheremainingsections,Irevisitandexaminetheextenttowhichwecanexplaintheassetreturnsbyrelaxingtheassumptionsofcompletecon-sumptioninsurance,perfectmarkets,andtime-separablepreferences.Asthereaderwillreadilyobserve—andIoffermyapologies—mychoiceofissuesiseclecticandmirrorsinpartmyownresearchinterests.InSectionIII,Ishowthatidiosyncraticincomeshocksconcentratedinperiodsofeconomicrecessionplayakeyroleingeneratingthemeanequitypremium,thelowrisk-freerate,andthepredictabilityofreturns.Iarguethatinsufficientattentionhasbeenpaidtothefactthattheannualaggregatelaborincomeexceedsannualdividendsbyafactorofover20.Laborincomeisbyfarthesinglemostimport