THE JOURNAL OF FINANCE.docx

THE JOURNAL OF FINANCE.docx

《THE JOURNAL OF FINANCE.docx》由会员分享，可在线阅读，更多相关《THE JOURNAL OF FINANCE.docx（48页珍藏版）》请在冰豆网上搜索。

THEJOURNALOFFINANCE

THEJOURNALOFFINANCE.

RationalAssetPrices

GEORGEM.CONSTANTINIDES*

ABSTRACT

Themean,covariability,andpredictabilityofthereturnofdifferentclassesof

financialassetschallengetherationaleconomicmodelforanexplanation.The

unconditionalmeanaggregateequitypremiumisalmostsevenpercentperyear

andremainshighafteradjustingdownwardsthesamplemeanpremiumbyintroducing

priorbeliefsaboutthestationarityoftheprice–dividendratioandthe~non!

forecastability

ofthelong-termdividendgrowthandprice–dividendratio.Recognition

thatidiosyncraticincomeshocksareuninsurableandconcentratedinrecessions

contributestowardanexplanation.Alsoborrowingconstraintsovertheinvestors’

lifecyclethatshiftthestockmarketrisktothesavingmiddle-agedconsumers

contributetowardanexplanation.

Acentralthemeinfinanceandeconomicsisthepursuitofaunifiedtheory

oftherateofreturnacrossdifferentclassesoffinancialassets.Inparticular,

weareinterestedinthemean,covariability,andpredictabilityofthereturn

offinancialassets.Atthemacrolevel,westudytheshort-termrisk-free

rate,thetermpremiumoflong-termbondsovertherisk-freerate,andthe

aggregateequitypremiumofthestockmarketovertherisk-freerate.At

themicrolevel,westudythepremiumofindividualstockreturnsandof

classesofstocks,suchasthesmall-capitalizationversuslarge-capitalization

stocks,the“value”versus“growth”stocks,andthepastlosingversuswinning

stocks.

Theneoclassicalrationaleconomicmodelisaunifiedmodelthatviews

thesepremiaastherewardtorisk-averseinvestorsthatprocessinformation

rationallyandhaveunambiguouslydefinedpreferencesoverconsumptionthat

typically~butnotnecessarily!

belongtothevonNeumann–Morgensternclass.

Naturally,thetheoryallowsformarketincompleteness,marketimperfections,

informationalasymmetries,andlearning.Thetheoryalsoallowsfor

differencesamongassetsforliquidity,transactioncosts,taxstatus,andother

institutionalfactors.

Thecauseofmuchanxietyoverthelastquarterofacenturyisevidence

interpretedasfailureoftherationaleconomicparadigmtoexplaintheprice

levelandtherateofreturnoffinancialassetsbothatthemacroandmicro

*UniversityofChicagoandNBER.IthankJohnCampbell,GeneFama,ChrisGeczy,Lars

Hansen,JohnHeaton,RajnishMehra,L’ubosˇPástor,DickThaler,andparticularlyAlonBrav

andJohnCochrane,fortheirinsightfulcommentsandconstructivecriticism.Finally,Ithank

LiorMenzlyforhisexcellentresearchassistanceandinsightfulcommentsthroughoutthisproject.

Naturally,Iremainresponsibleforerrors.

1567

TheJournalofFinance

levels.Acelebratedexampleofsuchevidence,althoughbynomeanstheonly

one,isthefailureoftherepresentative-agentrationaleconomicparadigmto

accountforthelargeaveragepremiumoftheaggregatereturnofstocksover

short-termbondsandthesmallaveragereturnofshort-termbondsfromthe

lastquarterofthe19thcenturytothepresent.Dubbedthe“EquityPremium

Puzzle”byMehraandPrescott~1985!

ithasgeneratedacottageindustry

ofrationalandbehavioralexplanationsofthelevelofassetprices

andtheirrateofreturn.

Anotherexampleisthelargeincreaseinstockpricesintheearlyand

middle1990s,whichFederalReserveChairmanAlanGreenspandecriedas

“IrrationalExuberance”evenbeforetheunprecedentedfurtherincreasein

stockpricesandprice–dividendratiosinthelate1990s.

Myobjectiveistorevisitsomeofthisevidenceandexploretheextenttowhich

therationaleconomicparadigmexplainsthepricelevelandtherateofreturn

offinancialassetsoverthepast100.

years,bothatthemacroandmicrolevels.

InSectionI,Ireexaminethestatisticalevidenceonthesizeoftheunconditional

meanoftheaggregateequityreturnandpremium.First,Idrawa

sharpdistinctionbetweenconditional,short-termforecastsofthemeanequity

returnandpremiumandestimatesoftheunconditionalmean.Iarguethat

thecurrentlylowconditionalshort-termforecastsofthereturnandpremium

donotlessentheburdenoneconomictheorytoexplainthelargeunconditional

meanequityreturnandpremium,asmeasuredbytheirsample

averageoverthepast130years.Second,Iarguethateventhoughonemay

introduceone’sownstrongpriorbeliefsandadjustdownwardsthesample-

averageestimateofthepremium,theunconditionalmeanequitypremiumis

atleast6percentperyearandtheannualSharperatioisatleast32percent.

Thesenumbersarelargeandcallforaneconomicexplanation.

InSectionII,Idiscusslimitationsofthecurrenttheorytoexplainempirical

regularities.Iarguethatpercapitaconsumptiongrowthcovariestoo

littlewiththereturnofmostclassesoffinancialassetsandthisimpliesthat

theobservedaggregateequityreturn,thelong-termbondreturn,andthe

observedreturnsofvarioussubclassesoffinancialassetsaretoolarge,too

variable,andtoopredictable.

Intheremainingsections,Irevisitandexaminetheextenttowhichwe

canexplaintheassetreturnsbyrelaxingtheassumptionsofcompleteconsumption

insurance,perfectmarkets,andtime-separablepreferences.Asthe

readerwillreadilyobserve—andIoffermyapologies—mychoiceofissuesis

eclecticandmirrorsinpartmyownresearchinterests.

InSectionIII,Ishowthatidiosyncraticincomeshocksconcentratedin

periodsofeconomicrecessionplayakeyroleingeneratingthemeanequity

premium,thelowrisk-freerate,andthepredictabilityofreturns.Iargue

thatinsufficientattentionhasbeenpaidtothefactthattheannualaggregate

laborincomeexceedsannualdividendsbyafactorofover20.Laborincome

isbyfarthesinglemostimportantsourceofhouseholdsavingsandconsumption.

Theshockstolaborincomeareuninsurableandpersistentand

arrivewithgreaterfrequencyduringeconomiccontractions.Idiosyncratic

RationalAssetPrices

incomeshocksgoalongwaytowardexplainingtheunconditionalmoments

ofassetreturnsandthepredictabilityofreturns.Theconstructofpercapita

consumptionislargelyirrelevantinexplainingthebehaviorofassetreturns

becauseidiosyncraticincomeshocksareaveragedoutinpercapitaconsumption.

InSectionIV,Ishowthatborrowingconstraintsoverthelifecycleplayan

importantroleinsimultaneouslyaddressingtheaboveissuesandthedemand

forbonds.Iarguethatinsufficientattentionhasbeenpaidtothe

consumers’lifecycleconsumptionandsavingsdecisionsinamarketwith

borrowingconstraints.Theseconsiderationsareimportantinaddressingthe

limitedparticipationofconsumersinthecapitalmarkets,theirrelevance

oftheconstructofpercapitaconsumption,andthedemandforshort-term

bondsbyconsumerswithmoderateriskaversion,giventhatequitiesearn

onaveragealargepremiumovershort-termbonds.

InSectionV,Idiscusstheroleoflimitedmarketparticipation.InSection

VI,Idiscusstheroleofhabitpersistenceinaddressingthesameclass

ofissues.InSectionVII,Iconcludethattheobservedassetreturnsdonot

supportthecaseforabandoningtherationaleconomictheoryasournull

hypothesis.Muchmoreremainstobedonetofullyexploittheramifications

oftherationalasset-pricingparadigm.

I.HowLargeIstheEquityPremium?

TheaveragepremiumofthearithmeticrateofreturnoftheS&PComposite

Indexovertherisk-freerate,measuredoverthelast130years,is

almost7percentandtheannualSharperatiois36percent.Iftheequity

premiumisastationaryprocess,thentheaveragepremiumisanunbiased

estimateoftheunconditionalmeanequitypremium.Onemayintroduce

one’sownpriorbeliefsandshaveabout1percentoffthepremium.Thepremium

andtheSharperatioarestilllargeandchallengeeconomictheoryfor

anexplanation.

InTableI,Ireportthesamplemeanoftheannualarithmeticaggregate

equityreturnandoftheequitypremium.Iproxytheaggregateequityreturn

withtheS&PCompositeIndexreturn.Iproxytheannualrisk-freerate

withtherolled-overreturnonthree-monthTreasurybillsandcertificates.

ThereportedrealreturnisCPI-adjustedforinflation.Overtheperiod1872

to2000,thesamplemeanoftherealequityreturnis8.9percentandofthe

premiumis6.9percent.Overtheperiod1926to2000,thesamplemeanof

theequityreturnis9.7percentandthatofthepremiumis9.3percent.Over

thepostwarperiod1951to2000,thesamplemeanoftheequityreturnis

9.9percentandthatofthepremiumis8.7percent.Thesesamplemeansare

large.Siegel~1998,1999!

IbbotsonAssociates~2001!

IbbotsonandChen

~2001!

Dimson,Marsh,andStaunton~2002!

FamaandFrench~2002!

Mehra

andPrescott~2002!

andseveralothersreportthesamplemeansoftheequity

returnandpremiumintheUnitedStatesandothercountriesandconclude

thattheyarelarge.Somedifferencesarisebasedontheproxyusedforthe

risk-freerate.

TheJournalofFinance

TableI

TheEquityReturnandPremium

Thistableshowsthesamplemeanandstandarddeviationoftheannualizedrealarithmetic

returnontheS&PCompositeIndextotalreturnseries,thesamplemeanoftherealrisk-free

rate,andthesamplemeanoftheequitypremium.Thearithmeticrateofreturnonequityfrom

thebeginningtotheendofyeartisdefinedasRt_1.

~Pt_1.

Dt_1.

Pt!

0Pt,wherePtisthereal

priceoftheaggregateequityatthebeginningofyeartandDt_1istheaggregate