希勒人类行为和金融系统的效率

*ThispaperwaspreparedforJohnB.TaylorandMichaelWoodford,Editors,HandbookofMacroeconomics.Anearlierversionwaspresentedataconference“RecentDevelopmentsinMacro-economics”attheFederalReserveBankofNewYork,February27–8,1997.TheauthorisindebtedtoRickyLamforresearchassistance,andtoMichaelKrause,VirginiaShiller,AndreiShleifer,DavidWilcox,andtheeditorsforhelpfulcomments.ThisresearchwassupportedbytheNationalScienceFoundation.1TheRoberts(1967)paperhasneverbeenpublished;thefameofhispaperapparentlyowestothediscussionofitinFama(1970).1HumanBehaviorandtheEfficiencyoftheFinancialSystembyRobertJ.Shiller*AbstractRecentliteratureinempiricalfinanceissurveyedinitsrelationtounderlyingbehavioralprinciples,principleswhichcomeprimarilyfrompsychology,sociologyandanthropology.Thebehavioralprinciplesdiscussedare:prospecttheory,regretandcognitivedissonance,anchoring,mentalcompartments,overconfidence,over-andunderreaction,repre-sentativenessheuristic,thedisjunctioneffect,gamblingbehaviorandspeculation,perceivedirrelevanceofhistory,magicalthinking,quasi-magicalthinking,attentionanomalies,theavailabilityheuristic,cultureandsocialcontagion,andglobalculture.Theoriesofhumanbehaviorfrompsychology,sociology,andanthropologyhavehelpedmotivatemuchrecentempiricalresearchonthebehavioroffinancialmarkets.InthispaperIwillsurveybothsomeofthemostsignificanttheories(forempiricalfinance)intheseothersocialsciencesandtheempiricalfinanceliteratureitself.Particularattentionwillbepaidtotheimplicationsofthesetheoriesfortheefficientmarketshypothesisinfinance.Thisisthehypothesisthatfinancialpricesefficientlyincorporateallpublicinformationandthatpricescanberegardedasoptimalestimatesoftrueinvestmentvalueatalltimes.Theefficientmarketshypothesisinturnisbasedonmoreprimitivenotionsthatpeoplebehaverationally,oraccuratelymaximizeexpectedutility,andareabletoprocessallavailableinformation.Theideabehindtheterm“efficientmarketshypothesis,”atermcoinedbyHarryRoberts(1967),1hasalonghistoryinfinancialresearch,afarlongerhistorythanthetermitselfhas.Thehypothesis(withoutthewords2efficientmarkets)wasgivenaclearstatementinGibson(1889),andhasapparentlybeenwidelyknownatleastsincethen,ifnotlongbefore.Allthistimetherehasalsobeentensionoverthehypothesis,afeelingamongmanythatthereissomethingegregiouslywrongwithit;foranearlyexample,seeMacKay(1841).Inthepastcoupleofdecadesthefinanceliterature,hasamassedasubstantialnumberofobservationsofapparentanomalies(fromthestandpointoftheefficientmarketshypothesis)infinancialmarkets.Theseanomaliessuggestthattheunderlyingprinciplesofrationalbehaviorunderlyingtheefficientmarketshypothesisarenotentirelycorrectandthatweneedtolookaswellatothermodelsofhumanbehavior,ashavebeenstudiedintheothersocialsciences.Theorganizationofthispaperisdifferentfromthatofotheraccountsoftheliteratureonbehavioralfinance(forexample,DeBondtandThaler,1996orFama,1997):thispaperisorganizedaroundalistoftheoriesfromtheothersocialsciencesthatareusedbyresearchersinfinance,ratherthanaroundalistofanomalies.Iorganizedthepaperthiswaybecause,inreality,mostofthefundamentalprinciplesthatwewanttostressherereallydoseemtobeimportedfromtheothersocialsciences.Nosurprisehere:researchersintheseothersocialscienceshavedonemostoftheworkoverthelastcenturyonunderstandingless-than-perfectly-rationalhumanbehavior.Moreover,eachanomalyinfinancetypicallyhasmorethanonepossibleexplanationintermsofthesetheoriesfromtheothersocialsciences.Theanomaliesareobservedincomplexrealworldsettings,wheremanypossiblefactorsareatwork,notintheexperimentalpsychologist’slaboratory.Eachoftheirtheoriescontributesalittletoourunderstandingoftheanomalies,andthereistypicallynowaytoquantifyorprovetherelevanceofanyonetheory.Itisbettertosetforththetheoriesfromtheothersocialsciencesthemselves,describingwhenpossiblethecontrolledexperimentsthatdemonstratetheirvalidity,andgiveforeachafewillustrationsofapplicationsinfinance.Beforebeginning,itshouldbenotedthattheoriesofhumanbehaviorfromtheseothersocialsciencesoftenhaveunderlyingmotivationthatisdifferentfromthatofeconomictheories.Theirtheoriesareoftenintendedtoberobusttoapplicationinavarietyofeveryday,unstructuredexperiences,whiletheeconomictheoriesareoftenintendedtoberobustinthedifferentsensethat,eveniftheproblemstheeconomicagentsfacebecomeveryclearlydefined,theirbehaviorwillnotchangeaftertheylearnhowtosolvetheproblems.Manyoftheunderlyingbehavioralprinciplesfrompsychologyandothersocialsciencesthatarediscussedbelowareunstableandthehypothesizedbehavioralphenomenamaydisappearwhenthesituationbecomesbetterstructuredandpeoplehavehadalotofopportunitytolearnaboutit.Indeed,therearepapersinthepsychologyliteratureclaimingthatmanyofthecognitivebiasesinhumanjudgmentunderuncertaintyuncoveredbyexperimentalpsychologistswilldisappearwhentheexperimentischangedsothattheprobabilitiesandissuesthattheexperimentraisesareexplainedclearlyenoughtosubjects(see,forexample,Gigerenzer,1991).Experimentalsubjectscaninmanycasesbecon-vinced,ifgivenproperinstruction,thattheirinitialbehaviorintheexperimentalsituationwasirrational,andth