Investor-Sentiment-and-the-Closed-End-Fund-Puzzle

InvestorSentimentandtheClosed-EndFundPuzzleCharlesM.C.Lee;AndreiShleifer;RichardH.ThalerTheJournalofFinance,Vol.46,No.1.(Mar.,1991),pp.75-109.StableURL:=0022-1082%28199103%2946%3A1%3C75%3AISATCF%3E2.0.CO%3B2-OTheJournalofFinanceiscurrentlypublishedbyAmericanFinanceAssociation.YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://@jstor.org.:422007THEJOURNALOFFINANCE.VOL.XLVI,NO1.MARCH1991InvestorSentimentandtheClosed-EndFundPuzzleCHARLESM.C.LEE,ANDRE1SHLEIFER,andRICHARDH.THALER*ABSTRACTThispaperexaminesthepropositionthatfluctuationsindiscountsofclosed-endfundsaredrivenbychangesinindividualinvestorsentiment.Thetheoryimpliesthatdiscountsonvariousfundsmovetogether,thatnewfundsgetstartedwhenseasonedfundssellatapremiumorasmalldiscount,andthatdiscountsarecorrelatedwithpricesofothersecuritiesaffectedbythesameinvestorsentiment.Theevidencesupportsthesepredictions.Inparticular,wefindthatbothclosed-endfundsandsmallstockstendtobeheldbyindividualinvestors,andthatthediscountsonclosed-endfundsnarrowwhensmallstocksdowell.FEWPROBLEMSINFINANCEareasperplexingastheclosed-endfundpuzzle.Aclosed-endfund,likethemorepopularopen-endfund,isamutualfundwhichtypicallyholdsotherpubliclytradedsecurities.Unlikeanopen-endfund,however,aclosed-endfundissuesafixednumberofsharesthataretradedonthestockmarket.Toliquidateaholdinginafund,investorsmustselltheirsharestootherinvestorsratherthanredeemthemwiththefunditselfforthenetassetvalue(NAV)pershareastheywouldwithanopen-endfund.Theclosed-endfundpuzzleistheempiricalfindingthatclosed-endfundsharestypicallysellatpricesnotequaltothepersharemarketvalueofassetsthefundholds.Althoughfundssometimessellatpremiatotheirnetassetvalues,inrecentyearsdiscountsof10to20percenthavebeenthenorm.Severalpaststudieshaveattemptedtosolvethepuzzlebypointingoutthatthemethodsusedtovaluethesecuritiesintheportfoliomightoverstatethetruevalueoftheassets.Threefactorsareoftencitedaspotentialexplanations:agencycosts,taxliabilities,andilliquidityofassets.Theagencycoststheorystatesthatmanagementexpensesincurredinrunningthefundaretoohighand/orthepotentialforsubparmanagerialperfor-mancereducesassetvalue.Thetaxexplanationarguesthatcapitalgainstax*LeeisAssistantProfessorofAccountingattheSchoolofBusiness,UniversityofMichigan.ShleiferisProfessorofEconomicsatHarvardUniversity.ThaleristheHenriettaJohnsonLouisProfessorofEconomicsattheJohnsonGraduateSchoolofManagement,CornellUniversity.WewouldliketoacknowledgehelpfulcommentsfromGregBrauer,EugeneFama,KenFrench,RaymondKan,MertonH.Miller,SamPeltzman,MarkReady,SySmidt,Ren6Stulz,LawrenceSummers,RobertVishny,RobertWaldmann,andananonymousreferee,andfinancialsupportfromRussellSageFoundationandSocialSciencesandHumanitiesResearchCouncilofCanada.ThanksalsotoGregBrauer,Nai-FuChen,ThomasHerzfeld,andJayRitterforprovidingdata,andSheldonGaoandErikHerzfeldforresearchassistance.7576TheJournalofFinanceliabilitiesonunrealizedappreciations(atthefundlevel)arenotcapturedbythestandardcalculationofNAV.Finally,becausesomefundsholdrestrictedorlettersecuritieswhichhavetradingrestrictions,theargumenthasbeenmadethatsuchassetsareovervaluedinthecalculationofNAV.Whileeachoftheseexplanationsislogicalandmayexplainsomeportionoftheobserveddiscounts,weshowbelowthatevencollectivelythesefactorsfailtoaccountformuchoftheexistingevidence.Ourprimarypurposeistoevaluateempiricallyanalternativeexplanationfortheclosed-endfundpuzzlepresentedbyZweig(1973)andDelong,Shleifer,Summers,andWaldmann(1990)(DSSW).Zweig(1973)suggeststhatdis-countsonclosed-endfundsreflectexpectationsofindividualinvestors.DSSWdevelopamodelinwhichrationalinvestorsinteractinfinancialmarketswithnoisetraderswhoarelessthanfullyrational.Animportantfeatureoftheirmodelistheexistenceofunpredictablefluctuationsinnoisetradersentiment,definedasthecomponentofexpectationsaboutassetreturnsnotwarrantedbyfundamentals.Investorsentimentcanrepresenttradingonnoiseratherthannews(Black(1986))ortradingonpopularmodels(Shiller(1984)).Inthecaseofclosed-endfunds,fluctuationsininvestorsentimentcanleadtofluctuationsindemandforclosed-endfundshareswhicharereflectedinchangesindiscounts.InadditiontoZweig'searlyideathatfunddiscountsreflectinvestorsentiment,theDSSWmodelexplainswhyfundscansellatdiscountsevenifinvestorsarenot,onaverag