A Theory of Dividends Based on Tax Clienteles Fran

ATheoryofDividendsBasedonTaxClientelesFranklinAllenAntonioE.BernardoIvoWelchNovember24,1999ABSTRACTThispaperexplainswhysomefirmsprefertopaydividendsratherthanrepurchaseshares.Wheninstitutionalinvestorsarerelativelylesstaxedthanindividualinvestors,dividendsinduce“ownershipclientele”effects.Firmspayingdividendsattractrelativelymoreinstitutions,whichhavearelativeadvantageindetectinghighfirmqualityandinensuringfirmsarewellmanaged.Thetheoryisconsistentwithsomedocumentedregular-ities,specificallyboththepresenceandstickinessofdividends,andoffersnovelempiricalimplications,e.g.,apredictionthatitisthetaxdifferencebetweeninstitutionsandretailinvestorsthatdeterminesdividendpayments,nottheabsolutetaxpayments.AllenisfromtheWhartonSchoolattheUniversityofPennsylvania.BernardoandWelcharefromtheAn-dersonSchoolatUCLA.TheauthorsthankLaurieHodrick,ShlomoBenartzi,CharlesCalomiris,MarkGrinblatt,KathleenFitzgerald,AvrahamKamara,JiangLuo,HamidMehran,AvanidharSubrahmanyam,RichardVinesandChrisHayden(GeorgesonInvestorServices)andseminarparticipantsatNorthwestern,NYU,UC/Davis,Columbia,UC/Riverside,Wharton,Yale,Atlanta,Duke,Chicago,andU.S.C.forhelpfulcomments.Althoughanumberoftheorieshavebeenputforwardintheliteraturetoexplaintheirpervasivepresence,1dividendsremainoneofthethorniestpuzzlesincorporatefinance.Inafrictionlessworldwithouttaxesortransactioncosts,dividendsandsharerepurchasesareequivalent.Ifdividendsaretaxedmoreheavilythancapitalgains,asisthecaseintheU.S.andmanyothercountries,sharerepurchasesareapparentlysuperiortodividends.Neverthe-less,dividendscontinuetobeasubstantialproportionofearnings—andpersonaldividendtaxescontinuetobeasubstantialsourceofincomefortheI.R.S.Forthe1973to1983period,dividendsforthelargest1,000firmsintheU.S.averaged44percentofearningswhilerepur-chasesaveragedonly6percent(seeAllenandMichaely(1995)).Although,asBagwellandShoven(1989)havestressed,repurchasesincreasedsignificantlyin1984andhaveremainedhigh,repurchaseswerenotasubstitutefordividends.From1984to1988,repurchasesin-creasedfrom6percentto38percentofearnings,butdividendsstillincreasedfrom44percentto51percent.TheI.R.S.StatisticsofIncomepublicationdocumentsthattaxabledividendsinadjustedgrossincomeamountedto$82billioninabout25millionindividuals’taxreturnsin1994.2Asecondpuzzlingfactaboutdividendsconcernstheinclinationoffirmsto“smooth”them,asoriginallydocumentedinLintner(1956).Ourpaperpresentsanintuitiveandnovelexplanationthatisconsistentwiththesetwopuzzles:theinclinationoffirmstopaydividendsratherthantorepurchaseshares,andtheinclinationoffirmstosmoothdividends.Ourhypothesisisbasedontwoassumptions.Ourfirstassumptionisthattherearegroupsofinvestorswhoaretaxeddifferentlyandwhohavedifferentincentivestobecomeinformedaboutcorporateaffairs.IntheUnitedStates,forexample,publicandcorporatepensionfunds,collegesanduniversities,laborunions,foun-dations,andothercorporationsareeitherfullyorlargelyexemptfromtaxes.Theproportionofstocksownedbythesegroupshasbeensignificantformanyyearsandhasincreasedsub-stantiallysince1980(seeAllenandSantomero(1998)).Forsimplicity,weassumetherearejusttwoclientelesandcallthem“untaxedinstitutions”and“taxedindividuals.”Becauseoftheirscale,the(untaxed)institutionshavegreaterincentivestobecomeinformedaboutthe1firm.Institutionsaremorelikelythanretailinvestorstoconduct“duediligence”tofindoutwhetheraparticularfirmiswellrunorpoorlymanaged.Theyarealsomorelikelytofacilitatemechanismsbywhichpotentialshortcomingsarecorrected.Suchinstitutionalactivitiescanbeintrinsic,suchastheirabilitytomakeaquicksaleofalargeblockofsharestoapotentialraider.Buttheycanalsobeproactive.Inrecentyears,institutionshavebecomeincreasinglyinvolvedincorporategovernance.Theirabilitytovoteinlargeblocksandinfluencecorporateleadershashadaninfluenceonsuchcorporatechoicesaschangesinthecorporatecharterandboardcompositionandmembership.(SectionIV.Cjustifiesthisassumptioninmoredetail.)Ineconomicterms,thepresenceofinstitutionalshareholdingscanbeassociatedwithhigherfirmvaluebecauseofsignalingeffects,agencyeffects,orboth.InSectionI,weassumethatbettermanagerstakeadvantageofthedetectionabilityofinstitutions;inSectionII,weassumethatinstitutionsaddvaluebyexpendingresourcestomonitorthecorporation.Oursecondassumptionisthatdividendsareonewayofattractinginstitutions.Thiscanbejustifiedbydirectlyappealingtocommoninstitutionalcharterandprudentmanrulere-strictionsthatmakeitmoredifficultformanyinstitutionstopurchaseinvestmentswithlowdividendpayouts(BravandHeaton(1998)).Yet,manytax-exemptinstitutionssuchasuniver-sitiesandcharitiesdonothavedirectrestrictionsandstillholdsignificantamountsofdividendpayingstock.Weshowthatthedividendpayingfirms’in-equilibriummarketpricesmakethemarelativelybetterpurchaseforinstitutionsthanforretailinvestors,becausedividendsaretaxedforindividualsbutuntaxedforinstitutions.Thiscomparativeadvantageresultsinanendogenouslyhigherfractionofownershipbyinstitutionsfordividendpayingstocks.3Ourtwoassumptionsimplythatfirmscanattractmoreinstitutionsasshareholdersbypayingdividendsandthatdividendpayingfirmswillperformbetterthan