控制权私人收益：一个国际比较

THEJOURNALOFFINANCE•VOL.LIX,NO.2•APRIL2004PrivateBenefitsofControl:AnInternationalComparisonALEXANDERDYCKandLUIGIZINGALES∗ABSTRACTWeestimateprivatebenefitsofcontrolin39countriesusing393controllingblockssales.Onaveragethevalueofcontrolis14percent,butinsomecountriescanbeaslowas–4percent,inothersashigha+65percent.Aspredictedbytheory,higherprivatebenefitsofcontrolareassociatedwithlessdevelopedcapitalmarkets,moreconcentratedownership,andmoreprivatelynegotiatedprivatizations.Wealsoana-lyzewhatinstitutionsaremostimportantincurbingprivatebenefits.Wefindevidenceforbothlegalandextra-legalmechanisms.Inamultivariateanalysis,however,mediapressureandtaxenforcementseemtobethedominatingfactors.THEBENEFITSOFCONTROLOVERcorporateresourcesplayacentralroleinmod-ernthinkingaboutfinanceandcorporategovernance.Fromamodelingdevice(GrossmanandHart(1980))theideaofprivatebenefitsofcontrolhasbecomeacenterpieceoftherecentliteratureincorporatefinance,boththeoreticalandempirical.Infact,themainfocusoftheliteratureoninvestorprotectionanditsroleinthedevelopmentoffinancialmarkets(LaPorta,Lopez-de-Salines,andShleifer(2000))isontheamountofprivatebenefitsthatcontrollingsharehold-ersextractfromcompaniestheyrun.Inspiteoftheimportanceofthisconcept,thereareremarkablyfewestimatesofhowbigtheseprivatebenefitsare,evenfewerattemptstodocumentempir-icallywhatdeterminestheirsize,andnodirectevidenceoftheirimpactonfinancialdevelopment.Alloftheevidenceonthislatterpointisindirect,basedonthe(reasonable)assumptionthatbetterprotectionofminoritysharehold-ersiscorrelatedwithhigherfinancialdevelopmentviaitscurbingofprivatebenefitsofcontrol(LaPortaetal.(1997)).Thelackofevidenceisnoaccident.Bytheirverynature,privatebenefitsofcontrolaredifficulttoobserveandevenmoredifficulttoquantifyinareliable∗DyckisfromtheHarvardBusinessSchoolandZingalesisfromtheUniversityofChicago.ChrisAllen,MehmetBeceren,andOmarChoudhryprovidedinvaluableresearchassistanceinpreparingthedata.WethankAndrewKarolyi,JohnMatsusaka,DavidMoss,TatianaNenova,KrishnaPalepu,MarkRoe,JulioRotemberg,AbbieSmith,DeboraSpar,PerStromberg,ReneStulz,ananonymousreferee,RichardGreen(theeditor),andseminarparticipantsfromGeorgetownUniversity,HarvardBusinessSchool,theNBERcorporatefinanceprogram,UniversityofChicago,theUniversityofPennsylvania(Wharton),andtheUniversityofSouthernCalifornia,andtheUniversityofTorontoforhelpfulcomments.WealsogratefullyacknowledgefinancialsupportfromtheDivisionofResearch,HarvardBusinessSchool,theCenterforResearchonSecurityPrices,andtheGeorgeStiglerCenterattheUniversityofChicago.Anyerrorsareourown.537538TheJournalofFinanceway.Acontrollingpartycanappropriatevalueforhimselfonlywhenthisvalueisnotverifiable(i.e.,provableincourt).Ifitwere,itwouldberelativelyeasyfornoncontrollingshareholderstostophimfromappropriatingit.Thus,privatebenefitsofcontrolareintrinsicallydifficulttomeasure.Twomethodshavebeenusedinattemptingtoquantifythem.Thefirstone,pioneeredbyBarclayandHolderness(1989),focusesonprivatelynegotiatedtransfersofcontrollingblocksinpubliclytradedcompanies.Thepricepershareanacquirerpaysforthecontrollingblockreflectsthecashflowbenefitsfromhisfractionalownershipandtheprivatebenefitsstemmingfromhiscontrollingpositioninthefirm.Bycontrast,themarketpriceofashareafterthechangeincontrolisannouncedreflectsonlythecashflowbenefitsnoncontrollingshare-holdersexpecttoreceiveunderthenewmanagement.Hence,asBarclayandHoldernesshaveargued,thedifferencebetweenthepricepersharepaidbytheacquiringpartyandthepricepershareprevailingonthemarketreflectsthedifferentialpayoffaccruingtothecontrollingshareholder.Infact,afteranadjustment,thisdifferencecanbeusedasameasureoftheprivatebenefitsofcontrolaccruingtothecontrollingshareholder.Thesecondmethodreliesontheexistenceofcompanieswithmultipleclassesofstockwithdifferentialvotingrights.Inthiscase,onecaneasilycomputethemarketvalueofavote(Lease,McConnell,andMikkelson(1983,1984),DeAn-geloandDeAngelo(1985),Rydqvist(1987)).Onanormaltradingdaymarkettransactionstakeplacebetweennoncontrollingpartieswhowillneverhavedirectaccesstotheprivatebenefitsofcontrol.Hence,themarketvalueofavotereflectstheexpectedpriceagenericshareholderwillreceiveincaseofacontrolcontest.Thisinturnisrelatedtothemagnitudeoftheprivateben-efitsofcontrol.Thus,ifoneiswillingtomakesomeassumptionsontheprob-abilityacontrolcontestwillarise,thepriceofavotingrightcanbeusedtoestimatethemagnitudeoftheprivatebenefitsofcontrol(Zingales(1994,1995a)).InthispaperweusetheBarclayandHolderness(1989)methodtoinferthevalueofprivatebenefitsofcontrolinalarge(39)crosssectionofcountries.Basedon393controltransactionsbetween1990and2000wefindthatonaveragecorporatecontrolisworth14percentoftheequityvalueofafirm,rangingfroma–4percentinJapantoa+65percentinBrazil.Interestingly,thepremiumpaidforcontrolishigherwhenthebuyercomesfromacountrythatprotectsinvestorsless(andthusismorewillingorabletoextractprivatebenefits).Thisandotherevidencesuggestthatourestimatescapturetheeffecttheinstitutionalenvironmenthasonprivatebenefitsofcontrol.Giventhelargenumberoftransactionsfr