FRM-Hedge Funds Performance Risk and Capital Forma

HedgeFunds:Performance,RiskandCapitalFormation*WilliamFungLondonBusinessSchoolDavidA.HsiehDukeUniversityNarayanY.NaikLondonBusinessSchoolTarunRamadoraiUniversityofOxfordandCEPRThisversion:March10,2006AbstractWeuseacomprehensivedatasetofFunds-of-Hedge-Funds(FoFs)toinvestigateperformance,riskandcapitalformationinthehedgefundindustryoverthepasttenyears.WeconfirmthefindingofhighsystematicriskexposuresinFoFreturns.Wedivideupthepasttenyearsintothreedistinctsub-periodsanddemonstratethattheaverageFoFhasonlydeliveredalphaintheshortsecondperiodfromOctober1998toMarch2000.Inthecross-sectionofFoFs,however,weareabletoidentifyFoFscapableofdeliveringpersistentalpha.WefindthatthesemoresuccessfulFoFsexperiencefargreater(andsteadier)capitalinflowsthantheirlessfortunatecounterparts.BerkandGreen’s(2004)rationalmodelofactiveportfoliomanagementimpliesthatdiminishingreturnstoscalecombinedwiththeinflowofnewcapitalleadstotheerosionofsuperiorperformanceovertime.Inkeepingwiththisimplication,weprovideevidencethatevensuccessfulFoFshaveexperiencedarecent,dramaticdeclineinrisk-adjustedperformance.JELClassifications:G11,G12,G23Keywords:hedgefunds,funds-of-hedge-funds,performance,alpha,factormodels,flow*WethankOmerSulemanandespeciallyAasmundHeenforexcellentresearchassistance.WethankJohnCampbell,HarryMarkowitz,LudovicPhalippouandseminarparticipantsattheLondonSchoolofEconomics,UniversityofOxford,StockholmInstituteforFinancialResearch,WarwickUniversity,theIXIS-NYUhedgefundconferenceandU.MassAmherstforcomments.WegratefullyacknowledgesupportfromtheBSIGammaFoundationandfromtheBNPParibasHedgeFundCentreattheLondonBusinessSchool.WilliamFung:[e-mail];DavidHsieh:[e-mail];NarayanNaik:[e-mail];TarunRamadorai:[e-mail]1Hedgefundsareconsideredbysometobetheepitomeofactivemanagement.Theyarelightlyregulatedinvestmentvehicleswithgreattradingflexibility,andtheyoftenpursuehighlysophisticatedinvestmentstrategies.Hedgefundspromise‘absolutereturns’totheirinvestors,leadingtoabeliefthattheyholdfactor-neutralportfolios.Theyhavegrowninsizenoticeablyoverthepastdecadeandhavebeenreceivingincreasingportfolioallocationsfrominstitutionalinvestors.1Accordingtopressreports,anumberofhedgefundmanagershavebeenenjoyingcompensationthatiswellinexcessofU.S.$10millionperannum.Howmuchofthehypeistrue?Weareawarefromavarietyofpastresearchpapersthattherearerisksinherentinhedgefundreturns.2However,inthecourseofactiveportfoliomanagement,riskexposuresareboundtochangewithmarketconditions.Canwecapturethetimevariationintheserisks?Canwecharacterizeinterestingdifferencesinrisk-adjustedperformanceoralphainthecross-sectionofhedgefunds?Canweusethesedifferencestopredictthefutureperformanceofcertainhedgefunds?Howhavecapitalflowsrespondedtocross-sectionalandtime-seriesmovementsinrisk-adjustedhedgefundperformance?Doestheentryofsophisticatedinvestorsheraldareductionintheperformanceofhedgefunds?Inthispaper,weinvestigatetheseimportantquestions.Indoingso,weprovideabodyofevidencethatstronglysupportsBerkandGreen’s(2004)modelofactiveportfoliomanagement.BerkandGreen’smodelhasthreekeyfeatures.First:investorscompetitivelyprovidecapitaltofunds.Second,managershavedifferentialabilitytogeneratehighreturns,butfacedecreasingreturnstoscaleindeployingtheirability.Third,investorslearnaboutmanagerialabilityfrompastperformanceanddirectmorecapitaltowardsfundswithsuperiorperformance.Thesefeatures,wedemonstrate,areevidentinthehedgefundindustry,providinguswithaninterestinglaboratorytotestthepredictionsoftheirrationalmodelofactivemanagement.1AccordingtotheTASSAssetFlowsreport,aggregatehedgefundassetsundermanagementhavegrownfromU.S.$72billionattheendof1994toover$670billionattheendof2004.2SeeAgarwalandNaik(2005)foracomprehensivesurveyofthehedgefundliterature.2TherearetwomainimplicationsoftheBerkandGreenmodel.Weprovideevidencethattheseimplicationsareindeedtrueinthecaseofhedgefunds.First,weshowthatmoresuccessfulfundsexperiencefargreatercapitalinflowsthantheirlessfortunatecounterparts.Asecondimplicationoftheirmodelisthatdiminishingreturnstoscalecombinedwiththeinflowofnewcapitalleadstoerosionofsuperiorperformanceovertime.Inkeepingwiththisimplication,weprovideevidencethatevensuccessfulfundshaveexperiencedarecent,dramaticdeclineinrisk-adjustedperformance.Beforeonecanusethedataonhedgefundsorhedgefundindexestoexaminetheseissues,onehastominimizethebiasesinthesedata.Thesebiasesarisefromalackofuniformreportingstandardsgiventhelackofregulation.Forexample,hedgefundmanagerscanelectwhethertoreportperformanceatall,andiftheydo,theycandecidethedatabase(s)towhichtheyreport.Theycanalsoelecttostopreportingattheirdiscretion.3Thisbiasesthereturnsreportedbyhedgefundindexesupwards.Furthermore,real-lifeconstraintsmakethehedgefundindexreturnsdifficulttoreplicate.Forexample,somehedgefundsincludedintheindexmaybeclosedtonewmoney,ormayevenbereturningmoney.Inaddition,hedgefundsoftenimposeconstraintsonthewithdrawalofmoney,usinglockupperiods,noticeandredemptionperiods.Moreimportantly,thehedgefundindexreturnsdonotreflect