A Quantitative Analysis of Hedge Fund Style and Pe

AQuantitativeAnalysisofHedgeFundStyleandPerformanceClausBangChristiansenPeterBrinkMadsenMichaelChristensen*AssociateProfessorsAarhusSchoolofBusinessDepartmentofFinanceFuglesangsAllé4DK-8210AarhusV.DenmarkAbstractInthisanalysisweidentifydynamichedgefundstrategiesquantitativelypursuingaPrincipalComponentAnalysisfollowingFung&Hsieh(1997).Weextractfivedominanthedgefundstrategieseachrepresentingsimilarinvestmentstylesandana-lysetheperformanceofeachstrategybyemployingamulti-factormodelcomprisingbothmarketindicesandpassiveoptionstrategiesalongthelinesofAgerwal&Naik(2000).Wefindthatthatsuchpassiveoptionstrategiesplayanimportantroleinex-plaininghedgefundreturns.Moreoverweshowthatthemajorityofthefiveho-mogenousstrategiesshowneutralperformance,butthisresultturnsouttobesensi-tivetoanypotentialsurvivorshipbiases.Keywords:Hedgefunds,dynamicstrategies,performance.JEL:G20,G23ToappearinBarrySchachter(ed.)(2004):IntelligentHedgeFundInvesting,RiskBooks.*Correspondingauthor:Phone:[phone],E-mail:[e-mail]collapseofLong-TermCapitalManagementin1998anincreasingfocushasbeendirectedtotheperformanceofhedgefunds.Inparticular,recentresearchhasbeencon-centratedonidentifyingdynamictradingstrategiesinordertomimictheactualtradesofhedgefunds.Contrarytomutualfunds,performanceanalysesofhedgefundsarequitediffer-entbecausethereturnsofhedgefundsusuallyhavelowcorrelationswithmarketindices,andthereforeatraditionalCAPManalysisusingtheJensenmeasureistypicallynotappropriate.Moreover,hedgefundspursuemanydifferentstyles,seeBrown,Goetzmann&Ibbotson(1999),andasshownbyFung&Hsieh(1997)thesehedgefundstylesoftengenerateoption-likereturns1.WhileFung&Hsieh(1997)wereprimarilyinterestedinidentifyingthehedgefundsstyles,thestudybyBrownetal.(1999)wasmainlyconcernedwiththeperformanceofhedgefundsincludingperformancepersistence.Theresearchintotheperformanceofhedgefundsis,how-ever,sparsecomparedtotheperformanceliteratureonmutualfundsandpensionfundsandatpresentthereisnogeneralconclusionastowhetherhedgefundsareabletogeneratepositiverisk-adjustedreturns.Brownetal.(1999)analysetheperformanceof399offshorehedgefundsfortheperiod1989to1995.TheyconcludethathedgefundsforthisperiodhavebeenabletooutperformtheS&P500indexintermsofhigherSharperatiosandpositiveJensenalphas.Furthermore,theyconfirmtheirconclusionusingvariousself-determinedbenchmarksbasedonindustryclassi-fications.Ackermann,McEnally&Ravenscraft(1999)analyseasampleof547hedgefundsusingmonthlyobservationsfortheperiod1988to1995,buttheyarenotabletoconfirmtheBrownetal.(1999)evidencethathedgefundshaveoutperformedstandardmarketindices.However,comparinghedgefundsandmutualfundsAckermannetal.findthathedgefundshavebeenabletosystematicallyoutperformmutualfunds.Liang(1999)formulatesamulti-factormodelcomprising8differentfactorsthatrepresentvariousmarketindiceswithinequities,bonds,exchangerates,commoditiesandthemoneymarket.Heanalyses385fundsfortheperiod1994to1996andfindsthatoutof16differentstrategyclassifications,7hedgefundgroupsshowasignificantlypositiveperformance,3whereas2hedgefundsshowasignificantlynegativeperformance,thusleavingtheremaining7fundswithaneutralperformance.Further,LiangconfirmstheAckermannetal.(1999)findingthatonaveragehedgefundsprovidebetterrisk-to-rewardratiosthanmutualfundsintermsofSharperatios.InLiang(2001)thesampleisextendedtocoveralongerperiodfrom1990to1999,andLiangconfirmsthathedgefundsonaverageprovidehigherSharperatios(0.41)thantheS&P500index(0.27)duringthisperiod.Similarly,Edwards&Caglayan(2001a)estimateamulti-factormodelcomprising6factors,andestimatethemodelforeightdifferentstrategyclassificationsforatotalsampleof836hedgefundsfortheperiod1990to1998.Theyfindthat25%ofthefundshaveyieldedsig-nificantlypositiveJensenmeasures.AsimilarconclusionisreachedbyCapocci&Hubner(2003),whoanalysetheperformanceofnearly2,800hedgefundsforaperiodaslongas1984to2000.Theyfindthatapproxi-mately25%ofthefundshaveobtainedsignificantlypositiveJensenalphasusingamulti-factormodelcomprising11differentfactors.However,thestudiesbyBrownetal.(1999),Ackermannetal.(1999),Liang(1999,2001),Edwards&Caglayan(2001a)andCapocci&Hubner(2003)donotexplicitlyconsiderdy-namictradingstrategiesintermsofoptions.Agerwal&Naik(2000)defineanumberofdif-ferentpassiveoptionstrategiesbasedonstandardmarketindicesandincludetheseoptionstrategiesaswellasthepassivemarketindicesintoamulti-factorstylemodelalongthelinesofSharpe(1992).Studyingtheperiod1990to1998for586hedgefundstheyareabletocon-firmthattheoptionstrategiesplayasignificantroleindeterminingthefundreturns.Thein-creaseinthegoodness-of-fitisremarkable,whenpassiveoptionstrategiesareincluded,andbasedontheextendedmulti-factormodeltheyconcludethathedgefundsgenerallyhavenotbeenabletogenerateabnormalreturns.Only35%ofthefundsanalysedobtainsignificantlypositiveJensenmeasures,while13%ofthefundsyieldsignificantlynegativeJensenmeas-ures.AlsoFung&Hsieh(2002)incorporateoptionstrategiesintoaSharpestylemodel,buttheyareprimarilyinterestedinexploringtheriskoffixedincomehedgefundstylesanddonotconsiderpe