2010金融專題課件及相關(guān)材料ms9behavioral asset manage

MSClassBehavioralAssetManagementYu-JaneLiu,2010AnExampleofBehavioralFullerFuller&ThalerAssetFuller&ThalerAssetIndependent,RegisteredInvestment100%employeeSharedequityownershipamongkeyExperiencedStrengthsinbothmoneymanagementandacademic$2.04BillionUnder$1.74billioninlong-onlyinstitutional$300millioninlong/shortIntroducestrategiesthatexploitbehavioralF&TInvestmentBenAnnualizedExcessNetofFeesJan-$730$640Small/Mid-$300Micro-Jan-Jul-$20-MarketNeutralSalomonBrothers3-MonthT-BillJan-$210ExpectationsDetermineStockStockpricesarebasedonexpectations-futureChangesinexpectationscausepricestoPeoplemaymakesystematicmentalmistakes--duetobehavioralbiases.BiasedExpectationsCauseStockInvestorscanmis-processnewinformationbyeitherover-reactingorunder-reactingtonewIftheconsensusexpectationregarding’sfutureisbiased,thestockwillbeWhatMakesOurStrategyOurSmall/Mid-CapGrowthstrategycapitalizesonbehavioralbiasesthatcauseunder-reaction.Overconfidence:Investors cetoomuchconfidenceintheirexistinginformationand/orknowledge.Anchoring:InvestorsaretiedtotheirpreviousvieworDispositionEffect:Investorstendtoholdtheirloserstoolongandselltheirwinnerstooearly.AnExampleofGiganticdepartmentManufacturer’ssuggestedprice Ournormaldiscount Today’ssale Startingwithahighpriceand“workingdown”capitalizesonanchoringbias.Thiscommonmarketingtechniquehasbeenusedforcenturies!F&TSmall/Mid-CapGrowthJanuary1,1992August31,IllustrationofaPermanentEarningsIllustrationofaTemporaryEarningstativeScreens(ThisisrelativelyIdentifygoodnews:alargeincreasein ysis(ThisisDeterminewhethertheearningsincreaseispermanentortemporaryinnature ysis(ThisisourcompetitiveDetermineifthemarketisunder-reactingtothesustainableincreaseinthefirm'sprofitability.Step tativeBeginwithaninitialuniverseof3,000-4,000IdentifycompaniesthathaverecentlyannouncedlargeandunexpectedincreasesinearningsIdentifystocksforwhichearningschangesarelikelytoSignificantbefore-andafter-taxearningstativescreensreducethenumberofcompanies y150perStep2: yzethesustainabilityoftheearningsImprovedgrossandoperatingFavorableproductmixandintroductionofnewSubstantialcostyzesignals Post-announcementcommentsonearningsnewproductAnnouncementofdividendincrease,stocksplitorstockInsidertrading ysisfurtherreducesthepoolto40-investmentcandidatesperStep3: Identifycompanieswithasustainableincreaseinearningsthatunrecognizedbysell- IdentifysourcesofdifferentialbetweenestimatesandactualReviewrecenthistoryofearningsestimate ysisfocusesonsigns ystsareunder-reactingthepermanentlyhigherListenfordisbelief Listen ysts'commentsfocusingon 'spastandnot ysisreducesthepoolto15-20companiesPortfolioPortfolioholdings:35-60Typicalholdingperiod:1Typicallynomorethan6%positioninanyonePortfoliosectorweightscontrolledrelativeto Cashposition:Lessthan5%;fullyinvestedmarketPortfolioMarketCapReturnonEquity

$1.6

Russell2500$1.3SectorRussell2500HealthSellSellanyholdingsforwhichtheearningsincreasewasonlytemporarySellafterconsensusStreetestimatesapproximateactualearningsSmall-CapValueStrategyCapitalizesonOverreactionresultsfromtwobiasesinhumanRepresentativeness--peopletendtoinferthatasingleobservationisrepresentativeoftheentireSaliency--Peopletendtoover-estimatetheprobabilityofalowfrequencyeventsuchasanearthquake(orearningsmiss)iftheyhaverecentlyheardthatsuchaneventhasoccurred.F&TSmall-CapValueJanuary1,1996–August31,GeneratesLargeFutureEarningsInvestmentLookforstocksDecliningpricepatternovermanyDiscountvaluationsrelativetoindustrySignalsfromInsiderCorporatestock PortfolioPortfolioholdings:80–90Annualturnover:40–TradingMarketimpact:ECNs,crossingnetworks,andblockTypicallynomorethan4%positioninanyoneSectorweightscontrolledrelativeto Cashposition:TypicallylessthanSector2000ConsumerConsumerHealthInformationmunication----WatchingoutforsystematicTheManyErrorsInvestorsRepeatedlyMakeinofInvestmentandProfitsWithoutPanic:InvestmentPsychology alWealth?JonathanMyersFoolsandPooledTheytradetooTheysell‘winners’tobuyTheychasetheBehavioralTradingValueExtremeEShortSalesWhyshouldtheopportunityexist?(temorWhatmarketfrictionsdoyouexplore?(margin,HowwelldoesthemodelforecastDoesthemodelforecastreturnseconomicallyHowdoesthemodelperformindifferentmarketsituations?(bull,bear)CanweexploititunderrealisticModelModelerrors(sampleestimations,factormodels,time-seriesmodels)Transactioncosts(marketAlpha-GooddataModelerrors-(sampleestimations,factormodels,time-seriesmodels):dynamaticadjustment,classificationsondetailedindustries,non-riskfactorsTransactioncosts(marketmicrostructure)-highfrequencydata,realtime ysis,abilitytoaccessmarketliquidityA.ValueInathreetofive-yearperiod,thereisatendency-forpreviouspoorperformerstobegintodowellandpreviousgoodperformerstobegintoperformlessInotherwords,valuestocksturnintoglamourandglamourstocksturnintovalue過度反

DebondtandThaler(1985,過去的Losers在之後的3-5年，其績效會(huì)比長期的winners Equating‘winners’withglamourstocksand‘losers’withThethirty-five‘loser’stocksintheportfoliosubsequentlyoutperformedthe‘winner’stocks,duringthenextthree-yearperiod,bytwenty-fivepercent.Investorswhospecializeinbuying‘loser’stocksmayEssentially,becausethesearegoodqualitystocksbuttheirpriceisdependentonmisperception,theymoveupanddowninacyclicalmanner.tativeTimingtheGoodnewsforValueBadnewsforGrowthPiotroskiValuestock+-+-Changeincapital↓↑+-EquityDidn’tOperating+-+-動(dòng)能動(dòng)能策

Jegadeeshand股價(jià)很差構(gòu)建投資組合方 構(gòu)建投資組合方構(gòu)造出投資組合后，每月都進(jìn)行相應(yīng) 操作在每個(gè)月t的開位置K個(gè)月。這個(gè)策略停止在第t－k個(gè)月建立的投資組合對于持有策略和每月重新調(diào)節(jié)策略，分別計(jì)算成功者，失敗者的益以及其收益差收益的計(jì)算方法是計(jì)算月份t中平均權(quán)重的從月份t－6到t－1生成的個(gè)分別的勝利者（失敗者）的平 收益六策，投組的成和有之免 差價(jià)壓。return或earningssurprise的策略。對於pricemomentum和earningsmomentum的投資策 市場上不論是好的或是不好的earningssurprise均會(huì)有#MomentumIfXYZIndustrieshasrecentlybeenapoorperformer,investorsbelieveitwillremainapoorperformeruntilthereisaweightofevidencetoconvincethemotherwise.Similarly,ifXYZIndustrieshasperformedexceedinglywellrecentmonths,investorsoftenbelieveitwillcontinuetodoTotheForexamplethePEratiomaybewayabovethesectoraverageandtheprojectedearningsmaynotjustifythehighpriceinvestorsarepaying.動(dòng)能的因Earningmomentum的持續(xù)性比pricemomentum來的DOINDUSTRIES TobiasJ.Moskowitz,MarkGrinblatt,Momentuminvestmentstrategiesaresignificantlylessprofitableoncewecontrolforindustrymomentum.FindingIndustrymomentuminvestmentstrategies,whichbuystocksfrompastwinningindustriesandsellstocksfrompastlosingindustries,appearhighlyprofitable,evenaftercontrollingfor:book-to-marketindividualstockcross-sectionaldispersioninmeanpotentialmicrostructuresPredictingmomentumreturnsandcontrarianeffectsJegadeeshandTitman3-12mths,pastwinnersoutperformpastDebondtand3-5yrs,L-TwinnersunderperformL-T RelatedS-TmomentumandL-TBarberis,Shleifer,Vishnyinitialunder-reactionthenover-ContextualFundamental ThroughthePredictionofExtremeReturnsM.C.RobinL.,Extremeperformersareyounger,lowerpriced,andsmallerfirmswithmorevolatilereturns,highertradingvolume,andmorepositivepricemomentum.Extremelosersappeartobeover-extendedgrowthfirms,withslowingsalesgrowth,shrinkingmargins,butsignificantlevelsofnewcapitalinvestments.InvestorMalcolmBakerandJeffreyJournalofFinance,Sentiment封閉型基金 周轉(zhuǎn)IPO第一天的權(quán)益增資股股利溢Howtomeasuresentiment封閉型基金折Delong,Shleifer,Summers,andWaldmann(DSSW;NYSE周轉(zhuǎn)Log（量/在NYSEFactBakerandStein(2004):在一個(gè)有放空限制、同時(shí)存在IPO的個(gè)IPO第一天的平權(quán)益增資股grossannualequityissuance/(grossannualequity+debt股利溢-通常為大型、有獲利、低成長機(jī)會(huì)的公司-每一個(gè)情緒 變數(shù)都包含情緒成分，同時(shí)也含有非關(guān)情緒的獨(dú)成分→使用主成份分析接下來就建構(gòu)第二個(gè)指數(shù)--在進(jìn)行主成份分析前先將企業(yè)循環(huán)變動(dòng)各 變數(shù)中移除Sentiment 股MethodologyIssues:DissectingEUGENEF.FAMAandKENNETHR.FRENCH,JFPreviousworkfindsthatnetstockissues,accruals,momentum,profitability,andassetgrowthareassociatedwithanomalousaveragereturns.Weexplorethepervasivenessofthesereturnanomaliesviasortsandcross-sectionregressionsestimated yonmicro-caps,smallstocks,andbigstocks.Likethepatternsinaveragereturnsassociatedwithnetstockissues,accruals,profitability,andassetgrowth,returnmomentumisleftunex inedbythethree-factormodelofFamaandFrench(1993)aswellasbytheCAPM.Werevisitthesize,value,profitability,growth,accruals,netstockissues,andmomentumStudythemtogethertoseewhichhaveinformationaboutaveragereturnsthatismissedbytheothers.Approachesusedtoidentify(i)sortsofreturnsonanomalyvariables(ii)regressionsthatuseanomalyvariables inthecross-sectionofaverageThemainadvantageofasimplepictureofhowaveragereturnsvaryacrossthespectrumofananomalyvariable.potentialpitfalls:acommonapproachistoformequal-weight(EW)decileportfoliosbysortingstocksonthevariableofinterest.Thoughthedetailedresultsfordecilesaretypicallyshown,itiscommontofocusonthehedgeportfolioreturnobtainedfromlong-shortpositionsintheextremedeciles.ApotentialThereturnsonEWhedgeportfoliosthatuseallstockscanbedominatedbystocksthataretiny(micro-caps,withmarketcapbelowthe20thNYSEpercentile).Micro-capscanbeinfluentialinEWhedgeportfolioreturnsfortworeasons.First,thoughmicro-capsareonaverageonlyabout3%ofthemarketcapoftheNYSE-Amex-NASDAQuniverse,theyaccountforabout60%ofthetotalnumberofstocks.Tocircumventthisvalue-weight(VW)hedgeportfolioreturnsareoftenshownalongwithEWreturns.ButVWhedgereturnscanbedominatedbyafewbigstocks,resultingagaininanunrepresentativepictureoftheimportanceoftheToattacktheseweexaminetheaveragereturnsfromseparatesortsofmicrocaps,smallstocks,andbigstocksoneachanomalyvariablewherethebreakpointsseparatingmicrofromsmallandsmallfrombigarethe20thand50thpercentilesofmarketcapforNYSEstocks.Sortshave Sortsareawkwardfordrawinginferencesaboutwhichanomalyvariableshaveuniqueinformationaboutaveragereturns.Multipleregressionslopesprovidedirectestimatesofmarginaleffects.Moreover,withourlargesamples,margin