CFA二级押题密卷-模块三.docx

1906高顿CFA押题密卷-LeVel2o。.UnPoBt5c.MM M- K wu。dnpopsBoUBUIJ-Vujkh<h宫惇槌AnnaGu,CFA,recentlyhasbeenpromotedasthechiefinvestmentofficer(CIO)oftheM3GloryFund,aprestigiousboutiquefundinChinamarket.GuisreadingananalysisreportwrittenbyhercolleagueAngelZhouabouttheChinese5Gcommunicationindustry.Zhou,sreportincludestheresultsofaregressionofthemonthlyreturnstoa5Gcommunicationequityindexfortheprevious100monthsonvariablesrepresentingthemonthlyreturnstotheShanghai-Shenzhen300Index(CSI300),andthemonthlyreturndifferencebetweenlong-termandshort-termChinesegovernmentbonds(SPREAD).AnnaGudoubtsthatCSI300andSPREADarereasonableindependentvariablesandwantstodetectmulticollinearity.Becausetheregressionhasonlytwoindependentvariables,GudecidestotestthesignificanceofthecorrelationbetweenCSI300andSPREADwithsignificancelevelof5%.UsingZhouzsdata,Gufindsthecorrelationcoefficientis0.18.GuisalsonotconvincedofthevalidityofZhou,smodelandthenrunsaregressionbyherself.GuzsregressionresultsarepresentedinExhibit1.Exhibit1:Gu,sRegressionModel5GCommunicationEquityIndexVariableCOeffidentt-statisticp-valueConstant0.02690.0180.95CSI30.46256.190<0.01SPREAD1.12644.280<0.01R20.40Durbin-WatsonStatistic0.84Durbin-Watsoncriticalvalues(5%significance)1.631.72wantstousetheregressionresultstotestthishypothesiswithasignificancelevelof5%.Selectedvaluesofthet-distributionareshowninExhibit2.Exhibit2SelectedValuesofthet-Distribution(degreesOffreedom=df,one-tailedprobabilities三p)Dfp=0.05p=0.02511.6601.984IlO1.6591.9821201.6581.98021.6531.972OO1.6451.960Guangzhouoffice.Zengbelievesnon-stationaritymaybeaproblemandconductsDickey-Fullertestsforaunitrootoneachofthetimeseries.TheresultsarereportedinExhibit3.Exhibit3ResultsoftheDidcey-FullerTestsTimeSeriesValueOftheTestStatisticStandardErrort-StatisticSignificanceoftSGcommunicationequityindex1.6600.00231.5910.1123CSI31.6590.0724-5.8460SPREAD1.6580.043-13.510ZengtellsGuitispossibletofindmorethanonemodelsevenwithsamedataset,thustheremustbeacriteriontoselectthemostaccurateone.Gustates,zzTherootmeansquarederror(RMSE)criterionistypicallyusedtoevaluatetheout-of-sampleforecastaccuracy7'.Zengreplies,z1fweusetheRMSEcriterion,themodelwiththesmallestRMSEistheoneweshouldjudgeasthemostaccurate/'EO。.UnPoBt5c.MMM-Kwu。doM0。UBUIJ-V工JSH三裾-BWnOOeISoIUS-SlUBlUSelSMU-MO-OJ一jlJolpzM、a<IJJPUeoosoUMlBqIUp一=Ooorol-oUoPSeg.#叶7珊翅Eo。.UnPoBt5c.MMM-Kwu。doM0。UBUIJ-V工JSH三裾dluHdSPUeOom-SO5O-IS×IoUSOPAIU=-O-l-nluPUe寸86roSs-S一-一sSneQaqroM-s-lup一七OSUO-IeaJJIO0山IIl=8.二让X8I.OU#叶7珊翅8.AccordingtoGu,sregressionresultsinExhibit1,ifthemonthlypredictedmonthlyreturn(%)tothe5GCommunicationEquityIndexEo。.UnPoBt5c.MMM-Kwu。doM0。UBUIJ-V工JSH三裾06SI9IH0IXSZ9NO+0IX-9ZII+69ZO0H()EnJ>-UOII#叶7珊翅Eo。.UnPoBt5c.MMM-Kwu。doM0。UBUIJ-V工JSH三裾-S-vLLJQe-dsUOlu-*ooUo-SSjicidlInoqe-ttS-SBqlOdAqSJDOI-l-Uo-SrIPUOJPUeORS-IeISzzx-SI-nsJUo万SBOIjoOIoo<6#叶7珊翅Durbin-WatsonstatisticreportedinExhibit1?TheDurbin-WatsonB.rejectsthenullhypothesisandconcludethereispositiveserialcorrelation.ThevalueoftheDurbin-WatsonstatisticisgiveninExhibit1as0.84.Thecriticalvaluesaregivenas1.63and1.72.Becausethevalue(0.84)islessthanthelowercriticalvalue(1.63),theDurbin-Watsontestrejectsthenullhypothesisandconcludethereispositiveserialcorrelation.Eo。.UnPoBbc.MMM-Kwu。doMBOUB=J-VHjSH三裾10Useop-q-JB>IUpuedP0SloOJ-un-q三xs-qro一e>IUePUBdePU-一U-q11J-JeAIU山pudPJoJaSnpUouU-100J-UneJoJIS山一-1s#叶7珊翅ThenullhypothesisisHO:gl=0versusthealternativehypothesisHa:gl<0.Ifthenullhypothesisfailstoberejected,thenthepossibilityexiststhatthetimeserieshasaunitrootandisnonstationary.Basedonthet-statisticandtheirsignificancelevelsinExhibit3,thenullhypothesisisrejectedforbothCSI300andSPREAD(i.e.,theindependentvariables),butthenullhypothesisisnotrejectedforthe5Gcommunicationequityindex(dependentvariable).Therootmeansquarederror(RMSE)criterionisusedtocomparetheaccuracyofregressivemodelsinforecastingout-of-samplevalues,andthemodelwiththesmallestRMSEisjudgedthemostaccurate.RosaSalazar7CFA,isafixedincomeanalystinAlitaFund.Sheisspecializedinthequantitativearea.Recently,sheconductsaresearchonTreasuryyieldsandbelievesthattheTreasurybondyieldsarecorrelatedwiththeFederalFundsrate(FFR)andtheunexpectedinflationrate(UIR).Aftertheanalysis,shecollects60monthlydatatosetupthequantitativemodel.TheregressionresultsarelistedinExhibit1andtheregressionmodelisasfollows:Treasuryyields=h0+b1×FFR+b2×UIR+Exhibit1ResultsfromRegressingTreasuryBondReturnsonFFRandUIRCoefficientStandardErrort-StatisticIntercept1.83210.118915.4087FFR1.01580.27633.6764UIR0.78510.15345.1180Afteranalysis,SalazarwonderswhethertheerrorsareconditionallyHeteroskedastic.Basedonhercalculation,theteststatisticis8.46.Then,sheperformsastatisticaltesttodetermineifconditionalHeteroskedasticityispresentatthe0.05significancelevel.Itisknownthattheteststatisticcriticalvaluewith2degreesoffreedomatthe0.05significancelevelis5.99.Beforeapplyinghermodel,SalazarasksthemanagerofAlitaFund,JamesCameron,toreviewthemodel'sspecificationandresults.CameronremindsherthatifthemodelsuffersfromconditionalKeteroskedasticity,theresultscanbeunreliable.ThenhemakestwostatementsabouttheconsequencesofconditionalKeteroskedasticity.Statement1"ConditionalHeteroskedasticitywillresultinconsistentcoefficientestimates,butboththet-statisticsandF-statisticwillbebiased,resultinginfalseinferences/'underestimatedandt-teststatisticscomputedwillbeinflated,suggestingthatthereisnosignificantrelationshipwhentherelationshipactuallyexists(TypeIlerror)/AfterdetailedanalysisforconditionalHeteroskedasticity,SalazarthenrunsaDurbin-Watsontesttotestwhethertheserialcorrelationexists.CameroninspectsSalazar,soutcomesandasksherhowtocorrecttheviolationsifbothconditionalHeteroskedasticityandtheserialcorrelationexist.SalazarrealizesthatapossibleremedialactionistoadjustthecoefficientstandarderrorsbyWhite'smethodorHansen'smethod.ShewonderswhichmethodismostsuitableforthecircumstancedescribedbyCameron.Finally,SalazarcalculatesthesamplecorrelationbetweentheFederalFundsrate(FFR)andtheunexpectedinflationrate(UIR)using60monthlydata,andgetstheresultfor0.52.Giventhesignificancelevelof5%,shewantstodeterminewhetherthecorrelationcoefficientisstatisticallysignificant.Thecriticalvalueforthetest7.TodetermineifconditionalHeteroskedasticityispresent,whichofthefollowingstatisticaltestsshouldbeperformedandwhatcanbeconcludedfromthetest?A.Breusch-Pagan2test.NoconditionalHeteroskedasticityexists.B.Breusch-Pagan2test.ConditionalKeteroskedasticityexists.C.F-test.ConditionalKeteroskedasticityexists.WhentestingforconditionalHeteroskedasticity,Breusch-Pagan2testshouldbeused.ThenthenullhypothesisofBreusch-Pagan2testisnoconditionalHeteroskedasticity.Becausetheteststatisticis8.46andthecriticalvalueis5.99,weshouldrejectthehypothesisofnoconditionalKeteroskedasticityatthe0.05level.Therefore,weneedtocorrectforconditionalKeteroskedasticity.Eo。.UnPoBt5c.MMM-Kwu。doM0。UBUIJ-V工JSH三裾0。S-TlU山UJSSA-Uo<p-JJ03S-Alp一pp-S0名±BU0三pu8Jooubj0WEecu0SIU一u-o<-Eroo¾uz».#叶珊翅Statement1iscorrectbecausetheconditionalKeteroskedasticityresultsinconsistentparameterestimates,butbiased(upordown)standarderrors,t-statistics,andF-statistics.Statere2S62BSncorrectbe<a0etbQnflatedt-teststatisticswillsuggestsignificantrelationshipswherenoneactuallyexist(TypeIerror).9.Whichremedialmethod(s)is/aremostappropriatetocorrecttheHansen,smethodcouldbeusedtocorrectforserialcorrelation,whichcouldalsobeusedtocorrectconditionalKeteroskedasticity.White'smethodcouldonlybeusedtocorrectforHeteroskedasticity.10.SalazarshouldconcludethatthecorrelationcoefficientbetweenFFRandUIRis:A.significantlydifferentfromzero,becausethenullhypothesiscanberejected.B.significantlydifferentfromzero,becausethenullhypothesiscannotberejected.C.notsignificantlydifferentfromzero,becausethenullhypothesiscannotberejected.Eo。.UnPoBbcMMM-辂置悻裾u。d三oMdec-U.2ZEo-lueej-uyss-nPUeM-M-5Uo-OOlPnPUoOUeuMPUept3jqUeoS-SlodAq=nuqlSo7UeIss&S-寸9.寸°U-S-IzdlsZMl.OlT-i_Zsd#叶珊翅o。.UnPoBt5c.MMM-Kwu。dnpopsBoUBUIJ-Vujkh<h宫惇槌YukiEndoisaseniorpartnerforHommaTrustandBrokerageCompany,asmalltrustmanagementcompanybasedinTokyo.Today,shehasnosoonercometoherofficethansheanswersaphonefromoneofherVIPclient.Inthephone,theVIPclientstatesthatheisworriedabouthisaccountbecauseoftheimpendingeconomicrecession.Heplanstoturntosmall-capstocksinsteadoflarge-capones,andinvestmoreinTreasurebondsandsellsomehigh-yieldbonds.Endogivesherprofessionalopinionforthisclient.AfterherdiscussionwiththisVIPclient,shebeginsameetingwithhertwoassistantportfoliomanagers,LouiseChenandWendyZeng.ChenandZenghavebeenaskedtodosomeresearchonwaystoimproveonthemethodscurrentlyusedbyHommaTrusttoevaluatesecuritiesanddevelopportfolios.Chensuggeststhatmultifactormodelsprovideabetterwaytomodelstockreturns.Shedevelopstwomodelsonawhiteboardwhilestating"Therearetwowaystomodelstockreturnsusingthefollowingmultifactormodel:iKKInthismodel,stockreturns(Ri)aredeterminedbysurprisesineconomicfactorssuchasGDPgrowthandthelevelofinterestrates.Here,stockreturns(Ri)aredeterminedbyfactorsthatarecompanyattributessuchasprice-earningsratioandmarketcapitalization.ZengnotesthatamultifactorArbitragePricingModel(APT)providesamuchbetterbasisthantheCAPMforcalculatingexpectedportfolioreturnsandevaluatingportfolioriskexposures.InordertoillustratetheadvantagesofthemultifactorAPTmodel,ZengprovidesinformationfortwoportfoliosHommaTrustcurrentlymanages.TheinformationisprovidedbelowinExhibit1.Thecurrentrisk-freerateis2percent.Exhibit1FactorSensitivitiesandRiskPremiumFactorRiskFactorFactorSensitivitiesFactorRiskPremium(%)PortfolioAPortfolioBBenchmarkConfidenceRisk0.810.040.54.5InflationRisk0.15-0.45-0.25-1.2BusinessCycIeRisk1.230.090.95.2"WecantellfromExhibit1thatPortfolioAisstructuredinsuchamannerthatitwillbenefitfromanexpandingeconomyandimprovingconfidencebecausethefactorsensitivitiesforconfidenceriskandbusinesscycleriskexceedthefactorsensitivitiesforthebenchmark.PortfolioBhasverylowfactorsensitivitiesforconfidenceriskandbusinesscycleriskbutmoderatelyhighexposuretoinflationrisk,thereforePortfolioBcanbereferredtoasapurefactorportfolioforinflationriskz51.DoesthisVIPclientmaketheappropriateplanonequityandInrecession,morelarge-capequityandmorelow-riskbondareappropriate.52.Withregardtothestatementonmultifactormodels,ChenisModel1isamacroeconomicfactormodel.Inthismodeltheinterceptvalueaiistheexpectedreturnonthestock.Model2isafundamentalfactormodel.Infundamentalfactormodelsthefactorsensitivitiesbiarestandardized,thustheinterceptisnotinterpretedasanythingmorethanaregressioninterceptthatensuresthatexpectedassetspecificriskequalszero.Itisnotinterpretedastheexpectedreturnforthestockasinthemacroeconomicfactormodel.53.BasedontheinformationinExhibit1,theexpectedreturnforB.10.2%E(RA)=2%+0.81X4.5%+-0.15X-1.2%+1.23×5.2%=12.21%ZengisincorrectaboutPortfolioB.Purefactorportfoliosbydefinitionwillhaveafactorsensitivityof1toaparticularfactorandzerosensitivityforallotherfactors.ForPortfolioBtobeapurefactorportfoliofortheinflationriskfactoritmusthavefactorbetaof1toinflationriskandzerofortheotherfactors.ElizabethRobbinisthenewchiefinvestmentofficeratTeacherRetirementSystems(TRS)zwhichoversees21fundsfrom11differentfirms.SheplanstoimplementnewperformancemeasurementtoolsforselectingandevaluatingTRS'smanagers.Robbinismeetingwithherstaff(Marano,Gladden,andWert)todiscusswhattheyknowaboutmanagerselectionandevaluation.Shestartsbydiscussionabouttheuseoftheinformationratio,Robbinexplainshowitmightbeusefulininvestmentmanagerselectionandinchoosingthelevelofactiveportfoliorisk.Sheaskseachanalysttomakeanobservationabouthisorherunderstandingoftheinformationratio.Gladden:BecauseTRS'sinvestmentpolicyprohibitsshortpositions,TRSwouldbeunabletotakeadvantageofanyoptimizedportfolioswithincreasedactiverisk.RobbinpointsoutthattheSharperatioandtheinformationratioarebothusefultoolsinevaluatingportfoliomanagersandasksGladdentoexplainsomeoftheimportantdifferencesbetweenthetwo.Gladdennotes,"TheinformationratioisameasureofrelativeexpectedorrealizedrewardtoriskwhereastheSharperatiomeasurestheabsoluterisk-returntrade-offofaportfolio.Sharperatioshelpinvestorsfocusontherelativevalueaddedbyactivemanagement.Althoughtheinformationratioisnotaffectedbytheadditionofcashorleverage,theSharperatioisaffectedbytheadditionofeither."RobbinthenintroducestheFundamentalLawofActiveManagementtoheranalysts,illustratingitwithagraphiccalledthe"correlationtriangle.""Thisgraphexplainshowamanager'sforecastedreturns,decisionsabouttheportfolio'sactiveweights,andrealizedactivereturnsarerelatedtoeachother/'shesays.高顿财经CFA:finance,高顿网校：Wertobserves,"Thatmakessense.Itisdifficulttoaddvalueifthemanager'sforecastsdonotcorrespondatleastsomewhattotherealizedactivereturns.Alsoziftheportfoliomanagerdoesnotoverweightsecuritiesforwhichhehasforecastedthebestrelativereturns,hewillnotgeneratepositiverelativereturns."1.astly,RobbinillustratestoherteamhowtheymightapplytheFundamentalLawofActiveManagementinevaluatingtheperformanceofKaribaInvestmentManagement.Robbinobserves,"Karibamaybeoverstatingitsexpectedactivereturn.BecauseKaribarebalancesweekly,itclaimsthatitsnumberofindependentdecisionsishigh.Howeverzsomeofthesesecurities(exchange-tradedfunds)mayclusterineconomicregionswherethesamegeneralanalysisappliestoseveralsecurities.ThiswouldmeanthatKariba,sbreadthisinfactmuchlowerthanstated.Furthermore,Karibaassertsthateachsecurityisindependentlyevaluated.Thatmaynotbetrueeither.Forexample,astrategythatfavorsaparticulareconomicregionwilllikelypersistforseveralmonths,andtherefore,theinvestmentdecisionsarenotindependent.Again7theresultwouldbealowerbreadth."Themeetingthenturnstoportfolioriskmanagement.RobbinasksMaranotoapplytheanalyticalmethodtoestimatetheVaRforAlphaAssetManagementsportfolio,whichisvaluedat$20million.Theportfoliohasanexpectedannualreturnof7.5%andastandarddeviationof22.4%.Anotherportfolioisapor