信用风险模型(1)

ModelingDefaultRiskModelingDefaultRiskKMVLLCiiReleaseDate:15-November-1993Revised:14-January-2002COPYRIGHT1997-2001KMVLLC,SANFRANCISCO,CALIFORNIA,USA.Allrightsreserved.KMVLLCretainsalltradesecret,copyrightandotherproprietaryrightsinthisdocument.Exceptforindividualuse,thisdocumentshouldnotbecopiedwithouttheexpresswrittenpermissionoftheowner.PortfolioManager™,PortfolioPreprocessor™,GCorr™,GlobalCorrelationModel™,SetAnalyzer™,ExpectedDefaultFrequency™,andEDF™aretrademarksofKMVLLC.KMV,theKMVlogo,CreditMonitor®,EDFCalc®andPrivateFirmModel®areregisteredtrademarksofKMVLLC.Allothertrademarksarethepropertyoftheirrespectiveowners.Publishedby:Author(s):KMV,LLCPeterJ.Crosbie1620MontgomeryStreet,Suite140JeffreyR.BohnSanFrancisco,CA94111U.S.A.Phone:[phone]FAX:[phone]email:[e-mail]::14-January-2002TableofContentsOverview.....................................................................................................................................................................1MeasuringDefaultProbability:TheProblem.....................................................................................................2MeasuringDefaultProbability:APracticalApproach......................................................................................7EstimateAssetValueandVolatility..................................................................................................................8CalculatetheDistance-to-default....................................................................................................................11CalculatetheDefaultProbability....................................................................................................................12PuttingitallTogether.......................................................................................................................................13ACloserLookatCalculatingEDFCreditMeasures.........................................................................................14CalculatingLong-TermEDFCreditMeasures...................................................................................................18SomeFrequentlyAskedQuestionsAboutKMV’sEDFCreditMeasures.....................................................19TestingtheDefaultMeasure’sPerformance......................................................................................................29SummaryandDiscussion......................................................................................................................................33ModelingDefaultRiskKMVLLC1ReleaseDate:15-November-1993Revised:14-January-2002OverviewDefaultriskistheuncertaintysurroundingafirm’sabilitytoserviceitsdebtsandobligations.Priortodefault,thereisnowaytodiscriminateunambiguouslybetweenfirmsthatwilldefaultandthosethatwillnot.Atbestwecanonlymakeprobabilisticassessmentsofthelikelihoodofdefault.Asaresult,firmsgenerallypayaspreadoverthedefault-freerateofinterestthatisproportionaltotheirdefaultprobabilitytocompensatelendersforthisuncertainty.Defaultisadeceptivelyrareevent.Thetypicalfirmhasadefaultprobabilityofaround2%inanyyear.However,thereisconsiderablevariationindefaultprobabilitiesacrossfirms.Forexample,theoddsofafirmwithaAAAratingdefaultingareonlyabout2in10,000perannum.AsingleA-ratedfirmhasoddsofaround10in10,000perannum,fivetimeshigherthanaAAA.Atthebottomoftheratingscale,aCCC-ratedfirm’soddsofdefaultingare4in100(4%),200timestheoddsofaAAA-ratedfirm.Thelosssufferedbyalenderorcounterpartyintheeventofdefaultisusuallysignificantandisdeterminedlargelybythedetailsoftheparticularcontractorobligation.Forexample,typicallossratesintheeventofdefaultforseniorsecuredbonds,subordinatedbondsandzerocouponbondsare49%,68%,and81%,respectively.Crossdefaultclausesindebtcontractsusuallyensurethatthedefaultprobabilitiesforeachoftheclassesofdebtforafirmarethesame.Thatis,thedefaultprobabilityofthefirmdeterminesthedefaultprobabilityforallofthefirm’sdebtorcounterpartyobligations.However,thelossintheeventofdefaultforeachoftheclassesofobligationscanvarywidelydependingontheirnature(security,collateral,seniority,etc.).Althoughingeneralapoorinvestmentstrategy,itispossibletoberewardedfortakingonlargeconcentrationsofriskinequitiesbecausetheseconcentrationsattimesproducelargereturns.However,overwhelmingevidenceoftheineffectivenessofthisstock-pickingstrategyhasbeenavailablesincetheearly1970sand,asaresult,themajorityofequityinvestmentsaremanagedindiversifiedportfolios.Unlikeequities,debthasnoupsidepotentialandthusthecaseformanagingdefaultriskinwell-diversifiedportfoliosisevenmorecompelling.Thelimitedupsidepotentialofdebtspreadsmeansthattherearenopossiblecircumstancesunderwhichaninvestororcounterpartycanberewardedfortakingonconcentrationsofdefaultrisk.Likeotherrareeventswithhighcosts,defaultriskcanonlybeeffectivelymanagedinaportfolio.Inadditiontoknowingthedefaultprobabilityandlossgivendefault,theportfoliomanage