标准普尔外部评级在新资本协议中的作用

BogieOzdemir,SeniorDirectorRiskSolutionsWDCFeb,2008TheRoleofExternalRatingsUnderBaselII2.Agenda•ExternalRatingscanplayimportantrolesinIRRsdesign,quantificationandvalidation–butweneedtobecarefulabouttheappropriateusage.•ExternalRatingscanbeusedin1.RiskRatingAssignmentUtilizingtheRatingsUtilizingtheMethodology2.RiskRatingQuantification3.ValidationBenchmarkingBecktesting3.UseoftheExternalRatingsinRiskRatingAssignment•AgencyRatingsareusedasinputstotheRiskRatingAssignmentprocessalongwithPDs/EDFsfromquantitativemodels,FinancialrationsandQualitativeFactors•Pro:AllavailableinformationisutilizedBUT..•Con:Doublecounting:FinancialRatiosandQualitativeFactorsarealreadyutilizedinExternalRatings.FinancialRatiosmayalsobeutilizedinquantitativePDs/EDFs.DetermineRiskRatingAgencyRatingsQuantitativePDmodelFinancialRatiosQualitativeFactorsRiskRatingtoPDMappingPDExampleRiskRatingandPDassessmentprocess4.UseoftheExternalRatingsinRiskRatingAssignment•Con-ConsistencyIssues:PhilosophicalDifferences;QuantitativePDs/EDFsaremorePITthanExternalRatings.•Wedon’thaveExternalRatingsandQuantitativePDs/EDFsforallobligors.BasedonwhereweareinthecycleandwhetherornotwehaveExternalRatingsandQuantitativePDs/EDFs,theratingsofsimilarobligorscanbeinconsistent•UtilizingratingmethodologyinsteadoftheratingsforLDPs?DetermineRiskRatingAgencyRatingsMoreTTCQuantitativePDmodelMorePITFinancialRatiosMaybePITbutlaggingQualitativeFactorsIntendedtobePITbutnotfrequentlyupdatedRiskRatingto(RiskRating)PDMappingPDassignedtotheRiskRatingLong-runUnconditionalPDassignedtotheObligorMorePIT(frequentre-gradingorotherwise)MoreTTC(nore-gradingorotherwisebasedonsystematicfactors)5.UseoftheExternalRatingsinRiskRatingQuantification•WeneedtoassignPDs,LGDsandEADstoIRRSbasedonhistoricalexperience•Internaldataislimitedandmaynotbesufficientlystressed•Externaldata(e.g.defaultratetimeseriesforexternalratings)aremuchlonger–Howcanwerobustlyutilizeexternaldata?ExternalRatingsInternalRatings12AA/AA-/A+3A/A-/BBB+4BBB/BBB-/BB+5BB/BB-/B+6CCC7+8+9DDAAA/AA+•Themappingsbetweeninternalandexternalratingsarenotalwaysrobust(usuallymethodologiesarenotmapped)•AdvantageofusingratingagencymethodologiesforLDPs:Internalratingsarecalibratedtoexternaldata•MiuandOzdemir(2007)’sapproach*enablesestimationsofLong-runPDsjointlyfrominternalandexternaldata*EstimatingandValidatingLong-RunProbabilityofDefaultwithrespecttoBaselIIRequirements6.•BenchmarkingistheexaminationoftheperformanceofriskratingsystemsrelativetothecomparableRiskRatingSystemsandModels•Benchmarkingperformingloans•Benchmarkingdefaultedloans•BacktestingistheexaminationoftheperformanceoftheRiskRatingSystembasedonitshistoricaldatacomparingrealizedandpredictedoutcomes.•Benchmarkingthebacktestingperformance.UseoftheIndependentModelsinValidation7.BenchmarkingPerformingLoansUseoftheExternalRatingsStandard&Poor'sratingInternalscale#ofcreditsDSDCCCCC-CCCCCC+B-BB+BB-BBBB+BBB-BBBBBB+A-AA+AA-AAAA+AAA1552033341114425101162622111737183968212136939130611102712051111316321121131411113522114412115612211611170182219112011Totals20000000012288122740251925106807S&PorCreditModelRatingsNumberofobligors•WebenchmarkinternalratingsagainsttheexternalratingswhenavailableandagainstS&PCreditEstimatesorCreditModelwhichproducesStandard&Poor'slettergraderatingsymbols,inlowercase.8.BenchmarkingPerformingloans,Monitoring•Animportantaspectofbenchmarkingistomonitorthestabilityofthemappingbetweeninternalandexternalratingsovertime.Thishelpstoidentifysystematictrendsandchangesintrends.Severalillustrativeexamplesareprovidedbelow.InternalRiskGradesExternalRiskGradesLessRiskyMoreRiskyInternalRiskGradesExternalRiskGradesInternalRiskGradesExternalRiskGradesInternalRiskGradesExternalRiskGradesConcentrationofObligorsDispersionTrend1:Increaseindispersion.Theagreementdecreases.Trend2:ParallelShift.IRRSbecomeslessconservative.Trend3:Non-ParallelShift.IRRSbecomeslessconservativeforlowriskgrades,andmoreconservativeforhighriskgrades.9.•Needtoexamineifrealizeddefaultandmigrationrates(&realizedLGDandEADs)areconsistentwiththerelevantindustryexperience.•WeusemodelsS&Pdatabasestoprovidethisbenchmarkinformation:CreditProprovidesDefaultRate,RatingMigration,andDefaultCorrelationstatisticscustomizedbyindustry,geographicregion,andtimeframebasedonStandard&Poor’shistoricalcorporateratinganddefaultdata.BenchmarkingDefaultedLoans10.UseoftheExternalRatingsinValidation:BacktestingDimensionsofvalidation,weneedtotest:DiscriminatoryPoweroftheriskratingsystem,Calibrationoftheriskratingsystem,RiskHomogeneityoftheobligorsineachoftheRiskRatingsandRealizationoftheRiskRatingPhilosophy:WemeasureourRiskRatingPhilosophyrelativetothatofS&PPurePIT,notachievablePureTTC,notachievableMorePITHybridPhilosophiesMoreTTCContinuousPDmodelsutilizingstockmarketinformationandleadingindicators(e.g.KMV’sCreditMonitor,S&P’sCRT)IntendedTTCIRRSS&P’sIntendedPITIRRS•allinformationrelatedtoanobligor’sdefaultlikelihoodisobservable,and•frictionlessre-gradingsystem•nochangesinratingsduetosystematicreasons,(ie.Relativerankingdoesnotchange)butonly