现有风险量化模型简介

量塞1993年行度念利率率行參數不度不G-30VaR量VaRJ.P.Morgan4.15J.P.Morgan1994年RiskMetricsVaR量金VaR量金數1995年塞行行量(ValueatRiskVaR)(110)(率95%~99%95%1.64599%2.33)量來度VaR率率VaR數量(1-α)(t)VaR3-1:金度金易率MarkettoMarket1來來1金來數理來1行DennisWeatherstone4:15度來2454來益金95%5%Xt95%5%X1095%5%X295%5%X1Var數95%5%Xt95%5%X1095%5%X295%5%X1Var數210t若來類料來理類量利VaR理了量來VaR度量料VaR精度料金量55來料度絶數度行VaRMarkettoMarket易量易料35年料VaR行量度不量VaR量兩參數來率參數Delta-NormalModelDelta-GARCHModelGamma-NormalModel來路歷羅行VaR理3-2:VaR異數異數Variance-CovarianceApproach羅MonteCarlosimulation歷istoricalsimulation理論ExtremeValueTheoryMethodBootstrapping利量度來來益不料率益不見料,利料,料來理56異數-異數(Variance-CovarianceApproach)異數異數數異數異數(Σ)Delta-Normal率率數(σ)異數數GARCH率便利率異(VaR)VaR(1-α)不ασ來率數金歷(HistoricalSimulation)歷列料行來利利率率歷料率理來率VaR利來歷歷念易了歷料度料歷料羅(MonteCarloSimulation)羅來立57路立羅數數理論VaR見量數亂數亂數金行不率羅亂數Efron1979年(Bootstrap)數數數量理論來利率3-1:行精度益諸不例不歷料羅歷料歷料來率度精度益不諸度羅歷不益益數異數料異數異數行精度益諸不例不歷料羅歷料歷料來率度精度益不諸度羅歷不益益數異數料異數異數料來李58量理塞勵金(InternalRatingsBasedIRB)行立易IRB了(CreditRating)念狀力量立易良20(Moody’s)(S&P)(FitchRatings)1997年424行理立(Moody’s)(FitchRatings)來立立立度易率(PD)率(LGD)率暴金(EAD)量料行J.P.MorganCreditMetrixKMV(KealhoferMcQuownVasicek)EDFExpectedDefaultFrequency率McKinsey&Co.CreditPortfolioViewCreditRisk+CreditSuisseFinancialProduct說:J.P.MorganCreditMetrixTMCreditMetrixJ.P.Morgan1996年RiskMetics量理利易()行VaR念路(MigrationAnalysis)量59降履行量料理理暴度易度KMVEDFTMEDFKMV1995年KMV率(ExpectedDefaultFrequency)來EDFTM不(朗)()T離(度)立離歷率離率率(PD)McKinsey&Co.CreditPortfolioViewCreditPortfolioViewMcKinsey&Co.1997年量McKinsey&Co.率理念率率來立量()易率不()率利歷料來量率CreditSuisseFirstBostonCFSBCreditRisk+CreditRisk+行(CreditSuisseFirstBoston)1996年量精立CreditRisk+數金列來量60没CreditRisk+利率連數來率不CreditMetricsEDFCreditPortfolioViewCreditRisk+3-2:量數率()(beta)(beta)率率數率數量量連EDF利率數率率數率離率連率CreditRisk+CreditProfolioViewEDFTMCreditMetricsTM數率()(beta)(beta)率率數率數量量連EDF利率數率率數率離率連率CreditRisk+CreditProfolioViewEDFTMCreditMetricsTM料來61參量不流不律落理行降念量識來金了錄料類ELUL數量率“”量料料不料歷料來率金流類3-3類類率不易聯塞不量量(AdvancedMeasurementApproachesAMA)(Qualitative)量(Quantitative)不率勵金料行量理降不易率不例量不易62流來類力不不行流易行契量料來類類不數暴量暴EI料類類率金暴率金率率度理量料羅HierarchicalMonteCarloTechniques量不易立4度3-3(BasicIndicator)利率(Standardized)類利利數量(InternalMeasurementApproach)料量量(LossDistribution)量例理度來63理Basel行年AMAABNAMROJ.P.MorganChaseBankofAmericaCreditSuisseCitigroup料量降TheRegulatoryApproach:FourIncreasinglyRiskSensitiveApproaches度料來理路料料漏復(DisasterRiskBased/lessRegulatoryCapital:BasicIndicatorLossDistributionRateBaseBankβ1EI1LOB1β2EI2LOB2LOB3βNEINLOBnBankStandardizedStandardizedApproachLossDistributionApproach•••InternalMeasurementApproachInternalMeasurementBankRateBaseRateBaseBaseRateBaseRiskType6Rate1EILOB1Rate2EILOB2BasLOB3RateEILOBnRiskType1•••••ExpectedLossRateofprogressionbetweenstagesbasedonnecessityandcapabilityProbabilityLossSevereCatastrophicUnexpectedUnexpectedLossLoss64RecoveryDR)(BusinessContinuityPlanningBCP)都落理金度塞1996年(BackTesting)來利力(StressTesting)來了聯連利率VaR參年(250)數理論數率量行塞理60數(α+β)αβα3250易數BaselⅡ不數不數數數3-4數β就越大數04綠度59令數數0.4~0.85數1065金數來量行行數金度力數不綠數度數數不綠數度數數料來力聯理連串菱行神225數暴1000兩都不不量說VaR率金不力來金參數66歷狀力度異不利3-4不流不利力率力料來IOSCO(1999)Breuer&Krenn(1999)RiskMetricsGroup(1999)臨兩類行率年度例行臨不力行臨力立狀行力理論類說:(SensitiveAnalysis)理利度度率更(ScenarioAnalysis)利行力:()利67()歷利烈什歷()理度行數理識(RiskIgnorance)不()理論(ExtremeValueTheoryEVT)連理論年來金理理論年來量利EVT理論VaR力論EVT量(ModelRisk)量料易度DataAvailability料度(DataCredibility)流料易料料不年來念理度立料不論理論量都念利料不料度量精度量不識更量數流力68不更度理量不量1994年J.P.MorganVaR來露金VaR理露精VaR行金金更率金率金率69