中国上市公司信用风险管理实证研究——KMV模型的应用

华中科技大学硕士学位论文中国上市公司信用风险管理实证研究——KMV模型的应用姓名：陈梅申请学位级别：硕士专业：企业管理指导教师：陈君宁20040229IIKMVJ.P.MorganCreditMetricsCreditPortfolioViewKMVGARCHEsBSMKMVKMVDemoAsEsAVAsEDFKMVEDFKMVKMVGARCHEDFIIIABSTRACTInourcountry,therehasbeentheadagethat“wewillbuildourdefensesbeyondchallenge”.Manyinternationalfinancialinstitutionstakethemanagementofcreditriskastheimportanttoolthatinsuresthemsurvivalanddevelopinthecomplexcircumstances.Establishingthepredictmodelwillbehelpfultothedecisionmaker,andtheycanmakedecisionbasetheobjectivedateto.Thebiggestriskthatfinancialinstitutionfacedisthecreditrisk,andhowtoimprovethemanagementlevelofcreditriskhasbecometheurgenttaskforthefinancialinstitutionsincewehavebecomethememberofWTO.Thispaperreviewsthedevelopmentandgeneralmethodsofcreditratingbrieflybutsystematically,andanalysestheprevailingcreditmodelssuchasCreditMetrics,CreditRisk+,andCreditPortfolioViewindetail.ItdiscussestheuseoftheOptionPricingTheorytomeasurecreditriskandthenresearchestheframe,theparameterestimationofKMVmodel.ThispaperproposesusingtheGARCHmethodstoestimateEsandmakinguseoftheBSMmodelandtherelationfunctionbetweenAsandEswhichispublicizedintheDemoofKMVtoestimateAsandAV,thendeducingtheExpectedDefaultFrequency(EDF).FinallyitdoessomecorrelativedemonstrationandparallelingresearchusingthedateofChinesestockmarket.KMVmodelhasitsspecialsignificanceinChinesemarketshortingofdata.ThispaperproposesthededucingprocedureofEDFandrelatingresolvemethodsonthebasisoftheresearchingresultofformer,anddoessomeempiricalresearchinthelistedcompaniesofChina.ThisresearchgivesanewwayofconceptualizingandthemethodfortheapplicationofOptionPricingTheorytomeasurecreditriskandanewframework.KeyWords:OptionTheoriesKMVGARCHEDFI_____1DefaultRisk90971.1123Caoutte,Altman,andNarayanan(1998)[1]1.1.125C5CCharacterCapitalCapabilityCollateralCycle1234561.1.2CreditRatingSystemOCCOCCOCC1-91-101.1.3Fitzpatrick3[2]Beaver[3]1954-1964797930/BeaverAltman,22[4~6]1946-198633Z-ScoreZ-Score1.812.991.812.9998%72%[78]AltmanDeekinAltman1010ProbitLogisticLogistic410(1)(2)Z-scoreZ-score(3)1.21.2.1ArtificialIntelligentSystemBP51.2.2Jonkhart(1979)IbenLiterman(1989)(1)(2);(3);(4)1.2.3,,,[9]1.3KMVEDFJ.PMorganCreditMetrics,(CreditSwissFirstBoston,CSFB)CreditRisk+,MckinseyConsultingCreditPortfolioView6KMV1.3.1CreditMetricsCreditMetricsJ.P.Morgan19974CreditMetrics[10]CreditMetrics1-11-1CreditMetricsCreditMetricsVAR7VARCreditMetricsCreditMetricsJ.P.MorganCreditManagement12341.3.2CreditRisk+CreditRisk+(CreditSwissFirstBoston,CSFB)1996[1112]CreditRisk+CreditRisk+8CreditRisk+CreditRisk+CreditRisk+CreditRisk+CreditRisk+ConstantTimeHorizonHold-to-MaturityCreditRisk+CreditExposureRecoveryRateCreditRisk+CreditRisk+CreditRisk+CreditRisk+Poissonn!)(nennPmm-=(1.1)n9∑=AApm1.2AP=AnCreditRisk+IncrementalCreditReserveICR=100%-1.3CreditRisk+123CreditRisk+CreditRisk+CreditRisk+CreditRisk+CreditRisk+1.3.3CreditPortfolioViewCreditPortfolioViewMckinsey1998[1314]CreditPortfolioViewt/itiYtieP,11,-+=tiY,i10∑=++=mjtitjijiitiuXaaY1,,,,0,,tjiX,,tCreditPortfolioView(1)Cholesky(2)1(3)(4)CreditPortfolioViewCreditPortfolioViewCreditPortfolioViewCreditPortfolioViewCreditPortfolioView//[15]11KMVKMV(expecteddefaultfrequency,EDF)[16]KMV2.1KMVBSM()()()30FisherBlackMyronScholesRobertMerton[17]2.1.1B1ackScholesMertonModiglianiMiller12DDD,DDDD,,D,,DV-DDDD02-1RRVVDD2-12-2B1ack-Scho1es(Merton,l974)[18]Black-Scholes132.1.2-DDDD()DD2-2()DDDKMV[1920]()2-3142.2KMVEDF()KMV2.2.115dZVVdVAAAAsm+=2.1AVAdVmAsdZ=f()2.2),,,,()()(21tsDrVfdNDedVNEArt=-=-2.3EDVrN21dd21ddtstsaarDVd)21()ln(21++=2.4tsadd-=122.516as2.2Vas2.2as=g2.6Black-Scholes2.3EsAEE,hs=AAdVdEAVss=2.7AE,hdVdE/DeltaDelta)(1dN),,,,()(1tsssDrVgEVdNAAE==2.82.32.8,EesVsNewton-Raphson⎥⎦⎤⎢⎣⎡--⎥⎥⎥⎥⎦⎤⎢⎢⎢⎢⎣⎡+⎥⎦⎤⎢⎣⎡=⎥⎦⎤⎢⎣⎡-EEAAAAAVgEVfVggVffVVsssdddsddddsdss),(),(1''2.92.2.2,DPDD17KMVKMV50%STDLTDDP=STD+0.5LTDDDDD)(1VEDPADPVEDDs-=)(1tstsmAAADPV)2(ln2-+=2.10m18”EDF2.2.3EDFDD95%d22.5%d22.5%ees,)(1⎥⎦⎤⎢⎣⎡≤--=ArtDPVEpp)1,0(N(2.11)EDFAVAVAsEDF,AVKMVEDF[21]EDF=ss22DPDPKMVEDF3-419EDF()[22]KMVEDFEDF20[23]N3.1SimpleMovingAverage,SWAntstt+n∑+==∧-=nTtTtttrrn)(12s3.1r=0ts∑+-=∧=tntiitrn121s3.21/n,n213.2ExponentiallyWeightedMovingAverage,EWMV∑∞=-∧-=02)1(iititrlls3.3)10(ll(decayfactor)EWMA∧--∧+-=21212)1(tttrslls3.4ll(MSE)RMSE3.3[24]yxx10bb+eebbbb+++++=ppxxxy...221103.5e),0(2sNb2s2212--=∧pnSes3.6∑∧-=2)(iieyyS∧-iiyyiy∧iy3.4ARCHARCH,,,,,,,(VolatilityClustering),(fattails),,R.F.EngleAutoRegressiveConditionallyHeteroscedasticmodels,ARCH[2526]Engle(1982){}tetΩARCH(q)),0(~1ttthN-Ωe3.7∑=-+=qiititaah120e3.823niaai,...,2,1,0,00=th0ARCHteqq1986BollerslevGARCH[27]EngleLilienRobins(1987)ARCH-in-meanNelson1991ExponentialGARCHEGARCHEngleGonzalez-Rivera(1989)ARCH(semi-parametricARCH)HamiltonSusmel(1994)SWARCH-L(k,q)ARCH3.5GARCHGARCH[2829]3.5.1GARCHBollerslevARCHGARCH(GeneralizedAutoRegressiveConditionalHeteroskedasticity),ARCHGARCH∑∑==--++=qipjjtjitithaa11202bes3.90th00aqiai,...,2,1,0=≥24pjj,...2,1,0=≥bGARCH(p,q)pqGARCHqpARCHGARCHGARCHARCHGARCH(1,1)212102--++=tttaabses0(a0)0≥b3.10abGARCH(1,1)),0(~21tttNse-Ω21)(tttVarse=Ω-10)1()(-