基于AAVS_CAViaR模型的股市风险测量研究

25320106JOURNALOFSYSTEMSENGINEERINGVo.l25No.3Jun.2010AAVSCAViaR王新宇1,宋学锋1,吴瑞明2(1.,221116;2.,200052):考虑金融资产收益与正负收益对分位数冲击的不对称性,建立含有不对称绝对值和斜率设定的AAVSCAViaR模型,采用混合最优化方法估计模型的参数并进行显著性检验.对19962008年期间沪深港股票指数的实证研究表明:沪深港股票市场均存在正负收益消息对市场风险冲击的不对称现象,且在不同VaR置信水平下存在差异;通过Granger因果检验发现沪深港股市的风险变化存在显著的互动传导关系;提出了VaR预测模型的评估准则,对所选样本数据而言AAVSCAViaR模型比间接GARCH模型更好地描述了市场风险的演化模式.:CAViaR;;:F224.7;F830.9:A:1000-5781(2010)03-0326-08MeasuringstockmarketriskbasedonAAVSCAViaRmodelWANGXinyu1,SONGXuefeng1,WURuiming2(1.SchoolofManagemen,tChinaUniversityofMiningandTechnology,Xuzhou221116,China;2.AntaiCollegeofEconomics&Managemen,tShanghaiJiaotongUniversity,Shanghai200052,China)Abstract:Consideringtheasymmetryofreturndistributionsandasymmetricimpactofpositiveandnegativereturnsonthequantiles,thepaperputsforthanewconditionalautoregressivevalueatriskbyregressionquantilesmodelwithasymmetricabsolutevaluesandslopequantilespecificationcalledAAVSCAViaRmode.lAmixedoptimizationalgorithmisusedtoestimateandtesttheparameters!significance,andthenanempiricalstudyontheevolutionpatternsofmarketriskinShangha,iShenzhenandHongKongstockmarketsisperformed.Itisfoundthattheasymmetricimpactsofpricenewsonthequantilesofreturnsexistinallthreemarkets,howevertheintensityofimpactsvarieswiththeconfidencelevelofvalueatrisk,andthatthesignificantinteractivetransmissionrelationshipsbetweenvariationsofmarketriskinthreestockmarketsarealsodiscoveredbyusingGrangercausalitytes.tFinallyaruletoselectproperVaRforecastingmodelissuggested,accordingtoi,tAAVSCaViaRmodelperformsbetterthanindirectGARCHmodelfortheselectedsample.Keywords:CAViaR;newsimpactcurve;measuringmarketrisk:2007-07-23;:2009-10-27.:(70601032).0VaR(valueatrisk),.,,VaR?,1978Koenker[1](quantileregression,QR).,,.QR,.,VaR,QRVaR,VaR,VaR.Engle[2]CAViaR(conditionalautoregressivevalueatriskbyregressionquantiles),VaR.Park[3].Chen[4]Nikkei225VaR,GARCHt.2005Koenker∀QuantileRegression#[5]..[6].[7]CAViaR.,Buchinsky[8-9].CAViaR,QRCAViaR.CAViaRCAViaR,,.1VaR1.1CAViaRCAViaR,VaR,CAViaR,,.,VaR,.{y1,∃,yT},ft()%ft(|t),t,.,VaRttVaR,,ft()=-VaRt.Engle[2]VaRt,VaRt=!0+&li=1!iVaRt-i+&mj=1∀j#(xt-j,∃)(1)lVaRt-i,i=1,2,∃,l,mxt-j(yt),j=1,2,∃,m.(1),(1)(2).Min&Tt=1%(yt-ft())=&Tt=1%(yt+VaRt)VaRt=!0+&li=1!iVaRt-i+&mj=1∀j#(xt-j,∃)t=1,2,∃,n(2)%(u)=u(-I(u)),I(u)=0,u∋01,u0Engle[2]NelderMeadquasiNewton.1.2AAVSCAViaRVaR,(asymmetricabsolutevalueandslope,AAVS),VaR(),3273:AAVSCAViaR(),VaR,(x)+=max(x,0),(x)-=min(x,0)AAVSCAViaRVaRt=0+1VaRt-1+2(yt-1-4)+-3(yt-1-4)-(3)(3),Engle[2]VaRt:1)2=3,4=0,(symmetricabsolutevalue);2)2=3,4(0,(asymmetricabsolutevalue);3)2(3,4=0,(asymmetricslope).2(3,4(0,.,Engle[2]GARCH(1,1)(indirectGARCH(1,1),IG)VaRt=(0+1VaR2t-1+2y2t-1)1/2(4)Engle[2],GARCH(1,1),.GARCH,.1.3yt=ft(|t)+&t(5)&tQ(&t|t)=0,ft(|t)yt.,Engle[2](2)^,^P,TA-1/2TDT(^-)dN(0,I)(6)AT=E(T-1(1-)&Tt=1)ft()ft())(7)DT=E(T-1&Tt=1ht(0|t))ft()ft())(8)ht(0|t)&t.c^T,∋∋^=T-1D^-1TA^TD^-1T(9)A^T=T-1(1-))f(^)f(^)PAT(10)D^T=(2Tc^T)-1&Tt=1I(|yt-ft(^)|c^T)∗)ft(^)ft(^)PDT(11)22.1CAViaRVaR,,VaR.,VaRVaR,,.,1996-12-16,2008-11-11(SSEC)(SZSC),,(HSI),.,500(2006-10-09).VaR(backtesting),[2]DQ(dynamicquantiletest)RQ(regressionquantilefunctionvalue)(^,VaR(hits);Christoffersen[10]:LRucLRindLRcc.Engle[11](newsimpactcurves)(),.,.2.22.2.1模型参数估计与分析1,32825JarqueBera.,VaR.1Table1DescriptivestatisticsofselectedsampleJarqueBeraHSI0.0000290.0003720.0191720.37980215.8961519352.020SSEC0.0002200.0004750.017841-0.1928557.6928852571.022SZSC0.0001720.0002640.019560-0.1886876.9949461867.16219962008,(2006-10-09),,AAVSCAViaR.2CAViaRRQVaR,DQ2∀P#.VaR0300.99%VaRAAVSCAViaR,1-3.,:1)VaR395%99%6RQ.RQGARCH(indirectGarch)RQ,.2)VaR,95%99%CAViaR,12DQ.AAVS3,GARCH6,DQ.3)95%99%VaR(backtesting)(^,2,GARCHAAVS5%1%;,95%,AAVSVaRIG.LRucLRindLRccAAVSIG99%;95%,LRind,IGLRucLRcc,AAVSHSILRucLRcc.4),(2(3),,VaR;,,(3(0).(yt-1-4)+-(yt-1-4)-,(yt-1-4)+∋0-(yt-1-4)-∋0.2,3.,1011,,VaR0/(1-1).40,99%VaR,95%VaR,VaR,.,30,203,VaR,,32,VaR,.,VaR,.2.2.2沪深港市场风险变化的传导性分析,3293:AAVSCAViaR,.,)VaRHSIt)VaRSSECt)VaRSZSCt,ADF-52.604,-22.871,-22.857,1%,.(Grangercausality)-.,FPE(finalpredictionerror)AIC(Akaikeinformationcriterion),3.2Table2Parameters!estimatesandsignificancetests99%VaR95%VaRAAAVSIGARCHAAAVSIGARCHHSISSECSZSCHSISSECSZSCHSISSECSZSCHSISSECSZSC00.1330-0.0029-0.00550.14600.01350.00390.10370.18630.14360.11850.67900.3348S.D.0.05730.01300.00970.14720.04630.07310.04340.10020.10700.04280.29930.1847P-0.01010.41150.28440.16050.38520.47860.00850.03150.08970.00280.01160.034910.89650.98490.97410.85800.98570.95700.91090.80530.84370.89210.74000.8564S.D.0.02490.00470.00660.01640.00390.00800.01660.06360.06840.00690.06740.0278P-00000000000000000000000000000000000000000000000000000000000020.09260.02420.06211.04340.08920.33390.00170.17060.12700.24750.36940.2293S.D.0.10550.00760.01640.81170.56540.32240.05200.04600.05100.07330.16590.6467P-0.19010.00070.00010.09930.43730.15020.48710.00010.00640.00040.01300.361530.44690.14850.17260.35250.33490.3514S.D.0.03820.08180.11020.04070.12330.0941P-000000.03470.0588000000.00330.00014-0.3539-1.2308-0.7676-0.6441-0.1326-0.3166S.D.0.21600.33840.53770.22500.