2012年第1期
投资组合风险测度———基于FIGARCH-EVT-Copula方法
·49·
EVT和Copula各自在测度风险的优越性:FIGARCH
能捕捉波动的长记忆性,这是一般GARCH族所不具备的;EVT只是对尾部建模,能准确地刻画分布厚尾性;Copula能灵活地度量边缘分布间的相关结构,是投资组合风险建模的有力工具。将上证指数和深成指数以等权重的方式构成投资组合,基于
结果表明:无论是上海股票市场还是深圳股票市场,其对市场冲击的影响具有持续性,前者对冲击的记忆时间略长于后者。比较FIGARCH-EVT-Copula与
GARCH-EVT-Copula模型预测VaR的效果,发现,
前者的预测效果比后者好,主要是因为前者更能准确地刻画收益率波动的长记忆性特征。
FIGARCH-EVT-Copula模型进行了实证研究。实证
参考文献:
[1]雷鸣,缪柏其.运用生存分析与极值理论对上证指数的研究[J].数量经济技术经济研究,2004(11):130-137.
[2]BaillieRT,BollerslevT,MIKKelsenHO.Fractionallyintegratedgeneralizedautoregressiveconditionalheteroskedasticity[J].JournalofEconometrics,1996,74(1):3-30.
[3]SibbertsenP.LongmemoryinvolatilitiesofGermanstockreturns[J].EmpiricalEconomics,2004,29(3):477-488.
[4]WuPT,ShiehSJ.Value-at-riskanalysisforlong-terminterestratefutures:fat-tailandlongmemoryinreturninnovations[J].
JournalofEmpiricalFinance,2007,14(2):248-259.
[5]AlouiC,MabroukS.Value-at-riskestimationsofenergycommoditiesvialong-memory,asymmetryandfat-tailedGARCHmodels[J].EnergyPolicy,2010,38(5):2326-2339.
[6]汤果,何晓群,顾岚.FIGARCH模型对股市收益长记忆性的实证分析[J].统计研究,1999(7):39-42.[7]王春峰,张庆翠.中国股市波动性过程中的长期记忆性实证研究[J].系统工程,2004,22(1):78-83.
[8]肖智,傅肖肖,钟波.基于EVT-POT-FIGARCH的动态VaR风险测度[J].南开管理评论,2008,11(4):100-104.
[9]EmbrechtsP,McneilA,StraumannD.Riskmanagement:valueatriskandbeyond[M].Cambridge:CambridgeUniversityPress,
1999:176-223.
[10]JondeauE,RockingerM.TheCopula-GARCHmodelofconditionaldependencies:aninternationalstockmarketapplication[J].
JournalofInternationalMoneyandFinance,2006,25(5):827-853.
——Copula-GARCH模型及其应用[J].系统工程,2004,22(4):7-12.[11]韦艳华,张世英.金融市场的相关性分析—
[12]MendesBVM,KolevN.Howlongmemoryinvolatilityaffectstruedependencestructure[J].InternationalReviewofFinancial
Analysis,2008,17(5):1070-1086.
[13]WangZR,ChenXH,JinYB,ZhouYJ.Estimatingriskofforeignexchangeportfolio:usingVaRandCVaRbasedon
GARCH-EVT-Copulamodel[J].PhysicaA,2010,389(21):4918-4928.
[14]应益荣,詹炜.资产组合ES风险测度的Copula-EVT算法[J].系统管理学报,2007,16(6):602-606.[15]傅强,邢琳琳.基于极值理论和Copula函数的条件VaR计算[J].系统工程学报,2009,24(5):531-537.[16]罗登跃,王玉华.上海股市收益率和波动性长记忆特征实证研究[J].金融研究,2005(11):109-116.
[17]NeftciSN.Valueatriskcalculation,extremeevents,andtailestimation[J].JournalofDerivatives,2000,7(3):23-38.[18]叶五一,缪柏其,吴振翔.基于Bootrap方法的VaR计算[J].系统工程学报,2004,19(5):528-531.
TheResearchonPortfolioRiskMeasurementBasedonFIGARCH-EVT-Copula
JIANGHongli,HEJianmin,ZHUANGYaming,ZHANGYuefeng
(SchoolofEconomics&Management,SoutheastUniversity,Nanjing211189,China)
Abstract:Itiswellknownthatfinancialequityhassharp-peaks,fat-tails,heteroskedasticityandlongmemory.Consideringthesethreefeatures,thisarticleconstructsariskmeasuremodelbasedontheFIGARCH-EVT-Copulaforfinancialportfolio.TheVaRandESriskmeasurebasedontheFIGARCH-EVT-Copulaisappliedontheportfolio,whichiscomposedbyShanghaiStockindexandShenzhenComponentIndexequalweight.TheempiricalresultsshowthatthereisapparentlongmemorypropertyinChinesestockmarket.Theresultsalsoshowthat,themodelofFIGARCH-EVT-Copulareallycancapturethepropertiesofsharp-peaks,fat-tails,heteroskedasticityandlongmemory,andprovesthatthemodelofFIGARCH-EVT-Copulaismoreefficiencythantraditionalmodelinmeasuretheportfolioriskwhosemarginaldistributionhasthepropertyoflongmemory.Keywords:FIGARCH;EVT;Copula;VaR;ES;portfolio
[责任编辑:孟青]