中国股市风格资产组合市场风险测度——基于EVT-t-Copula模型的实证研究

针对传统线性计量模型无法充分刻画中国股市风格资产收益系列存在"自相关"、"尖峰"、"厚尾"等非正态分布特征以及风格资产间存在复杂的非线性相关结构的局限。本文首先采用AR(1)-GJR(1,1)模型来刻画中国股市风格资产的边缘分布,接着结合各边缘分布的残差系列,引入Copula函数来分析这六种风格资产之间的相关结构,并结合极值理论和蒙特卡罗模拟方法来模拟大盘成长、大盘价值、中盘成长、中盘价值、小盘成长、小盘价值这六种股市风格资产投资组合的联合收益率分布函数,在此基础上求出各我国股市风格资产组合的市场风险(VaR与CVaR)。研究结果表明,根据极值理论得到的广义帕累托分布能够较好拟合风格资产日收益率序列的尾部特征,相比其他计算方法的VaR和CVaR值,基于EVT-t-Copula模型能够更准确度量中国股市风格资产组合的市场风险。因此,EVT-t-Copula模型有助于提高中国股市投资组合的风险管理效率。

The Risk Measurement of Style Asset Portfolio Market in China：The Empirical Research based on EVT-t-Copula Model

The traditional linear model cannot adequately describe the gains series of style assets in Chinese stock market and has autocorrelation,spike,fat tails and other non-normal distribution features;and there exists complex nonlinearly related structure between style assets.Firstly,we apply the AR （1 ）-GJR （1 .1 ）model to character-ize the marginal distributions of the style assets in Chinese stock market,and with marginal distribution of residuals series,we introduce copula function to analyze the related structure among these six kinds of style assets.By ex-treme value theory and Monte Carlo simulation method,we simulate the joint return distribution of various style as-sets,and eventually get the market risk of style assets portfolio in Chinese stock market （VaR and CVaR）.There-fore,EVT-t-Copula model helps to improve risk management efficiency of portfolio in Chinese stock market.