论文部分内容阅读
本文首先采用AR(1)-GJR(1,1)-SkT(v,λ)模型来刻画中国股市风格资产(大盘成长、大盘价值、中盘成长、中盘价值、小盘成长、小盘价值)的边缘分布,接着结合各边缘分布的残差系列,针对传统二元Copula函数面临的“维度诅咒”问题及多元Copula函数在刻画多变量联合分布时缺乏灵活性和精确性等局限,重点引入C-Vine Copula和D-Vine Copula函数来分析这六种风格资产之间的相依结构和联合分布,然后通过拟合效果比较选出最佳Vine Copula模型,运用蒙特卡洛模拟方法预测风格资产组合的VaR,最后通过Kupiec和Christoffersen返回检验方法测试其预测效果。研究结果表明:中国股市各风格资产之间的相依性存在结构性差异,最适合用D-Vine Copula模型来刻画各风格资产之间的相依结构。总体上来看,同类型的风格资产之间的相依程度比不同类型之间的相依程度要高;在同一类型的风格资产中,资产规模差距越大的风格资产之间的相依系数就越小。无条件的风格资产收益系列之间的相关性要显著大于有条件的风格资产收益系列之间的相关性;基于Student-t Copula、Clayton-Gumbel Copula具有尾部分布特征的Copula作为构建模块的D-Vine Copula模型能够有效地预测中国股市风格资产组合的VaR,有利于提升投资组合的风险管理能力。
In this paper, AR (1) -GJR (1,1) -SkT (v, λ) model is firstly used to characterize China’s stock market style assets (market growth, market value, midday market growth, midday market value, small- ), And then combined with the residual series of each edge distribution, aiming at the problem of “curse of dimensionality” faced by traditional Bivariate Copula functions and the lack of flexibility and accuracy of multivariate Copula function in describing multivariate joint distribution, The C-Vine Copula and D-Vine Copula functions are introduced to analyze the dependency structure and joint distribution between the six styles of assets. Then the best Vine Copula model is selected by the fitting effect and the style is predicted by Monte-Carlo simulation The VaR of the portfolio is tested by Kupiec and Christoffersen for the test results. The results show that there are structural differences in the interdependencies of various styles of assets in China’s stock market, and the D-Vine Copula model is most suitable for describing the interdependent structures of various styles of assets. Generally speaking, the dependences of the assets of the same type are higher than the dependency of different types of assets; in the same type of assets, the smaller the dependence coefficient is between the assets of the greater asset size gap. The correlation between unconditional style asset returns series is significantly greater than that of conditional style asset returns series. Based on Student-t Copula, Clayton-Gumbel Copula Copula with tail distribution features D-Vine Copula model can effectively predict the VaR of China’s stock market-style portfolio, which is conducive to enhancing the risk management ability of the portfolio.