股票资产组合VaR研究--基于混合D藤Copula模型

摘要

A mixed D-vines copula model based on Pair Copula Constructions method was used to construct the joint distri-bution function of dependency for studying risk of a portfolio of five stocks in Shanghai Stock Exchange by VaR. The model im-proves on traditional D-vines method and chooses the best families of copula functions for every Pair Copula by a rule. It not only takes into account of the impact of dimensions, but also can capture the difference of the correlation among portfolio factors, so it can describe the dependence structure of the multiple asset returns better and also can reflect the actual distribution of the portfo-lio's returns more precisely. Based on the mixed D-vines copula models , we apply the Monte Carlo methodology to study VaR of a portfolio of five stock assets in Shanghai Stock Exchange , and the validity of the model is proved by empirical analysis.%本文在对上证市场五种股票资产组合的风险分析中以VaR作为风险度量指标，采用基于Pair Copula高维建模理论的混合D藤Copula模型，建立了反应多个资产组合相关结构的联合分布模型。该模型对传统D藤Copula建模方法作了进一步的改进，通过一定的选择标准，确定了D藤中每个Pair Copula函数的最优函数族，这样使得所建立的模型不仅考虑到了资产维数的影响，而且还能捕捉到组合内部因子间相关结构的差异性，从而改进后的模型能更好地描述资产组合的相关结构，并且能更精确地反映资产组合收益的实际分布。最后，以混合D藤Copula模型为基础，利用Monte Carlo方法计算了上证市场五种股票资产组合的VaR，并通过实证研究进一步证明了该模型的有效性。