在研究金融资产的组合风险分析中,描述多个金融资产间的相关结构是选择最优组合权重的关键因素之一,如何准确地刻画金融资产间的非对称尾部相关结构,在定量研究组合资产的风险分析中尤其重要。利用多元阿基米德Copula捕捉多个金融资产间的相关结构,并用非参数核密度估计描述单个金融资产的边缘分布,建立Copula—Kernel模型。利用该模型和VaR风险测度,对中国股票市场的实际组合投资问题进行风险分析,并在风险最小原则下,给出不同置信水平下的最小VaR值及其对应的最优组合权重系数。
In modern portfolio optimization and risk management theory, it has been well known that the dependence among financial asset returns is one of the key factors in choosing optimal portfolio weights. Especially, it is very important to portray the asymmetric dependence among financial asset returns in studying the portfolio investment quantitatively. In this paper, the multivariate Archimedean Copula is used to analyze the asymmetric dependence structure among financial asset returns, whose marginal processes are captured by nonparametric kernel density estimation. Then, a Copula-Kernel model is built for risk analysis of portfolio investment. By this model and the risk measure VaR, empirical portfolio risk analysis is made in Chinese stock market. At last, the mini- VaR value of different confidence levels and the relative optimal investment weights are given under the principle of mini, risk.