运用均值-方差、均值-VaR与均值-CVaR模型对上证综指、香港恒生指数、台湾加权指数、标准普尔指数和日经指数的收益-风险进行了实证分析。结果表明,在收益服从正态分布下,均值-方差、均值-VaR以及均值-CVaR模型能用一个模型统一表示,三个模型的边界方程也能用一个方程统一表示,三个模型的有效前沿存在子集关系。在收益服从正态分布下三个模型的最优组合投资权重是等价的;在任意分布下三个模型的最优组合投资权重不是等价的。
Three models, mean-variance, mean-VaR and mean-CVaR, are often applied to portfolio selection for five stock market indices. The result shows that the throe models can be integrated into the same return-risk analysis frame under normal distribution assumption. Furthermore, the bound and efficiency frontier are analyzed in the integrated frame, and there is subset relationship among the efficiency frontiers of the three models. The optimal portfolio weights of the three model is not equivalent under other distribution assumption.