投资组合极值风险测度——基于T-Copula-SV-T-EVT模型

Risk Measurement of Financial Portfolio Based on T-Copula-SV-T-EVT Model

针对金融资产收益率的"典型事实"特征,结合Copula函数、极值理论与金融波动模型构建既能反映投资组合金融资产收益率分布特征又能反映其相关结构的风险测度模型,用Monte Carlo模拟方法测度投资组合风险,以东方策略成长基金为例进行实证检验。研究发现,边缘分布与Copula函数的选择均会对投资组合的风险产生影响,通过对不同边缘分布和Copula函数组成的投资组合模型风险测度的对比,发现T-Copula-SV-T-EVT模型更具优越性,同时返回式检验表明T-Copula-SV-T-EVT模型对风险的度量是合理而有效的。

According to the 'stylized facts'characteristic of financial asset returns, combining Copula functions, Extreme Risk Theory and SV model, this paper constructed the risk measurement model which could reflect both the characteristics and the correlation structure of the financial asset returns. The combination of Monte Carlo simulations was applied to measure the portfolio risks. Empirical studies of the Oriental Strategy growth fund were carried out. The results show that marginal distribution function selection and Copula marginal distribution choice have impact on portfolio risks. By comparing different risk measurement models that have different marginal distributions and Copula functions, we find that the T-Copula-SV-T-EVT model is more superior while test indicates this model can both effectively and reasonably measure financial market risks.