本文以中国上证综指、德国法兰克福DAX指数、美国S&P500指数为研究对象,分别运用DCC-GARCH及时变Copula-EVT模型建模,探讨了欧债危机爆发后股市间相依关系的变化状况。在此基础上,将三个股指收益两两组合,分别建立了各类模型假定下的资产组合预期损失ES模型,并通过后验分析方法,探讨了危机爆发后,各类ES风险模型测度精度的变化状况及对比结果。实证研究表明:欧债危机爆发后,时变Copula-EVT-ES的风险测度准确度明显高于DCC-GARCH-ES模型;边缘分布模型的选择对于时变Copula-EVT-ES模型的测度精度具有重要影响。综合对比分析发现,在金融市场极端波动的状况下,能够捕捉杠杆效应且善于刻画厚尾特征的时变tCopula-AR(1)-GJR(1,1)-EVT-ES模型能够取得相对较好的风险测度效果。
DCC-GARCH and time-varying Copula-EVT models are constructed respectively to discuss the changes of dependencies between stock markets after the outbreak of European Debt Crisis. 2-2 combinations of the stock index returns are made, the portfolio ES models are established under various models as- sumed, and the measurement accuracy of all ES models are compared and discussed after the crisis through backtesting analysis. China Shanghai Composite Index, the Frankfurt DAX index of Germany and S&P500 Index of the United States are used to make empirical experiment. Empirical studies show that after the crisis, the risk measure accuracy of time-varying Copula-EVT-ES models are significantly higher than DCC-GARCH-ES. Further, it is important to select marginal distribution for time-varying Copula-EVT-ES models. Finally,under extremely volatile financial market conditions, the time-varying t-Copula-AR (1)- GJR(1,1)-EVT-ES model, which is good at capturing the leverage effect and characterizing fat tails, a- chieves relatively better risk measure results.