摘要
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.
出处
《中国管理科学》
CSSCI
北大核心
2014年第5期8-15,共8页
Chinese Journal of Management Science
基金
国家自然科学基金资助项目(71071131
71090402
71371157)
高等学校博士学科点专项科研基金资助课题(20120184110020)
中央高校基本科研业务费专项资金资助项目(SWJTU11ZT30
SWJTU11CX137
SWJTU12CX120)
国家社会科学基金(12BGL024)
成都理工大学金融与投资科研创新团队(KYTD201303)
四川省软科学研究计划项目(2013ZR0068)
成都理工大学中青年骨干教师培养计划项目(JXGG201420)
四川省教育厅人文社科重点项目(14SA0039)
