统计研究 ›› 2018, Vol. 35 ›› Issue (3): 3-13.doi: 10.19343/j.cnki.11-1302/c.2018.03.001

• •    下一篇

相依结构、动态系统性风险测度与后验分析

王锦阳等   

  • 出版日期:2018-03-25 发布日期:2018-03-25

Dependence Structure, Dynamic Systemic Risk Measurement and Backtesting Analysis

Wang Jinyang et al.   

  • Online:2018-03-25 Published:2018-03-25

摘要: 本文利用Copula相依结构理论扩展和求解了现有的系统性风险测度CoVaR,以得到适用于不同类型常参数和时变参数Copula函数及不同分布假设的动态系统性风险测度。为了验证和评估模型设定的准确性与应用价值,我们构建了适用于该动态系统性风险测度CoVaR的严谨后验分析工具。除“无条件覆盖性”、“独立性”和“条件覆盖性”外,我们首次提出了“混合独立性”检验。基于中国14家上市商业银行的实证分析表明:中国上市商业银行与中国银行业之间的相依结构呈现多样化特征;无论是样本内还是样本外预测区间,我们的动态Copula-CoVaR模型能够有效地捕捉典型系统性风险事件;严谨的后验分析不仅需要检验系统性风险测度CoVaR,也需要检验条件事件的临界值VaR。

关键词: 相依结构, 条件在险价值CoVaR, 系统性风险, 后验分析

Abstract: Using the the Copula dependence structure theory, the paper extends the existing systemic risk measurement CoVaR, and obtains the dynamic systemic risk measurement which is suitable to different Copula functions and different distribution assumptions. At the same time, we construct a rigorous backtesting analysis tools for the dynamic systemic risk measurement CoVaR. In addition to the unconditional coverage property, independence property, and conditional coverage property, we first propose the cross independence property for backtesting the systemic risk measurement CoVaR. In the empirical study of 14 Chinese listed commercial banks, we show that: The dependence structure between the listed bank and bank system is diversified; Our dynamic copula CoVaR model can effectively capture the systemic risk events in the both in-sample period and out-of-sample period; We argue that one needs to apply the rigorous backtesting analyses not only to the systemic risk measurement CoVaR, but also to the critical value of the conditional event, VaR.

Key words: Dependence Structure, Conditional Value at Risk(CoVaR), Systemic Risk, Backtesting Analysis