系统性风险度量CoES的建模和检验

本文结合极值理论(Extreme Value Theory,EVT)和新的动态混合Copula(Dynamic Mixture Copula,DM-Copula)函数,提出了一种新的CoES估计方法DM-Copula-EVT。在EVT建模中,本文改进了阈值的选取方法以避免选择的主观性,并提出了一系列新的动态混合Copula以更好地刻画金融市场日益复杂的尾部关联性。此外,本文首次提出了检验CoES模型设定正确性的后验分析方法,包括无条件覆盖性检验和条件覆盖性检验。将本文建模和检验方法应用于我国金融市场,研究发现:相对于传统使用的t分布,EVT能更好地拟合指数的尾部分布;新的动态混合Copula函数能更好地刻画金融部门与系统之间的复杂关联性。

Modeling and Backtesting CoES for Systemic Risk Measure

In this paper, a new CoES estimation method DM-copula-EVT is proposed by combining extreme value theory (EVT) and the new dynamic mixture copula (DM-copula). In the modeling of EVT, this paper improves the threshold selection method to avoid the subjectivity of selection, and proposes a series of new dynamic mixture copulas to better describe the complex tail correlation of financial markets. In addition, this research proposes a rigorous backtesting framework for CoES for the first time, including unconditional coverage test and conditional coverage test. Applying the modeling and backtesting methods to China’s financial market, we find that EVT can better fit the tail distribution of financial market indices than the traditional t distribution, and new dynamic mixture copulas can better describe the complex correlation between the financial sector and financial system.