中文摘要：

将在刻画变量之间相关性方面具有优势的Copula函数与能够准确评估市场极端情况的压力测试两种方法相结合，通过度量市场危机发生的期望等待时闻，实证检验不同Copula函数对投资组合压力测试结果的影响。在实证过程中，首先选择上证A股、深成A股两个市场指数作为研究样本，构造能反映单变量极值的广义帕累托边际分布，然后采用不同形式的Copula函数刻画变量之间的相关性结构，利用压力测试的方法考察Copula函数的选择对刻画投资组合风险的影响。实证检验表明，在度量两个市场同时出现危机情况的等待时间时，拟合效果较好与拟合效果较差的两种Copula函数对风险的估计存在显著的差别。这一结论表明，在压力测试的应用中引入Copula函数为组合投资风险管理指出了一个新的方向，但仍需要在描述相关性过程中慎重考虑Copula函数的选择。

英文摘要：

In this paper, we combine Copula function by which the dependency structure can be well captured, with stress test which is the best risk measure in the extreme situation, and consider the effect of different Copula on the portfolio porffolio＇s stress test by measuring the expectant waiting time. In the empirical process, we choose shanghai A stock index and Shenzhen A stock index as the sample, use extreme value theory to model the marginal distribution, and compare how the different Copula functions affect the risk of portfolio through stress test. The stress test shows that when two markets are both in crisis time, the difference is significant between the Copulas with the best goodness of fit and the worst goodness of fit. This result indicates that the method that we combine the Copula function with the stress test shows us a new direction of using Copula. However when we use it, we should consider to select the best goodness of fit.