中文摘要：

为了分析由极端事件所引起的巨额损失变量之间的相依关系，本文引入了比一般copula函数更有效的极值copula和上尾copula。我们介绍copula的对角截面以确定上尾相依系数。基于极大似然法，讨论了关于这些copula函数类的半参数估计方法。通过构建Cramer—VouMises统计量对copula的拟合优度进行假设检验。在实证分析部分，我们通过具体的实例来说明，在应用研究中该如何选取最优的copula以描述变量之间的相关性。

英文摘要：

In order to anMyze the dependence between the catastrophe losses caused by extreme events, extreme value copulas and upper tail copulas are introduced, which play a more flexible role than copulas. We present the diagonal section of copulas for the purpose of determining the upper tail dependence coef- ficient. We study the semiparametric method of estimating these parametric copula families by maximum likelihood method. The goodness-of-fit testing procedures for them are also discussed by constructing Cramer-Von Mises statistic. In the empirical analysis, we illustrate the results by a numerical example, which tells us how to choose the fitted-best copula to model dependence in application studies.