Moment-Based Copula Tests for Financial Returns
提出一类基于矩的检验方法，用于参数化多元动态背景下金融收益率的Copula模型，考虑估计不确定性的影响，可生成尾部依赖检验等，帮助判断Copula是否准确刻画真实依赖结构，对风险管理等有实用价值。
We propose a class of moment-based tests for copulas in a parametric multivariate dynamic context for financial returns. The proposed method takes into account the effect of estimation uncertainty. This effect is quite important but often is ignored in related studies. Our method can be applied to generate various tests to detect copula misspecification in different directions. In particular, on the basis of the conditional probabilities of quantile exceedances (Kendall's tau), it generates the tail-dependence tests (the concordance test) that can be used to investigate whether the copula being tested is suitable for characterizing the true tail dependence (concordance) structure. Such tests may be useful for exploring the cross-dependence structures of financial returns that are essential for risk management and other purposes. The Monte Carlo simulation supports the validity of our method. As a demonstrative application, we also apply these tests to an empirical study of stock market relationships.