Variable Structure Correlation Analysis Based on a New Asymmetric Copula

Traditional dependence modeling has predominantly relied on symmetric copulas to capture variable relationships.However, empirical evidence reveals the prevalence of asymmetric dependence structures in real-world data.Addressing this limitation, contemporary research has progressively shifted toward asymmetric copulas for more precise dependence characterization.Building on this development, this paper develops an innovative construction method for asymmetric copulas by extending Archimedean copula theory.Employing a piecewise modeling framework, we empirically validate the proposed method using daily returns of the Shanghai Composite Index (SHCI) and Shenzhen Component Index (SZCI).The findings demonstrate that the mixed variable-structure copula model constructed using the novel methodology provides superior goodness-of-fit relative to a single-structure asymmetric copula model, which itself outperforms traditional symmetric Archimedean copulas.Furthermore, the paper explores the mutual transformation of tail copula functions through survival copulas, broadening the practical scope of asymmetric copula applications.