沪深股市相关结构之谜：基于贝叶斯 Copula 的研究

摘要

目前沪深股市相关结构的Copula模型选择差异很大，并没有形成统一的认识。在指出现有Copula检验要受到模型参数估计影响后，引入了贝叶斯估计方法将模型参数估计与拟合优度检验有效的分开。接着，沪深股市相关性的贝叶斯实证结果发现两市相关结构Copula模型具有时变特征，势必导致当前研究结果的不一致；同时也反映了Copula对样本区间选择有很强的依赖性。%There are different opinions about the correlation structure between shanghai and shenzhen stock mar -ket based on copula .After the fact that the goodness of copula fit test is affected by parameter estimation , the Bayesian copula selection is introduced , which can separate parameter estimation from the goodness of fit test . Then , the empirical results by the Bayesian copula selection show that the correlation structure between shanghai and shenzhen stock market is time-varying , which causes the difference of the results .