基于Pair-Copula贝叶斯网络模型的金融危机传播效应研究

为克服传统Copula函数建模的复杂性,采用Pair-Copula贝叶斯网络方法,将Copula函数与贝叶斯网络的优势相结合,有效降低了高维Copula函数参数估计的复杂程度,解决了连续型贝叶斯网络非正态下边际分布难以控制的问题,有利于捕捉网络结点间的非对称相依结构,从而构建了国际8个代表性股票市场在次贷危机爆发前、中、后各个时期的概率网络模型,对危机在主要国际市场上的传播效应问题及传播路径识别问题进行实证研究。研究结果表明:金融危机在国际金融市场上具有明显的区域性传播特征;香港市场在金融危机发生前后一直都是亚洲市场的连接枢纽,是风险防范的关键节点;危机发生后欧洲市场与亚洲市场间的紧密程度加强;金融危机减缓了全球一体化的进程,但经济全球化的趋势势不可挡。

Empirical Analysis on the Financial Crisis Spread Effects Based on Pair-copula Bayesian Network

In order to overcome the complexity of traditional modeling methods of Copula function , Pair‐copula Bayesian network method was adopted in this paper .Combining the advantages of Copula function and Bayesian network ,this method effectively decreases the complexity of the high‐dimensional Copula function parameter estimation ,solves the difficulty of the marginal distributions control of non‐normal continuous Bayesian networks , and makes asymmetric dependence structure between network nodes easier to capture .This method was applied on eight representative international stock markets in three periods（before ,during and after the financial crisis broke out ） .The probabilistic network model of them was used to analysis the propagation effects and the propagation path identification of the crisis .The results show s ：financial crisis has obvious regional propagation characteristics in the international financial markets ,Hong Kong market is always a hub which connects the Asian markets before and after the financial crisis and is a key node in risk prevention ,tightness between the European markets and the Asian markets strengthened after the crisis ,and financial crisis has slowed down the process of global integration but the trend of economic globalization is unstoppable .