摘要
沪深股市与香港股市近年来分割状况改善,出现"一体化"发展趋势.测定两市相依结构和波动溢出效应对于市场参与各方具有重要意义.结合传统的协整检验、Granger因果检验和VEC模型,引入混合copula函数对两市代表性指数——沪深300指数和恒生中国企业指数进行相关性分析.参数估计方面,利用非参数核密度方法估计边缘分布,结合EM算法计算混合copula的参数.实证结果表明两个市场具有显著相关性,且上尾相关性大于下尾相关性;混合copula函数与单copula函数分析结果一致,拟合效果优于单copula函数.
The gap between the mainland stock market and Hong Kong stock market has been reduced m recent years and there is a tendency toward their ＂integration＂. It is very important for all market participants to measure the degrees and patten of dependence and the volatility spillover between the two markets. Mixedcopula was established, as well as cointegration test, Granger causality test, the VEC model and also the single copula, to analyze the correlation between the representative indices of the two markets. The method of nonparameter kernel density estimation was used to estimate marginal distribution and EM algorithm was used to estimate the parameters of mixedcopula. The empirical results show that the two markets have larger upper tail dependence than lower tail dependence and that the mixedcopula has the best goodnessoffit test effects result among selected copulas.
