Tree structured DCC_multivariate GARCH model and its application in volatility correlation analysis of Shanghai, Shenzhen and Hong Kong stock markets

In oder to analyze the volatility asymmetry and volatility correlation of Mainland stock market and Hong Kong stock market, this paper attempts to apply MCMC algorithm to estimating the most probable tree structured DCC_Multivariate GARCH(1,1,1,1) model for daily returns of Shanghai, Shenzhen and Hong Kong stock markets. In the paper, the prediction performance of the most probable tree structured and general DCC_multivariate GARCH models were compared. The results show that the most probable tree structured DCC_multivariate GARCH model has better prediction performance. What's more, the predicted graphs of volatility and volatility correlation were analyzed, which indicates that asymmetry in the volatility of the three stock markets and volatility correlation among them does exist, Mainland stock market fluctuates more frequently than Hong Kong stock market and the correlation between Mainland stock market and Hong Kong stock market has been increasing since 2005.