An Analysis on the Structural Breaks in Dynamic Conditional Correlations Among Equity Markets Based on the ICSS Algorithm : The Case from 2015-2016 Chinese Stock Market Turmoil

By employing multivariate GARCH models to stock returns and AR-GARCH models with change points identified by ICSS algorithm to predicted dynamics conditional correlations, this paper seeks to analyze information shocks on daily return dynamics in three Chinese segmented stock markets, namely Shanghai, Shenzhen and Hong Kong, during the sample period 2014-2016. The paper finds the significant evidence of contagion in the time-varying correlation between Shanghai and Shenzhen stock market as well as the statistically and economically significant decrease in the correlation series of Shanghai-Hong Kong and Shenzhen and Hong Kong after the burst of the stock market bubble and the onset of Shanghai-Hong Kong Stock Connect, respectively, which proves that the constant conditional correlation model, even following the heteroskedasticity-adjusted method of Forbes & Rigobon (2002), may fail to capture the time-varying feature, to reflect the reality of dynamic market conditions or to arrive at a correct conclusion on financial contagion in the stock markets