Abstract: The reaction of Chinese stock market to economic uncertainty has always been an important issue for practitioners and researchers. In theory, stock market tends to be more volatile when economic environment is unstable. With the rapid development of stock market in China, we are concerned about how uncertainty from economics impact stock market. The GARCH-MIDAS introduced by Engle et al. (2013) is employed to investigate whether information contained economic uncertainty can help to predict long-term components of the Chinese A-share and B-share variance. Economic uncertainty used in this paper includes macroeconomic uncertainty and economic policy uncertainty (EPU). Our sample consists of daily stock returns and monthly economic variables from January 2003 through December 2016. ARIMA models are used to remove the trends, leaving residuals as economic uncertainty variables before incorporated into MIDAS specification. The empirical analysis indicates that economic uncertainty would impact stock market volatility in a slight way, and difference exists between A-share and B-share. IP (Industrial Production) growth rate and inflation rate contribute to the volatility of A-share and B-share, IP is the most significant factor. Neither monetary policy nor Chinese EPU contributes to A-share volatility, while they are factors of B-share volatility. American EPU doesn't significantly drive Chinese stock market fluctuate. Furthermore, variance decomposition verifies our conclusion. MIDAS approach is an appropriate way to investigate long-term component and short-term component of stock market volatility, which helps to identify economic factors driving Chinese stock market volatility.

Key words: stock market volatility, macroeconomic uncertainty, economic policy uncertainty, GARCH-MIDAS