The Predictability of Chinese Stock Market Volatility: An Analysis Using Non-linear GARCH Models

Abstract

This paper examines volatility forecasting for the Shanghai Stock Exchange Composite Index and Shenzhen Stock Exchange Component Index as well as separate analysis for Shanghai and Shenzhen A- and B-share markets over various subsamples covering 1992-2014. Four GARCH processes that are popular in empirical finance and capture different stylized facts about financial time series are compared based on their ability to accurately forecast volatility. Despite an earlier conclusion that the GJR model is not useful in Chinese volatility modeling, the paper finds that the GARCH, GJR and NGARCHK models all perform relatively well for the markets examined. The GARCH-M model significantly underperforms. Results also suggest that the GJR and GARCH models systematically perform better at different times and that there is a difference in the way volatility should be modeled between A- and B-share markets, even over the same time period.