Abstract: In this paper, ICSS: MV algorithm is used to detect structural breakpoints of daily return series of Shanghai composite index and Shenzhen component index from Dec 12, 1996 to Dec 31, 2010. By corresponding structural breakpoints and major events, all the policy events among the major events are selected, and then according to these policy events, the sample period is divided into thirteen sub-intervals. In order to avoid the shortfalls caused by misspecification using parametric models, nonparametric GARCH model is used to estimate the volatility. Finally, N-W kernel regression of the volatility and the daily return is conducted, and furthermore the policy factors that cause the structural volatility of stock market are analyzed. It is concluded that the adjustment of the benchmark interest rate of deposit and loan as well as the deposit reserve ratio, the reduction of the state-owned shares, the permission of institutional investors to trade mutual funds, the adjustment of the stamp duty and so on are policy impact factors that cause structural volatility of Shanghai and Shenzhen stock market.

Key words: stock market volatility, policy factors, ICSS：MV algorithm, nonparametric GARCH model, N-W kernel regression