The Study On Volatility Of Stock Index Futures Based On Markov Switching Models

Stock index futures officially listed on April16,2010in China. Before that, for the systemic risk of the stock market, investors can’t avoid, stock index futures can effectively avoid the systemic risk. Accurate analysis of the futures price fluctuation characteristics is the foundation of identifying the market risk and promoting healthy futures market in China. Based on the development process of GARCH model from univariate GARCH to multivariate GARCH model, we give the maximum likelihood method of estimation of the univariate GARCH model parameters.Firstly, we introduce the state variables on the basis of traditional GARCH model, and establish Markov-Switching model, with specific depict the nonlinear characteristics of nonstationary sequence, we improve the accuracy of the GARCH models parameter estimation. By analyzing of the CSI300stock index futures and the main statistical characteristic of stock index futures day’s closing price of logarithm return sequence, found that the day logarithmic return is more suitable for fitting by using the t-distribution and its volatility asymmetry exists. Based on this, the Markov state transition GARCH models was further improved, the thesis of MS-GARCH model is proposed Based on the t-distribution of MS-EGARCH model and MS-TGARCH mode, to study the distribution features of the stock index futures logarithm yield sequence, and promoting the futures return forecast and quantitative analysis of the risk identification.Secondly, this paper estimates parameters from GARCH mode, TGARCH(1,1,) model and EGARCH(1,1) model, through comparing the parameters estimation results and the fitting effect of different models, shows that stock index futures in China return sequence has obvious asymmetry, on this basis, the MS-GARCH Models is set up, the market volatility divided into high and low volatility state and the result of Parameter estimation result exist obvious difference, relative to the state of low volatility, high average of fluctuations of shorter duration, and the volatility Persistence is low. The empirical analysis illustrates that MS-GARCH models is superior to the conventional GARCH models in describing the volatility characteristics of stock index futures.