The Research Of Stock Index Futures’ Risk Measurement Based On Emi-parametric Model

Stock index futures is a new financial tool in China, with the asset allocation, risk aversion and price discovery functions. The CSI300stock index futures was officially launched on April16,2010.It’s a standardized futures contract based on stock index, the two sides trade with each other determined by the identified stock index, the launching of stock index futures has a positive effect in improving the price mechanism of the stock market, completing the function of stock market’s transaction, and reducing the transaction cost. But the stock index futures has the characters of big leverage ratio and high fluctuation, it is often used for speculation resulting many financial events. So it is more important to study the risk measurement tools.Although there are many risk measurement tools, they have respective defects. We introduce VaR method to measure CSI300stock index futures’risk, and then calculate VaR based on semi-parametric model. VaR(Value at Risk) has important applications in finance risk management, it is a popular market risk measurement tool in the international world.Generally, the methods to calculate VaR are parametric methods, non-parametric methods and semi-parametric methods. This paper describes the extreme value theory and quantile regression method in semi-parametric model, then establish the GARCH-GPD-VaR method and GARCH-quantile regression-VaR method, and then empirical analysis is made based on the real data.The empirical results show that the sequence of yield showing a peak, fat tail, volatility clustering phenomenon, and the traditional measurement model is no longer adequate to portray the return series of these features. However, the method of GARCH-GPD-VaR solves this problem effectively, and calculated the VaR in90%,95%,99%three different confidence levels. The empirical study finds that this model is good to measure the value at risk. Then VaR value is calculated by the GARCH-quantile regression-VaR model in the same confidence levels. The empirical study also finds that this model is good to measure the value at risk. Finally, the GARCH-GPD-VaR method can estimate VaR more accurately by comparing these two models. This shows that VaR method based on GARCH model and GPD model has good applicability for measuring the risk of our stock index futures market. It provides a feasible method to measure the risk of stock index futures in China.