Research On Risk Measurement Of Stock Index Futures Based On Quantile Regression

Stock index futures with highly leveraged features, is a financial derivative product with high risk. Followed by the April16,2010being officially listed for transaction of stock index futures of China, it is facing great challenges on market risk management of stock index futures. Therefore, after the stock index futures listing, market risk management and control is particularly important in order to effectively promote the healthy development of capital market of China. For the quantitative analysis and assessment of the risk is the core and foundation of financial risk management, study on the risk measurement of stock index futures is particularly important and urgent.It has been agreed that VaR is an effective means of risk management among the numerous methods of risk measurement of financial assets, so it is very necessary to use VaR for risk management of stock index futures in China. Value at Risk (VaR), which means "the value at risk", is defined as the maximum possible loss that a portfolio will lose under normal market fluctuations, with a given confidence level, over a certain time horizon. Despite its conceptual simplicity, the measurement of VaR remains a very challenging statistical problem, and none of the models developed so far gives recognized best solutions, for many models used to calculate the VaR each have their own advantages and disadvantages, research is continuing on the calculation of VaR model.In this paper, VaR is used as a tool to measure market risk of stock index futures in China. This paper selects the daily closing price data of Shanghai and Shenzhen300stock index futures from April8,2005to December31,2010, applies three parametric VaR models with conventional approach, three kinds of non-recursive quantile regression VaR models and three kinds of recursive quantile regression VaR models, a total of nine models, makes one-step ahead prediction of VaR for our stock futures market under95%and99%confidence levels respectively. Finally, under two different kinds of confidence levels, uses Kupiec test and Quantile Loss test to compare and rank those models and approaches from two different aspects, for providing some references for the different needs of management of which VaR model is the best for calculating the risk of stock index futures in China.After a series of empirical studies and analysis, this paper comes to the conclusion that VaR models with quantile regression outperform or as well as those parametric VaR models with conventional approach in almost all of the test criterion and confidence levels. Under Kupiec test, the non-recursive quantile regression VaR models is superior to the recursive ones, and of which, QR.EGARCH(1,1)-t is optimal. Under Quantile Loss test, the opposite is true, and of which, AS model is optimal under95%confidence level while SAV model is optimal under99%confidence level. So the proposal is to select VaR models for measuring the risk of our stock index futures market according to different needs of management. From the security point of view, to ensure the loss under control, focusing on the Kupiec test result, give priority to QR.EGARCH(1,1)-t; from liquidity and profitability point of view, focusing on the result of Quantile Loss test, give priority to AS model under95%confidence level while SAV model under99%confidence level.