Research On Dynamic Margin Setting Of Stock Index Futures In China

Futures margin is the core of futures market's risk management.If the level of margin is too low,it can't effectively cover the risk of market.If the level of margin is too high,it will increase the transaction costs of investors,affecting the enthusiasm of market participants.Therefore,setting the appropriate margin level,is the key of improving the efficiency of the futures market operation and achieving long-term stability and prosperity of the futures market.The methods of futures margin mainly include dynamic system and static system,the current international mature market mainly uses dynamic margin system,and China is using static margin system.Static margin system in the process of running a regulatory mechanism is facing rigid,margin set high and other defects,and dynamic margin can be adjusted according to market risk margin level,both effective control of market risk,but also can improve the efficiency of capital use.Therefore,the dynamic margin is the trend of the development of the deposit system,the study of dynamic margin setting method,has a very important practical significance.The key to setting the margin level of stock index futures is the measurement and calculation of VaR.This paper adopts the combination of theoretical analysis and empirical research.The theoretical part mainly introduces the method of calculating VaR by parameter method,nonparametric method and extreme value theory.The parameter method is based on the assumption that the yield sequence follows the normal distribution,including Delta-normal distribution,EWMA,GARCH model,The nonparametric method mainly refers to the historical simulation method,from the historical data to simulate the calculation of VaR.Extreme value theory mainly introduces the basic principle of POT model,and VaR calculation combining GARCH model and POT model.Based on the daily settlement data of stock index futures active contract from April 16,2010 to April 18,2010,five dynamic VaR models of Delta-normal distribution,EWMA,GARCH,historical simulation and GARCH-POT were established.The accuracy of the five models at different time windows and confidence levels was tested using the Kupiec test method.Before the establishment of the dynamic VaR model,the data of the normality,stability and volatility aggregation are tested.The test results show that China's stock index futures yield does not obey the normal distribution,has the characteristics of spike and thick tail,and the sample sequence is stable Sequence,with fluctuating aggregation and long memory.The empirical results show that the choice of historical time window has a greater effect on Delta-normal distribution and historical simulation,and the prediction result of 250 days VaR is more sensitive,and the breakdown frequency is lower in the case of accurate prediction.At the 95%low confidence level,only the GARCH-POT model has an overvalued risk,and the other four models are accurate.At the high confidence level of 97.5%and 99%,the Delta-normal distribution,EWMA and GARCH have a certain degree of risk underestimation,which is consistent with the actual situation where the yield is thick and thick.Historical simulation method Under these two high confidence levels,the VaR prediction is the most accurate.The GARCH-POT model is accurate at the 99%highest confidence level,which is determined by the basic principle and characteristics of the extreme value theory.