基于CAViaR和GARCH模型的沪深300股指期货动态风险测度

Risk Measurement of CSI300 Stock Index Futures——Based on CAViaR and GARCH Model

以我国期货市场上交易最为活跃的沪深300股指期货为例,分别采用CAViaR模型和GARCH模型对多头VaR和空头VaR进行风险建模,深入研究了股指期货的收益分布特征和波动形态规律,并运用严谨的后测检验的方法对比了各个模型的风险预测精度。实证结果表明:(1)沪深300股指期货具有明显的"尖峰厚尾"现象,却没有显著的有偏性和长记忆性;(2)基于杠杆效应的GJR模型和兼具长记忆性和杠杆效应的FIAPARCH模型并没有表现出比传统GARCH模型更高的预测精度,同时,先验GED分布对金融收益分布特征的刻画要优于正态分布和SKST分布;(3)半参数法的CAViaR模型相比GARCH族模型表现出绝对优异的预测能力。总之,CAViaR模型在股指期货的风险预测方面是相对更合理的模型选择。

By taking the most actively traded futures in China — CSI300 stock index futures as sample,this paper adopts CAViaR model and GARCH models to analyze the risk of long position and short position.Then it explores the characteristics of return distribution and volatility pattern.Finally it compares the accuracy of risk prediction of different risk models through rigorous backtesting methods.The results show that：firstly,CSI300 exhibits significant leptokurtic and fat tail,but it does not show an obvious skewed distribution and long memory.Secondly,the GJR model based on leverage effect and FIAPARCH model based on long memory and leverage effect do not perform better than GARCH model.Meanwhile,generalized error distribution（GED）is more appropriate for conditional return distribution than normal distribution and skewed student distribution.Thirdly,the CAViaR model based on semi-parametric method shows absolutely excellent predictive ability compared with the GARCH-type models.Overall,CAViaR model is a relatively reasonable model in predicting dynamic risk of stock index futures.