为了捕获资产收益正向冲击(利好消息)和负向冲击(利空消息)的非对称效应,将门限效应与状态相关的杠杆效应同时引入基本的随机波动率( SV )模型中,提出双杠杆门限SV (THSV-DL)模型对资产收益的波动率进行建模。继而,基于有效重要性抽样(EIS)技巧,给出了THSV-DL模型的极大似然( ML)估计方法。为了检验EIS-ML方法的精确性以及小样本性质,构建了蒙特卡罗模拟实验。模拟结果表明,EIS-ML方法是可靠和有效的。采用上证综合指数和深证成份指数的日收益数据为样本,运用THSV-DL模型对中国股市进行了实证研究。结果表明,中国股市具有很强的波动率持续性,并且存在显著的杠杆效应。更为重要的是,中国股市的波动率持续性、波动率的波动率以及杠杆效应都具有非对称性。具体而言,与利好消息相比,利空消息造成中国股市更高的波动率持续性以及更低的波动率的波动率和杠杆效应。最后,采用上证综合指数进行的实证研究表明,THSV-DL模型相比基本的SV、杠杆SV( SV-L)、THSV和杠杆THSV( THSV-L)模型具有更加均衡及优越的风险测度能力。
To capture the asymmetric effects of positive shocks( good news)and negative shocks( bad news) to asset returns,this paper incorporates both the threshold and state-dependent leverage effects into the basic stochastic volatility(SV)model,and proposes a threshold SV model with double leverage(THSV-DL)to model the volatility of asset returns. Based on the efficient importance sampling( EIS)technique,we use the maximum likelihood(ML)method to estimate the parameters of the THSV-DL model. Then,Monte Carlo simulations are presented to examine the accuracy and small sample properties of the proposed method. The experimental results show that the EIS-ML method performs very well. We apply the THSV-DL model to the daily returns of Shanghai stock exchange( SSE)composite index and Shenzhen stock exchange( SZSE)com-ponent index of China. Empirical results show that there exists a high persistence of volatility and a significant leverage effect in China’s stock market. More importantly,asymmetries in the volatility persistence,volatility of volatility and leverage effect are discovered in China’s stock market. Specifically,the volatility persistence tends to be higher,and both volatility of volatility and leverage effect tend to be lower following the bad news than following the good news. Finally,an empirical study on the accuracy of value at risk( VaR)estimates based on Shanghai stock exchange composite index is presented. The empirical results demonstrate that the THSV-DL model can yield more balanced and accurate VaR estimates than the basic SV,SV with leverage effect( SV-L),THSV,and THSV with leverage effect( THSV-L)models.