基于厚尾分布多分形模型的上证综指波动率预测

Forecasting Volatility of Shanghai Composite Index Using Markov-Switching Multifractal Model with the Fat-tailed Innovation Distributions

通过引入厚尾的、自由度参数范围更广的广义误差分布(GED)扩展标准马尔科夫转换多分形模型(MSM),探讨MSM-GED模型的参数估计和波动率预测问题,并利用上证综指日收益数据进行实证分析。实证结果表明,上证综指确实存在多分形性,MSM模型的波动预测能力强于(FI)GARCH模型,尤其是中长期波动率预测,MSM-GED能够提供更准确的波动率预测值。这为资产定价和风险管理提供一种新的波动建模选择。

The Markov-switching multifractal model of asset returns with generalized error distribution(MSM-GED henceforth) is introduced as an extension to the Markov-switching multifractal model of asset returns(MSM).Issues of parameters estimation and volatility forecasting were also discussed.The forecasting capability of MSM-GED is evaluated empirically in a forecasting analysis using daily returns of Shanghai Composite Index.The analysis shows that multifractility indeed emerge in daily return series of Shanghai Composite Index.Compared with GARCH and FIGARACH model,MSM performs better in volatility forecasting,especially in predicting the medium and long term volatility.Overall,MSM-GED dominates GARCH and FIGARACH model and is a good alternative volatility modelling for asset pricing and risk management.