GARCH类模型波动率预测效果评价——以沪铜期货为例

以沪铜期货为例,研究了GARCH、EGARCH、FIGARCH和FIEGARCH四种模型的波动率预测效果。以已实现波动率为模型评价衡量标准,分别采用M-Z回归和损失函数进行预测效果检验,结果表明,无论残差服从高斯分布还是t-分布,不同的GARCH类模型预测效果有显著差异,其中FIGARCH模型预测效果最好,其次是GARCH模型,EGARCH和FIEGARCH模型预测效果不佳。此结论说明我国铜期货市场具有显著的长记忆性,但不具有非对称效应。

Evaluation on Volatility Forecasting of GARCH-Type Models: Evidence from Shanghai Copper Futures

This paper studies volatility forecasting performance of GARCH,EGARCH,FIGARCH,and FIEGARCH models.Realized volatility is seen as a benchmark.M-Z regression and loss function are used to assess the effect of volatility forecasting.The results show that under the assumption of Gaussian distribution or t-distribution,GARCH-type models predicting effectiveness are significantly different： FIGARCH best,GARCH second,EGARCH and FIEGARCH worst.This conclusion shows that copper futures market has a significant long memory,but not an asymmetric effect.