中文摘要：

本文中,我们提出了一个FIGARCH模型的适应性预测方法.我们证明了一个适当的一阶或二阶的GARCH模型可以很好地对长记忆FIGARCH（1,d,0）模型进行一步预测.蒙特卡洛模拟结果显示,在波动率序列长记忆性不是太强的情况下,适应性预测的均方误差十分接近直接用原来FIGARCH模型进行预测的精度,但本文的方法计算量却小得多.实证方面,上证综指,恒生国企指数和欧元对美元汇率的波动率数据分析结果都表明,用低阶GARCH模型预测长记忆FIGARCH模型的有效性.

英文摘要：

In this paper we propose a prediction procedure for the fractionally integrated GARCH（FIGARCH） models.We show that a suitably adapted GARCH（1, 1） or GARCH（2,2） model performs satisfactorily in the one step ahead prediction of long memory volatility series with FIGARCH（1,d,0） structure.From the simulation results,we find that the MSE（mean squared error） of the proposed prediction method is only slightly higher than that of the prediction based on the original FIGARCH model when the volatility series is not very strongly persistent.Data examples of the Shanghai Composite Index,Hangseng China Enterprise Index and the exchange rate of the Euro to US dollar illustrate the usefulness of this low order GARCH adaptation to FIGARCH procedure.