GARCH族模型的预测能力比较:一种半参数方法

半参数GARCH模型无须设定条件分布的具体形式。本文首先将一种效率较高、易于实施的半参数方法——估计函数方法应用于10类常见的GARCH结构,并给出证据,显示该方法能显著提高GARCH族模型的波动率预测绩效。然后,应用估计函数方法,较为全面地比较各类GARCH结构的预测能力。为给出统计意义下的结果,并减少数据窥察问题,研究中分别使用OLS和SPA检验法进行绩效评价。结果发现,与其他GARCH类结构相比,EGARCH和APARCH模型能够较好地描述股市收益率的波动过程。

Comparing the Predicting Ability of GARCH-type Models:a Semi-parametric Approach

Semi-parametric GARCH does not need full specification on conditional distribution.We apply estimating function method,a semi-parametric approach to 10 GARCH-types model and show such method can significantly improve their predicting ability.Also employing such approach,we compare the predicting abilities of those GARCH structures.To present statistical result and reduce the problem of data snooping,we produce the results by OLS and SPA test. Finally,we find EGARCH and APARCH structure are more appropriate ...