非参数与参数GARCH类模型的比较

A Comparison of Nonparametric and Parametric GARCH Models

在对金融资产价格的波动进行定量建模研究时,使用最多的是参数GARCH类模型,这类模型有许多优点,但模型需要的设定条件却比较严格,当不清楚变量的分布形式时,利用实际样本数据建立的参数模型可能会由于错误的设定而得出错误的结论。与参数模型相比较,非参数GARCH类模型的建模方法是一种不同于参数方法建模的新思路,非参数模型不要求设定条件,不会由于错误的设定而得出错误的结论。比较了非参数与参数GARCH类模型的基本形式和适用条件,在对金融资产的波动性进行建模的时候应该根据数据的基本统计特征灵活选用非参数与参数GARCH类模型。

The GARCH models are frequently used in quantitative modeling study of fluctuations in financial assets prices. Despite of the advantages, their setting parameter is extremely strict, and a slight setting error may lead to wrong conclusion. Compared with parametric models, non- parametric GARCH model is a new modeling approach, which is free from setting parameters, thus avoiding setting errors. This paper compares the basic form and application condition of nonparametric and parametric GARCH models and suggests that the GARCH models should be selected flexibly according to the statistic features of the data in modeling fluctuations of financial assets.