基于广义可加模型的波动率估计方法  

A volatility estimation method based on generalized additive model

作　　者：王相宁[1] 邹佳[1] 

机构地区：[1]中国科学技术大学统计与金融系,安徽合肥230026

出　　处：《中国科学技术大学学报》2008年第11期1276-1281,1288,共7页JUSTC

基　　金：中国科学院知识创新重要方向项目(KJCX3-SYW-S02)资助

摘　　要：放宽了传统GARCH模型参数形式的假定,将广义可加模型引入条件方差的估计,改进了[Bhlmann P,McNeil A J.An algorithm for nonparametric GARCH modelling.Computational Statisticsand Data Analysis,2002,40(4):665-683]提出的迭代算法并将其用于可加GARCH模型的估计.通过能够控制真实波动率的数学模拟实验,分别用样本内和样本外估计的方法说明了可加GARCH模型在估计现有参数模型无法刻画的复杂序列波动率时具有更好的估计效果,而与非参数模型相比,可加GARCH模型也有较好的估计效果.通过一个推断波动特征的算例,表明了可加GARCH模型对研究新兴市场股市或者处于金融危机中的股市等存在复杂波动特征的金融市场波动率有着非常现实的意义.By introducing generalized additive model, the restrictions on the parametric form of the conventional GARCH model was broadened. An improved estimation algorithm based on [Bühlmann P, McNeil A J. An algorithm for nonparametric GARCH modelling. Computational Statistics and Data Analysis, 2002, 40 （4）： 6G5-683] has been proposed to estimate the additive GARCH models. The mathematic simulation experiments involving different volatility structures show that the additive GARCH models perform better than the traditional GARCH models employing both in sample and out of-sample methods. The additive GARCH models also exceed general nonparametric models. An interential empirical study shows that the additive GARCH model is practically meaningful in studying the volatilities of stock markets which involve complicated characteristics, such as emerging stock markets or stock markets in financial crisis.

关 键 词：金融时间序列 可加模型 GARCH 波动率 异方差性 

分 类 号：F830.9[经济管理—金融学]