基于半参数GARCH模型的我国外汇储备形态

The Forms of China's Foreign Exchange Reserve Based on Semi-parametric GARCH Model

基于CSMAR提供的1987年1月至2007年12月的月度外汇储备数据,首次使用半参数GARCH模型对高频月度外汇储备的动态增长机制进行了研究。研究表明半参数GARCH(1,1)模型要优于参数的GARCH(1,1)模型和其他非参数GARCH模型,半参数GARCH(1,1)能够准确地拟合我国外汇储备的分布形态和动态相依结构,同时半参数GARCH模型也提供了一种灵活描述数据分布非正态和高阶矩的途径。

This paper uses semi-parametric GARCH model to characterize the dynamic growth mechanism of foreign exchange reserve based on the monthly data from Jan.1987 to Dec.2007.We find that the semi-parametric GARCH（1,1） model which can adequately fit the dynamic inter-dependence structure of foreign exchange reserve is better than parametric GARCH（1,1） and other semi-parametric GARCH models.We firstly give the nonlinear and nonparametric modeling of the monthly data and obtain better results.Meanwhile,the smoothing function of semi-parametric GARCH models can conveniently characterizes the non-normality and higher order moments of the data distribution.