ARMA-GARCH模型及VaR方法在我国银行间同业拆借市场中的应用研究  

ARMA-GARCH Models and the Application of VaR Method to Chinese Inter-bank Borrowing Interest Rate

作　　者：王德全[1] 

机构地区：[1]北京大学软微学院,北京102600

出　　处：《系统工程》2009年第5期35-42,共8页Systems Engineering

基　　金：北京大学ACOM金融信息化研究中心项目(2008100601)

摘　　要：对我国银行间同业拆借市场利率建立不同分布假设下的12种ARMA-GARCH模型,并用V aR模型度量同业拆借市场的利率风险,得出结论:通过选择适当的滞后阶数,12种不同分布假设下的ARMA(m,n)-GARCH(p,q)模型均可以有效地刻画同业拆借利率序列的动态特性,且同业拆借利率的变化与其前两期的变化有较大关系;我国银行间同业拆借利率序列存在显著的反杠杆效应:利率向上变动时的波动性要大于利率向下变动时的波动性;t分布不适合描述同业拆借利率序列的尾部特征,正态分布和GED分布能较好地描述同业拆借利率序列的右尾特征,但不适合描述其左尾特征。This paper empirically studies the seven days diurnal weighted average of Chinese inter-bank borrowing interest rate with the ARMA-GARCH models from 3 March of 2002 to 30 June of 2008. Based on the models, we establish VaR Model of inter-bank borrowing interest rate. Finally we draw some conclusions. First, there are twelve ARMA（m, n）- GARCH （p, q） models with proper degrees in different distributions which are better fitted with the dynamic feature of inter-bank lending interest rate. Second, the leverage effect of inter-bank borrowing interest rate do exists, namely, the volatility is bigger when the interest rate gets soaring, and is smaller when the rate gets down. It is on the contrary to volatility behavior of the stock return. Finally, the models of t-distribution are not proper in measuring VaR of inter-bank borrowing interest rate, the models of n-distribution and GED distribution are more accurate than others in measuring VaR of short position and they are not proper in measuring VaR of long position.

关 键 词：ARMA—GARCH模型VaR 同业拆借利率 杠杆效应 

分 类 号：F823[经济管理—财政学]