我国外汇风险管理中的内部模型选择  

Choosing Internal Model in Risk Management of Exchange Rate

作　　者：田新时[1] 李耀[1] 

机构地区：[1]华中科技大学经济学院金融系,湖北武汉430074

出　　处：《运筹与管理》2003年第1期50-53,共4页Operations Research and Management Science

摘　　要：在VaR计算中 ,一般都有市场变量的日回报服从正态分布的假定 ,但汇率的变化呈非正态分布。随着我国加入了WTO后金融市场的进一步开放 ,选择适合我国外汇风险管理需要的统计模型已极为紧迫。本文提出了一种用有偏度的混合正态分布计算VaR的方法 ,使市场风险管理者可不受市场变量正态性假定的限制 ,其参数通过分位点到分位点的对应 ,参照GARCH模型进行更新 ,从而可以对回报分布的左右胖尾进行解释。本文中选取了八种与我国经济密切相关的外汇 ,分别计算其参数 ξ、q、α和β的最优值。运算结果显示 ,有偏度的混合正态分布对货币组合的模拟大大优于对个体货币的拟合。To calculate VaR,the daily returns of market variables are often assumed to be normally distributed.The empirical evidence however,indicates that exchange rates are rather non-normal.With China's entry into WTO,the financial market will be developed much quicker and it will be urgent to find a suitable model for exchange rate management.This paper proposes an approach which uses the skewed mixed normal distribution to calculate VaR,relieving users of the normality assumption of the market variables,with parameters of the distribution updated using schemes similar to GARCH model to fully take into account the fat left and right tails of the returns distribution.We use data of eight foreign currencies which are tightly relative to Chinese economy and find out the best fitted parameters ξ,q,α and β respectively.The result shows that the fit to the pool of the eight currencies is much better than the fits to the separate currency.

关 键 词：外汇风险管理 VAR 非正态分布 混合正态分布 中国 统计模型 

分 类 号：F832.21[经济管理—金融学] O212[理学—概率论与数理统计]