基于GAST分布的国际原油市场下行风险预测及后验分析  

Forecast of the Downside Risk of International Crude Oil Market under GAST Distributions and Backtesting Analysis

作　　者：吕永健[1] 王鹏[1] 

机构地区：[1]西南财经大学中国金融学院

出　　处：《国际金融研究》2016年第5期72-84,共13页Studies of International Finance

基　　金：国家自然科学基金(71473200);四川省教育厅创新团队建设项目(JBK130401);中央高校基本科研业务费专项资金(JBK1507085)的资助

摘　　要：本文基于包括GAST在内的多种统计分布建立风险预测模型,深入考察各种模型在原油市场下行风险预测中表现出的精确性差异,主要结论:(1)原油市场收益分布的左、右尾部的"厚度"显著不一致;(2)正态分布不能刻画原油市场分布的"尖峰和厚尾"和"有偏"等特征,在6种分布中表现出了最弱的风险预测精确性;(3)GAST分布不仅可以刻画原油市场收益分布的"有偏"特征,而且可以分别刻画收益分布左、右尾部的"厚尾"特征,并表现出了相对最高的Va R测度精确性。我们认为,就精确地预测原油市场下行风险而言,GAST分布可以作为相对合理的统计分布模型。We can accurately predict the downside risk of crude oil market by reasonably modeling the distribution of crude oil market. The paper establishes the risk prediction models based on 6 kinds of statistical distributions, such as the generalized asymmetric student distribution and skewed general error distribution, then makes comparison on the accuracy of these models in 3 different levels of quantile. Main conclusions include： （1） The ＂thickness＂ of the left and right tail of the crude oil market distribution are not consistent. （2） The normal distribution cannot describe some characteristics of the distribution of crude oil market, such as ＂leptokurtosis and fat tail＂ , and shows the weakest risk prediction accuracy in the 6 types of statistical distributions. （3） GAST distribution can not only describe the ＂Skewed＂ characteristics of the crude oil market, but also separately describe the ＂fat tail＂ characteristics of the left and right tails of return distribution. The risk prediction models based on GAST distribution show the highest VaR measure accuracy, and can be used as a relatively reasonable statistical distribution model choice.

关 键 词：下行风险 风险价值 国际原油市场 后验分析 

分 类 号：F831[经济管理—金融学]