基于BG分布的资产收益率分布拟合与尾部风险测度——以上海黄金市场为例  

Asset Return Distribution Fitting and Tail Risk Measurement Based on BG Distribution——Take Shanghai Gold Market as an Example

作　　者：姚萍[1] 王杰 杨爱军[1] YAO Ping;WANG Jie;YANG Ai-jun(College of Economics and Management, Nanjing Forestry University, Jiangsu Nanjing 210037, China)

机构地区：[1]南京林业大学经济管理学院

出　　处：《数理统计与管理》2019年第4期732-749,共18页Journal of Applied Statistics and Management

基　　金：国家自然科学基金(11501294);江苏省高校哲学社会科学基金项目(2018SJA0130);江苏省青蓝工程项目(2017)

摘　　要：本文以上海黄金市场为例,在GARCH模型下,系统性比较了基于正态分布、Logistic分布、HS分布、Laplace分布、t2分布和Cauchy分布的对称和非对称共12种BG分布在收益率分布拟合以及VaR和ES测度中的效果。研究结果表明,BG分布在收益率分布建模与尾部风险测度上的表现与原分布类型有关。当原分布为正态分布时,对称和非对称BG分布的效果都较差。当原分布为Logistic分布、HS分布、Laplace分布、t2分布和Cauchy分布时,对称和非对称BG分布的效果都较好,其中非对称BG分布效果在尾部分布拟合上优势更大。在所有分布中,基于t2分布和Cauchy分布的非对称BG分布表现最优。This article takes the Shanghai gold market as an example,under the GARCH model,systematically compare the performance of 12 distributions of symmetric and asymmetric BG distributions based on normal distribution,Logistic distribution,HS distribution,Laplace distribution,t2 distribution and Cauchy distribution,with respect to the return distribution fitting and VaR and ES measurement.The results show that,the performance of BG distribution in terms of return distribution modelling and tail risk measurement is related to the original distribution type.Both symmetric and asy mmetric BG distributions have poor effects when the original distribution is normally distributed.Symmetrical and asymmetric BG distributions are better when the original distribution is Logistic,HS,Laplace,t2,and Cauchy,the asymmetric BG distribution has a greater advantage in the tail return fitting.Among all distributions,the asymmetric BG distribution based on t2 distribution and Cauchy distribution have the best performance.

关 键 词：beta-generated分布 收益率分布 GARCH模型 VaR ES 

分 类 号：O212[理学—概率论与数理统计]