基于尖峰厚尾、有偏GARCH-copula模型的风险测度  

The Risk Measure of the Leptokurtic and Skewed GARCH-copula Model

作　　者：宫晓莉[1,2] 庄新田 刘喜华[2] GONG Xiao-li;ZHUANG Xin-tian;LIU Xi-hua(School of Business Administration, Northeastern University, Shenyang 110169, China;School of Economics, Qingdao University, Qingdao 266061, China)

机构地区：[1]东北大学工商管理学院,辽宁沈阳110169 [2]青岛大学经济学院,山东青岛266061

出　　处：《系统工程》2018年第1期31-38,共8页Systems Engineering

基　　金：国家自然科学基金资助项目(71671030;71571038;71771042)

摘　　要：考虑到证券投资组合中资产收益分布的尖峰厚尾属性和波动率集聚效应以及金融资产变量间的非线性相依结构,假设资产收益率分布服从广义误差分布(GED),以尖峰厚尾、有偏的GJR-GARCH-GED模型刻画资产收益率的边际分布,以copula函数描述变量间的相依性,构建起改进的GARCH-copula模型。以研究GED分布的GJR-GARCH模型与不同copula函数耦合对金融序列的拟合能力,进而测度投资组合风险。实证研究发现,沪深收益波动存在明显的非对称性,适宜采用GJR-GARCH-GED模型处理收益率的尖峰厚尾性,波动率的集聚性。考察上述模型与Clayton copula、Gumbel copula、Frank copula以及t copula耦合下的实际拟合表现,进而对投资组合风险VaR和CVaR检验发现,Clayton copula刻画风险相依性的效果最佳,适合应用于风险管理。Considering the leptokurtosis characteristics and clustering effects of the returns distribution in securities portfolio as well as the non-linear dependency structure between the multiple variables of the financial assets, the asset returns are assumed to follow the generalized error distribution （GED）. The leptokurtic and asymmetric GJR-GARCH- GED model is used to describe the marginal distribution of asset returns, and the copula function is used to describe the interdependence relationship between variables so as to construct the improved GARCH-copula model. The GJR-GARCH model is coupled with different copula functions to model the financial returns series under GED marginal distribution, and then measure the risk of financial portfolio. The empirical studies show that there is a significant asymmetry in the volatility of Shanghai and Shenzhen returns, and it is suitable to use GJR-GARCH-GED model to deal with the high peak and tail as well as the volatility clustering of the returns. This paper investigates the actual fitting performances of the above model with Clayton copula, Gumbel copula, Frank copula and t copula functions, and then analyzes the risk of portfolio VaR and CVaR. It is found that Clayton copula is best suited in describing the risk dependence in risk manage- ment.

关 键 词：尖峰厚尾 copula相依性 GJR-GARCH-GED模型 风险测度 

分 类 号：F830[经济管理—金融学]