基于稳定分布的EPGARCH模型  

EPGARCH Models with Stable Distributions

作　　者：武东[1] 汤银才[2] 

机构地区：[1]安徽农业大学理学院 [2]华东师范大学统计系,上海200062

出　　处：《工程数学学报》2008年第1期29-34,共6页Chinese Journal of Engineering Mathematics

基　　金：国家自然科学基金(10571057);安徽省高校青年教师科研资助计划(2006jql134)

摘　　要：本文对EGARCH模型进行了推广,得到了EPGARCH模型。对该模型的参数估计采用了带约束条件的非线性规划方法。利用直方图、时序图和Q统计量检验等方法对沪深指数收益序列进行了特征分析,得出收益序列具有高峰厚尾和波动聚集性。通过对沪深指数的VaR计算,得到在金融风险度量中基于稳定分布的EPGARCH模型比基于正态分布的EPGARCH模型更加有效。We proposed EPGARCH model, which is a generalization of EGARCH model. Nonlinear constrained programming method is adopted for parameter estimation of the model. Characteristics of return series of Shanghai and Shenzhen stock indices were analyzed with histogram, time series plot and Q statistic test. It shows that return series have some properties such as leptokurtosis, fat tail and volatility clustering. Calculated VaR for Shanghai and Shenzhen stock indices shows that EPGARCH with stable distribution is more efficient than EPGARCH with normal distribution in financial risk valuation.

关 键 词：稳定分布 EPGARCH VAR 高峰厚尾 波动聚集性 

分 类 号：O211.6[理学—概率论与数理统计] F830.9[理学—数学]