一种截断小样本的时变Copula模型的变结构点的诊断方法  

A Method for Diagnosing Variable Structure Points of Time-varying Copula Model Based on Small Samples

作　　者：杨湘豫[1] 徐晓环 YANG Xiangyu;XU Xiaohuan(College of Mathematics and Economics, Hunan University, Changsha 410082, China)

机构地区：[1]湖南大学数学与计量经济学院

出　　处：《湖南理工学院学报（自然科学版）》2019年第2期10-14,共5页Journal of Hunan Institute of Science and Technology(Natural Sciences)

基　　金：湖南省创新平台开放基金(16K017)

摘　　要：利用 Copula 函数对时间序列相关性的独特优势,进行二元正态 Copula-GARCH(1,1)建模,提出了将整体时间序列的样本截断成小样本,用t 检验判断变结构点的诊断方法.并以上证指数和深证成指为实证样本,研究两者发生显著变化的时刻.研究结果表明该方法能敏锐地捕捉金融市场的风险测度.Based on the uniqueness of the Copula function's correlation to time series, binary normal Copula-GARCH (1,1) modeling was performed. It was proposed to cut the sample of the overall time series into small samples and used the t-test to judge the diagnosis method of the variable structure points. The above Shanghai securities composite index and Shenzhen Stock Exchange Index were used as empirical samples to study the moment when the two have changed significantly. The results of this paper can sharply capture the risk measures of financial markets.

关 键 词：Copula-GARCH(1 1) t 检验 变结构点 

分 类 号：O211[理学—概率论与数理统计]