基于时变Copula模型的股指收益率相依关系研究  

Research on dependency of stock index returns based on time-varying Copula model

作　　者：胡月[1] 王甜甜 夏厚君 雷柳荣 姜燕霞 HU Yue;WANG Tiantian;XIA Houjun;LEI Liurong;JIANG Yanxia(School of Sciences,Zhejiang University of Science and Technology,Hangzhou 310023,Zhejiang,China)

机构地区：[1]浙江科技学院理学院,杭州310023

出　　处：《浙江科技学院学报》2022年第1期94-104,共11页Journal of Zhejiang University of Science and Technology

基　　金：浙江省科技计划项目(2015C33088);教育部产学合作协同育人项目(201901116048)。

摘　　要：为了研究股票市场之间的互动性与相关性,基于时变Copula模型研究上证指数、深证成指、香港恒生指数和美国道琼斯指数收益率间的相依关系。首先,对4个样本收益率序列建立自回归移动取平均-广义自回归条件异方差(autoregressive moving average-generalized autoregressive conditional heteroskedasticity,ARMA-GARCH)模型族以确定单个边缘分布;其次,利用常见的常系数Copula函数及时变Copula函数分别对股指收益率序列的相依关系建立模型,并对这2种模型进行对比;最后,基于正态Copula模型分别对两两股指间的相依关系做比较分析。研究结果显示,时变Copula模型的拟合效果要明显优于常系数Copula模型;沪、深股市相依性最强,相关系数接近0.9,A股与港股的相关系数接近0.5,与美股的相关系数在0.15上下波动,A股与港股的相依性要强于A股与美股之间的相依性,港股与美股的相关系数为0.27,与A股相比,港股与美股的相依性更强。本研究方法可应用于金融行业其他领域以了解资金的流向和市场效率。In order to study the interaction and correlation among the stock markets,the dependency among Shanghai Composite Index,Shenzhen Component Index,Hong Kong Hang Seng Index and US Dow Jones Index was explored on the basis of the time-varying Copula model.First,the autoregressive moving average-generalized autoregressive conditional heteroskedasticity(ARMA-GARCH) family of models was established for four sample returns series to determine a single edge distribution;next,the common constant coefficient Copula function and the time-varying Copula function were used to model the dependency of the stock index returns series respectively,results of which were then compared;finally,comparative analysis of the dependency between every two stock indexes was conducted respectively on the basis of the normal Copula model.The results show that the fitting effect of the time-varying Copula model is significantly better than that of the constant coefficient Copula model;Shanghai stock market has the strongest dependency on Shenzhen counterpart,with the correlation coefficient up to 0.9,in contrast to the correlation coefficient between A shares and Hong Kong stocks close to 0.5,and let alone the correlation coefficient between A shares and US stocks fluctuating around 0.15,which reveals that the dependency between A shares and Hong Kong stocks much stronger than that between A shares and US stocks.The correlation coefficient between Hong Kong stocks and US stocks is 0.27,which is stronger than that between A shares and Hong Kong stocks.This research method can be applied to other fields of the financial industry to understand the flow of funds and market efficiency.

关 键 词：时变COPULA 股指收益率 相依关系 GARCH模型族 

分 类 号：F831.5[经济管理—金融学]