摘要
本文利用沪深股票市场数据,研究了二者之间的相关结构,尤其是尾部相关情况。由于股票收益率序列存在着条件自相关和条件异方差,为避免这些对Copula参数估计的影响,我们先对收益率序列进行AR(4)-GJRGARCH(1,1)-t建模,得到的标准化残差经BDS检验为独立同分布(i.i.d.)序列,再进行Copula建模。实证结果表明,沪深股市存在很强的正相关性,以及对称的尾部相关。这与大多数国外学者认为股票市场之间存在非对称相关现象的结论不同。本文通过图形检测和解析方法相结合来选择对数据拟合最好的Copula函数,结果表明学生t-Copula可以很好地刻画沪深股市的相关性。
Using the data from Shanghai and Shenzhen stock markets,this paper studies their dependent structure,in particular,the tail-dependence.In empirical analyses,we firstly adopt the AR（4）- GJRGARCH（1,1）-t model to deal with the conditional auto-correlation and conditional heteroskedasticity of the sequence of stock return rates,so that the standardized residuals obtained from the model have passed through the BDS test,which means that the standardized residuals are independent and identically distributed（i.i.d.） sequences.Then we further proceed to the Copula modeling.Empirical results show that there exists a strong positive correlation between Shanghai and Shenzhen stock markets,as well as symmetry tail-dependence.This conclusion differs from the majority of foreign scholars who find that the stock markets show Asymmetry tail-dependence.In this paper,we combine Analytical Methods and Graphics Representation Methods to select the best fitting Copula function.Empirical results show that the Student t-Copula can be a good characterization of the relevance of Shanghai and Shenzhen stock markets.
出处
《数理统计与管理》
CSSCI
北大核心
2010年第5期890-898,共9页
Journal of Applied Statistics and Management
