论文部分内容阅读
研究了Copula函数对沪深股市的相关性建模问题.许多学者用Gaussian Copula建模,但是它无法捕捉到尾部变化,尾部相关系数不存在.用t-Copula度量中国股市的相关性,捕捉到了尾部变化,并计算出了尾部相关系数,克服了Gaussian Copula对相关性建模的不足,并通过AIC准则比较得到t-Copula优于Gaussian Copula.最后对3种Archimedean Copula进行比较,通过比较它们与经验分布函数的距离,说明Gumble Copula更加适用于中国的金融市场.
Copula function on the Shanghai and Shenzhen stock market correlation modeling problems.Many scholars use Gaussian Copula modeling, but it can not capture the tail changes, the tail correlation coefficient does not exist.Using t-Copula measure the correlation of the Chinese stock market, capture the Tails of the three kinds of Archimedean copulas were calculated and the tail coefficients were calculated to overcome the deficiencies of the Gaussian Copula for the correlation modeling and to compare t-Copula with Gaussian Copula by AIC criterion. Finally, The distance of the empirical distribution function shows that Gumble Copula is more suitable for China’s financial markets.