二元极值分布的独立性检验及在沪深股市分析中的应用

提出了二元极值分布的一个独立性检验统计量 T5,导出了它的渐近分布 ,得出了模拟分位点 ,并在小样本情况下 ,与其它已有的统计量进行比较 .结果说明本文给出的统计量 T5具有与似然比检验统计量 T3几乎相同的功效 ,且比其它检验方法有效 .最后对中国沪深两市的股票收盘指数的极值进行了独立性检验 ,认为具有显著的相关性 .

Testing for the Indepence of the Marginals in Bivariate Extreme Value Distribution and Application to the Analysis of the Stock Markets in Shanghai and Shenzhen

In this paper, we presented a new statistic testing for the independence between the marginal distributions in Bivariate Value Extreme Distribution. Its asymptotic behaviour and quantile table ware obtained. In small samples, we compared it with several present statistics. The result showed that our statistic has almost the same power as the likelihood ratio statistic, better than other methods. Finally, we applied these tests for independence to the extreme values of stock closing exponent betweeen Shanghai and Shenzheng in China. According to the outcome, we can consider that there is significant dependence.