中国股票市场的分形结构——股票收益长期记忆特征的实证研究

股票市场收益的长期记忆特征对于系统非线性结构的确定以及市场有效性的研究具有重要的意义.针对上海和深圳A股算术加权和流通市值加权市场指数的周收益序列以及上证180指数和深圳成份指数中选取的12只代表性股票的周收益序列,采用重标级差分析(R/S分析)和ARF IM A模型对其进行了实证研究.从统计结果来看,样本序列呈现出尖峰和肥尾等有偏特征,明显不满足正态分布的假设,表明收益序列可能具有长程相关或记忆性.进一步的研究发现,沪深两市A股市场指数收益序列和大多数个股(10只股票)存在明显的长期记忆特征,收益分布表现出持久性.从划分不同时段的分析结果来看,中国股票市场渐进趋于弱势有效.

The Fractal Structure of Stocks Market in China

Analysis of the effects of long-memory in security market returns has important sence in research of market efficiency and conforming framework of nonlinear system.Adopting rate of weekly return of A share arithmetic weighted index and A float cap weighed index in Shanghai and Shenzhen and 12 stocks which are selected from 180 index and Shenzhen market,the paper uses rescaled range(R/S) method and ARFIMA model to test the presence of long-memory effects in security market returns.In statistic sample series exhibit significant summit and fat tailed skewed characters with departure from hypothesis of normal distribution.The further empirical results display weekly return series of A share arithmetic index of Shanghai and Shenzhen and 10 stocks have significant persistency effects,and from the analysis of time sharing,we observe China security market tends to reach weak efficiency.