股票市场收益率波动长记忆性的分解及实证研究  

Empirical Research on the Long Memory of the Stock Market Volatility Series

作　　者：史代敏[1] 罗来东[2] 庞皓[1] 

机构地区：[1]西南财经大学统计学院 [2]中国人民银行成都分行

出　　处：《数量经济技术经济研究》2006年第8期136-141,151,共7页Journal of Quantitative & Technological Economics

基　　金：国家社科基金项目(05BJT009);教育部人文社会科学研究基地重大项目(02JAZJD790027);教育部"新世纪优秀人才支持计划"的资助

摘　　要：目前股票市场长记忆性检验和建模方法,不能很好地消除短期记忆的影响,针对这一问题,本文提出寻找序列的突变点,通过将序列分解为只包含长记忆性部分和不包含长记忆性部分的序列分解技术,来排除短期记忆的影响。对上证指数和深圳成分指数收益率波动的长记忆性进行实证研究发现,将序列分解以后进行长记忆性检验,不仅可以得出长记忆性检验更为精确的结论,同时可以检验序列分解过程的效果。The research methods for verifying and modeling the long memory on the stock market in recent years can not remove the influence of the short memory, how to remove the influence of the short memory becomes an important research. In this paper, we use ICSS algorithm to seek the breakpoints of stock market volatility series. The influence of the short memory is removed by decomposing the series into two parts： one only including long memory and the other not including long memory. We then use the rescale R/S to examine two series which have been decomposed. The result shows the success in decomposition process.

关 键 词：序列分解 结构转换 修正R/S分析 

分 类 号：F830[经济管理—金融学]