中国股市长记忆的修正R/S分析  

Modified R/S Analysis of Long Memory Property in China Stock Market

作　　者：胡彦梅[1] 张卫国[1] 陈建忠[2] 

机构地区：[1]宁夏大学,宁夏银川750021 [2]西北工业大学,西安710072

出　　处：《数理统计与管理》2006年第1期73-77,共5页Journal of Applied Statistics and Management

摘　　要：本文在比较各种长记忆检验方法优缺点的基础上,采用修正的R/S分析检验我国沪深两股市日收益和日绝对收益序列的长记忆性。结果表明在0.05的显著水平下,两股市的日收益序列均无长记忆,但深圳成指日收益序列的记忆长度比上证综指日收益序列的记忆长度长;以日绝对收益序列为代表的波动序列具有较强的长记忆性。On the basis of weighing advantages and disadvantages of various methods of testing long memory property, modified R/S analysis is adopted to examine long memory property for daily return and daily absolute return of our stock market. Under 5 percent significant level, the following results are obtained. For daily return series, no long memory exists in both of Shenzhen and Shanghai indexes,but memory length of Shenzhen is longer than that of Shanghai; while volatility series represented by daily absolute return series present stronger long memory property.

关 键 词：长记忆 修正R/S分析 HURST指数 

分 类 号：F830[经济管理—金融学]