基于LW估计的中国证券市场长记忆性实证研究

An Empirical Research on the Long Memory in China Security Market by local Whittle Estimation

摘要: 针对国内研究证券市场长记忆性主要采用的是时域方法,文章采用基于频域的半参数Local Whittle估计方法研究了中国证券市场的长记忆特性,并且和对数周期图回归(GPH)方法进行了比较。结果表明,LW方法具有不受时间频率的影响,能够有效消除时间序列中短期记忆和周期性对估计结果的影响等优点,明显优于GPH方法。实证结果表明,中国证券市场存在明显的长记忆性,并且长记忆行为在重大突发事件发生期间更加明显。文章还从长记忆的角度探讨了“政策市”的存在性和相关特点。

Abstract: Since there are some drawbacks in non-parametric methods,the paper applies methods of semi-parametric estimation based on frequency domain including local Whittle estimation and log-periodogram regression in China security market.LW estimation is proved to be more efficient than GPH regression by compare.Then LW estimation is applied in China security market,and some very important results are got.