基于局部Whittle法的长记忆参数及变点估计

利用最小二乘法判别变点存在性以及估计变点个数,同时利用改进的局部Whittle法对长记忆过程的变点及长记忆参数进行估计,并将其应用于上证指数的实证研究.结果表明:上证指数日收益率的波动率序列在不考虑变点的情况下具有显著的长记忆性,而在考虑变点的情况下,其长记忆性并不显著.因此,考虑变点能够避免伪长记忆的存在.

The Long Memory Parameter and Change Point Estimation Based on The Loca1 Whittle Method

In this paper,least square method is used to identify the existence of change points and estimate the number of change points,and the modified local Whittle method is proposed for estimating the long parameter and change points of long memory process.Finally,these methods are used to test for the Shanghai Complex Index.The results show that without considering change points,the long memory property of the volatility sequence of the Shanghai Composite Index’ daily return is significant,and the long memory property is not significant in case of considering change points.Therefore,the pseudo-long memory property can be avoided when considering change points.