摘要
文章比较了小波方法与传统滤波方法提取周期信息的滤波能力。利用AR(2)和随机游走过程生成具有不同周期特点的时间序列,再分别使用HP、BK及小波滤波,提取得到序列的周期成分,并通过构建统计量,对比三种方法的滤波效果。研究发现:当序列具有低频(长)周期、趋势主导时,三种方法的滤波效果均不理想;当序列具有高频(短)周期时,无论是趋势主导还是周期主导,三种方法都能有效地提取周期成分,且小波与BK方法显著优于HP滤波;此外,在提取我国经济周期的实证研究过程中,小波方法表现出良好的滤波能力,可以替代BK滤波和HP滤波,是一种更有效的提取周期信息的工具。
This paper compares the filtering ability of wavelet method and traditional filtering methods to extract cyclical information.Firstly,the paper uses AR(2)and random walk process to generate time series with different cyclical characteristics,then employs the HP,BK and wavelet filters respectively to extract cyclical components of series.Finally,the paper compares the filtering performance of three methods by constructing statistics.The research finds that if the series has low frequency(long)period and dominant trend,the filtering performance of three methods is unsatisfied,and that if the series has high frequency(short)period,three methods can effectively extract cyclical components whether it is trend-dominated or cycle-dominated,and the wavelet and BK methods are significantly superior to the HP filter.Furthermore,the wavelet method presents great filtering ability in the empirical study of extracting China’s economic cycle,which can replace BK and HP filters and is a more effective tool for extracting cyclical information.
作者
李乃乾
孙晨童
Li Naiqian;Sun Chentong(School of Economics,Dongbei University of Finance and Economics,Dalian Liaoning 116025,China)
出处
《统计与决策》
CSSCI
北大核心
2021年第1期29-34,共6页
Statistics & Decision
