The Characteristic Study Of Chinese Economic Cycle Fluctuation Based On Three Filtering Methods

Economics growth and cycle fluctuation have become the two mainstream problems of modern economics studies. In terms of cycle fluctuation, modern economic cycle research focuses on the pure wave3characteristics of economic fluctuations, that is, analyzing the metric characteristics of the time series after eliminating the trend part.In the analysis of the pure wave characteristics filtering techniques are the popular methods in foreign countries. Compared with that in foreign countries, the use of filtering methods in fluctuation characteristics of the economic cycle started late in China and the study mainly concentrated in the stylized facts of cycle fluctuations. There are two Deficiency of this kind of study. First, filtering methods were directly selected in data handling, without theoretical analysis. Actually, different filtering techniques are based on different principles, have different characteristics, and are suitable for samples with different backgrounds. Most of the foreign researches applied quarterly data, therefore, these filtering methods must be selected and tested with reasonableness, before they are used in actual economic data analysis. The second problem centralizes in the selection of the filter parameter. The vast majority of existing HP filtering method assumes the smoothing parameter X to be100,400,25, or even6.25, when dealing with annual data. It seems that the annual data of λ is assigned at random, without a relatively definite value. Moreover, it is much more complex to define the parameters, such like upper and lower bounds, and cut-off length for band-pass filters4.This paper begins from the above two problems. It firstly discusses the filtering method and parameter selection, which are suitable for the characteristics of annual data in China. Then, filtering methods selected are comprehensively utilized in analyzing the properties of China’s Economic Fluctuations during the period1978-2010, with which, the macroscopic properties of co-variation, stickiness, and volatility can be derived. The results shows:regarding the economic data after the Reform and Opening up, except that the filter results of CF are obviously different, the difference of HP (λ=100),HP (λ=6.25), BK (2,8) is not obvious, and the results of HP (λ=6.25),BK (2,8) are very similar. Nevertheless, due to the loss of observation points when applying BK filter, when dealing with annual macroeconomic series about2-8year cycle compositions after1978, the HP filter with the parameter of6.25is recommended in data processing.