用傅立叶和小波对我国产出缺口的分析

An analysis of China’s output gap based on Fourier and Wavelet techniques

本文采用二次时间趋势法、HP滤波、单变量状态空间和多变量状态空间四种方法对我国的产出缺口进行估算,并对估算结果应用傅立叶进行频域分析,应用小波进行时频分析。我们发现,我国的经济周期存在着5～12个季度的短周期、13～24个季度的中短周期、25～48个季度的中长周期,并且经济周期存在着互相叠加的关系。我们又通过设立波动性和准确性双重标准,研究表明在分析中长周期时,采用MV方法结果最好;在分析中短期周期时,采用UC方法稍好;在分析短周期时,HP的结果最好,MV的结果最差。

The authors use the measures of linear time trend filter （LT2）, Hodrick-Prescott filter （HP）, an unobserved components model （UC）, a multivariate time series filter （MV） to estimate China＇ s output gap, and then analyze the frequency and time domain of estimation results by applying Fourier and Wavelet techniques. The analysis shows that there are three types of cycles, namely, short-period of 5-12 quarters, medium-short period of 13-24quarters, medium-long period of 25-38quarters, which exist in China＇ s business cycle. These cycles overlay each other. By using fluctuation and veracity criterion, we find that MV method is the best for medium-long, UC is much better for medium-short and HP is the worst for short cycle.