制造业采购经理人指数季节调整的时域检验和频域分析  

Time Domain Test and Frequency Domain Analysis on Seasonal Adjustment of Manufacturing Purchasing Managers’Index

作　　者：孟文强[1] Meng Wenqiang(College of Economics&Management,Shandong University of Science and Technology,Qingdao Shandong 266590,China)

机构地区：[1]山东科技大学经济管理学院,山东青岛266590

出　　处：《统计与决策》2020年第24期28-32,共5页Statistics & Decision

基　　金：国家社会科学基金资助项目(15BJY070)。

摘　　要：我国自2005年开始发布制造业采购经理人指数PMI数据,由于初期数据序列较短,采用了一种短时序的季节调整方法,之后在2012年开始改用X-12方法。文章采用时域检验和频域分析方法,分析了现有PMI季节调整的表现、制造业采购经理人指数季节调整过程中线性滤波系统的选择,并考察了数据扩展预测带来的质量改善。研究表明:现有季调后序列仍存在明显的季节性,2012年成为PMI数据特征的分水岭,现有季调方法仍存在改进空间。文章选择3×5季节滤子,与传统二阶段选项相比,带来的频率阻带较为平稳,历史修正更小。Henderson趋势滤子选择9项为最佳,此时与PMI数据特点相符,既能去除短期波动,又不会造成趋势信息损失。由于现有PMI序列仍然较短,采用后向扩展两年的方法,可以有效提高季节调整的稳定性和对趋势的估计。无论时域检验还是频域分析,均得出了相似的结果,但频域分析更加直观和清晰。China has released manufacturing purchasing managers’index(PMI)data since 2005.Due to the initial short data series,a seasonal adjustment method of short sequence was adopted,and then the X-12 method was adopted in 2012.This paper uses time domain test and frequency domain analysis method to analyze the performance of the existing PMI seasonal adjustment and the selection of the linear filtering system in the process of seasonal adjustment of manufacturing PMI seasonal adjustment,and to investigate the quality improvement brought by the data expansion prediction.The research shows that there still exists obvious seasonality in the existing post-seasonal adjustment sequence,that the year 2012 has become the watershed of PMI data characteristics,and there is still room for improvement in the existing seasonal adjustment methods.The paper selects a 3×5 seasonal filter,which has relatively stable frequency stopband and smaller historical revision compared with traditional two-stage options.The Henderson trend filter selecting 9 items is the best,which is consistent with the characteristics of PMI data,able to remove short-term fluctuations without causing the loss of trend information.Since the existing PMI series is still short,adoption of backward two-year expansion method can effectively improve the stability of seasonal adjustment and the trend estimation.Both time domain test and frequency domain analysis come to the similar result,but frequency domain analysis is more intuitive and clearer.

关 键 词：PMI 季节调整 线性滤波系统 时域检验 频域分析 

分 类 号：C81[社会学—统计学] F064.1