制造业采购经理人指数PMI季节调整中的春节模型优化  

Optimization of Spring Festival Models in the Seasonal Adjustment of Manufacturing Purchasing Managers’ Index(PMI)

作　　者：孟文强[1] 陈中涛[2] Meng Wenqiang;Chen Zhongtao(College of Economics and Management,Shandong University of Science and Technology,Qingdao Shandong 266590,China;China Logistics Information Center,Beijing 100037,China)

机构地区：[1]山东科技大学经济管理学院,山东青岛266590 [2]中国物流信息中心,北京100037

出　　处：《统计与决策》2020年第21期57-61,共5页Statistics & Decision

基　　金：国家社会科学基金资助项目(15BMZ080)。

摘　　要：文章改进了美国普查局X-13A模型中的移动假日处理模型,在其内置的圣诞节模型和复活节模型基础上,建立并考察了春节效应的比例和指数时变模型,求解不同模型条件下的最优春节效应区间,并经过了滑移区间和修正历史检验。结果表明:现有PMI季节调整方法产生的序列仍能识别出明显季节性,经过季调后的序列不存在显著的季节性;所采用的圣诞节模型、复活节模型、比例时变、指数时变模型在拟合性和预测性方面优于现有季节调整方法。基于AICC和AAPE准则,选择春节效应的最优区间,圣诞节模型中,春节效应的影响区间为[-2,5]。复活节模型中最优影响区间为三段区间[-15,-3]、[-2,5]和[6,18];时变模型与均匀分布模型相比,具有预测误差小,对离群点的识别充分,季调稳定性好、趋势转折领先和调整结构准确的优点。特别是指数时变模型,在数据序列月份变化的修正中体现出较好的稳定性。This paper improves the mobile holiday processing model in the X-13 A model of the U.S. Census Bureau, based on the built-in Christmas model and Easter model to establish and investigate the proportion and exponential time-varying model of the Spring Festival effect. Then, the paper solves the optimal Spring Festival effect interval under different models, and has passed the slip interval and the revision history test. The results show that the series produced by the existing PMI seasonal adjustment method can still identify obvious seasonality, but there is no significant seasonality in the series after seasonal adjustment,and that the Christmas model, Easter model, proportional time-varying model and exponential time-varying model adopted in this paper are better than the existing seasonal adjustment methods. Based on AICC and AAPE criteria, the optimal interval of Spring Festival effect is selected. In the Christmas model, the interval of Spring Festival effect is [-2,5]. In the Easter model, the optimal impact interval is three intervals of [-15,-3], [-2, 5] and [6,18]. Compared with the uniform distribution model, time-varying model has the advantages of minor prediction error, sufficient identification of outliers, better adjustment stability, leading trend transition and accurate structure adjustment. In particular, the exponential time-varying model shows better stability in revising the monthly change of data series.

关 键 词：PMI 季节调整 春节效应 时变模型 稳定性检验 区间选择 

分 类 号：F064.1[经济管理—政治经济学]