德国BV4.1模型修正与中国CPI季节调整  

Correction of Germany BV4.1 Model and the Seasonal Adjustment of China'S CPI

作　　者：桂文林[1,2] 李玉玲[3] Gui Wenlin Li Yuling

机构地区：[1]暨南大学经济学院统计学系 [2]中国社会科学院数量经济技术经济研究所 [3]暨南大学经济学院

出　　处：《统计研究》2017年第7期104-117,共14页Statistical Research

基　　金：国家哲学社会科学基金项目"时间序列分解与中国经济下行压力下的风险识别及预警研究"(16BJY014);教育部人文社会科学基金青年项目"中国居民消费价格指数数据质量优化与通货膨胀治理"(13YJC910004);中央高校基本科研业务费"中国季节调整模型建构与环比增长率测算"(332201412614801);"中国CPI数据质量新框架下的系统优化"(332201512615309)的资助

摘　　要：居民消费价格指数(CPI)是衡量通货膨胀程度和经济活动水平的重要指标,通常要剔除季节性因素影响。本文对国际最新的BV4.1季节调整模型进行了系统的研究和软件开发,编写R程序增强了其实用性。首先考虑到了中国的节日因素,交易日因素和异常值,对2001年1月至2015年3月的CPI数据进行了预处理。在分离出季节成分以及日历成分之后,采用平滑区间和修正历史法进行模型诊断的研究。研究认为:CPI的趋势在短期内具有二阶多项式发展特征,节日因素、交易日影响和异常值不显著;实证结果表明,BV4.1的季节调整结果与其他模型如X-12-ARIMA和TRAMO/SEATS相比具有很强的稳定性。China＇s CPI is an important index of inflation and economic activity which usually exclude the seasonal factors. Then it is to introduce the latest seasonal adjustment method BV4. 1 which is systematically studied with full explortation. Besides, R programming is applied to increase its practicability. First of all, Chinese holidays, trading day effects and outlier detection are considered to make a preprocessing for CPI from Jan 2001 to Mar 2015. After separating seasonal and calendar components, sliding span as well as revised history is used for testing adjustment stability. Main conclusions ： （ 1 ） CPI has a quadratic polynomial trend in the short term, the impact of the Chinese holidays, trading day effects and outliers are not significant. （2） Seasonally adjusted series by BV4. 1 has a better stability than X-12-ARIMA and TRAMO/SEATS.

关 键 词：季节调整 BV4.1 中国CPI 修正历史法 

分 类 号：C81[社会学—统计学]