摘要
在时间序列季节调整的预调整过程中,需要识别和去除交易日效应的影响。文章考察了现有PMI季调方法的质量,基于PMI数据特征设定了未考虑交易日效应的参考模型,比较了采用不同交易日效应模型的季节调整结果,研究显示:参考模型季调结果明显优于现有季调后序列,参考模型季调后序列频谱图中存在交易日尖峰,需要对交易日效应做进一步处理;PMI数据中春节效应的影响明显大于交易日效应;单一变量交易日效应模型比多变量模型效果稍好,但二者季节调整质量均差于不考虑交易日效应的参考模型。设定和考察了考虑月份长度和常数项的交易日效应模型,与参考模型相比,该模型在季节调整后序列中不再存在交易日尖峰,季节调整质量较好。
In the pre-adjustment process of seasonal adjustment for time series,the influence of trading day effect needs to be identified and removed.This paper investigates the quality of existing PMI seasonal adjustment methods,and then sets a reference model without considering the trading day effect based on PMI data features.Finally,the paper compares the seasonal adjustment results of different trading day effect models.The study results are shown as follows:The seasonal adjustment results of the reference model are obviously better than those of the existing seasonally adjusted series.There is a peak of trading day in the seasonally adjusted series spectrum of the reference model,and the trading day effect needs to be further processed.The impact of the Spring Festival effect in PMI data is obviously greater than that of the trading day effect.The effect of the single-variable trading day effect model is slightly better than that of the multi-variable model,but the seasonal adjustment quality of both models is worse than that of the reference model without considering trading day effect.The paper also sets up and investigates a trading day effect model considering month length and constant term.Compared with the reference model,there is no trading day peak in the seasonally adjusted series in this model and the seasonal adjustment quality is better.
作者
孟文强
Meng Wenqiang(College of Economics and Management,Shandong University of Science and Technology,Qingdao Shandong 266590,China)
出处
《统计与决策》
CSSCI
北大核心
2021年第7期40-45,共6页
Statistics & Decision
基金
国家社会科学基金资助项目(15BJY070)
