摘要
文章从代表制造业＂晴雨表＂的PMI指数入手,探讨我国制造业在GDP混频预测中的作用。构建了混频数据抽样模型（MIDAS）,对GDP增长率的实时预测问题进行研究,将PMI指数及GDP增长率自身的滞后效应引入到GDP趋势预测中,实现高频数据PMI指数与低频数据GDP增长率的联动分析。结果表明：（1）自回归非限制混频数据抽样模型（U-AR（4）-MIDAS（3,K））的预测精度具有比较优势。（2）高频月度PMI指数对季度GDP具有正负两种效应,该效应会持续9个月,GDP自身之间也存在着相互影响,该影响会持续4个季度之久。（3）向前h步U-AR（4）-MIDAS（3,K,h）适合短期内利用最新公布的月度PMI对GDP进行实时预测和修正,预测结果显示对2017年第三季度的GDP增长率进行实时预测时,U-AR（4）-MIDAS（3,9,5）模型最优。
This paper starts from the PMI index, which represents the ＂barometer＂ of manufacturing industry to discuss the role of China＇s manufacturing industry in GDP mixed-frequency forecast. Firstly the paper establishes MIDAS model to make a research on the real-time forecast of GDP growth rate, and then introduces the PM1 index and hysteresis effect of GDP growth itself into the GDP trend prediction to relize the linkage analysis of the high frequency data PMI index and low frequency data GDP growth. The results show that 1） the model （U-AR（4）-MIDAS（3,K） has a comparative advantage; 2） the high frequency monthly PMI has both positive and negative effect on quarterly GDP, which effect lasts 9 months, and there is also an interaction among （;DP itself, which interaction lasts up to 4 quarters; 3） the model U-AR（4）-MIDAS（3, K, h） is suitable for short term use of the latcst monthly PMI for real-time prediction and correction, and the prediction resull shows that the U-AR（4）-MIDAS （3,9,5） model is optimal for real-time GDP growth forecast in the third quarter of 2017.
作者
丁黎黎
孙文霄
韩梦
康旺霖
Ding Lili;Sun Wenxiao;Han Meng;Kang Wanglin(a.Institute of Economics,b.Institute of Marine Development,Ocean University of China,Qingdao Shandong 266100,China;School 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
2018年第15期128-132,共5页
Statistics & Decision
基金
国家自然科学基金资助项目(71471105)
国家社会科学重大项目(15ZDB1717)
泰山学者工程专项经费资助(tsqn20161014)
