摘要
混频数据模型能有效避免传统计量经济模型对不同频率数据"预处理"带来的有效信息损失或无效信息虚增问题。文章将混频数据模型的演变发展分为3个阶段,系统梳理了MIDAS模型及其衍生模型的提出背景及国内外应用领域;分别从预测方向及时效、变量指标选择、研究对象的频率差、滞后阶数及权重函数确定、基准模型比较5个方面分类介绍了混频数据模型的应用研究。最后,提出混频数据模型的未来研究方向可考虑拓宽理论模型应用领域,将现有金融经济领域的最新研究方法拓展应用到其他行业;引入统计学中变量选择的方法,更科学地筛选混频变量指标。
Mixed-frequency data model can effectively avoid the problem of losing effective information or increasing invalid information caused by the"preprocessing"of data with different frequency by traditional econometric model.This paper divides the development of mixed-frequency data model into three stages,systematically reviews the background of MIDAS model and its derivative model as well as the application fields at home and abroad,and then introduces the application research of mixed-frequency data model from five aspects respectively:prediction direction and time limit,variable index selection,frequency difference of the research objects,determination of lag order and weighting function,comparison of reference models.Finally,it is suggested that the future research direction of mixed-frequency data model should consider broadening the application fields of theoretical models and extending the latest research methods in the current financial and economic fields to other industries,and that the method of variable selection in statistics can be introduced to screen mixed variable indexes more scientifically.
作者
吴培
李哲敏
Wu Pei;Li Zhemin(Agricultural Information Institute,Chinese Academy of Agricultural Sciences,Beijing 100081,China;Graduate School,Chinese Academy of Agricultural Sciences,Beijing 100081,China)
出处
《统计与决策》
CSSCI
北大核心
2021年第8期23-28,共6页
Statistics & Decision
基金
中国农业科学院创新工程项目(CAAS-ASTIP-2020-AII-02)
农业农村部农业科研杰出人才经费资助项目
关键词
混频数据模型
模型比较
变量选择
滞后阶数
权重函数
mixed-frequency data model
comparison of models
variable selection
lag order
weighting function