穿越周期的通胀预测:季节性模型及高频数据的适用性  

Inflation Forecasts and the Applicability of the Seasonal Model and High Frequency Data

作　　者：熊伟琪 张鹏[1] 

机构地区：[1]中信银行总行金融市场部

出　　处：《金融市场研究》2021年第6期90-98,共9页Financial Market Research

摘　　要：本文首先综述目前广泛应用的通胀模型,包括统计计量模型、结构化的经济模型,以及非结构化的季节性模型。其次,通过在季节性模型中加入宏观高频指标,实现对CPI、PPI环比指标的实时预测,并检验预测精准度。最后,本文实证表明,在以2020年为基期的新一轮基期轮换前后,季节性模型的预测表现稳定,仍然适用于通胀预测;而在其他时期,包含宏观高频数据的模型预测表现要优于季节性模型;此外,也不能忽视引入高频数据带来的模型过度拟合风险。This paper summarizes the in flation models widely used at present,including the statistical measurement model,the structured economic model and the unstructured seasonal model.By adding high-frequency macro indicators to the seasonal model,real-time forecasts of the CPI and PPI are achieved,and their accuracy can be verified.This paper shows that the seasonal model is stable before and after the new phase of base period rotation.With 2020 taken as the base period,this model is still suitable for inflation forecasting.In other periods,the predictive performance of the model containing high-frequency macro data is better than that of the seasonal model alone.Additionally,the risk of model over-fitting caused by the introduction of high-frequency data cannot be ignored.

关 键 词：CPI预测 PPI预测 宏观高频数据 基期轮换 

分 类 号：F822.5[经济管理—财政学]