统计研究

• 论文 • 上一篇    下一篇

基于多层因子模型的我国核心通货膨胀估计

高华川 赵娜   

  • 出版日期:2016-04-15 发布日期:2016-04-05

Estimation of China’s Core Inflation Based on Multi-level Factor Model

Gao Huachuan & Zhao Na   

  • Online:2016-04-15 Published:2016-04-05

摘要: 本文利用我国CPI分类数据,基于多层因子模型构建了一种新的核心通货膨胀指标。通过与单层因子模型方法的对比研究发现:单层因子模型会高估核心通胀的波动性,原因在于其无法识别仅影响大类价格变化的局部因子,并将局部因子的大部分作用归于具有普遍性影响的全局因子。而多层因子模型不仅能够识别局部因子,并且利用分离了局部因子之后的全局因子构造的核心通胀指标具有更小的波动性,且更符合核心通货膨胀的内涵。

关键词: 多层因子模型, 核心通货膨胀, 标题通货膨胀, 货币政策

Abstract: We use China’s CPI data to construct a new core inflation index based on multi-level factor model. Compared with the single-level factor model, we find that: single-level factor model could overestimate the volatility of core inflation, because it is unable to identify regional factors that only affect sector price, thus it attributes the bulk of impact of regional factors to global factor which has pervasive influence. While the multi-level factor model could identify regional factors, and the core inflation which is constructed by the global factor has smaller volatility, and is consistent with the concept of core inflation.

Key words: Multi-level Factor Model, Core Inflation, Headline Inflation, Monetary Policy