中国核心通货膨胀多变量动态测度  

Multi-variable Dynamic Measurement of China’s Core Inflation

作　　者：周德才[1] 汪小丁 张倩 Zhou Decai;Wang Xiaoding;Zhang Qian(School of Economics and Management,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学经济管理学院,南昌330031

出　　处：《统计与决策》2021年第24期119-123,共5页Statistics & Decision

基　　金：教育部哲学社会科学研究后期资助项目(19JHQ089);江西省高校人文社会科学研究项目(TJ19102);江西省学位与研究生教育教学改革研究项目(JXYJG-2018-015)。

摘　　要：核心通货膨胀测度的难点是从多个变量序列中同时时变提取共同因子、时变滤除噪音和时变识别异常因子。为此,文章沿用趋势定义,选取中国2001年1月至2020年4月的总体和八类CPI,构建加入异常值调整的多变量不可观测成分随机波动(MUCSVO)模型,将其长期趋势与暂时冲击项的因子载荷、波动率和异常因子,使用基于MCMC的Gibbs抽样法进行估计,测度中国总体和分类核心通货膨胀,并从统计性质分析、因果关系检验、预测能力检验和跨期相关性分析四个方面进行评价。结果表明:使用MUCSVO模型测度的中国总体和分类核心通货膨胀,能准确地反映通货膨胀的长期趋势;验证了使用MUCSVO模型测度中国总体和分类通货膨胀的合理有效性;中国通货膨胀的因子载荷、波动率和异常因子具有均值和分布差异明显的时变特征,较好地刻画了核心通货膨胀的多维动态变化。The difficulty of core inflation measurement is to extract common factors, filter noise and identify abnormal factors from multiple variable sequences simultaneously. To this end, this paper adopts the trend definition and selects the total and eight types of CPI in China from January 2001 to April 2020 to construct the multivariate unobserved components model with stochastic volatility and outlier adjustments(MUCSVO), and then estimates the factor load, volatility and abnormal factors of long-term trend and temporary shock by MCMC-based Gibbs sampling method to measure China’s overall and classified core inflation. The paper alsocarries out evaluation from the four aspects of statistical property analysis, causal relationship test, predictive ability test and inter-temporal correlation analysis. The results show that China’s overall and classified core inflation measured by the MUCSVO model can accurately reflect the long-term trend of inflation, that it is reasonable and valid to use MUCSVO model to measure China’s overall and classified inflation, and that the factor load, volatility and abnormal factors of China’s inflation have time-varying characteristics with obvious differences in mean and distribution, which can well describe the multidimensional dynamic changes of core inflation.

关 键 词：通货膨胀 MUCSVO模型 时变 异常因子 

分 类 号：F224[经济管理—国民经济]