Chinese Inflation: Measurement and Forecasting

Abstract

We propose a parsimonious semi-parametric model to uncover the nonlinear relationship between Chinese money growth and inflation, suggesting that the long-run nexus between these variables is significant only when money growth rates are high. We investigate if a model based solely on prices may explain inflation dynamics better in an era where money growth rates are lower and more stable. We further examine whether two unobserved components (UC) models by Stock and Watson (2016) are a good fit for China. We estimate these models with monthly CPI data during the December 2006-February 2023 period to dissect the persistent and non-persistent components of inflation. Then we run a forecasting competition among UC models, and other strong competitors from the literature, including the Bayesian vector autoregression models (BVAR) and time series models for the January 2015-February 2023 period. We find the multi-sector UC model provides successful forecasts across various horizons from 1- to 30-months ahead.

Keywords: E31 E37