Using the First Principal Component as a Core Inflation Indicator

This paper investigates the consequences of non-stationarity for the principal components analysis and suggests a data transformation that allows obtaining smoother series for the first principal component to be used as a core inflation indicator. The paper also introduces a theoretical model, which allows interpreting core inflation as a common stochastic trend to the year-on-year rates of change of the price indices of the basic CPI items. Finally, it is shown that the first principal component computed in real time meets the evaluation criteria introduced in Marques et al. (2000).