Description

Inflation is measured through price indices for a basket of diverse products and services. However, its various components may follow different paths depending on the pricing mechanisms in each market. This has led to building price sub-indices with a higher degree of homogeneity within them for analytical purposes. This paper focuses its attention on the behavior of prices of services in Chile. The latter are of special interest because they not only account for about half of the Consumer Price Index (CPI), but because their dynamics over time have differed significantly from the prices of goods. Our empirical analysis is based on the estimation of a set of Hybrid Neokeynesian Phillips Curves (HNKPC), which allows comparing the value and statistical significance of the estimated elasticities of various CPI components to those obtained for headline. The results suggest that: (i) HNKPC maintains a high degree of validity as a proxy for inflation determination, with significant values for key elasticities in all of its disaggregations; (ii) the disaggregated HNKPC estimate does not reduce the statistical significance of the results; (iii) Chile's inflationary history explains the high elasticity of lagged inflation and inflation expectations; (iv) the aggregate dynamics of services inflation differs from that of goods in both their adjustment and their response to shocks on its main components, a response that has varied unevenly in the recent period; (v) important differences in the behavior of indexed and non-indexed services by the nature of the covariates and the size of the basic HNKPC coefficients are found; (vi) the disaggregation of services raises its conceptual consistency without jeopardizing the forecasting ability of services inflation as a whole and is superior to other disaggregations of services, and (vii) the construction of indices by collecting prices at producer or provider level makes it difficult to compare the changes in purchasing power resulting from the application of specific taxes or subsidies. All these results suggest the usefulness of splitting the CPI between large components or sub-indices to improve inflation forecasts, including their insertion in more complex models for macroeconomic analysis.