A Data-Rich Measure of Underlying Inflation for Brazil

This paper proposes a new measure of underlying inflation for Brazil based on a generalized dynamic factor model (GDFM). The approach summarizes a wide set of indicators, which the Banco Central do Brasil (BCB) regularly monitors in its assessment of the inflation scenario, such as data on prices, activity, financial and monetary variables. Differently from most core inflation approaches, the model takes account of the time series dimension – by extracting the lower frequency component – as well as the cross-section dimension and is able to handle end-of-sample unbalances. To our knowledge, it is the first application of this procedure for Brazil. The resulting series exhibits lower variability, unbiasedness and a relatively good forecasting performance compared to various other measures of trend inflation. Overall, the findings suggest the novel underlying inflation measure may be an important complement to the information set used by the BCB.