Abstract

The paper examines the property of the core inflation indicator to predict future values of headline inflation, which is a critical task in preparing a medium-term inflation forecast to support decision-making on monetary policy. To test this property (on the example of the core inflation by Rosstat and inflation excluding fruits and vegetables, gasoline and housing and communal services), two approaches were applied. The first approach projected headline inflation based on actual core inflation and the historical relationship between these indicators, and then projected headline inflation based on actual headline inflation. The accuracy of the two headline inflation forecasts was then compared. As a result, it turned out that core inflation indicators, starting from 2011, better predict headline inflation than current headline inflation values over a horizon of up to a year. The second approach answer the question of whether the accuracy of the headline inflation forecast is improved if the core inflation forecast is used instead. To do this, within a certain class of models, a core inflation forecast was built, which was used as a headline inflation forecast, then its accuracy was compared with a forecast based on the same class of models estimated for headline inflation. As a result, it turned out that in the framework of VAR, MA(1) and ARIMA models, the use of a core inflation forecast instead of a headline inflation forecast increases the accuracy. We estimate VAR model and decompose core inflation by factors (output gap, ruble exchange rate, and monetary policy rate). It helps us to estimate the magnitude of the price shock in the 1st quarter of 2022 in Russian economy, which amounted to 6.2 p.p. Our inflation forecast based on the VAR model assumes its acceleration from 1.6% q/q SAAR in the 4th quarter of 2022 to 4.1% in the 1st quarter and 6.0% in the 2nd quarter of 2023, then it followed by stabilization at 7% QoQ SAAR by the end of 2023.