Slicing up inflation: analysis and forecasting of Lithuanian inflation components

ENIn this paper we model five Lithuanian HICP subcomponents in a medium scale Bayesian VAR framework. We deal with the parameter proliferation problem by setting the appropriate amount of shrinkage determined in the out-of-sample forecasting exercise. The main body of the paper consists of displaying the model’s performance in two applications: forecasting and analysis of inflation determinants. We find the model’s forecasts to be competitive against the univariate statistical models, particularly in the cases of predicting processed food and energy goods inflation. What is more, exercises based on conditional forecasting show that these two indices make the best use of accurate conditional information in terms of improving predicting accuracy. In the decomposition of the drivers of HICP components, we demonstrate that both, domestic and foreign factors can be prevalent inflation determinants in certain time periods. We also find some evidence on employees’ bargaining power playing a role in determining the Lithuanian consumer price inflation. Keywords: HICP subindices, Bayesian VAR, Bayesian shrinkage, inflation forecasting, structural decomposition.