Abstract:
This paper addresses the relative importance of monetary indicators for forecasting inflation in the euro area in a Bayesian framework. Bayesian Model Averaging (BMA)based on predictive likelihoods provides a framework that allows for the estimation of inclusion probabilities of a particular variable, that is the probability of that variable being in the forecast model. A novel aspect of the paper is the discussion of group-wise inclusion probabilities, which helps to address the empirical question whether the group of monetary variables is relevant for forecasting euro area inflation. In our application, we consider about thirty monetary and non-monetary indicators for inflation. Using this data, BMA provides inclusion probabilities and weights for Bayesian forecast combination. The empirical results for euro area data show that monetary aggregates and non-monetary indicators together play an important role for forecasting inflation, whereas the isolated information content of both groups is limited. Forecast combination can only partly outperform single-indicator benchmark models.