Adaptive Learning, Model Uncertainty and Monetary Policy Inertia in a Large Information Environment

Abstract

Due to the existence of imperfect information, central banks need to monitor a large variety of data series. This paper provides an attempt to model monetary policy-making in a large information environment. With a large information set, model uncertainty is likely to be very pervasive. We propose to model model uncertainty by means of Bayesian Model Averaging (BMA). We discuss some advantages of this technique over robust control or mere model selection. Parameters' estimates and models are updated over time through adaptive learning. In this enriched framework, we try to give an explanation of central banks' observed monetary policy inertia.

Keywords: Optimal monetary policy, Bayesian Model Averaging, leading indicators, model uncertainty, adaptive learning, interest-rate smoothing, inertia

JEL Classification: C11, C15, C52, E52, E58