Optimal Disinflation under Learning

Abstract

We examine how to design a disinflation when agents must learn the new policy rule. Erceg and Levin (2003) demonstrate that learning increases inflation persistence and the sacrifice ratio relative to what would occur under rational expectations. Accordingly, one might think that learning promotes gradualism. On the other hand, a gradual disinflation might retard the rate at which private agents learn, thus prolonging the transition and making it more turbulent. We sort out the competing forces in the context of a simple new Keynesian model. Despite the presence of sticky prices and learning, the optimal policy brings inflation down sharply and converges to the new target from below. This promotes rapid learning and minimizes the duration and turbulence of the transition