Abstract:
We study the impact of adaptive learning for the design of a robust monetary policy using a small open-economy New Keynesian model. We find that slightly departing from rational expectations substantially changes the way the central bank deals with model misspecification. Learning induces an intertemporal trade-off for the central bank, i.e., stabilizing inflation (output gap) today or stabilizing it tomorrow. The central bank should optimally anchoring private agents expectations in the short term in exchange of easier future intratemporal trade-offs. Compared to the rational expectations equilibrium, the possibility to conduct robust monetary policy is limited in a small open economy under learning for any exchange rate pass-through level and any degree of trade openness. The misspecification that can be introduced into all equations of the model is lower in a small open economy, and approaches zero at high speed as the learning gain rises.