Monetary Policy Design under Imperfect Knowledge: An Open Economy Analysis

This paper incorporates adaptive learning into a standard New-Keynesian open economy dynamic stochastic general equilibrium (DSGE) model and analyze under what conditions policymakers should target domestic producer price inflation (DI) versus consumer price inflation (CI). Our goal is to examine how monetary policy rules should adjust when agents’ information sets deviate from those assumed under the rational expectation paradigm. When agents form expectations using an adaptive learning mechanism, even though the central bank has no informational advantage, monetary policy can nonetheless facilitate the learning process and thus mitigate distortions associated with imperfect knowledge. We assume the policy-maker follows a forwardlooking Taylor rule and focus on analyzing the interplay between the source of the dominant shock and the extent of knowledge imperfection. We find that when agents have very limited knowledge and have to learn the dynamics governing both the relevant economic indicators and the underlying structural shocks, a DI targeting rule introduces fewer forecast errors and is better at stabilizing the economy. However, when agents can observe contemporaneous shocks and need only learn how key economic variables evolve (a situation akin to a post-structural-shift economy), targeting away from the dominant shocks helps anchor expectations and improve welfare. A CI target can then become the preferred policy rule when the economy is subject to large domestic shocks.