Two-sided Learning and Optimal Monetary Policy in an Open Economy Model

In this paper, we consider a dynamic New Keynesian model of the small open economy in the light of bounded rationality. This entails private agents and the central bank updating their beliefs about the laws of motion of inflation, the output gap and real exchange rate, when forming their optimal decisions. It is shown that when all players learn using recursive least-squares or stochastic-gradient adaptive algorithms, the optimal policy steers the economy towards a rational expectations equilibrium (REE) with probability one in some cases. This is also the case when only private agents are learning. However there also exists structural parameter values in the true model such that learning converges with probability zero to REE.