Sacha Gelfer (University of Oregon) presented current work involving the introduction of adaptive learning to households in a medium-scale New Keynesian model featuring many frictions and shocks. The boundedly rational agents form expectations about future aggregate variables using recursive OLS and choose from three different models or “perceived laws of motion” using the Bayesian Information Criterion. Some of these include the usage of a professional forecast while other do not.  Results indicate agents indeed react to changes in variables by adapting their perceived law of motion.  Simulations show that macroeconomic volatility decreases when an accurate professional forecast announced to agents is used in their perceived law of motion. However, if their is large noise around the dissemination of the forecasts to the agent macroeconomic volatility will increase.

Slides from the presentation are here.