Learning Dynamics: Complete and Incomplete Learning

Abstract

Rational expectations (RE) is currently the standard approach to modelling expectations in macroeconomics. However, refinements to modelling expectations formation have recently been developed and the literature on learning dynamics has grown rapidly. The basic idea in models of learning is that agents have limited information about the structure of the economy and they have to act inductively like scientists when they forecast the relevant aspects of its future course. Some forecasting is necessary since, as is common in economics, the decisions of individual agents depend on the future and these agents are forward looking in their decision-making. The RE hypothesis is weakened to the assumption that expectations follow a real-time learning rule which often will make expectations converge to RE over time.

This chapter is based on extensive collaboration with George W. Evans over many years. Certain other commitments save him from co-authorship. I thank Jacques Drèze for helpful comments. Financial support from the Academy of Finland and Yrjö Iahnsson Foundation is gratefully acknowledged.