Convergence of adaptive learning and the concept of expectational stability in linear rational expectations models with multiple equilibria

This paper analyzes the relationship between the expectational stability of rational expectations equilibria and the possible convergence of adaptive learning processes. Results obtained using recursive least squares lead to the conjecture that a correspondence between these both selection criteria exists. This paper demonstrates on the basis of a simple linear model and an alternative learning procedure that such a conjecture would be incorrect: there exist learning procedures that may converge to rational expectations equilibria which are not expectationally stable.