Coordination on Saddle Path Solutions: The Eductive Viewpoint. 1 - Linear Univariate Models

Abstract

We investigate local strong rationality (LSR) in a one step forward looking univariate model with memory one. Eductive arguments are used to determine when common knowledge (CK) that the solution is near some perfect foresight path is sufficient to trigger complete coordination on that path (I.e. the path is LSR). Coordination of expectations is shown to depend on three factors: the nature of the CK initial beliefs, the degree of structural heterogeneity and the information structure. Our sufficient conditions for LSR precisely reflect these features and provide basic consistent justifications for the choice of the saddle path solution.