Learnability and equilibrium selection under indeterminacy

This paper presents an equilibrium selection procedure for linear rational expectations models under equilibrium indeterminacy. In the procedure, a specific equilibrium is chosen such that the perceived law of motion for agents with adaptive learning rules converges to the particular solution. The procedure is applicable only under empirically implausible parameter settings if the agents perceive fundamental shock processes completely. However, if they misperceive the fundamental shocks in some degree, the procedure works under plausible parameters. It is also demonstrated that the selected particular solution exhibits different dynamics from that under the assumption often made in the literature.