ON PERFECT FORESIGHT MODELS OF A STOCHASTIC WORLD

Long-run analysis is often based on perfect foresight equilibrium relationships arbitrarily augmented to incorporate stochastic shocks. This paper examines whether perfect foresight models admit interpretations as limits of stochastic economies and reflect long-run behaviour and policy effects. It provides a novel interpretation of such models, specifies consistency criteria for long-run analysis, and proposes metrics for model convergence. Analytical and calibration exercises are conducted to show that convergence to perfect foresight need not occur, and even when it occurs, it does not ensure that relationships between long-run averages or low-frequency components are captured by perfect foresight models.