Firestorm: Multiplicity in Models with Full Information

Abstract

Dynamic stochastic models with full information and rational expectations (FIRE) are not as well determined as is commonly believed. If the assumption of causality is relaxed so that prices and decisions may anticipate future shocks, then FIRE models generally feature multiple equilibria. The multiplicity is due to the endogenous feedback from choices to information to choices, which in equilibrium may contain self-fulfilling news about future shocks. I demonstrate the multiplicity in several examples, including canonical asset pricing and business cycle models. To motivate relaxing the causality assumption, I also study examples with apparent non-causality, even if the model is fundamentally causal. Then I examine how the multiplicity arises in a dynamic programming problem with decentralized markets. Finally, I argue that the business cycle literature must adopt information frictions.

Keywords: News, Self-Fulfilling Equilibria, Multiplicity, Rational Expectations

JEL Classification: C62, D50, D84, E32