Selective Learning and Price Over- and Under-Reaction

Abstract

In this paper I propose a new theoretical framework that explains both over- and under-reaction of prices to new information. My model incorporates a behavioral investor who selectively chooses whether to incorporate new information. When the agent chooses to ignore information, the price under-reacts. On the other hand, when the agent chooses to incorporate information, the price over-reacts. While existing frameworks in the literature use multiple behavioral biases simultaneously to generate both over- and under-reaction, I rely on a single behavioral bias. My model generates novel empirical predictions. More specifically, over-reaction is stronger when the new information is extreme and when the agent faces higher uncertainty (and vice versa for under-reaction). I provide empirical support for these predictions using winner-loser portfolios. In addition, my model does not produce some existing empirical predictions that have been contested, e.g., the life-cycle hypothesis of momentum and reversal.

Keywords: Over-reaction, under-reaction, learning, beliefs

JEL Classification: G11, G12, G14, G41