NOISE, BAYES’ RULE AND THE RECENCY HEURISTIC

Abstract

This paper studies empirical market anomalies and the underlying model of market efficiency. Within a Bayesian framework, I find a closed form solution for the decision process under a random walk with noise model of market efficiency. The main conclusions are as follows:

1. A “bad market efficiency model problem” exists in addition to Fama’s (1998) “bad [asset pricing] model problem.” Noise is important and should be modeled explicitly. 2. Bayesian updating is a valid and insightful methodology. In order to support the psychologist’s claim that Bayes’ rule is invalidated by observing recency in decision making would require implausible levels of noise in market prices.

While providing a unified explanation of past market anomalies, the model endogenously derives the intuitive insight that markets are noisy in the short term and informative in the long term. JEL subject classifications: G12, G13, C14, G22

Key words: Bayesian updating, information extraction, behavioral finance, market efficiency, recency bias, noise.

[此贴子已经被作者于2004-10-27 23:18:48编辑过]