This paper presents a new theory of bubbles, or discrepancies between the market clearing price and the fundamental value of an asset. In the authors' setting, Bayesian traders, oriented towards long-term gains, receive private information ("news") and also make inferences from noisy price signals. Price exhibits higher variance than fundamental value (the latter defined.as fully-aggregated expected value) especially when news is informative but infrequent. The corresponding bubbles are self-limiting but may exhibit momentum and over-shooting. A parametric example, involving the exponential/gamma conjugate families, is provided. Copyright 1992 by Blackwell Publishing Ltd and the Board of Trustees of the Bulletin of Economic Research
