Announcements, Expectations and Stock Returns with Asymmetric Information

Abstract

Revisions of consensus forecasts of macroeconomic variables positively predict announcement day forecast errors, whereas stock market returns on forecast revision days negatively predict announcement day returns. A dynamic noisy rational expectations model with periodic macroeconomic announcements quantitatively accounts for these findings. Under asymmetric information, average beliefs are not Bayesian: they underweight new information and positively predict subsequent belief errors. In addition, stock prices are partly driven by noise, and therefore negatively predict returns on announcement days when noise is revealed and the market corrects itself.