Forecasting dynamics and convergence to market fundamentals

Abstract

Subjects trade shares of a fictitious asset in a computerized double-auction market and submit one-period ahead forecasts of the mean contract price. On average, the predictions of the Rational Expectations Hypothesis are not supported and forecasts tend to be both biased and inconsistent with the specification of the process which would generate prices in a Rational Expectations Equilibrium. Nevertheless, an empirical analysis of agents' forecasting dynamics indicates that learning does occur, and that agents' forecasts evolve in a direction consistent with a Rational Expectations Equilibrium.