We study the relationship between return uncertainty and behavioral .nance by aggregating multiple return forecasts for a single asset into an estimate of its unknown expected return. The combination of forecasts which minimizes the uncertainty of the estimated expected return is determined by the optimal information portfolio. This minimization provides an alternative explanation for biases in expected returns that have previously been attributed to psychology. Speci.cally, biases which appear similar to overcon.dence, biased self-attribution, representativeness, conservatism and limited attention arise from the information portfolio weights assigned to return forecasts. Higher dispersion across the return forecasts increases return predictability and the magnitude of these biases. However, our optimal information portfolio yields testable implications distinct from psychology, which we verify empirically using revisions in analyst earnings forecasts.
展开▼