For many benchmark predictor variables, short‐horizon return predictability in the U.S. stock market is local in time as short periods with significant predictability (“pockets”) are interspersed with long periods with no return predictability. We document this result empirically using a flexible time‐varying parameter model that estimates predictive coefficients as a nonparametric function of time and explore possible explanations of this finding, including time‐varying risk premia for which we find limited support. Conversely, pockets of return predictability are consistent with a sticky expectations model in which investors slowly update their beliefs about a persistent component in the cash flow process.

Forecasts of professional forecasters are anomalous: they are biased, and forecast errors are autocorrelated and predictable by forecast revisions. We propose that these anomalies arise because professional forecasters do not know the model that generates the data. We show that Bayesian agents learning about hard-to-learn features of the world can generate all the prominent aggregate anomalies emphasized in the literature. We show this for professional forecasts of nominal interest rates and Congressional Budget Office forecasts of gross domestic product growth. Our learning model for interest rates can explain observed deviations from the expectations hypothesis of the term structure without relying on time variation in risk premia.