We study the asset pricing implications of learning in an environment in which the true model of the world is a multivariate one, but agents update only over the class of simple univariate models. Thus, if a particular simple model does a poor job of forecasting over a period of time, it is discarded in favor of an alternative simple model. The theory yields a number of distinctive predictions for stock returns, generating forecastable variation in the magnitude of the value-glamour return differential, in volatility, and in the skewness of returns. We validate several of these predictions empirically.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below