Anomalies in Security Returns and the Specification of the Market Model



We examine the hypothesis originally advanced by Roll [12] that observed anomalies in excess returns can be explained by misspecification of the market model used to estimate systematic risk. We find substantial misspecifications in the model systematically related to size and period of listing of the securities in question. There is some evidence that these misspecifications are associated with systemic biases in measured betas used to construct excess returns.