Why Have Estimate Revision Measures Not Worked in Recent Years?

Abstract

Estimate revisions, whether measured using consensus or individual analyst data, continue to exhibit positive serial correlation. This suggests that the failure of estimate revision strategies in recent years was a result of investor overreaction to the widely known bias in analyst estimates. To measure this overreaction, the author presents a model that compares the persistence of estimate revisions expected by investors and the actual level of persistence warranted by analyst behavior. The results suggest that instead of simply selecting stocks with positive estimate revisions, the new task for active managers is to determine how much information in estimate revisions is already reflected in stock prices.