Research

WORKING PAPERS

[Internet Appendix]

Abstract: Active fund management models have two key features: sophisticated investor learning and decreasing returns to scale. I study mutual fund analyst reports and use dictionary-based and machine-learning approaches to measure report tone and examine the extent to which analysts are concerned with fund size. My three main findings are: (1) fund flows react to report tone; (2) high-tone funds outperform low-tone funds by an abnormal net-of-fee return of 0.94% per year; and (3) size-related vocabulary features more prominently when funds operate at inefficient sizes. These results are in principle consistent with the two key model features: (i) Some investors exert effort to learn about managerial skill from analyst reports; (ii) reports are informative, however, investors reallocate their assets slower than implied by active management models; and (iii) professional analysts qualitatively comprehend the concept of decreasing returns to scale. Overall, the results suggest that analysts and investors qualitatively understand key model features but quantitatively misjudge them.

[Internet Appendix]

Abstract: We recover forward-looking expected net-of-fee abnormal returns (alphas) for active equity mutual funds from analyst ratings. In contrast to the typical equilibrium implication of zero alphas, analyst alphas are negative for most funds, but positive for the largest funds. We compare analysts' subjective expectations with expectations from a rational expectations learning model. The model's rational learner believes that an increase in fund size leads to a decrease in returns, but we find no evidence that analysts believe so. Overall, analysts' expectations and the capital that follows analysts' recommendations are difficult to reconcile with existing rational expectations models of active management. 

WORK IN PROGRESS