Measuring the dispersion of beliefs by calculating the difference between a stock’s weighting in an active manager’s portfolio and its weighting in a benchmark index portfolio, the authors find that active holdings positively predict stock returns. The results are strongest for stocks with higher information asymmetry. They hypothesize that well-informed managers with positive private information increase the dispersion by placing large bets but are less able to act on negative private information because of binding short-sale constraints.
What’s Inside?
The authors look at actively managed mutual funds and examine how those managers’ holdings deviate from the appropriate benchmark, positing that those active holdings are a measure of dispersion of beliefs. They find that active holdings, in both the level of dispersion and changes in the degree of dispersion, positively predict future stock returns on a risk-adjusted basis. The authors hypothesize that differentially informed active managers who receive positive private information drive up the dispersion in active holdings, but because of short-sale constraints, they are unable to fully use negative private information, which leads to lower dispersion and a greater appearance of homogeneity in that situation.
How Is This Research Useful to Practitioners?
Mutual funds have become increasingly important as a substantial segment of the investing universe, and most mutual funds are actively managed. Thus, understanding the relationship between the beliefs and corresponding actions of active mutual fund managers and subsequent stock returns can be enlightening. The authors propose the idea of measuring the dispersion of beliefs about future returns on individual stocks by calculating the difference between the weight of an individual stock in the active mutual fund manager’s portfolio and its weight in the manager’s benchmark portfolio. Presumably, a higher weighting than the benchmark reflects an optimistic belief about the future price of the stock.
Having created this measure of dispersion, the authors then examine how the measure correlates with future stock returns at various levels of dispersion and what happens when the measure of dispersion increases or decreases. They also consider subsamples for various stock characteristics and within different industries. They find that the return-forecasting power is particularly strong for stocks with high information asymmetry, as measured by the adverse selection component of the bid–ask spread and idiosyncratic volatility, and within industries that tend to have high information asymmetry, such as information technology and health care.
The authors also consider how short-sale constraints affect the return-forecasting power of dispersion in active holdings and find that the results are in line with their hypothesis of differentially informed managers.
How Did the Authors Conduct This Research?
First, the authors construct a measure of dispersion that consists of active holdings defined as the difference between the weight of a stock in the active manager’s portfolio and its weight in the appropriate benchmark. Their sample contains 2,667 active equity funds from the first quarter of 1984 to the third quarter of 2008. From a selection of 19 commonly used benchmark indexes, they choose for each fund for each quarter the benchmark that minimizes the average distance between fund portfolio weights and the benchmark index weights.
Using both a portfolio-sorting approach and a multivariate regression approach, the authors relate the active holdings dispersion measure to future stock returns. The results show a strong relationship between level of dispersion and higher subsequent stock returns and an even stronger relationship between an increase in dispersion and future outperformance. The forecasting power of the level of dispersion is statistically significant for only the next quarter and does not show return reversal in the following three years.
These findings lead the authors to hypothesize that well-informed managers who receive positive signals about a particular stock tend to place large bets on that stock, which increases the level of dispersion among active fund managers. When well-informed managers receive negative signals, however, binding short-sale constraints prevent them from unrestrainedly using that private information, which reduces the level of dispersion in active holdings.
Abstractor’s Viewpoint
The authors have devised a simple yet elegant way of inferring the beliefs of active mutual fund managers and relating those inferred beliefs to future stock returns. They take a variety of approaches to examining that measure, from looking at it from different angles to considering the ramifications of their findings. It will be interesting to see the results of their future research on dispersion of beliefs and aggregate market returns.