Publicly disclosed mutual fund portfolio holdings contain valuable information about stock fundamentals and can be used to predict future stock returns. The authors’ analysis suggests that investors can exploit the stock selection information of fund managers without directly investing in their funds.
The authors develop a measure for predicting stock returns that aggregates the portfolio holdings of all actively managed U.S. domestic equity mutual funds. They use this model to show that a subset of funds exhibits some ability to outperform and can be accessed gross of fees. The goal of the approach is to improve on simple copycat strategies that mimic the holdings of a single mutual fund. The approach takes advantage of a cross-sectional copycat strategy that invests long in stocks that are held in common by many top-performing funds and that are not held by very many bottom-performing funds and invests short in stocks for which the reverse is true.
How Is This Research Useful to Practitioners?
An increase in the transparency of financial institutions is a cornerstone goal of recent regulatory changes, such as the Dodd–Frank Act. The implication is that expanded disclosure requirements level the playing field for the average investor, but these changes also allow closer examination of the value of active mutual fund management. A burgeoning number of studies have explored how required disclosure of holdings can provide useful information to investors seeking to precisely measure manager skill and to evaluate fund investment style, strategy, and risk taking. The research is notable for the development of new approaches to measuring and attributing precisely the performance of mutual funds as well as devising winning strategies for investing in these vehicles.
The authors suggest that mutual fund managers, in aggregate, have better stock selection abilities than other investors and conclude that the stock selection information produced by fund managers through fundamental research is distinct from the information contained in publicly available accounting information and traditional quantitative signals. Notably, they find that the return-forecasting power remains significant after they control for well-known and publicly available quantitative predictors common in empirical asset pricing, such as size, book-to-market ratio, and momentum, and in linear factor models, such as Fama–French and Carhart.
How Did the Authors Conduct This Research?
Using historical disclosures that funds must file every quarter, the authors introduce a technique that systematically harvests the best stock selections in the fund industry. The objective is to examine the aggregate holdings of mutual funds to measure the aggregate value of active fund management and use this information to estimate future stock returns rather than future fund returns. Stocks that are heavily held in common by multiple fund managers exhibiting past skills might be expected to outperform. Using data from Thomson Reuters and CRSP, the authors study a sample period from 1980 to 2006; this period was characterized by tremendous growth in the number of actively managed domestic equity mutual funds.
Making adjustments to reflect the fact that observation of mutual fund holdings is delayed, the authors form portfolios that favor stocks that appear mainly in market-beating funds while shorting stocks that appear mainly in lagging funds. They introduce a methodology to obtain the expected value of future stock alphas that they refer to as “generalized-inverse alpha,” which is in reference to an unusual statistical approach necessitated by the fact that there are not enough funds to solve for each stock’s alpha because there are more stocks than mutual funds.
A portfolio’s weight measures the size of a bet and is a proxy for the amount of information possessed by the manager, whereas the fund’s past alpha estimates the precision of the private information. The authors find strong evidence that fund managers possess stock selection information and alternative skill proxies across a wide spectrum of characteristics. This evidence is consistent with the view that fund managers may be uncovering information about corporate fundamentals, but the return-predictive information possessed by fund managers with persistent skills is relatively short lived.
Explaining cross-sectional differences in asset expected returns is one of the great challenges of modern finance. The authors’ work contributes to studies of the efficiency of securities markets, as well as the role of institutional investors in setting stock prices and the impact of their trades on stock markets. The results are particularly interesting in light of recent studies that have found improved stock market efficiency over time. If managers’ “high conviction” best-idea investments outperform, it would speak to the suitability of concentrated equity positions as a vehicle for higher returns and the need to suppress powerful institutional incentives to over-diversify and dilute performance.