Forecasting Stock Returns through An Efficient Aggregation of Mutual

Fund Holdings We develop a stock return-predictive measure based on an efficient aggregation of the portfolio holdings of all actively managed U.S. domestic equity mutual funds, and use this model to study the source of fund managers ’ stock-selection abilities. This “generalized-inverse alpha ” (GIA) approach reveals differences in the ability of managers to predict firms ’ future earnings from fun-damental research. Notably, the GIA’s return-forecasting power is not subsumed by publicly avail-able quantitative predictors, such as momentum, value, and earnings quality, nor is it subsumed by methods shown in past research to forecast stock returns using fund holdings or trades. 2 The value of active investment management is a long-standing controversy. While it has been well documented that actively managed US equity funds, on average, underperform passive bench-marks, recent studies on fund returns find that a subset of funds exhibit some ability to outperform, especially when evaluated gross of fund expenses. Another body of literature uses fund portfoli