Acting on the Most Valuable Information: "Best Idea" Trades of Mutual Fund Managers

Abstract

I propose a proxy for how attractive a given trade is to the fund manager who makes it and find evidence that mutual fund managers exhibit substantial skill in trades they ex ante identify as more promising. The proxy for ``best ideas'' does not use any performance-related variables. Instead, it is based on common trades of managers from the same management company. Such managers likely have access to similar information. The more important the new information they receive, the more likely it is that multiple managers will act on it and engage in similar trades. Best ideas outperform benchmarks and other fund trades by as much as 47 basis points per month. This effect is already apparent for trades at least two same-company managers make and its magnitude is increasing in the number of managers trading. The outperformance is visible not only in a subset of fund trades, but also in funds' after-fee returns: Funds which participate in best idea trades beat other funds by up to 0.3% per quarter. The remaining fund trades (not best ideas) fail to beat passive benchmarks even before expenses, e.g., their average characteristic-adjusted return is statistically insignificant 3 basis points per month. The performance of best ideas is not explained by herding. In fact, there is some evidence that trades made by multiple unrelated managers (managers from different companies) underperform passive benchmarks even before expenses. The effect is also robust to other alternative explanations (e.g., short-term liquidity pressures, changes in index composition, or funds' reaction to analyst revisions), and does not revert in the long term.