‘Best Idea’ Trades of Mutual Fund Managers [STUDY]
Lukasz Pomorski
University of Toronto – Rotman School of Management
March 17, 2009
Any given portfolio manager is likely to have a number of investment ideas. They are not all equal and the manager will rank them according to their attractiveness. This premise, coupled with a limited amount of money each idea can potentially absorb, has become a vital part of our understanding of the fund industry, both through theoretical (e.g., Berk and Green, 2004) and empirical work (e.g., Chen, Hong, Huang, and Kubik, 2004, Pollet and Wilson, 2008). However, managers’ rankings are not observable, and, consequently, we know relatively little about about how attractive the best ideas” are compared to the benchmarks and other trades, how frequently they occur, and how much they affect overall fund returns.
This paper addresses such questions using a proxy for the ranking of attractiveness of funds’ trades. Importantly, this proxy is not based on any performance-related information. While fund managers’ best ideas indeed turn out to generate abnormal pro¯ts, under this proxy the ranking of trades is determined before performance is evaluated. Distinguishing “best ideas” from other transactions offers a novel perspective on the ability of portfolio managers and sheds light on what the “upper limit” of their skill is (i.e., how much they could outperform in the best of circumstances).
To identify trades that managers consider particularly valuable I study management companies that sponsor multiple funds. Managers employed by such companies share resources, such as buy-side research, and exchange information in their company’s communication networks. If the information they share includes a valuable news item, multiple funds are likely to act on it and buy or sell similar stocks. Thus, when multiple funds that belong to the same company trade the same stock in the same direction, this trade is classi¯ed as a best idea.” This proxy is fundamentally di®erent from simply counting possibly unrelated funds that trade a stock in that managers from the same company have access to similar information. Making the same trade in multiple portfolios is a vote of confidence for a particular item from their information set. In contrast, managers from unrelated companies have different information sets. Even if they receive the same news, it may be less valuable than other information they already have.
Thus, when multiple unrelated managers make the same trade, it may not be the best idea” of any of them.
In the quarter subsequent to when best ideas” are identified, these trades beat benchmarks and other fund trades by as much as 0.33% per month. This effect is apparent already in trades at least two same-company funds make. However, the results are stronger when three-or-more-fund trades are considered, and the strongest when at least four funds trade the same stock.
It is crucial that best ideas are generated from information shared by same-company managers. In contrast, trades made by multiple managers from unrelated companies, if anything, under-perform the benchmarks. Consequently, the difference in performance of best ideas and trades repeated by multiple unrelated funds reaches 0.47% per month.
The number (or capacity) of best ideas is fairly low: They account for only about 30% of a typical management company’s dollar volume. The remaining trades, given the proxy proposed in this paper, rank lower in managers’ list of ideas. Unfortunately for fund investors, these additional trades do not outperform passive benchmarks even before transaction costs. For example, the 95% confidence interval for the characteristic-adjusted returns on such trades is from about -4 to 8 basis points per month. Even the highest values from this interval are likely too small to cover transaction costs and other fund expenses. Since such additional trades account for most of fund volume, this helps explain why overall after-fee returns of mutual funds are, on average, negative.
‘Best Idea’ Trades of Mutual Fund Managers [STUDY] via SSRN
