The Harm In Selecting Mutual Funds That Have Recently Outperformed
California Institute of Technology
Rayliant Global Advisors; Research Affiliates, LLC; University of California, Los Angeles – Anderson School of Business
The American College; Pennsylvania State University
February 13, 2016
In this paper, we empirically investigate the performance of commonly used fund manager selection strategies which involve hiring outperforming managers and firing underperforming managers using U.S. mutual fund data. Based on portfolios constructed using typical 3-year holding and evaluation periods, we find that the excess return to investors who chose mutual funds with poor recent performance is higher than the excess return to investors who chose mutual funds with great recent performance. Our results pose a challenge for asset owners. If the results are accepted at face value, then if past performance is used at all for hiring and firing managers it is the best performing managers that should be replaced with those who have performed more poorly. Despite our findings, a policy of firing successful managers and replacing them with poor performers is not likely to gain widespread acceptance. Instead, the practical implication of our paper is that asset owners should focus on factors other than past performance when selecting managers. We offer possible criteria that could be employed in this context.
The Harm In Selecting Mutual Funds That Have Recently Outperformed – Introduction
Dating back to Jensen (1968), the immense literature on investment performance has focused on whether or not managers possess stock picking or market timing talent that allows them to consistently produce positive risk-adjusted performance. The extensive early literature, up through Carhart (1997) and Wermers (2000) found little evidence that managers could consistently outperform the market on a risk-adjusted basis. While this result is depressing to the individuals engaged in manager selection, it is perhaps unsurprising. First of all, smart active managers competing fiercely would suggest few opportunities for a free lunch. Additionally, rational active managers ought to extract the full rent of their talent given that alpha skill is scarce and capital is plenty. Investors would be naïve to expect managers to not charge sufficient fees or attract sufficiently large assets to effectively capture the dollar alphas they generate (see Berk and Green (2004)).
However, the academic discussions of manager skill and measurements of manager outperformance ignore the reality that many trillions of dollars are already delegated through a beauty contest process, which focuses substantially on the recent three-year performance of the manager being examined for hiring or firing. For better or worse, based on December 2015 data from Morningstar Direct, investors choose to allocate twice as many assets towards actively managed funds than passively managed funds. Thus, by revealed preference, they have already made up their minds regarding the both the possibility of and the method for selecting active managers who will outperform. Given modern manager selection heuristics, it is less interesting to ask whether managers can outperform net of fees, it is more interesting to examine whether selecting managers based on recent performances can lead to outperformance for investors.
Because our paper focuses on the implications of modern manager/fund selection heuristic employed by industry practitioners, we define outperformance precisely as excess return over the stated benchmark.1 We are specifically interested in determining whether the common manager selection methodology based on recent manager excess return over benchmark would lead to future excess return over benchmark for investors.
The large literature on mutual fund flows suggests that investors often employ a simple algorithm—usually they focus only the recent 2-3 year excess return over the stated benchmark. Papers including Chevalier and Ellison (1997), Sirri and Tufano (1998), Wermers (2003), Lamont and Ellison (2008), all report that flows are positively correlated with past performance. Anecdotally, investment consultants and fiduciaries acknowledge that past outperformance is “a” if not “the” dominant manager selection criterion, because it is intuitive and thus defensible to investors. Selecting a manager based on his recent outperformance does seem perfectly rational. Past outperformance, the thinking goes, can either be due to luck or skill. If it is due to luck, hiring or firing based on past performance has no impact. If it is due to skill, moving mutual funds to more successful managers will improve future probability for outperformance. Therefore, relying on past track records may be beneficial, but in any event it will not be harmful.
In addition, from the standpoint of investment consultants and advisors, hiring managers with great recent performance and firing manager with poor recent performance is preferred because doing the opposite would be unacceptable to most investors. Indeed the performance measurement employed is almost exclusively the simple excess return over the stated benchmark without further risk adjustment because the average investor does not have the knowledge to understand what is meant by risk adjustment and risk models much less why such adjustment would be appropriate2. Some argue that in the event of dispute between an investor and his investment advisor, a selection criteria dominated by simple recent performance would be the most defensible.3 However, putting aside the agency conflict between an investment advisor and his client, the investor, what if managers with better recent outperformance were more likely to underperform subsequently over the standard investment evaluation horizon? Intuitively, if there is mean-reversion over the horizon of interest, which is roughly three years for the average institutional investors4, the modern hiring/firing practice could possibly lead to a worse outcome than the apparently paradoxical strategy of investing in managers with poor recent performance and firing the recently successful ones! While such a manager selection strategy may seem daffy, there is method in the madness.
See full PDF below.