Is selecting mutual funds based on strong performance over the last three years helpful (discovering fund manager skill) or harmful (signaling imminent fund strategy mean reversion)? In the February 2016 version of their paper entitled “The Harm in Selecting Funds that Have Recently Outperformed”, Bradford Cornell, Jason Hsu and David Nanigian investigate future mutual fund performance based on recent past performance relative to stated benchmarks. They focus on a past performance interval of three years because: institutional consultants cite this measurement as one of the most important criterion for fund selection; and, Morningstar’s rating algorithm emphasizes three-year past performance. Specifically, every three years they:
- Rank funds by expense ratio and exclude the highest tenth as likely poor choices.
- Define Winner, Median and Loser funds as the tenths of the rest with the highest, middle (centered on the 50th percentile) and lowest benchmark-adjusted returns the past three years.
- Track the performance of the equally weighted and monthly rebalanced Winner, Median and Loser groups over the next three years.
Using benchmark-adjusted returns for actively managed U.S. equity mutual funds during January 1994 through December 2015, they find that:
- Median funds outperform Winner funds based on average annual return (9.8% versus 8.0%), annual Sharpe ratio (0.48 versus 0.29) and annual four-factor alpha, which controls for market, size, book-to-market and momentum factors (-2.74% versus -0.44%).
- Loser funds beat both Median funds and Winner funds, with average annual return 10.4%, annual Sharpe ratio 0.51 and annual four-factor alpha +0.16%.
- Findings are robust to fund selection based on past adjusted return thresholds rather than rankings. Specifically, funds that underperform their benchmarks by 1% (3%) per year the last three years outperform other funds the next three years with annual Sharpe ratio 0.48 versus 0.37 (0.48 versus 0.39).
- Findings are robust to using two-year ranking and holding intervals. Annual Sharpe ratios for Winner, Median and Loser funds for this scenario are 0.32, 0.44 and 0.46, respectively.
- Findings are robust to pre-screening for top long-term performers. Picking with perfect foresight (after first excluding the tenth of funds with the highest fees) the quarter of funds with the highest annualized returns over the entire sample period generates Sharpe ratio 0.61. Selecting past three-year Winners and Losers from these known top performers as specified above generates annual Sharpe ratios 0.43 and 0.68, respectively.
- Findings are consistent with belief that fund performance derives from persistent investment strategies subject to mean reversion, not fund manager skill.
- Based on past research, investors should instead focus on the following indicators of future fund outperformance:
- Theoretical soundness of fund manager’s strategy.
- Presence of performance-linked compensation/bonuses for the fund manager.
- High level of fund manager ownership.
- Fund board of directors ownership.
- High Active Share.
- No affiliation with an investment bank.
- Outsourced shareholder services.
- Presence of a short-term redemption fee.
- Ph.D.s in key portfolio management roles.
- Strong positive firm culture.
In summary, evidence indicates that fund performance is mean reverting at horizons of two or three years, so buying recently outperforming funds and selling recently underperforming funds is exactly wrong.
Cautions regarding findings include:
- The test methodology involving large portfolios of funds with monthly rebalancing is impractical.
- Some of the recommended fund selection factors are vague. Some others are difficult to estimate.
See also “Mutual Fund Hot Hand Performance” and “Mutual Fund Hot Hand Performance Robustness Test”, both based on a one-year fund ranking interval.
