How much difference does it make to calculate mutual fund alphas with exchange-traded funds (ETF) rather than ideal (frictionless) indexes/factors? In their November 2012 paper entitled “Mutual Fund’s Net Economic Alpha: Definition and Evidence” Sharon Garyn-Tal and Beni Lauterbach investigate how benchmarking mutual funds with ETFs differs from traditional benchmarking with ideal performance models based on one to five factors (market, fund style, size, book-to-market ratio and momentum). They calculate traditional alphas via regressions against a specified number of factors. They calculate net economic alphas by adding to the traditional alphas costs of implementing associated factors with one or several actual ETFs. Net economic alpha therefore represents the actual value of a fund to investors relative to mimicking ETF alternatives. While accounting for ETF expense ratios, they ignore trading frictions associated with periodic (monthly) rebalancing sets of ETFs to maintain alignment with multi-factor models. They also ignore mutual fund redemption fees and loads, hoping that ETF and mutual fund cost omissions cancel. They focus on post-2000 data because factor-implementing ETFs are not available earlier. Using returns and style designations for over 1,000 open-end, non-specialized U.S. equity funds and values for traditional performance model factors during 2001 through 2009 (segmented into three equal subperiods), they find that:
- Average traditional alphas (based on ideal benchmarks) for the sample of mutual funds are mostly negative, with the average fund annual expense ratio of about 1.3% largely explaining fund underperformance.
- Average net economic alphas (based on ETF benchmarks) are between 0% and -1% per year, typically less than 0.8% per year higher than corresponding traditional alphas.
- Using net economic rather than traditional alphas makes little difference in ranking mutual funds. For example, the separate use of net economic and traditional alphas to select the top 20 mutual funds results in overlaps of 19-20 funds across factor models.
- Average net economic alphas increase across the three subperiods, becoming positive during 2007-2009, suggesting that mutual funds are becoming more competitive with ETFs (or that they manage risk more effectively than ETFs during crises).
In summary, evidence indicates that U.S. mutual funds on average just slightly underperform risk-matched sets of ETFs and that this underperformance may have disappeared.
Cautions regarding findings include:
- As noted, the study ignores: (1) the costs of periodically rebalancing the ETFs used to implement multi-factor models; and, (2) mutual fund redemption fees and loads. It is not obvious that these omissions offset.
- Some results in the paper reflect a mutual fund survivorship bias estimated at less than 0.5% per year, and some do not.
- It would be interesting to know whether results differ for no-load versus load mutual funds.
