Skill Exists, But It’s Hard To Find
Previously, I reviewed several academic studies that evaluate whether luck or skill is playing a part in mutual fund performance, hoping to glean some insight on evaluating active mutual fund managers. Today we’ll continue our look at the research with a focus on a seventh and eighth study on the topic of luck versus skill.
Skill And The 6-Factor Model
Our seventh study is the April 2016 paper “Skill and Persistence in Mutual Fund Returns: Evidence from a Six-Factor Model.” The authors, Bradford Jordan and Timothy Riley, contribute to the literature by extending the analysis on fund performance to a six-factor model, adding profitability and investment to the Carhart four-factor model (market beta, size, value and momentum). Their data set covers the period 2000 through 2014. Following is a summary of their findings:
- In the best-performing ventile (top 5%) of funds, the average alpha is -0.52% per year (t-stat = -0.35) using the four-factor model, but increases to 3.73% per year (t-stat = 3.44) with the addition of RMW and CMA.
- In the worst-performing ventile of funds, the average alpha decreases from -2.14% per year (t-stat = -1.92) when using the four-factor model to -2.99% per year (t-stat = -3.28) when using the six-factor model.
- The difference in six-factor alpha between the best and worst ventiles is economically large, at 6.73 percentage points per year, and highly statistically significant (t-stat = 4.70).
Based on these findings, Jordan and Riley concluded: “Using the four-factor model, we find very limited evidence of positive alpha after fees beyond that attributable to luck. However, the addition of RMW and CMA to the model leads to a very different conclusion. Using the six-factor model, we show that a significant percentage of mutual fund managers display skill by delivering positive alpha after fees that cannot be attributed to luck. Beginning at about the 85th%ile of fund performance, the t-stats associated with after-fee fund alphas begin to exceed what we would expect from luck alone.”
While this sounds like great news for investors in the top-performing funds, another finding was quite surprising—one with negative consequences not highlighted by the authors. In fact, Jordan and Riley state: “The best performing funds create on average about $3 billion per year in value for investors.”
Unfortunately for investors in these top-performing funds, Jordan and Riley also found that the top performers have large negative exposures to RMW, CMA and HML (high-minus-low, or value), while the worst-performing funds have positive exposures. Negative exposures to factors that have premiums reduce returns earned by investors. The negative loadings resulted in these surprising results:
- An equal-weighted portfolio of all funds produced an average return of about 7% per year.
- The average annual return for the portfolio of the worst performers was only slightly less than average, at 6.8%.
- The average annual return for the portfolio of the best performers was essentially the same, at 6.9%.
Factor Exposures Are Important
A seeming puzzle is how the worst performers, who produced negative six-factor alphas of almost 3% per year, only slightly underperform the best performers, who produced positive six-factor alphas of about 3.7% per year.
The answer is that the while the top performers produced large alphas, their negative loadings on RMW, CMA and HML had large negative impacts on returns. During this period (2000‒2014), the return to RMW was 5.86%, and the average loading of the top performers was -0.43. Thus, the impact on returns was -2.52% (5.86 x -0.43).
The return to CMA was 6.02%, and the average loading of the top performers was -0.35. Thus, the impact on returns was -2.11%.
The return to HML was 4.99%, and the average loading was -0.18. Thus, the impact on returns was -0.90%. The total impact was a drag on returns of -5.53%, more than offsetting the 3.73% alpha.
On the other hand, the worst performers had positive loadings on the RMW (0.05), CMA (0.14) and HML (0.07) factors, which had positive impacts on their returns, partly offsetting the negative alphas. The positive loadings combined to improve returns by 1.48 percentage points.
The difference between the +1.48% return benefit the worst performers received from their positive loadings on RWM, CMA and HML, and the -5.53% cost from the negative loadings on these factors that the top performers experienced is a cumulative 7.01%.
That is slightly more than the 6.7 percentage point difference in their alphas. (Note that there were also slight differences in exposures to the market beta (0.03 percentage points lower for the top performers) and size (0.01 percentage points higher for the top performers), factors that had very minor impacts on the differences in returns; and the momentum loadings were identical).
Unfortunately, because investors cannot spend alpha—only returns—even if you knew ahead of time which funds would produce alpha, you would not have benefited in terms of returns. In fact, your returns would have been slightly worse than the average fund.
Making matters worse is that we know the vast majority of actively managed funds underperform passively managed funds such as index funds. Thus, you certainly would have been better off using passively managed funds such as index funds.
These results provide this important reminder: When choosing mutual funds to implement an asset allocation strategy, you should not only consider a fund’s expense ratio but also how much exposure it has to factors that have persistently generated premiums.
As a good example, over their lifetimes, many of the passively managed funds of Dimensional Fund Advisors have outperformed Vanguard’s index funds in the same asset class despite having higher expense ratios.
The reason is that their fund construction methodology provides deeper exposure to the factors that have delivered premiums. In other words, it’s cost per unit of exposure that matters, not just cost.
The study “An Analysis of the Expense Ratio Pricing of SMB, HML, and UMD Exposure in U.S. Equity Mutual Funds” by my colleagues Sean Grover and Jared Kizer, which appears in the October 2016 issue of The Journal of Portfolio Management, explores this issue.
A Fund-Picking Exercise
Our eighth and final study is the December 2015 paper “Picking Funds with Confidence,” by Niels Gronborg, Asger Lunde, Allan Timmermann and Russ Wermers. Their data set included more than 2,000 U.S. equity mutual funds over the 32-year period June 1980 through December 2012.
They adopted a new approach to select the set of mutual funds with superior performance. The approach undertakes a series of pairwise tests to sequentially eliminate funds with inferior performance.
If at least one fund with significantly inferior performance can be identified, the fund with the “worst” performance is eliminated, and the elimination process is repeated on the reduced set of funds. The procedure continues until no more funds with inferior performance can be identified and eliminated. Following is a summary of their findings:
- Consistent with prior research, the median fund has a negative alpha ranging from -61 bps per year to -74 bps per year across the three different versions of the four-factor (beta, size, value and momentum) model used.
- The bottom 5% of funds ranked by alpha performance have a negative alpha estimate around -35 bps per month, or just under -4% per year—a number that again does not vary much across the three model specifications.
- The top 5% of funds have alpha estimates ranging from 21 bps per month to 25 bps per month, approximately 3% per year—a figure that is again similar across the three models.
Thus, they concluded: “Funds identified as being superior go on to earn substantially higher risk-adjusted returns than top funds identified by conventional ranking methods.” They also found that “superior funds change their industry concentration and shift their risk loadings significantly over time.”
There are some concerns we need to address. Their study used use the model confidence set approach to identify which and how many funds to select each month from a subset of funds that are in the top 5%, 15% or 25% of funds ranked on estimated alpha.
Their results depended on which starting set was used, which calls into question the robustness of the results. Some of the four-factor alphas are reliably positive when funds are selected from the top 15%, but most are not reliable when funds are selected from the top 5% or the top 25%.
In addition, among the paper’s considerations is how the results impact an investor’s portfolio. The paper’s method can lead to selecting funds with concentrated security holdings. For example, the funds identified as superior in 2005 had a 50% sector weight in computers and electronics. Between 2008 and 2011, a single media and telecommunications fund is identified as being superior. Portfolios should be broadly diversified to manage risk and help increase the reliability of outcomes.
Another issue is that, to capture the strategy’s alpha, a high level of fund turnover is required. For example, average monthly fund turnover is between 11% and 19%. That annualizes to 132% and 228%.
Finally, there’s the issue we identified in the Jordan and Riley paper: The paper only reports the alpha for their fund identification strategy. Most investors are concerned with net returns, and high alpha is not the same as high returns.
As we have discussed, some funds will do well and others badly purely based on chance. The empirical challenge is whether the good (and bad) managers can be identified in advance with a degree of confidence that is different from what we would expect by chance.
In 1998, at a time when about 20% of actively managed mutual funds were outperforming their risk-adjusted benchmarks, Charles Ellis called active management a loser’s game. What he meant was that, while it certainly was possible to win the game by selecting funds that would outperform, the odds of doing so were so poor that it wasn’t prudent to try.
Today the combination of academics having converted what once was alpha into beta (a common factor explaining returns)—thus eliminating potential sources of alpha—and increasingly skilled competition have raised the hurdles such that now only about 2% of actively managed mutual funds are generating statistically significant alpha, about what we would randomly expect. And that’s even before the impact of taxes on taxable investors.
All the evidence presented above demonstrates that rather than trying to outguess market prices, a better approach relies on skillful implementation and uses daily information in prices to target factors that have higher expected returns.
The choice is yours. You could try to beat overwhelming odds and attempt to find one of the few active mutual funds that will deliver future alpha while also generating higher returns. Or you could accept market returns by investing passively in the factors to which you desire exposure. The academic research shows that investing in passively managed funds is playing the winner’s game.
Larry Swedroe is the Director of Research for The BAM Alliance.
This commentary originally appeared May 17 on ETF.com
By clicking on any of the links above, you acknowledge that they are solely for your convenience, and do not necessarily imply any affiliations, sponsorships, endorsements or representations whatsoever by us regarding third-party Web sites. We are not responsible for the content, availability or privacy policies of these sites, and shall not be responsible or liable for any information, opinions, advice, products or services available on or through them.
The opinions expressed by featured authors are their own and may not accurately reflect those of Beacon Hill Private Wealth or the BAM ALLIANCE. This article is for general information only and is not intended to serve as specific financial, accounting or tax advice.
© 2017, The BAM ALLIANCE