Different Types of Skill
I recently came across an interesting paper, published in March 2018 on SSRN, titled “How Skilled Are Security Analysts?”, that found a large majority of the nearly 5,000 equity analysts it examined between 1993 and 2015 appeared skilled. Consistent with prior literature, the authors, Alan Crane and Kevin Crotty, of Rice University, also found:
- Analyst revisions are more informative than recommendations.
- Analysts issuing higher numbers of recommendations per year are negatively associated with the conditional probability of being a skilled analyst.
- Analysts with a higher fraction of buy recommendations also are less likely to be skilled.
- Less favorable analysts (those with fewer buy recommendations as a fraction of their total recommendations) are more accurate in terms of recommendations and revisions.
- Analysts with consistently good picks are less likely to win The Wall Street Journal’s “Best on the Street” designation, while analysts with more volatility in the success of their recommendations are more likely to win.
- Tenured analysts are more skilled.
The apparent prevalence of analyst skill stands in stark contrast to research demonstrating that only a small fraction of actively managed mutual funds outperform on a risk-adjusted basis.
For example, in their study, “Luck versus Skill in the Cross-Section of Mutual Fund Returns,” published in the October 2010 issue of The Journal of Finance, Eugene Fama and Kenneth French found that fewer active equity managers (about 2%) than would be expected by chance were able to outperform their three-factor (market beta, size and value) model benchmark.
Stated differently, the very-best-performing traditional active managers have delivered returns in excess of the Fama-French three-factor model. However, their returns have not been high enough to be confident in concluding that such managers have enough skill to cover their costs, or that their past performance will persist.
Fama and French concluded: “For (active) fund investors, the simulation results are disheartening.” They did concede that active managers’ results appear better when looking at gross returns (returns without the expense ratio). But gross returns are irrelevant to investors unless they can find an active manager willing to work for free.
A more recent study, an August 2016 research paper by Philipp Meyer-Brauns of Dimensional Fund Advisors titled, “Mutual Fund Performance through a Five-Factor Lens,” reached similar findings.
His data sample included 3,870 active funds over the 32-year period 1984 to 2015. Benchmarking active fund returns against the newer Fama-French five-factor model, which adds profitability (RMW, robust minus weak) and investment (CMA, conservative minus aggressive) to the three-factor model, he found an average negative monthly alpha of 0.06% (with a t-stat of 2.3). Moreover, he found that about 2.4% of the funds had alpha t-stats of 2 or greater, which is slightly fewer than what we would expect by chance (2.9%).
Meyer-Brauns also found that the distribution of actual alpha t-stats had shifted to the left of what would be expected from chance if all managers were able to produce excess returns over the five-factor model sufficient to cover their costs.
He concluded: “There is strong evidence that the vast majority of active managers are unable to produce excess returns that cover their costs.” He added, “funds do about as well as would be expected from extremely lucky funds in a zero-alpha world. This means that ex-ante, investors could not have expected any outperformance from these top performers.”
Meyer-Brauns extended his work in March 2017 in his paper, “Luck vs. Skill Across Different Fund Categories.” He examined four separate categories of U.S. equity mutual funds (large-cap value, large-cap growth, small-cap value and small-cap growth) over the period from January 2000 through June 2016. His goal was to determine whether active managers’ ability to outperform the Fama-French five-factor model varied across fund categories. Following is a summary of his findings:
- The best-performing funds perform no better than would be expected by chance alone in a zero-alpha world. For example, the by-chance distributions indicate that if all funds could cover their costs, slightly more than 2% should be expected to have alpha t-stats larger than 2. Looking at the actual distributions across fund categories, he found that in two of the four categories, large-cap value and large-cap growth, not a single fund had an alpha t-stat above 2. For the two other categories, small-cap value (1.8%) and small-cap growth (1.1%), the percentages were lower than would be expected by chance.
- The reverse is true when looking at the number of funds with reliably negative five-factor intercept t-stats. Substantially more than 2% of funds reliably underperformed the five-factor benchmark: 18.8% of large-cap value funds, 8.2% of large-cap growth funds, 10.3% of small-cap value funds and 11.4% of small-cap growth funds all had alpha t-stats below -2.
Meyer-Brauns concluded: “Taken together, this evidence across different fund categories suggests that the vast majority of active managers have been unable to produce excess returns, with respect to the Fama/French five-factor model, that cover their costs.”
How does one explain the seeming contradiction that a large majority of security analysts have skill, but fund managers don’t outperform?
Explaining The Contradiction
It is important to understand that Crane and Crotty were not asking if security analysts added value. Instead, they sought to determine whether they have skill. Those are two very different questions.
As Crane and Crotty note, there is a distinction between analysts and fund managers. Analysts do not have to deploy capital following their stock picks. Fund managers do not have this ﬂexibility. Analysts don’t have expense ratios to overcome. And they do not incur trading costs (bid/offer spreads and market impact).
Further, they don’t face yet another problem that successful active managers do—cash flows based on successful past performance sow the seeds of future failure due to well-documented diseconomies of scale. These hurdles dramatically erode the proﬁtability of analyst recommendations.
The bottom line is that the evidence demonstrates the market is sufficiently efficient: While prices are not “perfect,” they are correct to the extent that once you consider the costs of implementation, the odds of successfully generating alpha (risk-adjusted outperformance) are extremely low. That is why active management is a loser’s game—the costs of exploiting mispricings are greater than the pricing errors.
This commentary originally appeared May 7 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 the BAM ALLIANCE. This article is for general information only and is not intended to serve as specific financial, accounting or tax advice.
© 2018, The BAM ALLIANCE