Rob Arnott, Amie Ko, and Lillian Wu went back to the 1980s to grab the title of their recent article: “Where’s the Beef?” That’s fitting, in that they address an age-old problem that the investment industry never seems to solve. It was even around when Clara Peller starred in those Wendy’s advertisements and the saying became a cultural phenomenon.
Where’s the alpha?
As noted in the abstract, Peller’s famous question sets up an analogy:
Many investors in today’s so-called smart beta strategies may well be asking a similar question, “Where’s the alpha?”
But let’s not restrict its application; the challenge goes far beyond smart beta to quantitative strategies in general, and to traditional investing as well:
The problem is data mining and performance chasing, the nemeses of all investors. Yes, academics, “quants,” and investment professionals are all subject to those same temptations, very nearly to the same extent as retail investors.
That is something that the individual investors who hire professionals don’t realize — and professionals don’t admit. While organizations have the data to document the chasing, few try to analyze it.
The problem of noise
Performance records are noisy — and that noise lasts longer than the three-to-five year evaluation periods of managers and strategies that are the industry standard. Here are the authors on what that means for quantitative strategies:
Selection bias guarantees that random positive noise will be overwhelmingly the norm when performance is the basis for selection. As a result, among many factors tested, some will appear to be statistically significant by random luck. Moreover, the contrast between pre- and post-publication outcomes is stark, with the excess return following publication falling far short of in-sample published results.
To consider the issue more broadly, here is an altered version of the last two sentences from that excerpt, recast in a way to characterize the major hurdle in manager selection:
Among many asset managers evaluated, some will appear to be better than others by random luck (perhaps even in a way that appears to be statistically significant). Moreover, the contrast between before- and after-selection outcomes is often stark, with the excess return following selection falling far short of the previous results.
Since very few countercyclical choices are made by allocators, because they primarily look for (historically) top-quartile managers, their most common error is extrapolating that performance and expecting it to continue. Perhaps there is a way to gauge the risk inherent in doing so.
Revaluation alpha
The authors’ core point is that investors are fooled by what they call “revaluation alpha,” that part of relative performance that stems from changes in the relative valuation of whatever (security, manager, strategy, asset class) is being analyzed. While they address this phenomenon in regards to factor portfolios, it applies more generally. If performance is accompanied by a change in valuation, which it often is, “then lofty past performance may presage future underperformance.”
In contrast, the substantive part of outperformance is referred to as “structural alpha” — that which is not due to changes in valuation — and charts are included in the article that display the two concepts in regard to some of the most widely-used quantitative factors. This fits with a previous article by Arnott and others, in which they
urged academia to demand that journal articles on new factors examine revaluation alpha so that academics are not feted for finding a “new factor” that merely “worked” by becoming more expensive.
A similar goal ought to guide everyone responsible for manager and strategy selection. Granted, the level of difficulty varies considerably depending on the particular situation. Assessing where mean relative valuations ought to be is an appropriate topic of debate, but assessing the direction of the changes in them is the first step. If the components of a strategy have benefited in a significant way from revaluation, then caution is warranted unless a manager actively changes exposures in response to such moves.
Performance evaluation practices ought to highlight revaluation effects but they rarely do. Thus, one of the few edges available to allocators requires both analytical work (gauging revaluations, which can be difficult) and behavioral commitment (being willing to avoid taking those risks at times, which can also be difficult).
Other considerations
The article also addresses implementation shortfall, “the difference between a paper portfolio’s performance and the realized performance of a live portfolio,” which is a significant issue with quantitative strategies. “Portfolio concentration, universe coverage, turnover, and capacity” are often ignored by academic studies of potential approaches — and are downplayed by managers that bring them to market.
Another topic covered is the “expectations gap,” in this instance related to pension plans (one manifestation of a more universal problem). The authors’ stark conclusion is that “the likelihood a pension fund can deliver 7% or more in the coming decade is under 1%.”
As important as those ideas are, if you are looking through a due diligence magnifying glass, revaluation is the concept that needs to be brought into view.
Published: October 17, 2022
Subscribe
To comment, please send an email to editor@investmentecosystem.com. Comments are for the editor and are not viewable by readers.