The efficient market hypothesis, AI, and good-bet searches for alpha
Markus Mäkelä
“Fire your investment adviser.” – Burton G. Malkiel¹
“I will spend [my Nobel Prize money] as irrationally as possible.” – Richard H. Thaler²
The efficient market hypothesis says that prices fully reflect all available information and that as such, no one can create abnormal returns – bar by fleeting luck. Finance theorists and economists tend to apply the “EMH” to security markets, where many factors are relatively standard and measurable, and transaction volumes high, so that empirical research is feasible.
Markets are all around, however, and not just where cash, securities, and currencies are being traded. We all routinely buy products and services from a market; markets for goods and services extend from executive skill to raw materials and from ideas even to individual attention.3
All exhibit price-like mechanisms for discovering and allocating value with varying degrees of efficiency. Per Eugene Fama’s canonical review, we talk of three of these degrees: weak (historical prices alone comprise the information that determines price), semi-strong (all public information does), and strong (all public information and private, such as acquired via painstaking proprietary analysis).⁴
How efficient are markets, then, in capital-markets investing? Quite.5
In closer detail, one can consider them semi-strongly efficient in many ways and even bordering on strongly efficient in some, with a generally low percentage of market-beating analysts and a rapid diffusion of at-first proprietary insights. (But a caveat is in order. Many opinions and viewpoints offer a richness of complementary or contrary arguments. For example, humans are not fully rational and anomalies persist, as has been shown by Robert Shiller6 and further top business scholars, including our school's Matti Keloharju. Nature did not endow us humans with all the information processing capacity that rationality today demands, as per the fifth economics Nobel laureate cited here, Herbert Simon.7 Nor did nature shape our emotions to steer us measuredly and stably in all we do, like in stock picking and trading.8)
But by and large, as is widely known, it is difficult for a capital-markets investor to consistently achieve meaningful net-of-costs alpha, abnormal return. If newest information technology like AI will have a meaningful impact on this, one would surmise that it will just soon lead to yet-strengthened market efficiency.9 Public capital markets are indeed quite efficient.
Most of markets are far from efficient nonetheless – case buyouts
Elsewhere than in the thick and liquid capital markets of the Western stock exchanges, things can be very different. Many markets are notably inefficient, with some notoriously so. Inefficient markets include many in business life, and offer a commensurately large opportunity.
Abnormal returns can, in fact, be made over a term of many years, sometimes more than a decade.
Witness buyout investing, for example. It has retained its edge over public equity, despite a trend toward market efficiency.10 Investing with established, socially "certified" firms can certainly yield abnormal risk-adjusted returns. Buyouts can deliver excess returns in part because the market for selecting private investments is not efficient. Skill levels of target pickers show a trend where they would even out only along a very long period which we still are in the midst of.
However, there is much more to this, and probably much more importantly: the skill levels of private equity firms in selecting, supporting, and supervising operative management and approving business strategy – non-finance skills whose development PE firms for long de-emphasized – will display a notably larger variation still. The opportunity is large and this particular market is slow at canceling that opportunity by luring operative talent in into the PE industry and its target companies.
Have you noticed how Warren Buffett has for decades emphasized buyouts more in more, at the cost of public equity, until he ran into difficulty of simply finding suitable buyouts with defensible prices?11 The numerous buyouts – subsidiaries – in Berkshire Hathaway's portfolio explain a large deal (not just a pun) of the market-beating success that Buffett and Charlie Munger have had during this century.
Buffett has long faced, as have others, an increasingly efficient public market.
But he found a solution. And that is no longer to be the public-equity investment advisor of others, as he essentially was as Buffett Partnership’s fund manager in the mid-20th century.
PE firms' quest at finding skilled operating partners is actually more of a difficult gauntlet for the average firm. A great efficiency deficit remains in the market for management – and will remain so, lest AI change it.12 Management – leading and administering business processes, their systematic development and digital transformation, organization designs, and a corps of other leaders in a global organization – is a thoroughly challenging and multifaceted skill domain and profession.
Beyond industries, can functional capabilities “dig you a moat” and generate alpha?
One of the most difficult aspects of management is strategy, which by itself is complex and infused into a veritable plethora of processes – some of which even span company boundaries in larger value systems, such as modern platform ecosystems. Strategy is possibly also the aspect of management that is most conducive to generating abnormal returns in skilled and experienced hands.
To elaborate, let us consider the following. Progressive and effective skills in the various subdomains of management provide companies with operational capabilities that are usually a basic necessity for any abnormal returns – a basic hurdle, the clearing of which is the first step toward competitiveness.
Effective management capabilities will not alone distinguish a business strategically, that is.13 Neither can they distinguish its financial longer-term returns, therefore.
In precise terms, the return on invested capital of effectively-managed companies may rise to match their cost of capital.14
That is, such fundamental management effectiveness is simply table stakes from strategy’s viewpoint; it can accelerate the pace on your “treadmill of business,” to use an old metaphor, to match your competitors’ already-transpiring and similar basic efforts to perform.
Fundamental management effectiveness cannot render a moat, to use some favorite language of Buffett and others like Morningstar. (A “moat,” as they use the term, and various others terms, are synonyms to each other. These include “results of a successful strategy,” “what above-par competitiveness leads to,” “dampened market efficiency,” and sustainable competitive advantage.) A moat is the essential outcome of a successful strategy and is not on the level of fundamental management.
Strategy is how the field of business administration describes how to business-wise achieve and sustain a competitive advantage. It is qualitatively different from management overall.
What does empirical research have to say about strategy's role and effectiveness? Studies have observed and in rather crisp terms communicated how company-specific characteristics – and they much more than any industry characteristics – explain a large swath of profitability differences between comparable companies within any sector.15 Durable differences are essentially the structural effects of competitive strategies.
May AI change some of that, by the way?
AI is an up-and-coming technology for strongly supporting companies’ strategy processes, although its impact is not yet properly felt in the substantive contents of strategy, as managers and boards do not yet have an understanding of how AI will shape which strategic choices are a possible and relevant set of complementary courses of action for them.
For both reasons, AI will certainly influence the chances for alpha from a well-formulated and -executed business strategy. Whether it will do so in ways that will increase variation in profits among comparable companies or decrease it is something that remains to be seen.
One is tempted to put forth that at first, the former may happen, but that in the end, a type of base effect will lead to the latter outcome.
But only time will tell.
Your investment advisor could only guess, too. This specific market is quite public.
Markus Mäkelä is AFA’s past president and an expert in strategy and AI.
Notes
Burton G. Malkiel, The Random Walk Guide to Investing (New York: W. W. Norton, 2005), 5.
See, e.g., Lee Borton, Irrationality explains much about economies, Wall Street Journal, Opinion (Oct 16, 2017, online article, wsj.com).
See Gary S. Becker, The Economic Approach to Human Behavior (Chicago: University of Chicago Press, 1976). For example, academic research or software product development can be examples of markets for ideas.
Eugene F. Fama, Efficient capital markets: A review of theory and empirical work, Journal of Finance 25(2) (1970): 383-417. As Fama recites, it was Harry Roberts who first suggested two of the forms. The specific degree of capital-market efficiency and drivers behind it have always been controversial.
See Eugene F. Fama, Two pillars of asset pricing, American Economic Review 104(6) (2014): 1467-1485; Eugene F. Fama and Kenneth R. French, A five-factor asset pricing model, Journal of Financial Economics 116(1) (2015): 1-22.
Robert J. Shiller, Speculative asset prices, American Economic Review 104 (6) (2014): 1486-1517. This and Fama’s above-cited 2014 article embody the 2013 Nobel Prize lectures of the respective gentlemen.
Herbert A. Simon, A behavioral model of rational choice, Quarterly Journal of Economics (1955): 99-118. Here, the polymath explains his concept of bounded rationality, though he appears to have only later crystallized this label itself.
Nature molded and shaped these human factors for the evolutionary environment that hunter-gatherer humans faced tens of thousands of years ago. For example, our emotions are mostly what evolution “intended,” only that not for our era.
Before that, however, it is possible that a few large investors can successfully develop proprietary asset-picking AI tools. They would gain informational advantages that I believe would prove relatively short-lived.
Robert S. Harris, Tim Jenkinson, Steven N. Kaplan, and Rüdiger Stucke, Has persistence persisted in private equity? Evidence from buyout and venture capital funds, Journal of Corporate Finance 81 (2023), 1-16.
Nicole Friedman, Buffett’s hunt for big purchases fails, Wall Street Journal (Feb 22, 2019).
In the long scheme of things, it would be very surprising that an information technology would change this. However, here AI can be so powerful and exceptional that it actually stands a very good fighting change of doing just that.
Michael E. Porter, What is strategy? Harvard Business Review (Nov-Dec 1996), 61-78.
As I often note, finance theory is what lays out the gold-standard yardsticks for measuring business results – of course. But when something returns abnormally many “yards” recurringly, strategy is what gives the business explanation for the why of that.
Jaime A. Roquebert, Robert L. Phillips, and Peter A. Westfall, Markets vs. management: What “drives” profitability? Strategic Management Journal 17(8) (1996): 653-664; Richard P. Rumelt, How much does industry matter? Strategic Management Journal 12(3) (1991): 167-185.