Radar Perene / Archive / science

Efficient market hypothesis: the theory and its critics

◦ Index methodology v2.2 (working papers with DOI). See the methodology.

Science

Two economists are walking down the sidewalk when one of them spots a twenty-dollar bill on the ground. He reaches for it; the other stops him: "Don't waste your time — if it were a real bill, someone would have picked it up already." The anecdote has circulated for decades in the discipline's corridors because it captures, with the cruelty typical of good jokes, both the strength and the weak spot of the most influential idea in academic finance.

The efficient market hypothesis states that a market's prices reflect, at every moment, the available information — and that, therefore, no consistent gain above the market is to be expected from using that same information. What is known would already be built into the price; only genuine news moves it, and news, by definition, cannot be anticipated.

Formalized by Eugene Fama in a 1970 article that organized two decades of evidence, the hypothesis is less a thesis about the wisdom of markets and more a thesis about the speed of competition: where many informed people fight over every cent, twenty-dollar bills do not stay on the ground for long.

What the hypothesis says — in three strengths

The 1970 article split the claim into three versions, each more demanding than the last. In the weak form, prices already incorporate all the information contained in past prices themselves — which, if true, empties chart reading of content and connects the hypothesis to the random walk seen in the previous step of this trail. In the semi-strong form, they incorporate all public information — statements, news, indicators; analyzing what everyone can read would yield no advantage. In the strong form, they incorporate even private information — a version Fama himself treated as a limiting ruler, useful for measuring how far reality falls short of it.

The distinction matters because the three versions have had different empirical fates. The weak form accumulated considerable support; the semi-strong generated a vast and quarrelsome literature; the strong form serves, above all, to remind us that privileged information exists — and it is precisely because it moves prices that exploiting it is a crime in most jurisdictions.

What the critics answer

The most elegant attack came from inside the logic itself. In 1980, Sanford Grossman and Joseph Stiglitz pointed out a paradox: if prices perfectly reflected all information, nobody would have any incentive to spend resources gathering information — and, with nobody gathering it, prices would have nothing to reflect. Perfect efficiency would destroy itself. The modern reading of the hypothesis lives with that amendment: prices would be efficient enough that the remaining edge barely pays the cost of chasing it.

The second attack came from the data. In 1981, Robert Shiller showed that stock prices swung far more than later revisions of dividends could justify — excess volatility for a series that merely processed facts. Behavioral economics added to that bill an inventory of biases documented in flesh-and-blood people. And the empirical body of finance itself recorded regularities — the anomalies, the factors — that resisted too long to be dismissed out of hand, although part of them has the habit of shrinking once published.

The 2013 Nobel recorded the stalemate with fine institutional irony: the prize was shared between Fama and Shiller — the author of the hypothesis and its most persistent critic, laureled on the same day by the same committee. Few disciplines admit so frankly that their central question remains open.

The coherence a publication owes the theory

This debate is not decorative for anyone publishing market readings — it defines what is honest to promise. If informational advantage is rare, costly and short-lived, as even the moderate versions of the hypothesis maintain, then selling "signals" to the reading public would require explaining why such a valuable edge would be up for sale by subscription.

This house's position is public and verifiable on every page: Radar Perene sells no signals, publishes no recommendations and announces no forecasts — the archive organizes precedents and describes regimes, and stops there. That editorial choice, prior to any theoretical sympathy, is what the efficient market hypothesis seems to demand of anyone who takes it seriously: between promising the improbable and recording the observable, the house records. The research output that sustains this stance is deposited in an open repository, with public DOIs — six working papers whose question is never "what will go up", and one of which concludes, against its own tested premise, that the edge under examination does not hold.

That is not proof of the hypothesis — a publisher proves no theory. It is the more modest and rarer case: a practice that does not contradict what it claims to believe.

Frequently asked questions

Does the efficient market hypothesis claim nobody can beat the market?

Not exactly. It claims no consistent gain is to be expected from using available information — which admits gains from luck, from bearing more risk and, under the Grossman-Stiglitz amendment, thin margins that pay for the cost of research. What it denies is the cheap, durable shortcut.

Do bubbles and crashes refute the hypothesis?

They are the critics' main material, but the refutation is less simple than it looks: identifying a bubble with confidence before it bursts is precisely what the hypothesis says is hard — and the celebrated episodes were named bubbles, almost all of them, in hindsight. The technical debate remains open; the practical consensus is narrower: overconfidence in either direction ages badly.

Are smaller markets, like Brazil's, less efficient?

The literature documents gradations: thinner liquidity, sparser analyst coverage and higher costs leave more rough edges unsanded. Rough edges, however, are not twenty-dollar bills — exploiting them costs money, and that cost is part of the explanation for why they survive.

If the hypothesis is true, what is market research for?

For the questions that do not depend on predicting: how much risk a position carries, how similar episodes ended, which past regime the present resembles. That is exactly the terrain where a descriptive reading has something to offer — and the only one this trail occupies.

If efficiency leaves few crumbs, what to make of the "premia" the literature documented — value, momentum — and the industry turned into product? The next step separates the empirical fact from the sales material: Risk premium and factors: empirical fact, not a buy tip →

House readings: today's note, in the Daily · the precedents, in the Atlas.

Discussing how closely a specific market approaches efficiency is workbench conversation — the kind the house keeps on request.