Introduction: The Most Useful Wrong Idea in Finance
In 2013, the Nobel Prize in Economic Sciences was shared by Eugene Fama and Robert Shiller. Fama had spent his career arguing that asset prices efficiently incorporate all available information. Shiller had spent his career demonstrating that they do not. Both were right, and the committee knew it.
The Efficient Market Hypothesis (EMH) is the central organising idea of financial economics. It is also, in its strong forms, empirically false. The interesting question is not whether markets are efficient — they are not perfectly so — but how nearly efficient they are, and what the deviations tell us about how prices are actually formed.
This guide covers the theory, the three forms, the anomalies that broke it, the behavioural response, and the deep problem — the joint hypothesis problem — that makes the EMH unfalsifiable in principle.
1. What the Hypothesis Actually Claims
Eugene Fama’s 1970 survey in the Journal of Finance, “Efficient Capital Markets: A Review of Theory and Empirical Work,” gave the hypothesis its canonical statement: a market is efficient if prices fully reflect all available information.
The mechanism is competition among informed traders. If a stock is underpriced relative to what public information implies, traders buy it. Their buying pushes the price up. The mispricing disappears. The speed of this correction is the speed of the market’s information processing, and in liquid markets it is measured in milliseconds.
The critical implication is about predictability. If prices already reflect all information, then price changes must be driven by new information — and news, by definition, is unpredictable. Prices therefore follow something close to a random walk:
Pt+1 = Pt + εt+1
where ε is unforecastable. Samuelson (1965) proved this formally: properly anticipated prices fluctuate randomly. The random walk is not an assumption of the EMH; it is a consequence of it.
Two common misreadings, both of which cost exam marks:
- The EMH does not claim that prices are always correct. It claims that deviations from correct are unpredictable.
- The EMH does not claim that no one beats the market. In a population of thousands of fund managers, some will outperform by chance. It claims you cannot identify them in advance.
2. The Three Forms
Weak form
Prices reflect all information contained in past prices and trading volumes.
Implication: technical analysis — chart patterns, momentum indicators, head-and-shoulders formations — cannot generate risk-adjusted excess returns.
Status: broadly supported, with one enormous and embarrassing exception, discussed below.
Semi-strong form
Prices reflect all publicly available information: past prices, earnings announcements, macroeconomic data, analyst reports, filings.
Implication: fundamental analysis of public data cannot generate excess returns. Prices should adjust to news instantaneously and completely, with no drift afterward.
Status: the central battleground.
Strong form
Prices reflect all information, public and private, including insider knowledge.
Implication: even corporate insiders cannot profit from their private information.
Status: false. Insiders demonstrably earn abnormal returns on their own trades, which is precisely why insider trading is illegal. The existence of securities enforcement is an admission that strong-form efficiency does not hold.
3. The Evidence For: Why the Hypothesis Survived
Event studies
Fama, Fisher, Jensen and Roll (1969) pioneered the event study methodology, examining stock splits. They found that abnormal returns accumulated before the split announcement — as the market anticipated the associated dividend increase — and ceased immediately upon announcement. The market had priced the news before it was news.
Thousands of event studies since have found the same pattern for earnings surprises, merger announcements and regulatory decisions: rapid, near-complete price adjustment within minutes.
The performance of active management
The most powerful evidence for practical efficiency comes not from theory but from the fund industry’s own record.
Fama and French (2010), in the Journal of Finance, examined the cross-section of US mutual fund returns. Using a bootstrap simulation, they compared the actual distribution of fund performance against the distribution that would arise if no manager had any skill and all variation were luck. The actual distribution was close to the luck distribution — and after fees, the aggregate portfolio of active funds underperformed passive benchmarks by roughly the amount of the fees charged.
Sharpe (1991) made the point without any data at all, in an argument known as the arithmetic of active management. Before costs, the return on the average actively managed dollar equals the return on the average passively managed dollar, because together they constitute the whole market. After costs, and since active management costs more, the average actively managed dollar must underperform. This is not a hypothesis. It is arithmetic, and it holds regardless of whether markets are efficient.
4. The Evidence Against: The Anomalies
Excess volatility
Robert Shiller’s 1981 paper in the American Economic Review, “Do Stock Prices Move Too Much to be Justified by Subsequent Changes in Dividends?”, is the most damaging single piece of evidence against the EMH.
Under the present-value model, a stock’s price is the rational expectation of the discounted stream of future dividends. Shiller constructed the ex post rational price — the price that would have prevailed had investors known the actual future dividends — using realised dividend data.
Because a price based on an expectation should be less volatile than the realised variable it forecasts (the variance of a conditional expectation cannot exceed the variance of the underlying variable), actual prices should be smoother than the ex post rational price series.
Shiller found the opposite. Actual stock prices gyrated wildly around an ex post rational series that was remarkably smooth. Prices moved far more than any plausible news about future dividends could justify.
The rebuttal, offered by Fama and others, is that Shiller assumed a constant discount rate. If required returns vary over time — rising in recessions when investors are risk-averse, falling in booms — then price volatility can be rational. This defence is coherent. It is also the beginning of the joint hypothesis problem.
Momentum: the anomaly that will not die
Jegadeesh and Titman (1993), in the Journal of Finance, documented that stocks which performed well over the past three to twelve months continue to outperform over the subsequent three to twelve months. Buying past winners and shorting past losers earned substantial abnormal returns.
This is a direct violation of weak-form efficiency — the weakest and supposedly best-supported version. It uses nothing but past prices.
Momentum has survived out-of-sample testing across international equity markets, across asset classes including bonds, currencies and commodities, and across nearly a century of data. Asness, Moskowitz and Pedersen (2013) documented it as a pervasive cross-asset phenomenon.
Fama and French, in constructing their five-factor model, conspicuously omitted momentum — Fama has described it as the anomaly that most troubles him. No consensus risk-based explanation exists.
Post-earnings-announcement drift
Firms reporting positive earnings surprises continue to earn abnormal positive returns for weeks afterward. The market under-reacts. This violates semi-strong efficiency using entirely public information, and has been documented continuously since Ball and Brown (1968).
Value and size
Fama and French (1992, 1993) themselves documented that small-capitalisation stocks and high book-to-market (value) stocks earned higher average returns than the CAPM predicted. Their resolution was to reinterpret these not as anomalies but as compensation for additional risk factors. This move is elegant, and it is also unfalsifiable in a specific way we now examine.
5. The Joint Hypothesis Problem
This is the deepest idea in the field, and the one that most students never encounter.
To test whether a return is “abnormal,” you must know what a normal return is. That requires an asset pricing model — CAPM, Fama-French three-factor, five-factor, or another.
Therefore any test of market efficiency is simultaneously a test of two propositions:
- Markets are efficient.
- The asset pricing model used to define normal returns is correct.
If you find abnormal returns, you cannot tell which proposition failed. Fama stated this explicitly in 1970 and has repeated it ever since. Market efficiency is not independently testable.
The consequence is that every anomaly admits two interpretations. Value stocks outperform because the market irrationally underprices boring companies (Lakonishok, Shleifer and Vishny, 1994) — or because value stocks carry a risk the CAPM does not capture, and investors are compensated for bearing it (Fama and French, 1993).
There is no experiment that distinguishes these. A sufficiently creative theorist can always construct a risk factor to rationalise any observed return pattern, and a sufficiently determined behavioural economist can always construct a bias to explain it. The debate is, in a strict Popperian sense, not resolvable by data alone.
6. The Behavioural Response and the Limits of Arbitrage
The standard defence of efficiency is that irrational traders are exploited by rational arbitrageurs and driven from the market. Friedman (1953) made this argument forcefully.
Shleifer and Vishny (1997), in “The Limits of Arbitrage” (Journal of Finance), demolished it.
Real arbitrage is conducted by professional managers using other people’s money. Suppose a manager correctly identifies an overpriced asset and shorts it. If the mispricing worsens before it corrects, the manager posts losses. Investors, unable to distinguish a wrong manager from an early one, withdraw capital — forcing the manager to close the position at exactly the worst moment.
The implication is severe: arbitrage is riskiest precisely when mispricing is largest. The mechanism that is supposed to enforce efficiency is weakest exactly when it is most needed.
De Long, Shleifer, Summers and Waldmann (1990) formalised a related result: noise trader risk. Because rational arbitrageurs face the risk that irrational sentiment moves further against them, they limit their positions. Noise traders can therefore survive indefinitely, and may earn higher expected returns than rational traders by bearing the risk they themselves create.
The empirical exhibits are hard to argue with. Closed-end fund discounts, where a fund trades persistently below the market value of the assets it holds. The 3Com/Palm carve-out of 2000, in which the market valued 3Com’s stake in Palm at more than the entire market capitalisation of 3Com — implying a negative value for the rest of the company. These are not subtle statistical anomalies requiring a risk model to detect. They are arithmetic impossibilities that persisted for months.
7. The Grossman-Stiglitz Paradox
Grossman and Stiglitz (1980), in the American Economic Review, identified a logical impossibility at the heart of the EMH.
Information is costly to acquire. If prices fully reflected all information, no one would have any incentive to pay for information, since they could learn everything for free by observing the price. But if no one acquires information, prices cannot reflect it.
Therefore perfectly efficient markets are impossible. Markets must be inefficient enough that gathering information earns a return sufficient to cover its cost. The equilibrium degree of inefficiency is exactly the amount that compensates informed traders.
This resolves the apparent paradox that active management persists despite underperforming on average. Active managers are the mechanism by which prices become efficient. They must, collectively, be paid for this. That payment is the market’s inefficiency.
8. What Should a Student Actually Conclude?
A defensible synthesis:
- Markets are highly efficient with respect to easily accessible public information. Trading on last quarter’s earnings, or a chart pattern, will not make you rich.
- Strong-form efficiency is false. Private information has value.
- Persistent, documented anomalies exist — momentum most stubbornly — and no consensus explains them.
- Deviations from efficiency are limited by arbitrage, but arbitrage itself is limited by capital constraints and horizon risk.
- Perfect efficiency is logically impossible (Grossman-Stiglitz), so the question is one of degree.
- The practical investment implication is nonetheless the one the EMH predicts: because of Sharpe’s arithmetic, and because identifying skill ex ante is nearly impossible, low-cost index funds outperform the average investor. This conclusion survives even if the EMH is false.
Point 6 is worth emphasising, because it is counterintuitive. You do not need markets to be efficient in order for indexing to be the right strategy. You need only costs to be positive and skill to be hard to identify in advance. Both are true.
Summary
The Efficient Market Hypothesis is best understood not as a description of reality but as a null hypothesis — a disciplined default that any claim of predictable profit must overturn. Most such claims fail. A few survive, and those few have generated an entire industry of factor investing whose theoretical status remains contested.
Fama and Shiller shared a prize because financial economics needed both: the discipline to reject easy stories about mispricing, and the honesty to acknowledge that prices sometimes do things no rational model can explain.
Exercises for Further Thought
1. The joint hypothesis problem means that any evidence of abnormal returns can be reinterpreted as evidence that the asset pricing model was wrong. Fama and French responded to the value anomaly by adding a value factor to their model — at which point value stocks no longer earned abnormal returns by construction. Is this a scientific advance, or is it an unfalsifiable manoeuvre? Specify what evidence, if any, could in principle refute the efficient market hypothesis, and if you conclude that none could, explain what intellectual role the hypothesis then serves.
Suggested reading: Fama, E. F. (1991). “Efficient Capital Markets: II.” Journal of Finance, 46(5), 1575–1617. Note in particular Fama’s own discussion of why the joint hypothesis problem is not, in his view, fatal.
2. Grossman and Stiglitz proved that if information is costly, markets cannot be perfectly efficient — there must be enough mispricing to pay informed traders for their research. Now consider the rise of passive index investing, which by 2020 accounted for a majority of US equity fund assets. If passive investors free-ride on the price discovery conducted by active investors, and passive share continues to grow, what happens to the equilibrium level of market efficiency? Is there a share of passive investment beyond which price discovery breaks down, and how would we know if we were approaching it?
Suggested reading: Grossman, S. J., & Stiglitz, J. E. (1980). “On the Impossibility of Informationally Efficient Markets.” American Economic Review, 70(3), 393–408.
References
- Asness, C. S., Moskowitz, T. J., & Pedersen, L. H. (2013). Value and Momentum Everywhere. Journal of Finance, 68(3), 929–985.
- Ball, R., & Brown, P. (1968). An Empirical Evaluation of Accounting Income Numbers. Journal of Accounting Research, 6(2), 159–178.
- De Long, J. B., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990). Noise Trader Risk in Financial Markets. Journal of Political Economy, 98(4), 703–738.
- Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. Journal of Finance, 25(2), 383–417.
- Fama, E. F. (1991). Efficient Capital Markets: II. Journal of Finance, 46(5), 1575–1617.
- Fama, E. F., & French, K. R. (1993). Common Risk Factors in the Returns on Stocks and Bonds. Journal of Financial Economics, 33(1), 3–56.
- Fama, E. F., & French, K. R. (2010). Luck versus Skill in the Cross-Section of Mutual Fund Returns. Journal of Finance, 65(5), 1915–1947.
- Grossman, S. J., & Stiglitz, J. E. (1980). On the Impossibility of Informationally Efficient Markets. American Economic Review, 70(3), 393–408.
- Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers. Journal of Finance, 48(1), 65–91.
- Lakonishok, J., Shleifer, A., & Vishny, R. W. (1994). Contrarian Investment, Extrapolation, and Risk. Journal of Finance, 49(5), 1541–1578.
- Samuelson, P. A. (1965). Proof That Properly Anticipated Prices Fluctuate Randomly. Industrial Management Review, 6(2), 41–49.
- Sharpe, W. F. (1991). The Arithmetic of Active Management. Financial Analysts Journal, 47(1), 7–9.
- Shiller, R. J. (1981). Do Stock Prices Move Too Much to be Justified by Subsequent Changes in Dividends? American Economic Review, 71(3), 421–436.
- Shleifer, A., & Vishny, R. W. (1997). The Limits of Arbitrage. Journal of Finance, 52(1), 35–55.