Efficient Capital Markets: A Review of Theory and Empirical Work
Source: Fama, E. F. (1970) · The Journal of Finance 25(2), 383–417 · DOI: 10.2307/2325486
TL;DR
The defining statement of the efficient market hypothesis (EMH): in an efficient market, prices "fully reflect" available information, so consistently beating the market on a risk-adjusted basis is impossible. Fama organizes efficiency into three nested information sets — weak (past prices/returns), semi-strong (all public information), and strong (all information, including private/monopolistic) — formalizes them with "fair game" / submartingale / random-walk models, and surveys the evidence, concluding that "with but a few exceptions, the efficient markets model stands up well."
The idea
A market is efficient with respect to an information set Φₜ if prices already incorporate it, so trading on that information cannot earn abnormal returns. Fama makes the slogan testable by embedding it in an expected-return ("fair game") model: E(r̃ⱼ,ₜ₊₁ | Φₜ) is set by some equilibrium theory of risk, and the realized "excess" of price over its conditional expectation is a fair game (unforecastable) given Φₜ. Special cases include the submartingale (expected return ≥ 0, justifying that filter/technical rules can't beat buy-and-hold) and the random-walk model (i.i.d. returns) — the random walk is a sufficient but stronger condition than efficiency requires. The three-way weak / semi-strong / strong partition (attributed in the paper to Harry Roberts) classifies tests by how broad the information set Φₜ is:
Weak form: Φₜ = past prices/returns → technical analysis cannot beat the market.
Semi-strong form: Φₜ = all public information → fundamental analysis of public data cannot beat the market; event studies (e.g., stock splits) test this.
Strong form: Φₜ = all information including private → not even insiders/specialists can profit.
Evidence
Fama's survey reads the literature as broadly supportive:
Weak form: serial-correlation and runs tests find return dependence too small to exploit after costs; filter-rule studies (Alexander, Fama-Blume) don't beat buy-and-hold net of transaction costs.
Semi-strong form: the Fama-Fisher-Jensen-Roll stock-split event study and earnings-announcement work show prices adjust quickly and on average completely to public news.
Strong form: the main documented exceptions — corporate insiders and NYSE specialists with monopolistic order-book information appear to earn abnormal returns — so strong-form efficiency is rejected, but these are narrow, expected exceptions.
Why it matters
The intellectual benchmark against which every anomaly and trading strategy is judged.
Establishes that any test of efficiency requires a model of equilibrium expected returns to define "abnormal" — the kernel of what later (Fama 1991) is sharpened into the joint-hypothesis problem: a rejection can mean the market is inefficient or the asset-pricing model is wrong, and the two cannot be cleanly separated. This is central to honest evaluation of any backtest.
The "fair game" and event-study machinery here became the standard empirical toolkit of asset pricing.
Caveats
The joint-hypothesis issue makes EMH effectively untestable in isolation.
"Fully reflect" is, by Fama's own admission, too general to test without auxiliary assumptions (stating equilibrium in terms of expected returns elevates expected value beyond what efficiency per se implies).
Strong-form efficiency is empirically rejected (insiders/specialists).
Behavioral finance and later-documented anomalies (momentum, value, PEAD, excess volatility) challenge semi-strong efficiency; Fama (1991, 1998) revisits the debate.
Key references
Fama, E. (1970) — Efficient Capital Markets: A Review of Theory and Empirical Work — Journal of Finance
Fama, E. (1991) — Efficient Capital Markets: II — Journal of Finance
Fama, Fisher, Jensen & Roll (1969) — The Adjustment of Stock Prices to New Information — International Economic Review
Grossman, S. & Stiglitz, J. (1980) — On the Impossibility of Informationally Efficient Markets — American Economic Review
Provenance: verified/generated from the paper's full text.
