Efficient Market Hypothesis

Definition

A capital market is efficient if security prices "fully reflect" all available information, so that prices provide accurate signals for resource allocation and no trading rule based on the available information set can earn expected returns in excess of equilibrium expected returns (Fama 1970). The hypothesis is graded by which information set prices are claimed to reflect.

Key Ideas

• Three forms (by information subset $\Phi_t$):
  • Weak form — $\Phi_t$ = the history of prices/returns. Implies past prices carry no exploitable information (technical analysis is useless); tested via serial correlation, runs, and filter rules.
  • Semi-strong form — $\Phi_t$ = all publicly available information (earnings, splits, news). Implies prices adjust quickly and unbiasedly to public announcements; tested via event studies (e.g. Fama-Fisher-Jensen-Roll 1969 on stock splits).
  • Strong form — $\Phi_t$ = all information, including private/monopolistic. Tested via insider trading and mutual-fund performance (Jensen).
• The "fair game" formalization. With excess return $z_{j,t+1}=r_{j,t+1}-E(r_{j,t+1}\mid\Phi_t)$, efficiency requires $E(z_{j,t+1}\mid\Phi_t)=0$ — $\{z_{j,t}\}$ is a fair game with respect to $\{\Phi_t\}$. Any trading system on $\Phi_t$ then has zero expected excess value.
• Special cases. The submartingale model ($E(p_{t+1}\mid\Phi_t)\ge p_t$, i.e. non-negative expected returns) and the random walk (successive returns i.i.d. — a stronger claim about the whole distribution, not just the mean) are the fair game's tractable specializations used in the empirical literature.
• Sufficient (not necessary) conditions. No transaction costs, freely available information, and agreement among investors on the implications of information for prices — all sufficient but not required for efficiency.

How It Works

Efficiency is not tested directly; one specifies an equilibrium expected-return model $E(r_{j,t+1}\mid\Phi_t)$, forms the residual $z$, and checks whether it is unforecastable from $\Phi_t$. This is the joint-hypothesis problem: a rejection can mean the market is inefficient or that the assumed expected-return (asset-pricing) model is wrong — the two cannot be separated by the test alone.

Why It Matters

• Foundational to asset pricing and forecasting. EMH is the null against which return predictability, anomalies, and factor models are judged; the joint-hypothesis problem is why "anomalies" are always ambiguous between mispricing and missing risk factors — the motivation for multi-factor models like Fama-French.
• Discipline on forecasting claims. Weak-form efficiency is the reason excess-return forecasting from past prices is presumed hard; apparent predictability must clear the equilibrium-return bar.
• Links to no-arbitrage. Efficiency (no expected excess returns) is the statistical cousin of no-arbitrage pricing (no riskless profit); both formalize "prices already embed what is knowable."

Open Questions

• The joint-hypothesis problem is unavoidable: efficiency and the expected-return model are only jointly testable.
• Whether documented anomalies (value, momentum, size) reflect inefficiency or compensation for risk — the debate that produced the factor-model literature.
• Behavioral critiques: over/under-reaction and limits to arbitrage as departures from the fair-game benchmark.

Related

• Fama-French Five-Factor Model
• No-Arbitrage
• Capital Asset Pricing Model
• Forecasting
• Fama (1970) Efficient Capital Markets: A Review of Theory and Empirical Work
• Eugene F. Fama