Grossman & Stiglitz (1980)

Citation: Grossman, S. J., & Stiglitz, J. E. (1980). “On the Impossibility of Informationally Efficient Markets.” American Economic Review.

Grossman and Stiglitz’s 1980 paper presents a famous paradox: if markets were perfectly informationally efficient—if prices fully reflected all available information—then no one would have an incentive to gather information. But if no one gathers information, prices cannot become informed. Therefore, perfectly efficient markets are impossible.

The argument is deceptively simple. Gathering and processing information is costly. If prices already reflect all information, then an investor gains nothing from doing research—they could simply trade on prices and achieve the same outcome for free. But if all investors reason this way, no one does research, and prices become uninformed. This creates an equilibrium where prices must be partially but not fully efficient.

flowchart TD
  subgraph Efficient Market Assumption
    E[Prices fully reflect information]
  end
  E --> N[No return to gathering info]
  N --> S[No one gathers info]
  S --> U[Prices become uninformed]
  U --> C[Contradiction]

The paper formalizes this intuition with a model in which some traders are informed (they pay a cost to acquire a signal about asset value) and others are uninformed (they observe only prices). In equilibrium:

1) Informed traders must be compensated for the cost of information. This means prices cannot fully reveal their information—otherwise they’d earn no return for their effort.

2) Noise is necessary for informed trading to be profitable. If prices perfectly revealed information, uninformed traders could free-ride. The presence of noise traders (or liquidity traders) allows informed traders to profit from their knowledge.

3) The degree of market efficiency depends on information costs. When information is cheap, more traders become informed, and prices become more efficient—but never perfectly so.

flowchart LR
  IC[Information cost] -->|Low| MI[More informed traders]
  IC -->|High| FI[Fewer informed traders]
  MI --> PE[Prices more efficient]
  FI --> PL[Prices less efficient]
  PE --> LR[Lower returns to research]
  PL --> HR[Higher returns to research]
  LR --> FI
  HR --> MI

Key implications:

1) The efficient market hypothesis is an approximation, not a literal truth. Markets can be “nearly efficient” in that prices reflect most available information, but they cannot be perfectly efficient because that would eliminate the incentive to produce the information prices are supposed to reflect.

2) Information asymmetry is an equilibrium feature. Markets do not eliminate information advantages; they produce an equilibrium level of asymmetry that balances the cost of information against the profits from being informed.

3) Liquidity and efficiency are linked. Markets need noise or liquidity trading to provide profits for informed traders. Paradoxically, the “irrationality” of noise traders helps make prices informative.

4) Price discovery is costly. Someone has to do the work of making prices accurate, and they have to be paid for it. This has implications for market design, disclosure requirements, and the role of analysts, rating agencies, and other information intermediaries.

The Grossman-Stiglitz result is foundational for understanding why prices are informative but not omniscient, why information asymmetries persist in markets, and why intermediaries who produce information (analysts, auditors, rating agencies) can earn rents. It also provides a framework for thinking about any system that relies on prices to aggregate dispersed information: the aggregation works, but only if someone is incentivized to provide the raw inputs.

Perfectly efficient markets are a logical impossibility. Information has costs, and those who bear them must be compensated through imperfectly efficient prices. The efficiency of prices depends on someone being paid to make them accurate.