Overview
The folk theorem is a collection of results in repeated-game theory showing that when the same strategic interaction (the “stage game”) is played repeatedly and players are sufficiently patient, a very wide range of long-run outcomes can be sustained as equilibria through credible rewards and punishments. In plain terms: if parties expect to interact again and can observe past behavior, they can discipline one another to achieve cooperative outcomes that would be impossible in a one-shot setting. This insight underpins relational contracts, tacit coordination in oligopoly, and the role of reputation in platforms and supply chains.
Origins and Credit
The result was long informally known (“folk wisdom”) before formal proofs. Key milestones include Friedman (1971) on trigger strategies sustaining collusion, Aumann (1981) and Rubinstein (1979–1980s) on repeated games, and Fudenberg and Maskin (1986) who provided general “folk theorems” for infinitely repeated games with discounting. Green and Porter (1984), Abreu (1988), and subsequent work extended the results to imperfect monitoring (only noisy public signals observed). The label “folk theorem” now refers to several related theorems differing by monitoring and equilibrium concepts.
Core Idea and Mechanics
A repeated game plays a fixed stage game (e.g., pricing, effort choice) over many periods. Players discount the future at factor δ (0 < δ < 1): a payoff next period is worth δ relative to today. An equilibrium is a set of strategies—rules mapping past history to current actions—such that no player wants to deviate given others’ strategies. A subgame perfect equilibrium (SPE) requires credibility after every possible history.
The folk theorem states, roughly, that any feasible payoff profile giving each player at least their minmax (the worst they can be held to by others’ punishment) can be supported by some equilibrium of the repeated game, provided players are patient enough (δ sufficiently close to 1). The logic is simple:
- Cooperate today to generate high current payoffs.
- If someone deviates, trigger a credible punishment (e.g., reverting to a low-payoff action profile, such as aggressive pricing) long enough to make deviation unprofitable.
- When δ is high, the threat of losing future cooperative rents outweighs the one-time gain from cheating.
Under perfect monitoring (actions are observed), strategies like grim trigger (“cooperate until a deviation, then punish forever”) or finite-length punishments can sustain cooperation. With imperfect public monitoring (only a noisy public signal is observed), cooperation can still be supported using public strategies (e.g., price wars triggered by low sales signals à la Green–Porter). With private monitoring (each player observes only a private signal), sustaining cooperation is harder but still possible under additional conditions.
Key Assumptions and Conditions
- Indefinite or infinite horizon, or an uncertain stopping time that avoids a known last period (to prevent end-game unraveling).
- Players are sufficiently patient (high δ), valuing future gains from cooperation relative to current deviations.
- Observability/monitoring:
- Perfect monitoring: actions are publicly observed; strongest results.
- Imperfect public monitoring: a common noisy signal is observed; folk theorems hold under regularity conditions.
- Private monitoring: each player sees a private, noisy signal; additional structure is needed.
- Ability to condition strategies on history (credible punishments and rewards).
- Rationality and common knowledge of the game and discount factors.
Implications
- Self-enforcing cooperation: Long-term relationships can substitute for formal contracts when performance is hard to verify, as future interaction disciplines opportunism.
- Relational contracts: Bonus/penalty schemes and discretion work when supported by reputational consequences; formal pay alone may be insufficient.
- Oligopoly conduct: Even without explicit collusion, repeated interaction can sustain higher prices or capacity restraint if firms can detect deviations and care about the future.
- Multiple equilibria: Many outcomes—cooperative, punitive, or mixed—can be equilibria. Selection, norms, and expectations become central to predicting behavior.
- Monitoring design: Better, timely, and public performance measures strengthen cooperative equilibria; noisy or delayed metrics necessitate more conservative triggers and can cause inefficient “price wars.”
- Legal and policy lens: Antitrust agencies recognize that repeated interaction and transparency can facilitate tacit coordination; market design can alter monitoring and patience to deter it.
Example in Practice
Supplier quality in a relational contract. A manufacturer buys components from a supplier. High quality costs the supplier more today but increases the manufacturer’s future value. Courts cannot easily verify subtle quality dimensions, so formal enforcement is weak.
They adopt a relational agreement: the manufacturer pays a premium as long as quality complaints remain below a threshold; if quality slips (detected via audits/returns), the manufacturer switches volume to a backup supplier for six months. The supplier trades off a one-time saving from cutting corners against losing future premium business. If both sides are sufficiently patient (δ high), and monitoring is timely enough to detect cheating with reasonable probability, the threat of volume loss deters shirking. This self-enforcing arrangement implements an outcome superior to any feasible one-shot contract with unverifiable quality—an application of the folk theorem logic.
Limitations and Common Misunderstandings
- Patience and horizon matter: If players are impatient or the relationship is short/likely to end, punishments lack bite; cooperation unravels.
- Monitoring frictions: Imperfect or delayed signals can trigger mistaken punishments (“price wars”) or allow undetected cheating, reducing achievable surplus.
- Multiplicity ≠ predictability: The theorem guarantees existence of many equilibria; it does not select among them. History, norms, and communication shape which outcome prevails.
- Renegotiation: Very harsh punishments may be time-inconsistent; once off the path, both sides may prefer to renegotiate rather than carry out punishments. Renegotiation-proofness can restrict sustainable outcomes.
- Legal constraints: Using repeated interaction to sustain high prices can violate antitrust law; firms must avoid coordination that constitutes collusion.
- Finite games: With a known final period, backward induction kills cooperation (unless uncertainty about the end injects an “effective infinity”).
Related Concepts
Repeated games; Subgame perfect equilibrium; Grim trigger and tit-for-tat; Relational contracts; Imperfect public monitoring; Reputation models.
