Skip to content
Back to search results

Subgame Perfect Equilibrium

Introduced by
Reinhard Selten
Year
1965
Type
algorithm
Family
Game-theoretic
Font size

What does it do?

Subgame Perfect Equilibrium (SPE) is a refinement of Nash Equilibrium for sequential games, introduced by Reinhard Selten in 1965.

It requires that strategy profiles constitute a Nash Equilibrium in every subgame, eliminating non-credible threats and incredible promises. Backward induction is the primary computational method for finding SPE in finite games.

The intuition behind the method

A subgame perfect equilibrium ensures that no player would want to deviate from their planned strategy even if they suddenly found themselves at any point in the game tree. Think of it as a fully detailed battle plan where every contingency is accounted for: not only are decisions optimal at the start, but they remain optimal when actually reached. Backward induction works by starting at the end of the game tree and computing optimal decisions at each node, working back to the start, guaranteeing that threats and promises made early are credible.

History

Reinhard Selten introduced Subgame Perfect Equilibrium in 1965 as a refinement of Nash's concept to address the implausibility of non-credible threats in sequential games. His work earned him the Nobel Prize in Economics in 1994 (shared with John Nash and John Harsanyi). The concept became central to dynamic game theory and has profoundly influenced economic modeling of markets, negotiations, and strategic competition.

When to use it

Apply SPE analysis to sequential games with incomplete information resolved over time: negotiations, auctions with multiple rounds, entry and exit decisions, or sequential moves in competition. Essential when understanding the order of moves and credibility of commitments matters. Works best for games with perfect or complete information; imperfect information requires refinements such as Perfect Bayesian Equilibrium.

Limitations and considerations

  • Restricted to finite games with perfect recall (no information asymmetries across decision nodes)
  • Exponential complexity as game tree size grows, making large sequential games intractable
  • May yield multiple equilibria in games with simultaneous moves or imperfect information
  • Backward induction is sensitive to game tree representation; different extensive forms can yield different equilibria

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Eliminates implausible equilibria based on non-credible threats or promises
  • Computationally tractable for finite games using backward induction
  • Uniquely applicable to sequential games where move order and timing are critical
  • Provides a clear decision procedure via explicit algorithm, enabling direct implementation

Applications

  • Industrial organization: modeling sequential entry and pricing decisions in oligopolistic markets
  • International relations: analyzing sequential bargaining and threats in trade negotiations
  • Labor economics: modeling employer-employee wage negotiations with sequential offers and counteroffers
  • Environmental economics: studying repeated interactions between regulators and polluting firms

How it works

How it works

1

Construct the extensive-form game tree with nodes, actions, players, and payoffs at terminal nodes.

2

Starting from the final decision nodes, solve for the optimal action at each node by comparing each player's payoffs from available moves.

3

Work backward through the game tree, using optimal strategies from later nodes to determine optimal play at earlier nodes.

4

Check that the resulting strategy profile forms a Nash Equilibrium in every possible subgame, not just the overall game.

Common pitfalls

  • Confusing SPE with Nash Equilibrium; many Nash Equilibria are not subgame perfect due to out-of-equilibrium threats
  • Misapplying backward induction to games with imperfect information, which requires belief updating
  • Overlooking that SPE refinement requires checking optimality at every subgame node, not just terminal payoffs
  • Assuming unique equilibrium exists; many sequential games have multiple SPE with different outcomes

Frequently asked questions

What is the difference between a Nash Equilibrium and a Subgame Perfect Equilibrium?

Every SPE is a Nash Equilibrium, but not vice versa. SPE requires Nash equilibrium play at every subgame, eliminating equilibria sustained by non-credible out-of-equilibrium threats that would never be carried out.

Why does backward induction sometimes seem to predict unrealistic behavior?

Backward induction assumes common knowledge of rationality and perfect information. Real players may have limited foresight, bounded rationality, or imperfect information, making SPE predictions less reliable in practice.

How does SPE handle games with imperfect information?

SPE alone does not directly apply; instead, Perfect Bayesian Equilibrium extends SPE by requiring players to update beliefs about others' types using Bayes' rule and then play optimally given those beliefs.

Related methods

Sources

  1. Selten, R. (1965). Spieltheoretische Behandlung eines Oligopolmodells mit Nachfrageträgheit. Zeitschrift für die gesamte Staatswissenschaft, 121, 301-324.

    Open publication reference
    Copied.
  2. von Stackelberg, H. (1934). Marktform und Gleichgewicht. Julius Springer.

    Open publication reference
    Copied.