Game Theory Lecture: SPNE, One-Shot Deviation, Infinite Bargaining
Backward induction remains reliable for every subgame, even when extensive‑form games contain “complicated” deviations. The logic holds because each subgame can be analyzed independently, preserving optimality from the end of the game back to the start.
Multistage Games
Multistage games consist of several stages that may be finite or infinite. Within each stage, players move simultaneously, and all actions from the previous stage are observed before the next stage begins. Typical examples include Cournot entry games, where firms decide whether to enter a market and then choose quantities, and price‑haggling scenarios that unfold over repeated rounds. The Boston game serves as a standard simultaneous‑move illustration of these dynamics.
One‑Shot Deviation Principle
A one‑shot deviation means a player changes his action at exactly one history or contingency and then returns to the original strategy thereafter.
Theorem: A strategy profile is a Subgame‑Perfect Nash Equilibrium (SPNE) if and only if no player has a profitable unilateral one‑shot deviation at any history (h).
This result simplifies verification because it eliminates the need to examine complex, multi‑history deviations. The principle requires the game to be “continuous,” a condition usually satisfied when future payoffs are discounted.
“A one‑shot deviation means I only deviate once. Or more precisely, I only deviate at one history or one contingency.”
“The upshot is that, to verify a subgame‑perfect Nash equilibrium, it’s enough to check the one‑shot deviations.”
Infinite‑Horizon Alternating‑Offer Bargaining
The bargaining model reduces negotiations to utility pairs ((x_1, x_2)) drawn from a feasible set (X) with a disagreement point at ((0,0)).
The discount factor (\delta) (where (0<\delta<1)) captures each player’s patience; a higher (\delta) indicates greater willingness to wait for future gains.
Equilibrium strategy: The proposer offers the respondent a share of (\frac{\delta}{1+\delta}); the proposer retains (\frac{1}{1+\delta}). The respondent accepts because the offer equals the discounted value of becoming the proposer in the next period.
“If the world ends tomorrow, then you as player 2 have to accept whatever I give you.”
“The first mover, player 1, is going to have an advantage in this game because they have the power of moving first.”
When (\delta \to 1) (players become very patient), the split converges to an even 50/50 division. When (\delta \to 0) (players are impatient), the proposer captures nearly the entire surplus.
Hard Facts & Numbers
- Discount factor: (\delta) ( (0<\delta<1) )
- Proposer’s share: (\frac{1}{1+\delta})
- Respondent’s share: (\frac{\delta}{1+\delta})
- Disagreement point: ((0,0))
These formulas illustrate how patience directly translates into bargaining power and how the one‑shot deviation logic underpins the verification of SPNE in both finite and infinite settings.
Takeaways
- Subgame‑Perfect Nash Equilibrium refines Nash equilibrium by requiring optimal play in every subgame, and backward induction remains valid for each subgame despite complex deviations.
- The One‑Shot Deviation Principle states that a strategy profile is an SPNE precisely when no player can profit from a single unilateral deviation at any history, eliminating the need to examine multi‑step deviations.
- The principle relies on game continuity, typically ensured by discounting future payoffs, so that checking one‑shot deviations suffices for verification.
- In an infinite‑horizon alternating‑offer bargaining model, the proposer offers the respondent a share equal to δ/(1+δ), giving the proposer 1/(1+δ), and the respondent accepts because the offer matches the discounted value of waiting.
- As the discount factor δ approaches 1, the equilibrium split converges to an even 50/50 division, while as δ approaches 0 the proposer captures almost all surplus, illustrating the impact of patience on bargaining power.
Frequently Asked Questions
How does the discount factor affect the equilibrium split in infinite‑horizon alternating‑offer bargaining?
The discount factor δ determines each player’s patience and directly sets the share each receives: the proposer gets 1/(1+δ) and the respondent gets δ/(1+δ). Higher δ leads to a more even split, while lower δ lets the proposer keep most of the surplus.
What technical condition must hold for the One‑Shot Deviation Principle to be valid?
The game must be continuous, meaning payoffs change smoothly with actions, which is typically achieved by discounting future payoffs. Continuity ensures that eliminating profitable one‑shot deviations guarantees the absence of any profitable multi‑step deviation.
Who is MIT OpenCourseWare on YouTube?
MIT OpenCourseWare is a YouTube channel that publishes videos on a range of topics. Browse more summaries from this channel below.
Does this page include the full transcript of the video?
Yes, the full transcript for this video is available on this page. Click 'Show transcript' in the sidebar to read it.
Helpful resources related to this video
If you want to practice or explore the concepts discussed in the video, these commonly used tools may help.
Links may be affiliate links. We only include resources that are genuinely relevant to the topic.