Skip to content
Tutorials Center
My learning

Topics

Lesson 33 of 37 ยท Applications

Bargaining: The Nash Solution and Rubinstein Offers

Divide a surplus by negotiation: the Nash bargaining solution, its axioms, disagreement points, and Rubinstein's alternating-offers model solved by backward induction in Python.

Two parties can create a surplus together: a worker and a firm, a buyer and a seller, two countries signing a treaty. They must agree on how to share it, and if they fail, both get a fallback. Bargaining theory predicts the split. Two approaches complement each other. The axiomatic approach (Nash, 1950) says what a fair outcome should satisfy. The strategic approach (Rubinstein, 1982) models the negotiation as a sequential game and solves it with backward induction.

The bargaining problem

A bargaining problem has a set S of feasible utility pairs (u1, u2) and a disagreement point d = (d1, d2), the utilities if talks fail. The solution picks a point in S where both gain relative to d.

The Nash bargaining solution is the unique point satisfying four axioms:

  1. Pareto efficiency: no wasted surplus.
  2. Symmetry: identical players get identical outcomes.
  3. Invariance to utility scaling: changing the units of utility does not change the outcome.
  4. Independence of irrelevant alternatives: removing options other than the chosen one does not change the choice.

The theorem: the solution is the point that maximizes the Nash product (u1 - d1)(u2 - d2) over feasible points with u >= d.

Nash bargaining

Disagreement point What each gets if talks fail
Surplus Gains over the disagreement point
Maximize the product (u1 - d1) times (u2 - d2)
Unique fair split Satisfies the four axioms

Threat points matter: a better fallback means a better deal.

1 / 2

Splitting a pie

Divide 100 units of money. If the players have linear utility and disagreement points d1 and d2, the Nash solution gives each their fallback plus half of the remaining surplus: u1 = d1 + (100 - d1 - d2) / 2.

Python Nash product on a money pie

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The outside option is crucial: a player who can walk away with more captures half of the extra. And a more risk-averse player gets less in the Nash solution, because concave utility lowers the gain from each extra unit.

Rubinstein alternating offers

In Rubinstein's model, two players divide a pie of size 1. Player 1 proposes a split. Player 2 accepts, or rejects and proposes in the next period. Each period of delay shrinks the pie's value: player i's utility from receiving share x in period t is delta_i^t times x. There is no deadline.

The unique subgame-perfect equilibrium has the first proposer offer exactly the amount that makes the other indifferent to waiting, and the offer is accepted immediately. Player 1's share is

x1 = (1 - delta2) / (1 - delta1 * delta2)

With equal patience delta, this is 1 / (1 + delta), a bit more than half: the proposer has the first-mover advantage, which vanishes as delta goes to 1.

Finding it by backward induction

Consider a game with a finite number of rounds T. In the last round the proposer takes everything (the responder accepts any positive share). Working backward, the proposer in each earlier round offers the discounted value of what the responder would earn as the next proposer. As T grows, the result converges to Rubinstein's formula.

Python Finite horizon converges to the Rubinstein split

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

With an odd number of rounds, player 1 also proposes last (a large share), and with an even number player 2 does (a small share). Both sequences converge to the Rubinstein value (0.2 / 0.28 = 0.714 here) as T increases. The more patient player gets the larger share: a patient bargainer can wait, and an impatient one concedes.

The two approaches meet

When time between offers becomes very short, so that delta goes to 1 for both players, Rubinstein's solution tends to the asymmetric Nash solution, with the weights determined by the players' relative patience (their discount rates). When both players are equally patient, it tends to the symmetric Nash solution (an equal split). So the strategic and axiomatic models support each other.

Python Rubinstein shares as patience grows

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Practical lessons

  • Improve your outside option: your fallback raises your share.
  • Patience is power; urgency is costly. Deadlines and costs of delay matter.
  • A believable commitment (see Stackelberg and commitment) can shift the split, though it risks a breakdown.
  • In real negotiations, information asymmetries, multiple issues, fairness norms and delays create richer behavior than these clean models, but their logic stays visible.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What does the Nash bargaining solution maximize?

    It maximizes (u1 - d1)(u2 - d2) among feasible points.
  2. 2. With a pie of 100 and linear utility, player 1's outside option is 20 and player 2's is 0. What does player 1 get?

    The surplus above the outside options is 80, split equally: player 1 gets 20 + 40 = 60.
  3. 3. In Rubinstein's model with equal discount factor delta, how does the first proposer's share compare with a half?

    The share is 1 / (1 + delta): the first mover has an advantage that fades as players become patient.
  4. 4. Who gets a larger share in alternating-offers bargaining?

    Patience lets a player wait for a better offer, which improves their bargaining position.

Summary

  • The bargaining problem has a feasible set and a disagreement point; Nash's solution maximizes the product of gains.
  • The Nash solution is characterized by Pareto efficiency, symmetry, scale invariance and independence of irrelevant alternatives.
  • Better outside options and less risk aversion raise a player's share.
  • Rubinstein's alternating offers give a unique subgame-perfect split, agreed immediately: x1 = (1 - delta2) / (1 - delta1 delta2).
  • Patience is bargaining power, and both models agree as delays become short.

Next: Voting and Arrow's Theorem