Example 1(a), seller half — the seller's strategy is a best response
ProvedChatterjeeSamuelson.UniformEquilibrium.seller_best_responsebargaininggame-theoryincomplete-informationp2o-batch-p200ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Let and , let both reservation prices lie in , and let each player's belief about the other's value be uniform on . Let be measurable offer strategies such that
- for ;
- for ;
- for ;
- for .
Then for every seller value and every real ask ,
where is the seller's expected profit against under the uniform belief. That is, is a best response to for every seller type.
Together with the buyer half, this is the statement that is an equilibrium (Example 1(a)).
Formalization Note Measurability of and is added: the page leaves on the upper range and on the lower range free up to an inequality, and a non-measurable choice would make expected profits meaningless. The printed coefficient is read as .
Preamble
import Mathlib import Definitions.Def_ChatterjeeSamuelson_Shared_unif import Definitions.Def_ChatterjeeSamuelson_Shared_buyerProfit import Definitions.Def_ChatterjeeSamuelson_Shared_sellerProfit import Definitions.Def_ChatterjeeSamuelson_Shared_sellerLinear import Definitions.Def_ChatterjeeSamuelson_Shared_buyerLinear open MeasureTheory ProbabilityTheory Set
Formal statement
namespace ChatterjeeSamuelson.UniformEquilibrium
/-- Example 1(a), seller half (Chatterjee & Samuelson, *Bargaining under Incomplete
Information*, Oper. Res. 31(5) 1983, §3, Example 1(a), p. 842 [PDF 8]): with both values
uniform on `[0, v̄]` (`F_s(v) = F_b(v) = v/v̄`), the seller's strategy `S` of Example 1(a) is a
best response to the buyer's strategy `B` of Example 1(a): for every seller value
`v ∈ [0, v̄]` and every real ask `s`, `π_s(s, v) ≤ π_s(S(v), v)`, where the seller's belief
about the buyer's value is uniform on `[0, v̄]`.
Hypotheses: `0 ≤ k ≤ 1`, `0 < v̄`, `S` and `B` measurable, and the four lines of
Example 1(a) on their printed ranges:
1. `S v = v/(2 − k) + ((1 − k)/2) v̄` for `0 ≤ v ≤ ((2 − k)/2) v̄`;
2. `S v ≥ v/(2 − k) + ((1 − k)/2) v̄` for `((2 − k)/2) v̄ < v ≤ v̄`;
3. `B v ≤ v/(1 + k) + (k(1 − k)/(2(1 + k))) v̄` for `0 ≤ v < ((1 − k)/2) v̄`;
4. `B v = v/(1 + k) + (k(1 − k)/(2(1 + k))) v̄` for `((1 − k)/2) v̄ ≤ v ≤ v̄`.
`S`, `B` are unconstrained outside `[0, v̄]`.
*Formalization Note.*
`Measurable S`, `Measurable B` are added: the page leaves `S` above
`((2 − k)/2) v̄` and `B` below `((1 − k)/2) v̄` free up to an inequality, and a
non-measurable choice would make an expected profit a junk Bochner integral; the paper's
offer densities `g_b`, `g_s` presuppose measurable strategies. The printed coefficient `k(1 − k)/2(1 + k)` is read as
`k(1 − k)/(2(1 + k))` (see `buyerLinear`). -/
theorem seller_best_response (k vbar : ℝ) (hk0 : 0 ≤ k) (hk1 : k ≤ 1) (hvbar : 0 < vbar)
(S B : ℝ → ℝ) (hSmeas : Measurable S) (hBmeas : Measurable B)
(hS_eq : ∀ v, 0 ≤ v → v ≤ (2 - k) / 2 * vbar → S v = Shared.sellerLinear k vbar v)
(hS_ge : ∀ v, (2 - k) / 2 * vbar < v → v ≤ vbar → Shared.sellerLinear k vbar v ≤ S v)
(hB_le : ∀ v, 0 ≤ v → v < (1 - k) / 2 * vbar → B v ≤ Shared.buyerLinear k vbar v)
(hB_eq : ∀ v, (1 - k) / 2 * vbar ≤ v → v ≤ vbar → B v = Shared.buyerLinear k vbar v) :
∀ v ∈ Icc (0 : ℝ) vbar, ∀ s : ℝ,
Shared.sellerProfit k (Shared.unif vbar) B s v ≤ Shared.sellerProfit k (Shared.unif vbar) B (S v) v := by sorry
end ChatterjeeSamuelson.UniformEquilibrium
Source
Chatterjee & Samuelson, Bargaining under Incomplete Information, Oper. Res. 31(5) (1983), https://doi.org/10.1287/opre.31.5.835, p. 842 [PDF 8], Example 1(a) (seller's best-response half)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.