Section 7: maximal joint surplus excludes Pareto collusion
ProvedTeamDelegation.Optimal.surplus_implies_collusion_proofcollusion-proofnessdelegationjoint-surplusp2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1
Consider delegation at any two success bonuses. If in each state both agents working with the correct method maximises their total expected one-period payoff over every feasible pure stage outcome, then no state-contingent probability distribution over feasible joint outcomes Pareto dominates the desired payoff vector:
The result turns the statewise surplus comparison into the collusion condition used in Proposition 7.
Formalization Note The conclusion quantifies over all feasible state-dependent outcome lotteries. This is stronger than the paper's footnote 30, which quantifies only over other subgame-perfect equilibria.
Preamble
import Mathlib import Definitions.Def_TeamDelegation_Optimal_Model
Formal statement
namespace TeamDelegation.Optimal
/-- §7, p. 29: if the desired outcome maximizes total agent surplus
state by state, no joint feasible outcome lottery Pareto improves upon it. -/
theorem surplus_implies_collusion_proof (P : Params) (hS : P.Standing)
(C : Costs) (wA wB : ℝ)
(hmax : SurplusMax P C .deleg wA wB) :
CollusionProof P C .deleg wA wB := by sorry
end TeamDelegation.Optimal
Source
Upton, Implicit Incentives and Delegation in Teams, TRR 266 Working Paper No. 67 (SSRN 3991736, 17 Dec 2021), p. 29 (PDF 30), §7, paragraph immediately following Proposition 7; footnote 30 on p. 28 (PDF 29)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.