§4 p. 312 and §6 'First' pp. 314–315 — an optimal monopoly assignment is below the competitive one except at the top
ProvedMussaRosen.Mono.monopoly_quality_below_competitiveSetting. Consumer types are distributed on () with a density that is positive and differentiable on the whole interval; and . The unit cost of quality satisfies , is twice differentiable, and , for all . Under competition type buys the quality with when , and does not buy when . Let be an optimal monopoly assignment: nondecreasing, nonnegative and piecewise differentiable on , maximizing
among all such assignments.
Claim. Then
- every type served under competition, other than the top type, gets a strictly lower quality from the monopolist:
that is, wherever ; 2. types not served under competition are not served by the monopolist: for with ; 3. the top type gets the competitive quality: if , then converges, as , to a limit with , i.e. ; if , then .
This is the paper's main conclusion: the monopolist lowers the quality sold to every consumer who buys under competition, except the consumer with the highest valuation of quality ("no distortion at the top").
Formalization Note The competitive assignment is never formed: "" is written as , which is equivalent because is strictly increasing on . The value at the top is stated as a left limit, because the profit integral does not depend on the single value , which an optimal assignment may raise freely. The paper writes "" on p. 312; the intended reading is , hence . encodes that quality means not buying, so that no cost is charged for unserved types.
import Mathlib import Definitions.Def_MussaRosen_Mono_Setting open MechanismDesign.Screening Filter Topology
namespace MussaRosen.Mono
theorem monopoly_quality_below_competitive {θlo θhi : ℝ} (D : TypeDistribution θlo θhi)
(hfd : DifferentiableOn ℝ D.f (Set.Icc θlo θhi))
(C : ℝ → ℝ) (hC0 : C 0 = 0) (hCd : Differentiable ℝ C) (hCd2 : Differentiable ℝ (deriv C))
(hC1 : ∀ q : ℝ, 0 ≤ q → 0 < deriv C q) (hC2 : ∀ q : ℝ, 0 ≤ q → 0 < deriv (deriv C) q)
(q : ℝ → ℝ) (hq : IsOptimal D C q) :
(∀ θ ∈ Set.Ico θlo θhi, deriv C 0 < θ → deriv C (q θ) < θ) ∧
(∀ θ ∈ Set.Ico θlo θhi, θ ≤ deriv C 0 → q θ = 0) ∧
((deriv C 0 < θhi → ∃ L : ℝ, Tendsto q (𝓝[<] θhi) (𝓝 L) ∧ deriv C L = θhi) ∧
(θhi ≤ deriv C 0 → Tendsto q (𝓝[<] θhi) (𝓝 0))) := by sorry
end MussaRosen.Mono
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.