Proposition 6(a), p. 1350 — with σ_ε = 0, the QF contract with flexibility ψ and transfer price c̄(ψ) of (4) is system-efficient
ProvedTsayQF.EffQF.proposition_6aLet , , be the cost data, and let the demand signal have law , a probability measure on with finite variance whose distribution function is differentiable and strictly increasing; with market demand equals . Let satisfy , i.e. . Take QF parameters , , total flexibility , and the transfer price
Let . Under the QF contract :
- a forecast maximizes the retailer's expected profit over if and only if ;
- given the forecast , a production maximizes the EM's expected profit (2) over if and only if ;
- the resulting expected system profit is at least the central planner's expected profit at every production level:
Together, the equilibrium of the QF contract puts the efficient quantity into the system and makes all of it available to the market, which is what the paper calls system efficiency. A menu of efficient contracts is obtained by varying and charging .
Formalization Note. is not used as a function: is a real number with the hypothesis , which determines it uniquely since is strictly increasing, and (4) is kept in its printed form. The retailer's forecast objective fixes the EM's production at , as §6.1 does after Proposition 2; clause 2 checks that the EM indeed wants to build exactly that amount at this contract. The full game in which the retailer could anticipate other EM responses is not modelled. The hypothesis makes explicit the paper's presumption that demand is almost certainly nonnegative (§3.3); without it no forecast could produce . The bound is Proposition 3's standing case, where is finite. can exceed when ; no restriction is imposed.
import Mathlib import Definitions.Def_TsayQF_EffQF_Model open MeasureTheory ProbabilityTheory
namespace TsayQF.EffQF
/-- Proposition 6(a), p. 1350: when `σ_ε = 0`, the QF contract `{c̄(ψ), (α, ω)}` with
`ψ = (1 + α)/(1 − ω)` and `c̄` given by (4) is system-efficient: the retailer's unique optimal
forecast is `q̂ = Q*_CC/(1 + α)`, the EM's unique optimal production is then `q̂(1 + α) = Q*_CC`,
and the resulting expected system profit equals the centralized optimum. -/
theorem proposition_6a (D : Data) (ν : Measure ℝ) [IsProbabilityMeasure ν]
(hΘ : StrictMono (cdf ν)) (hΘd : Differentiable ℝ (cdf ν)) (hν : MemLp id 2 ν)
(α ω : ℝ) (hω0 : 0 ≤ ω) (hω1 : ω < 1) (hα : -ω ≤ α)
(Qcc : ℝ) (hQcc : cdf ν Qcc = kS D) (hpos : 0 < Qcc) :
let c := cbar D ν (psi α ω) Qcc
let qhat := Qcc / (1 + α)
(∀ q, 0 ≤ q → (IsMaxOn (forecastProfit D ν c α ω) (Set.Ici 0) q ↔ q = qhat)) ∧
(∀ Q, qhat * (1 + α) ≤ Q →
(IsMaxOn (emProfit D ν c ω qhat) (Set.Ici (qhat * (1 + α))) Q ↔ Q = qhat * (1 + α))) ∧
(∀ Q, ccProfit D ν Q ≤
forecastProfit D ν c α ω qhat + emProfit D ν c ω qhat (qhat * (1 + α))) := by sorry
end TsayQF.EffQF
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.