§4.1: at a fixed reserve, (corrected: , )
ProvedAuctionsOfAuctions.revenue_ranking_fixed_reserveauctionsfirst-price-auctionintermediationmechanism-designp2o-correctedp2o-gran-per-chapterp2o-plan-paperp2o-strengthenedp2o-v1reserve-price
For every reserve :
- for every , the bids the publisher faces satisfy
- the publisher's revenues satisfy
with of (8), of (18) and of (4).
At any fixed reserve the competing-exchange arrangement is less profitable for the publisher than direct bidding and more profitable than a monopoly exchange. With the optimality of the direct reserve this gives the publisher's half of Proposition 2.
Differences from the source:
- reading: valuations drawn iid from a distribution with support (p.7) → , with continuous on , on , on and ; reason: the paper writes , divides by for and differentiates its revenue integrals in (A18), which uses a continuous density; and are allowed
- encoding: Stieltjes integrals in (3), (4), (5), (13), (18) → (interval integrals); reason: is absolutely continuous with density ; same instances
- generalization: the standing assumption that is regular (p.7) → not assumed; reason: the statement holds for every distribution of the model; regular distributions are instances
- correction: "for every , " → for every ; reason: at all three bids equal (any , any ), so the strict inequalities fail there
- correction: "for any fixed reserve price" → ; reason: at (and any ) the impression is never sold and all three revenues are , so the strict ranking fails
Preamble
import Mathlib import Definitions.Def_AuctionsOfAuctions_Model
Formal statement
namespace AuctionsOfAuctions
/-- **Revenue ranking at a fixed reserve**, Zeithammer–Choi, *Auctions of Auctions*, Management
Science (2024), doi:10.1287/mnsc.2024.05233, §4.1, text after (18), p.16 (PDF 17): "for every
`M ≥ R`, `R = γ^mono(M; R) < γ^ex(M; R) < γ^dir(M; R)`. It follows that, for any fixed reserve
price, the exchange mechanism is less profitable for the publisher than the direct mechanism, but
more profitable than the monopoly benchmark".
For every reserve `R ∈ [0, 1)`:
(1) for every `M ∈ (R, 1]`, `R < σ̃(M; R) < β(M; R)` (the bids the publisher faces with a monopoly
exchange, `γ^mono = R`, with competing exchanges, `γ^ex = σ̃`, and directly, `γ^dir = β`);
(2) `π^mono_P(R) < π^ex_P(R) < π^dir_P(R)`, with `π^mono_P` of (8), `π^ex_P` of (18) with
`γ^ex = σ̃`, and `π^dir_P` of (4).
**Formalization Note.** At `M = R` all three bids equal `R`, so the pointwise ranking is strict
only for `M > R`. At `R = 1` the item is never sold and all three revenues are `0`, so the strict
ranking is stated for `R < 1`. -/
theorem revenue_ranking_fixed_reserve (D : ValuationDist) (R : ℝ) (hR : R ∈ Set.Ico (0 : ℝ) 1) :
(∀ M ∈ Set.Ioc R 1, R < D.sigmaTilde M R ∧ D.sigmaTilde M R < D.beta M R) ∧
D.revMono R < D.revEx R ∧ D.revEx R < D.revDir R := by sorry
end AuctionsOfAuctions
Source
Zeithammer–Choi, Auctions of Auctions, Management Science (2024), doi:10.1287/mnsc.2024.05233, §4.1, text after eq. (18), p.16 (PDF 17)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.