[Paper Review] Strengthening Bulow-Klemperer-Style Results for Multi-Unit Auctions

Moshe Babaioff, Yiding Feng+1 more|arXiv (Cornell University)|Feb 12, 2026

Auction Theory and Applications0 citations

TL;DR

The paper shows that with stronger distributional assumptions (MHR and λ-regularity) or via a prior-independent supply-limiting VCG variant, far fewer additional buyers are needed for the VCG mechanism to match or closely approximate Bayesian-optimal revenue in multi-unit auctions, with precise finite and asymptotic guarantees.

ABSTRACT

The classic result of Bulow and Klemperer (1996) shows that in multi-unit auctions with $m$ units and $n\geq m$ buyers whose values are sampled i.i.d. from a regular distribution, the revenue of the VCG auction with $m$ additional buyers is at least as large as the optimal revenue. Unfortunately, for regular distributions, adding $m$ additional buyers is sometimes indeed necessary, so the "competition complexity" of the VCG auction is $m$. We seek proving better competition complexity results in two dimensions. First, under stronger distributional assumptions, the competition complexity of VCG auction drops dramatically. In balanced markets (where $m=n$) with MHR distributions, it is sufficient to only add $(e^{1/e} - 1 + o(1))n \approx 0.4447n$ additional buyers to match the optimal revenue -- less than half the number that is necessary under regularity -- and this bound is asymptotically tight. We provide both exact finite-market results for small value of $n$, and closed-form asymptotic formulas for general market with any $m\leq n$, and any target fraction of the optimal revenue. Second, we analyze a supply-limiting variant of VCG auction that caps the number of units sold in a prior-independent way. Whenever the goal is to achieve almost the optimal revenue, this mechanism strictly improves upon standard VCG auction, requiring significantly fewer additional buyers. Together, our results show that both stronger distributional assumptions, as well as a simple prior-independent refinement to the VCG auction, can each substantially reduce the number of additional buyers that is sufficient to achieve (near-)optimal revenue. Our analysis hinges on a unified worst-case reduction to truncated generalized Pareto distributions, enabling both numerical computation and analytical tractability.

Motivation & Objective

• Investigate whether stronger distributional assumptions reduce the number of extra buyers needed for VCG to beat or approximate the Bayesian-optimal revenue in multi-unit auctions.
• Characterize worst-case distributions under MHR/λ-regularity that determine competition complexity.
• Develop and analyze a supply-limiting, prior-independent variant of VCG to improve revenue guarantees without distributional knowledge.
• Provide exact finite-market results and asymptotic formulas across balanced and general markets (varying m ≤ n).

Proposed method

• Reduce the problem to a worst-case one-parameter family of truncated λ-generalized Pareto distributions that captures the revenue gap between VCG and the Bayesian optimum.
• Use both numerical grid search and analytical arguments to compute competition complexity for small markets (n ≤ 593) and derive asymptotic bounds for large markets (n → ∞).
• Prove a uniform upper bound on competition complexity under MHR (and λ-regular) distributions, approaching (e^{1}/e − 1)·n ≈ 0.4447·n for balanced markets.
• Extend the worst-case reduction to general λ-regular distributions and imbalanced markets, deriving closed-form expressions for asymptotic competition complexity as a function of supply-to-demand ratio α and target Γ.
• Introduce and analyze a Supply-Limiting VCG Auction that sells a fixed, prior-independent fraction of units to improve revenue guarantees for Γ < 1.
• Show that the worst-case distributions remain within truncated λ-generalized Pareto class for all variants (VCG and supply-limiting).

Experimental results

Research questions

• RQ1 Question 1: Under stronger distributional assumptions (λ-regular/MHR), is it sufficient to add far fewer buyers to guarantee that VCG outperforms the Bayesian-optimal mechanism?
• RQ2 Question 2: Can a prior-independent, supply-limiting variant of VCG achieve comparable or better revenue guarantees with fewer additional buyers than standard VCG?
• RQ3 How do results extend from balanced markets to general markets with m ≤ n and varying α (supply-to-demand ratio)?
• RQ4 What is the asymptotic competition complexity for achieving a given revenue fraction Γ under λ-regular/MHR distributions?

Key findings

• Under MHR distributions in balanced markets (n = m), fewer than n additional buyers suffice to beat the Bayesian-optimal revenue; for n ≥ 23 this is already below half of n.
• Asymptotically, the competition complexity for MHR distributions in balanced markets is at most (e^{1}/e − 1)·n ≈ 0.4447·n, and this bound is tight.
• In general markets with λ-regular distributions, for any α ∈ [0,1] and any Γ ∈ (0,1], the worst-case distribution lies in the class of truncated λ-generalized Pareto distributions, enabling closed-form asymptotics (Theorem 4.2, Theorem 5.2).
• In large markets, for the Supply-Limiting VCG Auction, achieving a target Γ < 1 strictly improves upon standard VCG by requiring significantly fewer extra buyers; when Γ approaches 1, the advantage diminishes.
• Optimal supply in the Supply-Limiting VCG Auction is approximately a Γ-fraction of total units; for balanced markets, the asymptotic CC of Supply-Limiting VCG is strictly lower than that of standard VCG for Γ < 1.
• The worst-case reductions and analyses rely on a unified reduction to truncated λ-generalized Pareto distributions, enabling both numerical and analytical treatment.