[Paper Review] Beckmann's approach to multi-item multi-bidder auctions

А. И. Колесников, Fedor Sandomirskiy+2 more|arXiv (Cornell University)|Mar 14, 2022

Auction Theory and Applications4 citations

TL;DR

This paper establishes a strong duality between revenue-maximizing multi-item, multi-bidder auctions and Beckmann's continuous optimal transportation problem, showing that optimal auction design reduces to finding a vector field minimizing transportation cost under supply-demand imbalance. The key contribution is a novel duality framework that enables existence of optimal mechanisms and provides a guess-and-verify approach using complementary slackness.

ABSTRACT

We consider the problem of revenue-maximizing Bayesian auction design with several bidders having independent private values over several items. We show that it can be reduced to the problem of continuous optimal transportation introduced by Beckmann (1952) where the optimal transportation flow generalizes the concept of ironed virtual valuations to the multi-item setting. We establish the strong duality between the two problems and the existence of solutions. The results rely on insights from majorization and optimal transportation theories and on the characterization of feasible interim mechanisms by Hart and Reny (2015).

Motivation & Objective

• To address the long-standing challenge of designing revenue-maximizing auctions with multiple bidders and multiple items under independent private values.
• To resolve the lack of general solutions in multi-item, multi-bidder settings, where even single-bidder cases are notoriously complex.
• To establish a formal connection between auction design and continuous optimal transportation theory, specifically Beckmann’s 1952 formulation.
• To prove strong duality between the primal auction problem and a dual Beckmann-type optimal transport problem, enabling new solution techniques.
• To demonstrate the existence of optimal mechanisms and optimal vector fields (generalizing ironed virtual valuations) using majorization and functional analysis.

Proposed method

• Formulate the revenue-maximization problem in Bayesian multi-item, multi-bidder auctions as a primal optimization problem over interim allocation rules.
• Derive the dual problem as Beckmann’s continuous optimal transportation problem, where the flow field $ c(x) $ represents generalized ironed virtual valuations.
• Use the divergence constraint $ \mathrm{div}[\rho \cdot c] = \pi_p - \pi_c $ to model supply-demand imbalance across bidder type space.
• Apply the cost function $ \Phi(c(x)) $ to represent transportation cost, with $ \rho \equiv 1 $ in the main formulation.
• Leverage the equivalence between Beckmann’s problem and a Monge-Kantorovich-type problem with measures on curves, enabling a dynamic, Lagrangian interpretation.
• Use Dacorogna–Moser interpolation to construct flows and establish existence of solutions via variational methods and compactness arguments.

Experimental results

Research questions

• RQ1 Can the complex problem of revenue-maximizing multi-item, multi-bidder auctions be reduced to a known problem in optimal transport?
• RQ2 Does strong duality hold between the auction design problem and Beckmann’s optimal transportation problem?
• RQ3 Can the existence of optimal mechanisms be established using tools from majorization and optimal transport theory?
• RQ4 Can complementary slackness conditions derived from duality be used to verify candidate solutions in specific cases?
• RQ5 What is the structural role of the vector field $ c(x) $ in generalizing ironed virtual valuations to multi-item settings?

Key findings

• Strong duality holds between the revenue-maximizing auction problem and Beckmann’s optimal transportation problem, with the optimal revenue equal to the optimal value of the dual transport problem.
• The optimal vector field $ c(x) $ in Beckmann’s problem generalizes the concept of ironed virtual valuations to multi-item settings, representing the marginal revenue allocation across types.
• Existence of an optimal auction and an optimal flow field $ c $ is proven using compactness and convexity arguments, relying on Hart and Reny’s characterization of feasible interim mechanisms.
• The guess-and-verify approach is validated: given a candidate mechanism, complementary slackness allows construction of a dual solution that acts as a certificate of optimality.
• The framework recovers known results, such as Jehiel et al. (2007)’s proof that separate sales of independent items are never optimal under continuous distributions.
• In specific cases—such as one bidder with two i.i.d. uniform items—the method recovers the optimal mechanism found by Manelli and Vincent (2006), confirming the approach’s practical utility.