Berzack et al. prove optimal auctions differ under buyer constraints

A new paper from Berzack et al. tackles auction design when buyers operate under real-world constraints like “no overbidding” rules or an inability to tolerate temporary debt. These constraints fundamentally break the classical tools of auction theory—the revelation principle, the envelope theorem, and Myerson’s lemma—allowing a much wider range of implementable mechanisms. The authors focus on monotone constraints that limit how buyers can bid or spend, rather than capping their total budget. Their key finding is a clear separation between revenue-aligned objectives (e.g., seller revenue, social welfare, any linear mix) and consumer-aligned objectives (e.g., consumer surplus). For revenue-aligned goals, they prove that despite the expanded design space, the classic Myerson-style auctions remain optimal—essentially a robustness result. For consumer-aligned objectives, the seller can actually exploit the buyers’ strategic limitations to design auctions that strictly outperform any classical incentive-compatible mechanism.

The paper then dives into the specific case of buyers who cannot go into debt, even temporarily, designing the optimal deterministic auction for a wide class of instances. This work provides a systematic theoretical foundation for an area that has mostly been tackled ad-hoc. It shows that a rigorous, measure-theoretic approach can still yield general insights even when the usual assumptions fall apart. The practical implications span any multi-unit auction setting with constrained bidders, from ad exchanges to spectrum auctions to decentralized finance, where participants often operate under hard constraints that Myerson’s framework doesn’t capture.