Classical mechanism design relies on a clean, elegant assumption: agents are purely self-interested. In this traditional view, bidders care entirely about maximizing their own surplus and are completely indifferent to the payoffs of their competitors. However, real-world strategic environments frequently violate this assumption. As noted by Sharma et al. and Waniek et al., bidders often exhibit competitive social preferences—most notably spite, a desire to minimize an opponent's welfare even at a personal financial cost.
When spite enters the equation, standard economic models break down. For instance, in the classic "Dollar Auction"—a game-theoretic model used to study conflict escalation—the introduction of spiteful players shifts the equilibrium from rational withdrawal to total resource depletion. As shown by Waniek et al., when agents value an opponent's loss as much as their own gain, their behavior undergoes a fundamental shift toward total escalation, prioritizing mutual destruction over profit.
Understanding how spite alters bidding behavior is not just a psychological curiosity; it fundamentally reshapes how we must design auctions to maximize revenue and efficiency.
How Spite Breaks Revenue Equivalence
One of the foundational pillars of auction theory is the Revenue Equivalence Theorem. It states that under standard conditions, first-price sealed-bid auctions (FPAs) and second-price auctions (SPAs) yield the exact same expected revenue for the seller. Spite completely shatters this equivalence.
To understand why, we must look at how a spiteful bidder behaves under different pricing rules. In an FPA, the winner pays exactly what they bid. If a spiteful bidder wants to reduce the winning rival's surplus, they must bid closer to their true valuation, as analyzed by Brandt et al.. However, they are strictly constrained: if they bid too high, they risk winning the auction themselves at an unprofitable price.
In an SPA, however, the winner pays the second-highest bid. This decoupling of what you bid from what you pay creates a powerful mechanism for a losing spiteful bidder: price-pushing. A spiteful loser can deliberately overbid past their own private valuation to artificially inflate the price the winner must pay. Because the loser only pays if they accidentally cross the winner's threshold, they can aggressively push the price upward.
Brandt et al. and Chen et al. demonstrate that because the second-price mechanism allows losing bidders to directly manipulate the winner's payment, SPAs generate strictly higher expected revenue than FPAs when spiteful dynamics dominate.
The Geometry of Optimal Auctions with Spite
If bidders naturally want to hurt each other, how should an auctioneer structure the rules to extract the maximum possible revenue? Tang et al. generalized Myerson's landmark optimal auction framework to accommodate spiteful agents, revealing a highly counter-intuitive structural insight: the revenue-optimal mechanism involves subsidizing losing bidders.
Why would a profit-maximizing seller ever pay the losers? In a spiteful environment, a losing bidder experiences negative utility simply from watching a rival succeed. By providing a targeted subsidy to the losers, the auctioneer mitigates this negative utility. This reduction in the loser's spite-induced loss changes the strategic balance of the entire game. Because the losers are cushioned, the auctioneer can extract a significantly higher payment from the winning bidder, who is willing to pay a premium to maintain their relative utility gap over the field.
Engineering Spite via Coupon Design
What if the bidders in an auction are not inherently spiteful? Can an auctioneer still leverage the mathematics of price-pushing to increase revenue?
The modern breakthrough in this space bridges the gap between psychology and platform engineering through coupon design in digital advertising systems. In a series of papers, Tang et al. and Tang et al. demonstrated that giving optional subsidies (coupons) to lower-value participants in an SPA or VCG auction artificially replicates the price-pushing dynamics of spite.
When an advertising platform offers a coupon worth $c$ to a bidder, that bidder's dominant strategy shifts from bidding their true value $v$ to bidding exactly $v + c$.
Standard SPA:
[Bidder A: v_A] ---> Wins, pays Bidder B's bid (v_B)
[Bidder B: v_B]
SPA with Loser Coupons:
[Bidder A: v_A] ---> Wins, pays Bidder B's inflated bid (v_B + c)
[Bidder B: v_B + c] ---> Loses, but forces Winner to pay more externality cost
Crucially, if this subsidized bidder loses, their inflated bid directly increases the market-clearing price paid by the high-value winner. By strategically distributing these coupons to lower-value participants, the platform increases the externality costs imposed on the winner.
This mechanism acts as a highly effective form of price discrimination without the downside of rigid reserve prices, which often discourage participation. Finally, as Liu et al. show, by mathematically incorporating the cost and impact of these subsidies directly into the payment rules, platforms can actively manipulate item relevance and boost revenue while preserving the auction's incentive compatibility—ensuring advertisers still find it optimal to bid truthfully relative to their subsidized valuations.
Key Foundational Work
Spiteful Bidding in Sealed-Bid Auctions (2007) — contributed 7 concept elements to this direction.
Bayesian Auctions with Friends and Foes (2009) — contributed 5 concept elements to this direction.