Multiplicative Pacing Equilibria in Auction Markets

Budgets play a significant role in real-world sequential auction markets such as those implemented by Internet companies. To maximize the value provided to auction participants, spending is smoothed across auctions so budgets are used for the best opportunities. This paper considers a smoothing procedure that relies on pacing multipliers: for each bidder, the platform applies a factor between 0 and 1 that uniformly scales the bids across all auctions. Looking at this process as a game between all bidders, we introduce the notion of pacing equilibrium, and prove that they are always guaranteed to exist. We demonstrate through examples that a market can have multiple pacing equilibria with large variations in several natural objectives. We go on to show that computing either a social-welfare-maximizing or a revenue-maximizing pacing equilibrium is NP-hard. Finally, we develop a mixed-integer program whose feasible solutions coincide with pacing equilibria, and show that it can be used to find equilibria that optimize several interesting objectives. Using our mixed-integer program, we perform numerical simulations on synthetic auction markets that provide evidence that, in spite of the possibility of equilibrium multiplicity, it occurs very rarely across several families of random instances. We also show that solutions from the mixed-integer program can be used to improve the outcomes achieved in a more realistic adaptive pacing setting.