Conference Agenda

Procurement managers have various payment schemes available when awarding contracts to suppliers. Among these, cash auctions are the most commonly used type of procurement auction. In a cash auction, the contract is awarded to the supplier who bids the lowest cash amount for the project. This method is straightforward and easily comparable across different bids, making it a preferred choice in many scenarios.

Another option available to suppliers is to propose a profit-sharing model in what is known as a share auction. Here, instead of a fixed cash payment, suppliers request a percentage of the profits generated from the project they are involved in.

In our project, we provide a theoretical and experimental analysis of share and cash auctions. The theoretical analysis shows that share auctions lead to better buyer outcomes, i.e., lower prices. Furthermore, it shows that the buyer should specify a high project value in the contract to foster competition.

Our experimental data substantiates the theoretical predictions. We observe that cash auctions yield lower prices than cash auctions. This effect is stronger, when the buyer specifies a high project in the contract. Even though share auctions may appear more complex, we do not find any difference in efficiency.

The transition of display ad exchanges from second-price to first-price auctions has raised questions about its impact on revenue, but evaluating these changes empirically proves challenging. Automated bidding agents play a significant role in this transition, often employing dynamic strategies that evolve through exploration and exploitation rather than using the static game-theoretical equilibrium strategies. Thus revenue equivalence between first- and second-price auctions might not hold.

Research on algorithmic collusion in display ad auctions found that first-price auctions can induce Q-learning agents to tacitly collude below the Nash equilibrium, which leads to lower revenue compared to the second-price auction.

Our analysis explores widespread online learning algorithms' convergence behavior in both complete and incomplete information models but does not find systematic deviance from equilibrium behavior. Convergence for Q-learning depends on hyperparameters and initializations, and algorithmic collusion also vanishes when Q-learning agents are competing against other learning algorithms.

The objective of bidding agents in these auctions is typically to maximize return-on-investment or return-on-spend, but not necessarily payoff maximization. The revenue comparison under such utility functions is an open question.

Analytical derivations of equilibrium are challenging, but learning algorithms allow us to approximate equilibria and predict the outcome when agents have such non-quasilinear objectives.

Our analysis shows that if learning agents aim to optimize such objectives rather than payoff, then the second-price auction achieves higher expected revenue compared to the first-price auction. Understanding the intricate interplay of auction rules, learning algorithms, and utility models is crucial in the ever-evolving world of advertising markets.