October 2
@
12:00 PM
–
1:00 PM
UW-ISyE looks forward to welcoming Bhaskar Ray Chaudhury from Illinois State University.
Beyond Nash in Non-Rival Markets: Response-Aware Rational Equilibria
We study pricing equilibria in markets with budget-constrained buyers and non-rival goods: selling a good to one buyer does not diminish its availability to others. Our motivating application is the growing market for data to train and improve AI systems, illustrated by partnerships between Google and Reddit, OpenAI and News Corp, and OpenAI and the Financial Times.
What should stable prices mean in these markets? We show that pure Nash equilibrium typically does not exist: even with two sellers and equal buyer budgets, its probability of existence tends to zero as the number of buyers grows when linear valuation coefficients are drawn independently from a uniform distribution. This motivates revisiting the counterfactual underlying Nash stability. A price cut may appear profitable when competitors’ prices are held fixed, yet lose its appeal when competitors can respond—for example, by matching the new price.
We introduce response-aware rational equilibrium (RARE). A seller evaluates a deviation against responses that leave each competitor at least as well off as immediately after that deviation. The seller proceeds only if the deviation remains profitable against every such individually rational response. RARE requires that no seller have such a deviation.
For strictly increasing, strictly concave buyer utilities, we establish existence with two sellers and arbitrary buyer budgets, and with any number of sellers when buyer budgets are equal. We also introduce monotone revenue ascent, deterministic dynamics in which each update increases the moving seller’s revenue without reducing anyone else’s. These dynamics reach approximate RARE and converge to exact equilibrium prices under an additional positive-demand condition satisfied by positive-weight CES utilities.
Starting from a pricing question specific to data markets, these results suggest a broader direction: understanding stable behavior among agents that anticipate one another’s responses. We conclude by discussing the possibilities—and open questions—for such an equilibrium perspective in markets populated by AI agents.
Bio: Bhaskar Ray Chaudhury is an Assistant Professor in the Department of Industrial and Enterprise Systems Engineering at the University of Illinois Urbana–Champaign, with an affiliate appointment in the Siebel School of Computing and Data Science. His research lies at the intersection of theoretical computer science, economics, and machine learning, with a focus on fair division, equilibrium computation, and the foundations of data economics. He has received several honors, including the NSF Career Award, best paper awards at the ACM Conference on Economics and Computation, and an oral presentation at ICML.