Learning the Stackelberg Equilibrium in a Newsvendor Game

Abstract

We study a repeated newsvendor game between a supplier and a retailer who want to maximize their respective profits without full knowledge of the problem parameters. After characterizing the uniqueness of the Stackelberg equilibrium of the stage game with complete information, we show that even with partial knowledge of the joint distribution of demand and production cost, natural learning dynamics guarantee convergence of the supplier and retailer's joint strategy profile to the Stackelberg equilibrium of the stage game. We also prove finite-time bounds on the supplier's regret and asymptotic bounds on the retailer's regret, where the specific rates depend on the type of knowledge preliminarily available to the players. Finally, we empirically confirm our theoretical findings on synthetic data.