Title: Distributionally Robust Competitive Newsvendors
Game-theoretic models have been successfully applied to newsvendor problems to study retailers' strategic decisions, typically assuming players have complete knowledge of the demand distribution. However, these models do not apply to more practical settings where the distribution is unknown and retailers rely only on limited information to make pessimistic decisions against distributional ambiguity. This ambiguity-averse behavior introduces significant complexity in analyzing potential equilibria as retailers may perceive different worst-case distributions–a challenge that remains underexplored in the literature. To address this gap, we build on the competitive newsvendor model of Lippman and McCardle (1997) in the context of a duopoly with distributionally robust retailers, that are provided only the mean and variance of the distribution and adopt the maxmin expected utility criterion. Employing distributionally robust optimization techniques, we analytically characterize the unique Nash equilibrium in closed form, offering new managerial insights. Our equilibrium is also shown to inherit several key properties of the equilibrium under the expected utility criterion with perfect distributional knowledge that has been studied in the literature. Overall, this work broadens the understanding of competitive newsvendor models with insufficient distributional information and suggests several potential avenues for future research.