Robust newsvendor games with ambiguity in demand distributions

In classical newsvendor games, vendors collaborate to serve their aggregate demand whose joint distribution is assumed known with certainty. We investigate a new class of newsvendor games with ambiguity in the joint demand distributions, which is represented by a Fr\backslash'echet class of distributions with some, possibly overlapping, marginal information. To model this new class of games, we use ideas from distributionally robust optimization to handle distributional ambiguity and study the robust newsvendor games. We provide conditions for the existence of core solutions of these games using the structural analysis of the worst-case joint demand distributions of the corresponding distributionally robust newsvendor optimization problem.