In this paper, we establish sufficient conditions on the domain of preferences and agents' behavior in order to characterize the existence of stable assignments in many-to-one matching problems with externalities. The set of stable matchings depends on what agents believe other agents will do if...