Key points are not available for this paper at this time.
In markets with avoidable fixed costs, the core may be empty. I argue that a reason for collusion is to impose an equilibrium where none exists. I set up a framework for deriving testable implications from the model and compare those implications with those of a cartel model. I compare the two models empirically with data from liner shipping conferences, legal collusive agreements that have been in force for over a century, and I conclude that the evidence supports the theory of the core.
William Sjostrom (Sun,) studied this question.