CSIO working paper / Northwestern University, Center for the Study of Industrial Organization 0035
In this paper I present a theory of the boundary of the firm that accounts for some important characteristics of real-world multidivisional firms: Operative decisions are in the hands of middle managers who are rewarded with incentive contracts based on the performance of their units; Managers' decisions are subject to approval and intervention by the top management of the firm; and managers are better informed regarding the affairs of their divisions than their superiors in the firm's hierarchy. In this setup, the integration of a producer of an intermediate input and its buyer as separate divisions within a single firm is unambiguously desirable, as long as the choice of trading partners can be credibly delegated to the divisions' managers. I show that this is satisfied not only under the assumption of full commitment by the general office of the firm, but also interestingly, if it has no commitment power at all. At the time of trade, the uninformed general office prefers to delegate the choice of trading partners to the divisions whose decision is ex-post optimal. An explanation of the boundaries of the firm emerges only if we assume that the general office retains some limited commitment power. The general office may then mandate internal trade in order to encourage the divisions to specialize towards one another before the trade. In the context examined, I show that the general office faces a 'time-consistency' problem. It tends to mandate internal trades in more instances than would have been optimal with full commitment, adversely affecting the levels of investment taken by the divisions' managers. Whenever such inconsistency arises, it may be optimal to have the trade conducted between independent, non-integrated parties.