Incentives in a Divisionalized Firm

This paper examines the problems of coordinating and controlling divisions of a large firm. Divisions are typically interdependent and the corporate headquarters plays an important role in coordinating such decisions as pricing, allocation of funds to the divisions, and determining the level of corporate-wide research and development. One important problem is how to provide division managers with incentives both to act in congruence with the goals of the overall firm and to transmit accurate information to the corporate center so that the center's decisions may be enhanced. The analysis proceeds by assuming that each division manager possess perfect knowledge of his own division's technology and the market conditions it faces. Each division's profits, however, depend upon certain coordinating decisions taken by the center. The center is assumed to be at least partially ignorant about the operations of the divisions and to rely upon information from the divisions in making coordinating decisions. A control structure for a divisionalized firm is defined to be a set of rules for evaluating division managers and a decision-rule for central coordinating decisions. Division managers are assumed to act to maximize their evaluation measures so that managerial shirking is not considered. In this setting where perfect information is available, problems of risk aversion and risk sharing do not arise, and a minimum test for optimality of a control structure is given. Specifically, when each division manager seeks to maximize his own evaluation measure and when the center uses its decision-rule for coordination, overall firm profits should be maximized. A control structure is proposed in which the center acts as if it is receiving "truthful" information from the division managers. This naive decision rule is justified by the accompanying evaluation measures that provide division managers with incentives to communicate in an honest manner. With the proposed evaluation measures, each division manager is rewarded on the basis of his division's contribution to overall profits. If the division's message to the center has no impact on the center's coordinating decisions, then the manager is evaluated on the basis of his own division's realized profits. If the division's message changes the coordinating decision, then the manager is evaluated on the basis of his division's realized profits less the reported impact of the change in coordinating decisions (resulting from his message) on the profits of the other divisions. It is shown that with this evaluation measure, a division manager is best off communicating truthfully, regardless of the messages of all others. The proposed control structure, therefore, provides an alternative to traditional profit-sharing. An application of the control structure to a resource allocation problem is also given.