We develop a theory of firm scope in which integrating two firms into one facilitates the allocation of resources, but leads to weaker incentives for effort, compared with non-integration. Our theory makes minimal assumptions about the underlying agency problem. Moreover, the benefits and costs of integration originate from the same problem – to allocate resources efficiently, the integrated firm's top management must obtain information about the possible use of resources from division managers. The division managers' job is to create profitable investment projects. Giving the managers incentives to do so biases them endogenously towards their own divisions, and gives them a motive to overstate the quality of their projects in order to receive more resources. We show that paying managers based on firm performance in addition to individual performance can establish truthful upward communication, but creates a free-rider problem and raises the cost of inducing effort. This effect exists even though with perfect information, centralized resource allocation would improve the managers' incentives. The resulting tradeoff between a better use of resources and diminished incentives for effort determines whether integration or non-integration is optimal. Our theory thus provides a simple answer to Williamson's “selective-intervention” puzzle concerning the limits of firm size and scope. In addition, we provide an incentive-based argument for the prevalence of hierarchically structured firms in which higher-level managers coordinate the actions of lower-level managers.