We examine how owners of productive resources (e.g., public enterprises or financial capital) optimally allocate their resources among wealth-constrained operators of unknown ability. Optimal allocations exhibit: (1) shared enterprise profit—the resource owner always shares the operator's profit; (2) dispersed enterprise ownership—resources are widely distributed among operators of varying ability; (3) limited benefits of competition—the owner may not benefit from increased competition for the resource; and, sometimes, (4) diluted incentives for the most capable—more capable operators receive smaller shares of the returns they generate. Implications for privatizations and venture capital arrangements are explored.

We extend the standard procurement model to examine how an agent is optimally induced to acquire valuable planning information before he choose an unobservable level of cost‐reducing effort. Concerns about information acquisition cause important changes in standard incentive contracts. Reward structures with extreme financial payoffs arise, and super‐high‐powered contracts are coupled with contracts that entail pronounced cost sharing. However, if the principal can assign the planning and production tasks to two different agents, then all contracting distortions disappear and, except for forgone economies of scope, the principal achieves her most preferred outcome.

We extend William Rogerson's (2003) intriguing analysis of simple procurement contracts to settings where the supplier's innate production cost is not necessarily distributed uniformly. Although the simple contract that Rogerson analyzes performs remarkably well when the smaller cost realizations are relatively likely, it can perform poorly when the larger cost realizations are relatively likely. We show that in all settings under consideration, a simple pair of contracts – one that involves linear cost sharing and one that involves full cost reimbursement – can always secure more than 73 percent of the gain achieved with a fully optimal contract.