Allocating Resources with Nonmonotonic Returns

The literature on resource allocation under adverse selection has focused on models in which the resource being allocated is such that the privately informed agent always prefers more of it to less. We analyze a firm’s optimal resource allocation mechanism when this assumption does not hold and show that the resulting mechanism has a number of novel characteristics. For example, first best may be achievable even with nontrivial information asymmetry; when first best cannot be achieved, it is always optimal to overinvest relative to first best, and the most efficient agent may not earn rents, even when a less efficient agent does. This paper was accepted by Suraj Srinivasan, accounting.