Optimal resource allocation in an engineering design team with asymmetric information

Engineering design is the process by which systems and products are built to satisfy the needs of customers in an efficient and reliable manner. I examine the issue of allocating resources to manage uncertainty in engineering design in light of potential strategic behavior by the participants in the design process. This thesis addresses the problem of developing mechanisms that will yield efficient resource allocations in a design team and induce optimal effort levels from engineers given informational asymmetries between the engineers and the manager. First, I address, through a mixed-integer, non-linear programming formulation, the perfect-information problem in which the manager knows everything that her subsystem engineers know about the effect of resources on the performance of their subsystem. The solution to this problem provides the basis to which subsequent mechanisms are compared. I then address the asymmetric-information problem in which the engineers have private information about the impact of resource allocations and about their ability and effort. I use an approach that achieves truthful reporting of private values as an ex-post Nash equilibrium and selection of the manager's preferred effort levels by risk-neutral engineers. This mechanism generalizes a Vickrey-Clarke-Groves payment scheme to incorporate interdependent subsystem performances and privately known agent ability levels and unobservable, but costly, effort. Three key assumptions are made: the engineers (agents) are risk-neutral, effort and resource are not substitutes, and the admitted types of subsystem performance-versus-resource allocation functions are limited. Numerical examples are used to illustrate the effects of these assumptions, and extensions are presented that overcome the limitation on the admitted types of performance functions by including intrinsic motivation on the part of the engineers and a repeated contracting game. This dissertation highlights the difficulty of achieving optimal resource allocations in an engineering design team with informational asymmetries between the engineers and the manager. In these cases, traditional optimization models that assume that the manager has perfect information will yield sub-optimal allocations. The methods developed in this dissertation can, in many cases, overcome these difficulties to yield the best possible resource allocation.