Resource allocation to interrelated risky projects using a multiattribute utility function

Two major contributions are made to the problem of resource allocation to indivisible projects with uncertain outcomes: the concepts of preferential independence and utility independence for the space of discrete projects are defined, and a multiattribute utility theory is linked with mathematical programming for allocation. The utility function, as shown by Keeney, can be either multiplicative or additive. An overall procedure for structuring the model is outlined, a branch-and-bound algorithm for solving the multiplicative model is developed, and a numerical example is presented.