The preceding section described models for evaluating projects in isolation. However, the problem faced by most companies is selecting a portfolio of projects. Portfolio selection models have been developed to address this problem. They fall into four major categories:
Mathematic programming models use quantitative optimization techniques to identify optimal portfolios.
Portfolio diagrams facilitate portfolio selection decisions by displaying the distribution of available projects with respect to a set of criteria.
Strategic frameworks impose a high-level structure to guide resource allocation decisions.
Procedural approaches use a variety of approaches, but stress the value of interaction over the content of the approach.
Mathematical programming models structure the portfolio selection problem as an objective function to be maximized subject to a set of constraints. For their inputs, these models generally rely on measures of project value generated by economic value models or scoring models. For example, a very basic model might define the objective function as the sum of the economic values of the selected projects and use R&D budget as a constraint. Hundreds of variants, some quite sophisticated, have been proposed to deal with the complexities of the portfolio selection problem (Schmidt and Freeland, 1992). Families of approaches that appear frequently in the literature include linear programming, nonlinear programming, ...