A dynamic model of some multistage aspects of research and development portfolios

Many models of research and development (R&D) projects do not consider the intermediate outcomes and decisions that may be foreseen to arise during their evolution through the technical and commercial stages. Consideration of the sequential aspects of allocating scarce resources to a set of projects may make a great difference to the solution of the problem in terms of the optimum subset of projects to work on in the immediate future. The solution of this problem is important to R&D management. An approach is described based on the application of a stochastic linear-programming formulation to a portfolio of projects, each planned using a form of the decision-tree structure called a `project tree.´ A series of elementary examples are presented as a means of gaining insight into the method. The general formulation is then presented in detail and applied to a small problem.