Resource allocation and project selection: Control of r & d under dynamic process of data improvement

This paper deals with the problem of optimal control of R&D. The fundamental aspects of this problem are viewed as: uncertainty about economic parameters of new technology (or new activity, e.g. a search for new oil fields) explored by a group of parallel projects attempting to make advances in this technology as well as improve economic data; allocation of investments among rival projects under a given total budget; timing of the ultimate selection of new technology (activity) according to the most successful project. The process of data improvement (PDI) is described as a (random) process of reducing the intervals of uncertainty, the rate of data improvement being controlled, in effect, by the investment policy. A stationary dynamic model with infinite horizon of planning and with the criterion of total discounted costs is developed. The model is carefully investigated from the economic point of view, and some new effects are discovered. It is demonstrated that uncertainty may be the source of additional gain; the bigger the initial uncertainty, the bigger the gain may be under optimal control. This effect is based on PDI, the economic background of which is presented as multiplication of the expected gain over time, which is typical for the classical macromodels of economics.Optimal continuous control is constructed at every time moment, and answers to two main questions are given:(1) either to prolong R & D or to make ultimate selection of one project, (2) how to allocate resources among the rival projects in the case of R&D prolongation. The last question is answered on the basis of profitability indices that relate efficiencies of PDI for particular projects to the costs of information.