Performance Utility and Optimal Job Assignment

The Army is currently conducting a long-term research project to improve the ability to predict enlisted soldier performance. Also, a new Enlisted Personnel Allocation System (EPAS) is being developed to enhance the effectiveness with which performance information is used to match recruits with Army MOS. To maximize the payoff to the Army of the available manpower pool, the assignment system must consider not only the level at which a recruit will perform in different jobs, but also the utility to the Army of each potential combination of performance level and job. This paper describes a series of simulations undertaken to assess how different models of performance utility will affect the manpower distributions produced by an optimal assignment system. The results show that (a) the use of performance utility in optimal assignment can produce performance gains equal to those achieved when utility is ignored, while simultaneously providing a more balanced distribution of performance across jobs; and (b) specifications in which performance value depends on the mean level of performance seem to produce better results than those that assume constant marginal value. (RWJ)