The Accuracy of Simple Enlisted Force Forecasts

This Note presents an analysis of the historical accuracy with which enlisted force manpower strengths can be forecast using simple and widely used modeling assumptions. The accuracy of one-, three-, and five-year forecasts is presented for enlisted personnel groups from all four military services for fiscal years 1971 through 1980. Estimating the accuracy of forecasts and analyzing the pattern of errors allow an assessment of the need for more sophisticated techniques, help in developing a strategy for disaggregating enlisted groups when forecasting, and form a basis for the design of an improved enlisted forecasting system. The author finds that the models tested provide reasonably accurate short-term forecasts of the level and structure of enlisted personnel strength. However, long-term forecasts show very large errors for certain enlisted groups. The error pattern, which is stable across the services, shows a distinct structure when estimated by year of service.