The Robust Separation Projection Method for Predicting Monthly Losses of Air Force Enlisted Personnel

The Short-Term Aggregate Inventory Projection Model (SAM) is the component of the Enlisted Force Management System that makes monthly projections (for the rest of the current fiscal year) of the "aggregate" force (the total enlisted force across all specialties). Module 1 (the separation projection module) of SAM is designed to forecast the monthly loss and reenlistment flows that would occur in the absence of early release and early reenlistment programs. This Note describes the robust separation projection method of predicting monthly losses and reenlistments. The robust method uses data on past losses and reenlistments to estimate separation rates for a model that predicts separation flows one month at a time for each of a mutually exclusive set of about 1,000 cohorts. After these flows are predicted for a projection month, the inventory is updated and the models are applied to the updated inventories to predict the flows for the following month. This process is repeated until the inventory for the last month of the fiscal year is projected. The authors describe the required data, the robust model, and present technical specifications as well as a test and evaluation of the model.