Aviation MOSterpiece: Markov-Based Forecasting of Aviation MOS Inventories During Platform Divestment

Abstract

Platform divestment introduces complex personnel transitions that challenge traditional manpower forecasting models developed for steady-state environments. This thesis examines how enlisted aviation manpower inventories evolve during the sundown of the Marine Corps F/A-18 community. Using personnel data from the Marine Corps Total Force System covering Fiscal Year (FY) 2009–FY2025, this study develops and validates Markov models to forecast transitions across five F/A-18 maintenance military occupational specialties (MOSs). The analysis evaluates two model structures: an annual fiscal year–to–fiscal year model and a six-month transition model, each implemented with MOS-level and MOS-by-rank state definitions. The analysis uses rolling validation windows to estimate transition matrices and assess predictive accuracy using Weighted Absolute Percentage Error (WAPE) and aggregate bias metrics. Results show that a six-month transition model incorporating rank structure and using a longer historical training window provided the best predictive performance. Forward projections through FY34 indicate that personnel inventories decline more slowly than institutional planning targets assume during the F/A-18 sundown, producing temporary overages in most MOS and grade combinations while persistent shortfalls emerge in the E-5 cohort. These findings demonstrate that Markov transition modeling can improve visibility into manpower inventory dynamics during platform divestment.