Retention Analysis Modeling for the Acquisition Workforce

Abstract

The Department of Defense (DoD) Acquisition Workforce (AWF) Strategic Plan FY 2016 FY 2021 describes the efforts and activities that, after more than twenty years of almost uninterrupted contraction in the AWF, the DoD has undertaken since 2010 to strategically restore and reshape it. Combined with recommendations from the Section 809 Panel, the mandate of the AWF leadership is to reform and reshape the workforce to improve acquisition and delivery of world-class warfighting capabilities for the U.S. military. This technical research report is the first in a proposed series of three linked studies to provide a cutting-edge modeling and simulation tool that leverages the increase in availability of AWF data and the exponential increases in computing power in the last three decades. The tool is designed to support the continuing mission of recruiting, hiring, and developing a diverse, agile, highly qualified, and motivated workforce of acquisition professionals. This report accomplishes two foundational tasks that will support the long-term effort to build, estimate, and simulate a custom-designed dynamic retention model (DRM) for the AWF that uses the principals of a powerful mathematical/econometric technique called dynamic programming. The first step in building the DRM is understanding the observed behavior of the AWF employees by studying the available data on AWF. The summary statistics, long-run trend analysis, and survival analysis via a Cox proportional hazard model of a subset of the AWF data suggest that the career paths of those employees do not differ substantially by gender or ethnicity. However, we find that there are some features, such as education level and prior military experience that have profound effects on the retention behavior of AWF workers. The correct identification of those factors is fundamental for customizing the DRM for the AWF. Subsequent to the empirical analysis, we proceed to develop the proof-of-concept DRM that could eventually assist the AWF leadership in reshaping the entire force as desired. This model takes a complex, multi-period problem and breaks down the solution into a series of much simpler, one-period sub-problems in a recursive manner. We use this technique to model the lifetime labor market decisions of an acquisition worker. This single-worker model is then aggregated up to simulate the entire AWF to examine the effects of various policies to changes the size and shape of the workforce. We describe how the model simulates a rational employee who makes stay or leave decisions over his/her entire life. We also show how that employee would react to different monetary incentives such as salary raises or bonuses at certain points in his/her career. Overall, we find that the proposed model, though simple at this stage, already replicates the main patterns of the observed AWF employee retention behavior. Complementing this analysis, we study how the bathtub problem would evolve over time with no active and forward-looking intervention by the AWF leadership. We finalize the report by explaining how this multiyear research effort can be extended over time to make the DRM custom fit the specific manpower needs of the AWF, including a proposal to develop a custom web app the leadership could use to run simulations in real-time to assist in decision making processes.