The U.S. Army seeks to understand how it can better select, shape, train, and retain the force it wants. Using vast personnel records and machine learning techniques for survival analysis, IDA developed the Retention Prediction Model – Army (RPM-A). The RPM-A produces high-fidelity retention forecasts for a wide range of forecast horizons, enabling the Army to anticipate which individuals are likely to exit service. We use this model to examine the magnitude of expected economic rents the Army could have anticipated when it implemented its 2007 Menu of Incentives Program (MOIP) for retaining Army Captains.