Research for forecasting

An understanding of the relationship between changes in compensation and changes in reenlistment behavior is crucial to shaping the force. A common measure of this relationship is the pay elasticity of reenlistment, the percentage change in reenlistment associated with a 1-percent increase in pay. The literature on Navy enlisted personnel has produced widely varying estimates of this relationship; with changes in both analytic approach and in the Sailors being studied, the reasons for these differences are unclear. Our analysis suggests that most of the variation in these estimates can be explained by the use of different analytic models. Different specifications yield different estimates that span the range found in previous research. Because each specification uses the same data, these different estimates reflect differences in the degree to which these models attribute differences to pay, not differences in the behavior of enlisted personnel. In contrast, there is little variation in the pay elasticity over time; the only significant changes occur during the drawdown. We choose a preferred specification by examining its ability to accurately predict reenlistment behavior. For both in-sample and out-of-sample predictions of reenlistment, our baseline model, with a pay elasticity of 1.5, provides the best fit of the data.