Empirical Biases and Some Remedies in Estimating the Effects of Selective Reenlistment Bonuses on Reenlistment Rates

Researchers have, for decades, been attempting to estimate the effects of Selective Reenlistment Bonuses (SRBs) on the probability of reenlistment for the military services. SRBs are targeted to specific military occupations for which reenlistment rates are lower (or expected to be lower) than what is needed. This article first identifies four primary sources of biases affecting these models: reverse causality from supply shifts (a negative bias), the endogeneity of the decision point causing coded SRBs to be higher for reenlisters than leavers (a positive bias), measurement error (a likely negative bias), and excess supply preventing the full effect of an SRB change to materialize (a positive or negative bias). The report proceeds to develop a model that attempts to address the first two biases. With U.S. Navy data from FY2001-FY2008, I examine the extent to which these two biases are affecting the estimated SRB effects. Despite these corrections, the difficulty of addressing the other biases calls into doubt studies that examine the effects of retention bonuses or even the effects of the structure of military pay in general.     

Ridge regression analysis of the defence‐growth tradeoff in the United States

An examination of past analyses suggests that today's sailors may be less responsive to compensation changes than previous generations. Such a change could make recruiting and retaining high‐quality sailors more difficult. However, variation in researchers' decisions over time may simply have created the appearance of such a change. Our results suggest there is little variation in the pay elasticity over time. In contrast, the different reduced‐form models we use to measure this relationship can explain most of the variation in the literature. Therefore, the evidence suggests that while sailors may have changed over time, their response to compensation has not.