The Effect of Military Service of Civilian Labor Market Experience

Abstract

In this paper, the random assignment of the risk of induction generated by the draft lottery is used to estimate the effect of military service on civilian wages, earnings and weeks worked- Data from the National Longitudinal Survey of Young Men in 1981 offer no conclusive evidence of an effect on earnings or weeks worked- However, marginally significant negative wage effects are found for white veterans, while positive wage effects are found for black veterans- Conventional ordinary least squares estimates which do not exploit the randomization of the draft lottery fail to identify these effects, suggesting the presence of selection bias in conventional estimates, Finally, an attempt is made to gauge whether instrumental variables estimates which do not exploit the lottery generate similar inferences regarding the effects of military service- Two sets of conventionally available instruments result in estimates which differ greatly from those constructed using lottery based instruments- However, both the least variance ratio and the generalized method of moments tests of over- identifying restrictions provide some help in isolating the most misleading conventional specifications.