A study of Navy first-term reenlistments

Abstract

It occasionally happens in economic analyses that the correctly specified model contains variables for which no observed data has been collected. When the data in a linear regression model are crosssectional it is possible, under certain conditions on the nature of the variables, to estimate the independent effects of a specific set of explanatory variables on the dependent variable. A procedure for doing this is presented. A commonly used model of reenlistment behavior, for which the data base is cross-sectional, satisfies the requisite conditions. This permits the estimation of the independent effect of the military wage on reenlistment rate, as an illustration of the proposed procedure.