Assessing Changes in Wage Gaps: A New Approach

Little work has been done on directly estimating differences in pay gaps. Studies that estimate pay di erentials generally compare them across different sub-samples or rely on decompositions that are based on the assumption of independent errors across samples. Both methods contain serious drawbacks that we overcome by proposing an extension of the Oaxaca-Blinder decomposition. Our proposed method overcomes both the index number and the indeterminacy problem of standard Oaxaca-Blinder decompositions. In addition, like the standard decomposition, our proposed approach can be extended beyond the mean by using linear unconditional quantile regressions and can be decomposed in detail. We present two empirical applications to illustrate the methodology.