Abstract
It is well-known that the standard Oaxaca decomposition method on wage differentials produces arbitrary results when calculating the coefficients effect of sets of dummy variables: the estimated sum of coefficients effects of sets of dummy variables is not invariant to the choice of reference groups. We generalize this identification problem in the detailed Oaxaca decomposition and its solution based on a normalized equation to a decomposition analysis for differences in the first moment where linear or non-linear (e.g., probit or logit) models can be used for estimation. We provide a practical and simple algorithm to derive the normalized equation, and show that using normalized equation does not change the size and inference of the overall characteristics and coefficients effects.