Decomposing Gender Wage Differentials in Urban Ethiopia: Evidence from Linked Employer–Employee (LEE) Manufacturing Survey Data

Abstract

This study estimates the magnitude of gender wage differentials for a sample of workers from the Ethiopian manufacturing sector using the traditional Oaxaca–Blinder and an augmented Cotton–Neumark methodologies. In doing so, it separates part of the estimated log of gender wage differential explained by differences in human capital characteristics between men and women from that which is not explained by such differences. The latter is known in the literature as “treatment” component or “discrimination” due to differing pay structures for the two gender groups. Accordingly, it is found that in Ethiopia's manufacturing sector men on average get up to 30% more than women depending on the measure used. However, once we control for a number of individual and establishment level characteristics, the level of wage premium for men over women is close to 5% or around 12 Ethiopian cents per hour. Out of this, both decomposition procedures estimate that close to 60% of the premium is a result of discrimination (different treatment of men and women in the labour market). Using an augmented decomposition technique, it is found that out of the 60% “discrimination component” close to 13% is due to men's treatment advantage in the labour market and the remaining 47% is due to women's treatment disadvantage. Also it is found that firm level characteristics are important contributors to the total discrimination component. Without controlling for establishment level characteristics, the discrimination component would have been around 27% indicating that ignoring establishment characteristics in decomposition exercises would result into a biased estimation, and in this case it would have underestimated the level of discrimination by close to 50%.