This paper investigates inter-industry wage differentials in Belgium, taking advantage of access to a unique matched employer-employee data set covering the period 1995-2002. Findings show the existence of large and persistent wage differentials among workers with the same observed characteristics and working conditions, employed in different sectors. The hypothesis that workers with better unmeasured abilities are over-represented in high-wage sectors may not be rejected on the basis of Martins’ (2004) methodology. However, the contribution of this explanation to the observed industry wage differentials appears to be limited. Further results show that ceteris paribus, workers earn significantly higher wages when employed in more profitable firms. Our instrumented wage-profit elasticity stands at 0.063 and Lester’s range of pay is about 41 per cent of the mean wage. This rent-sharing phenomenon accounts for a large fraction of the industry wage differentials. We find indeed that the magnitude, dispersion and significance of industry wage differentials decreases sharply when controlling for profits.
Industry wage differentials Unobserved heterogeneity Rent-sharing Matched employer-employee data Quantile regressions