Trade unions are consistently found to compress the wage distribution. Moreover, unemployment affects in particular low-skilled workers. The present paper argues that an extended Right-to-Manage model can account for both of these findings. In this model unions compress the wage distribution by raising wages of workers in low productivity industries (or low-skilled workers) above market clearing levels. Our analysis suggests that the most direct way to test this model would be via a test for stochastic dominance. We also allow for capital adjustments and compare union and non-union wage distributions in a general equilibrium framework.
This paper establishes theoretical and empirical linkages between union wage setting and the structure of the wage distribution. Theoretically, we identify conditions under which a right-to-manage model implies compression of the wage distribution in the union sector relative to the nonunion sector as well as first-order stochastic dominance. These implications are investigated using quantile regressions on the 2001 GSES, a large German linked employeremployee data set which contains explicit information on coverage by collective agreements. The empirical results confirm that, in case of industry-wide collective agreements, log union wage effects decline in quantiles, implying union wage compression. This finding, however, cannot be corroborated for wages determined at the firm level. Stochastic dominance is confirmed, as predicted by the theoretical model, for both types of collective agreements.