Previous research on union wage effects has underestimated the potential fo
r unions to raise member wages since the data used do not enable difference
s across bargaining units to be properly accounted for. This study addresse
s this deficiency by utilizing matched employer-employee data that permit w
orkplace-specific union wage effects to be identified Results from the esti
mation of wage equations indicate that, while there is only a very small in
tra-workplace union wage effect, differences across workplaces are consider
able. This differential, however, only exists at workplaces where there is
substantial coverage by collective agreements.