Unions, Wages, and Skills

Prior research has focused on the relationship between union wage premiums and worker skills. A standard result is that premiums are highest for workers with low levels of measured skills, and lowest among workers with high measured skills. Evidence from matched Current Population Survey (CPS) panels for 1989/90-1993/94, for example, produce wage level estimates of union premiums ranging from .23 log points among high school dropouts to .09 among college graduates. Longitudinal estimates controlling for worker-specific abilities, however, are .10 to .13 log points among all education groups, indicating an inverse relationship between measured premarket skill levels and relative union-nonunion unmeasured ability. Direct evidence from the NLSY on union-nonunion differences in AFQT aptitude scores further supports our thesis that union workers with high measured skills have low unmeasured skills, relative to nonunion workers. Also explored are alternative strategies to limit measurement error in the union status variable. Our findings provide support for the thesis that there is substantial homogeneity in union workforces, likely to result from union wage standardization policies, union success in organizing establishments where workers have homogeneous preferences and skills, employer selection on quality, and employee sorting such that few highly-skilled workers are in the union queue.