A New Look at the Union Wage Premium during the Early Years of the AFL

This study provides new evidence showing that the union wage premium in the late nineteenth century in the United States was lower than previously believed. Analysis of wage and productivity data from an 1890 survey of individual workers in Maine yields a 9.2% union-nonunion wage gap, once correction is made for self-selection bias (the disproportionate representation in unions of workers who, because of skills and other attributes, would probably gain above-average wages even in the absence of unions). The author argues that business cycle conditions and distinctive union dynamics can affect the results of analyses that employ the Heckman procedure, a common procedure designed to correct for self-selection bias.