Wage Determination in the Union and Nonunion Sectors: A Sample Selectivity Approach

This paper re-examines the question, recently raised in this journal by Bloch and Kuskin, of whether wages are determined differently in the union and nonunion sectors. Whereas Bloch and Kuskin employed ordinary least squares to estimate separate wage equations for the two sectors, this study uses a methodology proposed by Heckman and by Lee to correct for the possibility that wage differences may determine the union status of workers as well as vice versa. The authors find that union status is strongly related to the predicted union-nonunion wage differential, but their evidence nevertheless reinforces Bloch and Kuskin's empirical finding that the union earnings function is less sensitive than the nonunion earnings function to changes in nearly every observable attribute of workers, such as education and experience. The authors also conclude that previous studies using separately estimated union and nonunion wage equations may have understated the success of unions in raising the relative wages of their members.