Measuring the Union–Nonunion Wage Gap Using Propensity Score Matching

This paper examines the union wage effect among private sector wage and salary workers using the Panel Study of Income Dynamics data. We estimate the average union premium by utilizing semi (non) parametric matching techniques, as well as by linear regression model. Our analysis reveals: (1) conditioning linearly on covariates understates the union wage effect. In addition, matching results under the alternative support conditions indicate that selection on unobservables has at most a modest impact on average. (2) Log wage estimations lead to an overstatement of the union wage effect. Solving these problems yields the union membership premium as 21.5 percent.