Wage Dispersion in a Partially Unionized Labor Force
John T. Addison
Wage Dispersion in a Partially Unionized Labor Force This paper critiques Card’s (2001) method for analyzing wage dispersion in a partially unionized labor market based on a disaggregation of the population into skill categories. We argue that disaggregation is a good idea, the use of skill categories less so. We offer a modified model in which each worker is assigned a union-membership probability, a predicted union wage, and a predicted nonunion wage. Our model provides a natural three-way decomposition of variance, and is also suited to counterfactual analysis. By way of an application, we examine the effect of de-unionization on wage dispersion in the United Kingdom between 1983 and 1995, reporting that the decline in membership accounts for only about one-fifth of the observed increase in wage dispersion.
JEL Classification: D3, J31, J51.
Keywords: wage dispersion, three-way variance decomposition, bivariate kernel density smoothing, union membership, deunionization.