July 2018

Study measures effect of labor market concentration on wages

Peter C. Fisk

Azar, Marinescu, and Steinbaum use standard measures—including the Herfindahl-Hirschman Index (HHI)—to quantify labor market concentration. Doing so allows for easy comparison with regulatory guidelines that use that same metric in evaluating concentrations on both the buying and selling sides of markets. The study’s analysis of labor markets found an average HHI of 3,157, which is above the 2,500 threshold for high concentration as defined by the horizontal merger guidelines of the Department of Justice and Federal Trade Commission. The results show concentration varying by occupation and city, with larger cities tending to have less concentration.

The authors document a negative correlation between labor market concentration and average posted wages in given markets. Using quarterly panel data from 2010 to 2013, the authors run two types of regressions—ordinary least squares (OLS) and instrumental variables (IV)—of posted wages on concentration at the market level. In the baseline OLS specification, the elasticity of the real wage with respect to the HHI is −0.038, while in the baseline IV specification the elasticity is −0.127.

Going from the 25th to the 75th level of concentration decreases posted wages by 5 percent in the baseline OLS specifications and by 17 percent in the baseline IV specification. The study includes several robustness checks, such as controlling for labor market tightness in order to address the potential concern that the effect of market concentration on wages could be an endogenous result of the relationship between the number of vacancies and concentration.

Having documented high concentrations on the employer side of labor markets and an association between increasing concentration and lower wages, the authors suggest that these findings and the study’s methods could be important factors for consideration in antitrust analysis and regulation. Azar, Marinescu, and Steinbaum do acknowledge some potential limitations of the study. For instance, only 20 percent of job listings on CareerBuilder.com include wage information. Additionally, the data come from a single website, and although CareerBuilder lists about a third of U.S. job vacancies, it does not contain all vacancies in the occupations in the study’s sample. Therefore, the study may be overestimating labor market concentration.