Concentration in U.S. local labor markets: evidence from vacancy and employment data

This paper characterizes the cross-sectional and time-series properties of concentration in employment, job creation, and vacancy flows across U.S. local labor markets. We proceed in three steps: first, we derive conditions for indices of labor market concentration to be appropriate proxies for monopsony power. Then, we compute Herfindahl-Hirschman Indices at the local labor market level using data on the universe of online vacancies (BGT) and the universe of employers (LBD). Finally, we document that labor market monopsony does not manifest itself only through a negative effect on the level of wages, but also through a positive effect on the demand for skills. We find that (i) in the last decade, at most 5% of new U.S. jobs are in moderately concentrated local markets; (ii) local labor market concentration decreased over time, dropping by at least 25% since 1976. We reconcile our findings to previous studies on increasing national concentration through a statistical decomposition which implies that the covariance between a local labor market's size and its concentration level decreased over time. When it comes to the effects of monopsony, we find that a 1% increase in local labor market concentration is associated with a 0.14% decrease in average hourly wages, and an increase in the number of jobs requiring cognitive and social skills equal to 10-13% of the mean (``upskilling''). We conclude that our evidence is consistent with the presence of employers' market power and discuss how upskilling constitutes a policy challenge not readily addressed by increases in the minimum wage.