NBER WORKING PAPER SERIES LABOR MARKET POWER David W. Berger Kyle F. Herkenhoff Simon Mongey Working Paper 25719 http://www.nber.org/papers/w25719 NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambridge, MA 02138 March 2019, Revised May 2021 We thank Costas Arkolakis, Chris Edmond, Oleg Itskohki, Matt Notowidigdo, Richard Rogerson, Jim Schmitz, Chris Tonetti, and Arindrajit Dube for helpful comments. We thank seminar participants at the SED 2018, SITE, Briq (Bonn), FRB Minneapolis, Princeton, Queens, Stanford, UC Berkeley, USC, UT Austin, Rochester, BFI Firms andWorkersWorkshop, Paris School of Economics, Toulouse School of Economics, EIEF, Barcelona GSE Summer Forum, Stanford, MIT, Boston University, NBER Summer Institute. We thank Chengdai Huang for excellent research assistance. This research was supported by the National Science Foundation (Award No. SES-1824422). Any opinions and conclusions expressed herein are those of the author(s) and do not necessarily represent the views of the U.S. Census Bureau. All results have been reviewed to ensure that no confidential information is disclosed. The views expressed in this study are those of the author and do not necessarily reflect the position of the Federal Reserve Bank of New York, the Federal Reserve System, or National Bureau of Economic Research. NBER working papers are circulated for discussion and comment purposes. They have not been peer- reviewed or been subject to the review by the NBER Board of Directors that accompanies official NBER publications. © 2019 by David W. Berger, Kyle F. Herkenhoff, and Simon Mongey. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source. Labor Market Power David W. Berger, Kyle F. Herkenhoff, and Simon Mongey NBER Working Paper No. 25719 March 2019, Revised May 2021 JEL No. E2,J2,J42 ABSTRACT To measure labor market power in the US economy, we develop a tractable quantitative, general equilibrium, oligopsony model of the labor market. We estimate key model parameters by matching the firm-level relationship between labor market share and employment size and wage responses to state corporate tax changes. The model quantitatively replicates quasi-experimental evidence on (i) imperfect productivity-wage pass-through, (ii) strategic behavior of dominant employers, and (iii) the local labor market impact of mergers. We then measure welfare losses relative to the efficient allocation. Accounting for transition dynamics, we quantify welfare losses from labor market power relative to the efficient allocation as roughly 6 percent of lifetime consumption. An analytical decomposition attributes equal parts to dead-weight losses and misallocation. Lastly, we find that declining local concentration added 4 ppt to labor’s share of income between 1977 and 2013. David W. Berger Simon Mongey Department of Economics Kenneth C. Griffin Department of Economics Duke University University of Chicago 419 Chapel Drive 1126 E. 59th Street Social Science Building 231 Chicago, IL 60637 Durham, NC 27708 and NBER and NBER [email protected] [email protected] Kyle F. Herkenhoff University of Minnesota Department of Economics 4-101 Hanson Hall 1925 Fourth Street South Minneapolis, MN 55455 and IZA and also NBER [email protected] Figure 1: Cross-market distribution of concentration: Longitudinal Business Database, 2014. Notes: Panel A plots the distribution of number of firms in markets. Panels B plots the across market distribution of the payroll wn Herindahl index (HHIj ). Bins are determined by the following bounds: f0, 0.10, 0.25, 0.50, 0.75, 0.99, 1g. Horizontal axis gives the mean in each bin. Blue line (solid squares) gives the distribution of markets, red line (dashed crosses) gives the distribution of total wage payments. Data is Census LBD for the whole US economy in 2014. Market is defined as a commuting zone and NAICS 3-digit industry. See AppendixC for additional details. Table A2 provides additional data on employment HHI’s. In the average local labor market in the U.S., there are many firms but employment and wages are concentrated in only a few. Defining a labor market as a commuting zone and three-digit industry, the average number of firms is over 100, while the weighted average level of market concentration is 0.11, the same level of concentration one would observe with only 9 equally sized firms (Figure 1).1 This has led to the growing concern that these firms may exert “labor market power” over their workers, generating large welfare losses.2 In this paper, we measure the amount of oligopsony power in labor markets and quantify its consequences for welfare. We do so by developing a tractable, quantitative, general equilibrium model with differentially concentrated local labor markets in which firms behave strategically under an oligopsony equilibrium. These novel features allow the model to match empirical regularities in the labor literature such as incomplete wage pass-through and strategic competitor wage responses that a standard monopsony model misses. The model delivers a structurally consistent for- mulation of labor market power and a framework for understanding the mechanisms behind potential welfare losses. Our benchmark oligopsony model features two sources of market power. The first is classical monop- sony: atomistically small firms face upward sloping labor supply curves due to preference heterogeneity, which they internalize (Burdett and Mortensen, 1998; Manning, 2003; Card, Cardoso, Heining, and Kline, 2018; Lamadon, Mogstad, and Setzler, 2019). Optimal wages are a markdown relative to competitive wages, i.e. the marginal revenue product of labor. Second, motivated by Figure 1 and the focus of this paper, is oligopsony: firms are non-atomistic and compete strategically for workers, further internalizing how they expect other employers to respond to their hiring and wage policies. This strategic interaction 1Appendix Table F2 reports 113 firms per market across all industry codes. AppendixC provides additional market level summary statistics. 2For example: Azar, Marinescu, and Steinbaum(2020), Benmelech, Bergman, and Kim (2020), Card, Cardoso, Heining, and Kline (2018), and Lamadon, Mogstad, and Setzler (2019). 1 leads to large equilibrium markdowns at the most productive firms and provides a second source of wel- fare loss. Understanding the welfare consequences of labor market power requires understanding how these markdowns vary across firms. In our model, the markdown is an exact function of the structural labor supply elasticity that a firm faces in equilibrium which—via a closed-form—depends on the firm’s observable labor market share and parameters that determine how easily labor is reallocated across- (q) and within- (h) markets. We estimate the model on U.S. Census data, and derive three main results. First, the framework is quantitatively consistent with documented empirical regularities suggestive of oligopsony: incomplete wage pass-through, strategic competitor wage responses, and size-dependent post-merger wage dynam- ics. A monopsony version of our model cannot qualitatively match these empirical regularities. Second, the model implies substantial welfare losses from labor market power, both across steady states and along the transition path to an efficient allocation. Welfare losses are large, ranging from 4 to 9 percent of lifetime consumption depending on wealth effects. A representative agent counterpart to our economy delivers equilibrium aggregate prices and quantities and decomposes welfare loss into two components: (1) a dead-weight loss due to average markdowns, (2) a misallocation effect due to wider markdowns at more productive firms. While the former exists under monopsony, the latter does not. We show that both channels account equally for welfare losses. Third, despite these large losses, we find that labor market power has not contributed to the declining labor share. Despite the backdrop of stable national concen- tration, we find that the model-consistent measure of local concentration, which we measure for the first time, has declined over the last 35 years, indicating that most local labor markets are more competitive than they were in the 1970s.3 In terms of the general equilibrium theory of the model, we prove two properties that are central to our main applications. First, we show that our model is block recursive, meaning that local labor market equilibrium is independent of aggregates. This allows us to estimate the model quickly and decompose welfare for arbitrary aggregate preferences. Second, we provide a closed-form relationship between labor’s share of income and local payroll concentration. Our model-relevant measure of payroll concentration is new to the literature. We use our formula to measure the contribution of changes in local payroll concentration on labor’s share of income. In terms of estimation of the model, strategic interaction complicates the identification of the key parameters by violating exclusion restrictions that are otherwise applicable in monopsonistically com- petitive models. We address this issue by integrating into our structural estimation the first reduced- form estimates of the size dependence of employment and wage responses to state corporate taxes. We estimate our key parameters using U.S. Census Longitudinal Business Database (LBD) micro data (see Figure2). Given a quasi-experiment that yields an identified shock to labor demand,