An axiom of economic thought is that competition makes for healthier markets. With regards to product markets (markets for buying and selling goods and services), more competition generally leads to lower prices and more choice, enhancing consumers’ bargaining power.1 This, in part, is why U.S. antitrust authorities, including the U.S. Department of Justice (DOJ) and the Federal Trade Commission (FTC), consider whether proposed mergers are likely to diminish competition enough to allow firms the ability to exploit consumers.2
Recently though, there has been growing interest in competition in labor markets and in the anticompetitive effects of increased employer concentration, especially on workers. Theoretically, as competition decreases within a labor market, firms’ abilities to exploit market power (to diminish the quality of work conditions and to reduce wages in that market) increases.3 If workers cannot easily switch jobs—because there are few places for them to apply their particular set of skills and expertise—chances are they will work for whatever wage they can get, under whatever conditions are minimally viable.
Indeed, there is growing evidence to suggest that lack of competition within labor markets leads to less favorable (e.g., more hazardous) working conditions and, critically, to lower wages.4 In 1991, Professor Shulamit Kahn documented anticompetitive labor market effects in the Kentucky coal mining industry, finding that the number of employers (i.e., greater employer choice) was associated with lower occupational accident rates.5 Small, remote coal mining towns of the 19th and early 20th centuries—places like Lynch, Wheelright, and Coal Run, Kentucky—were so notoriously monopolized they were dubbed “company towns.” One historian went so far as to call them “exploitationvilles” because the inhabitants fully depended on a single company for most of life’s necessities, including housing, food, employment, and law enforcement.6
Recent studies have pointed to wage depression as an effect of labor market concentration. For example, using online job posting data from 2016, a team of researchers found a significant, negative relationship between employer concentration and wages, such that a 10.0-percent increase in concentration was associated with around a 0.3-percent decrease in wages.7 This finding was replicated with different methods and datasets three separate times in 2022 alone.8
To that end, this article explores a new measure of labor market concentration using the near-complete universe of U.S. employment and wage data from the U.S. Bureau of Labor Statistics (BLS) Quarterly Census of Employment and Wages (QCEW) between 2002 and 2023. This article finds that the average labor market in the United States is “highly concentrated” according to the 2010 DOJ and FTC antitrust review guidelines and finds a significant, robust, and negative association between labor market concentration and wages. This article also provides a novel merger-impact analysis, showing that employer power could be meaningfully and significantly enhanced, according to antitrust review guidelines, in thousands of local labor markets if the top two employers in those markets were to merge.
Data
The statistics on labor market concentration presented in this article are derived from the QCEW microdata. The QCEW program publishes a quarterly count of employment and wages based on reports by private and public sector employers to state unemployment insurance (UI) programs. Nearly all employers in the United States are required to pay UI taxes to insure their workers against unemployment. To calculate an employer’s UI tax rate and bill, the state needs to know how many workers that employer has and how much said workers are paid. Employers report this information to their state, and the state then shares the data with BLS to use for statistical purposes.9
In 2023, the QCEW contained over 11.8 million employer reports, covering more than 153 million jobs and $11.0 trillion in wages. This coverage constituted a virtual census of U.S. employment, covering almost 95 percent of jobs in all 50 states, the District of Columbia, Puerto Rico, and the U.S. Virgin Islands.10
QCEW employment and wage data are reported to BLS at the establishment (i.e., worksite) level. Employment is reported monthly and represents the number of covered workers who worked during, or received pay for, the month.11 Wages constitute total compensation paid during the calendar quarter, regardless of when the services were performed. Total compensation includes bonuses, stock options, severance pay, profit distributions, the cash value of meals and lodging, tips and other gratuities, and, in some states, employer contributions to certain deferred compensation plans (such as a 401k). In this article, monthly employment and quarterly wage reports are aggregated into annual data. For employment, I use 12-month averages, and for wages I use 4-quarter sums.
Although the QCEW program publishes data on government employment and wages, I exclude government from this analysis. However, government employment makes up a relatively small share of the labor economy. In 2023, the public sector accounted for 14 percent of annual average employment (21.8 million jobs). Of this 14 percent, over a third (7.6 million jobs) was in public administration, an industry sector reserved specifically for government employers, and the remaining two-thirds were mostly concentrated into two cross-ownership industry sectors: education and healthcare. In other words, in most private industries, public sector employment is so small that its exclusion has little impact on the measurement of market concentration. Accordingly, this article measures and reports on private labor market competition.
QCEW employment and wage reports include an abundance of additional information that allow for aggregation along many dimensions. Relevant here are the geographic classification, industrial classification, and firm identification information provided with each establishment’s quarterly report.
With respect to geographic classification, each establishment is coded to the state in which it reports and the county in which the worksite is physically located. This code allows for state-based and county-based geographic aggregation. For example, BLS routinely publishes county-based employment and wage aggregates for U.S. Office of Management and Budget metropolitan statistical areas (MSAs) and combined statistical areas (CSAs).
With respect to industrial classification, each establishment is assigned a six-digit North American Industry Classification System (NAICS) industry code on the basis of its primary economic activity. This code allows for aggregation at the sector, subsector, and industry-group levels.12
Finally, with respect to firm identifiers, QCEW establishment reports come with legal or trade name information, UI account numbers, and, importantly, federal tax employer identification numbers (EINs). These fields allow for firm-level (as opposed to establishment-level) analysis of employment and wages.13 In 2023, less than 1 percent of private sector establishments in the QCEW—accounting for less than 1 percent of private sector employment and wages—were missing an EIN. For these establishments, state UI account numbers were used as a substitute employer identifier.14
Methods
The following section explains how the data and conceptual underpinnings are combined to produce empirical results.
Herfindahl-Hirschman Index (HHI)
The Herfindahl-Hirschman Index (HHI) is an existing measure of market concentration that is widely used to evaluate market competition. It is defined as the sum of the squares of the market shares of participants in a given market, and can be represented by the following equation:
where 
is the market share of participant 
(expressed as a number between 0 and 100), and 
is the number of participants. Given the squaring of market shares, the HHI accounts for the relative size distribution of participants in a market. When a market is occupied by a large number of participants that are relatively equal in size, the HHI approaches zero. When a market has just one participant, the HHI reaches a maximum of 10,000. Thus, the HHI is a measure of market concentration on a scale from zero (“perfectly competitive”) to 10,000 (“perfectly monopolistic” or, in the case of labor markets, “perfectly monopsonistic”).15
Using EINs to identify market participants (i.e., firms), one can construct HHIs for various labor markets in the United States on the bases of both employment and wages. In fact, with a slight modification to the HHI formula, the concentration of employment or wages in a labor market with geographic area 
and industry 
during year 
can be defined as follows:
where 
is the market share of firm (expressed as a number between 0 and 100) in year , area , and industry , and where is the total number of firms—as identified by EIN—operating in the market.
Market concentration taxonomy
In the DOJ and FTC’s 2010 horizontal merger guidelines, markets are classified into three general categories: (1) unconcentrated markets, in which the HHI observed is less than 1,500; (2) moderately concentrated markets, in which the HHI observed is between 1,500 and 2,500; and (3) highly concentrated markets, in which the HHI observed is greater than 2,500.16 Antitrust authorities use the taxonomy primarily to identify mergers that are likely to raise competitive concerns. On this point, the guidelines state that “mergers resulting in highly concentrated markets that involve an increase in the HHI of more than 200 points will be presumed to be likely to enhance market power,” while “mergers resulting in highly concentrated markets that involve an increase in the HHI of between 100 points and 200 points potentially raise significant competitive concerns,” as do “mergers resulting in moderately concentrated markets that involve an increase in the HHI of more than 100 points.”17 For the purposes of this article, this taxonomy is useful for categorizing markets more qualitatively.
Labor market definition
Being that it constitutes a nearly universal set of employment and wage data for the United States, the QCEW is positioned to provide perhaps the most precise picture possible of U.S. labor market competition and concentration. First, though, it is important to define the concept of a labor market.
In the DOJ and FTC’s merger guidelines, a market is defined along two dimensions: (1) a section of the economy, and (2) a line of commerce. With respect to labor markets, the section of the economy is best understood in terms of geography. For example, the state of Montana is a section of the larger U.S. economy, as is the city of Los Angeles, California. The other dimension, the line of commerce, is perhaps most readily thought of in terms of an industry or occupation. For instance, there are markets for mining (industry) and lawyers (occupation).
Putting the two components together may be used to define, for example, markets for mining in Montana and lawyers in Los Angeles (and vice versa). Thus, a labor market can be thought of as a geography-industry (or geography-occupation) combination. Workers within a geography-industry/occupation (e.g., miners, lawyers) sell their labor in exchange for wages and benefits, and businesses operating in the geography-industry/occupation (e.g., mines, law firms) use that labor as an input into their business activities.
Multiple studies from the early 2020s suggest that the most appropriate market definition afforded by the QCEW is the MSA-by-industry-group labor market, so this definition is used for this article’s baseline analysis.18 With respect to the section-of-economy dimension, the use of MSAs helps to account for potential intercounty/inter-state labor pools. Undoubtedly, many labor markets bleed beyond county or state lines because of commuting. In fact, MSAs were specifically designed to capture commuting ties between geographic areas. In 2023, there were 388 MSAs represented by the QCEW, and nearly 9 out of every 10 private sector employees worked in one of these areas.
Although product markets tend to be national or even global in scope, and though one may think that the MSA is too small of a labor pool in an era of heightened remote work, the data on job holding and job seeking behavior strongly suggest that labor markets are localized, often even hyperlocalized to within a few miles of a worker’s residence. With respect to job holders, the Survey of Working Arrangements and Attitudes estimated that 87.3 percent of full-time workers in the U.S. ages 20 to 64 worked onsite some or all of the time in December 2023.19 Also, the U.S. Census Bureau estimated that the average one-way commute time of U.S. workers in 2022 was 26.7 minutes, basically unchanged from 10 years prior.20 With respect to jobseekers, a 2018 study by Professors Ioana Marinescu and Roland Rathelot found that people are 35 percent less likely to apply to a job that is 10 miles away from their ZIP Code of residence than to a job within their ZIP Code.21 That study also found that 4 out of every 5 job applications are sent to employers within a job seeker’s metropolitan or micropolitan statistical area of residence. Job taking (i.e., acceptance) behavior likely exhibits even further “distaste for distance” because of relocation costs, both monetary and nonmonetary, which have been found to be substantial enough to significantly inhibit cross-MSA migration.22
With respect to the line of commerce, the use of industry groups as part of the baseline definition of a labor market helps to strike a balance between too broad and too narrow of a market. At higher levels of industrial aggregation (such as sectors and subsectors), the line of commerce may be too broad. Similarity of jobs within the market—and the education, skills, and expertise needed to hold such jobs—may be overestimated, and therefore the opportunities available to workers with the requisite knowledge and skills may also be overestimated. This, in effect, spreads market shares out across somewhat similar but altogether noncompeting firms, leading to underestimated HHIs. On the other hand, at the finest levels of industrial aggregation (e.g., single industries), job prospects outside of the market for workers with the requisite knowledge and skills may be plentiful, meaning the line of commerce is probably too narrow. In that case, HHIs are likely to be overestimated, as they discount market shares of firms actually operating in the market.23
Taken together, an MSA-by-industry-group labor market definition is that which most closely approximates a genuine labor market, as defined by the QCEW. In such markets, participants—workers and firms—experience relatively high switching costs with respect to both the line of commerce (industry) and the section of economy (geography).
Results
The following section examines the state of labor market competition in the United States, both past and present, and probes the relationship between competition and wages.
The average labor market in the United States is highly concentrated
Chart 1 reports summary statistics for the HHI based on annual average employment in 2023 under various labor market definitions. Under the baseline market definition (MSA by industry group), the average HHI across 93,252 markets was 3,734 and the median HHI was 2,604, meaning that both the average and the median U.S. labor markets were “highly concentrated” according to the DOJ and FTC antitrust guidelines. These results are consistent with several studies from the early 2020s.24
As expected, and in line with the previous discussion surrounding how to define labor markets, the average MSA-bounded labor market is more concentrated than the average state-bounded labor market, but less concentrated than the average county-bounded labor market; the average industry-group-bounded labor market is more concentrated than the average subsector-bounded labor market, but less concentrated than the average industry-bounded labor market.
One out of every eight labor markets in the United States is controlled by a single employer
Chart 2 reports the distribution of HHIs for MSA-by-industry-group labor markets in 2023. Overall, 51 percent of labor markets were highly concentrated, while another 15 percent were moderately concentrated. Thirteen percent of labor markets were nearly or perfectly monopsonistic (HHI > 9,500). In other words, more than 1 out of every 8 labor markets in the U.S. is dominated by a single employer. These results are highly consistent with other, contemporary research.25
Highly concentrated labor markets account for over 15 percent of private sector employment and payrolls in the United States
The employment-weighted average HHI under the baseline market definition was 1,186 in 2023. (See chart 1.) This level of concentration was much lower than both the unweighted average HHI (3,734) and the median HHI (2,604). The difference between the unweighted and weighted average HHIs reflects the fact that concentrated markets tend to have smaller workforces. Indeed, chart 3 shows that highly concentrated markets accounted for 15 percent of private employment in 2023 at the MSA-by-industry-group level (17.4 million jobs), while chart 4 shows they accounted for a slightly higher share of payrolls (18 percent). Moderately concentrated markets accounted for another 9 percent of employment (9.9 million jobs) and 9 percent of payrolls.
The average local labor market became 7 percent less concentrated between 2002 and 2023
Chart 5 shows the trend in national- and local-level labor market concentration between 2002 and 2023. At the local level (across MSA-by-industry-group labor markets), labor markets have tended to become less concentrated over time. Between 2002 and 2023, the average local labor market became about 7 percent less concentrated. However, over that same period, the average nation-by-industry-group labor market became more concentrated by about 46 percent.26 In both cases, though, the relative level of concentration remained unchanged. For local labor markets, the average HHI remained between 1,100 and 1,300 (unconcentrated), while for national labor markets, the average HHI remained between 100 and 200 (unconcentrated).27
Higher labor market concentration is significantly associated with lower wages
Chart 6 shows a scatter plot of the residualized logarithm (log) of annual average wages and the residualized log of the annual average employment HHI in MSA-by-industry-group labor markets in 2023.28 Markets are aggregated into 20 quantile-spaced bins, which reduces noise and makes the association easier to visualize. There is a clear negative correlation between market concentration and average annual wages after residualizing out the year and market fixed effects. This relationship is consistent with multiple contemporaneous studies.29
To gauge the impact of labor market concentration on wages, I performed ordinary least squares (OLS) regressions of average annual pay on concentration at the market level for MSA-by-industry-group markets between 2002 and 2023. The baseline specification (model 1 in table 1) is as follows:
where 
is the log of the annual average wage in a market constituted by metropolitan area a and private industry group d in year t, 
is the log of the annual average employment HHI in that same market and year, 
is a constant, 
and 
are year and market fixed effects, and 
is the error term.
The baseline model demonstrates that higher labor market concentration is significantly associated with lower wages such that a 10.0-percent increase in concentration is associated with about a 0.3-percent decrease in average wages. Practically speaking, this wage elasticity implies that a shift in employer concentration from the upper-bound threshold for unconcentrated markets (HHI = 1,500) to the lower-bound threshold for highly concentrated markets (HHI = 2,500)—a 66.7-percent increase in concentration—is associated with a 2.0-percent decrease in average wages. Stated differently, average wages in highly concentrated markets are expected to be at least 2.0-percent lower than average wages in unconcentrated markets, all else being equal. The effect size is moderated after removing monopsony markets (HHI = 10,000) from consideration, but the relationship remains negative and statistically significant. (See model 2 in table 1.)
Table 1. Effect of labor market concentration on log annual average wages, OLS regression, 2002–23| Variable | Model 1 | Model 2 |
|---|
Log employment HHI | -0.0293[1] | -0.0049[1] |
|---|
| (0.0015) | (0.0014) |
Fixed effects |
|---|
Year | Yes | Yes |
|---|
Market (MSA by industry group) | Yes | Yes |
|---|
Observations | 2,007,284 | 1,759,161 |
|---|
R-squared | 0.8436 | 0.8940 |
|---|
The effect size is also modulated by the geographic and industrial parameters applied, but the overall finding remains robust to other reasonable labor market definitions. (See table 2.) For example, the relationship remains negative and statistically significant for MSA-by-sector, MSA-by-subsector, and MSA-by-specific-industry labor markets (see models 3, 4, and 5, respectively, in table 2) as well as county-by-industry-group and state-by-industry-group markets (models 6 and 7, respectively, in table 2). However, there seems to be no association between market concentration and wages at the nation-by-industry-group level. (See model 8 in table 2.)
Table 2. Effect of labor market concentration on log annual average wages, OLS regression, alternative market definitions, 2002–23| Variable | Model 3 | Model 4 | Model 5 | Model 6 | Model 7 | Model 8 |
|---|
Log employment HHI | -0.0138[1] | -0.0183[1] | -0.0445[1] | -0.0559[1] | -0.0129[1] | -0.0015 |
|---|
| (0.0037) | (0.0023) | (0.0010) | (0.0008) | (0.0033) | (0.0144) |
Geographic level | MSA | MSA | MSA | County | State | Nation |
|---|
NAICS level | Sector | Subsector | Industry | Industry group | Industry group | Industry group |
|---|
Observations | 163,502 | 690,647 | 5,102,978 | 9,921,196 | 339,692 | 6,726 |
|---|
R-squared | 0.9497 | 0.8818 | 0.8160 | 0.8159 | 0.9128 | 0.9830 |
|---|
Overall, these results are highly consistent with the modeling done in previous studies that use different datasets and definitions of labor markets. The baseline wage elasticity of concentration (see model 1 in table 1), −0.0293, is highly consistent with the wage elasticities found in similar models in four of the leading studies on the subject, thereby adding to the mounting evidence of employers’ wage-setting power in the United States. (See table 3.)30
Table 3. Summary of related research on the effect of labor market concentration on wages| Article | Model | Wage elasticity | Market definition | Data source |
|---|
Azar, Marinescu, Steinbaum, and Taska (2020) | Table 2, panel A, model 3 | -0.0286 | CZ by six-digit SOC occupation | Burning Glass job posts (2016) |
|---|
Azar, Marinescu, and Steinbaum (2022) | Table 4, panel A, model 2 | -0.0347 | CZ by six-digit SOC occupation | CareerBuilder job posts (2010–2013) |
|---|
Benmelech, Bergman, and Kim (2022) | Table A2, model 7 | -0.0110 | CZ by four-digit SIC manufacturing industry | LBD (1978–2016) |
|---|
Rinz (2022) | Table 2, model 2 | -0.0561 | CZ by four-digit NAICS industry | LBD (2005–2015) |
|---|
These regression analyses have a few limitations though. The first has to do with the QCEW’s wage measure, the outcome variable of interest. QCEW average wages are not always representative of typical wages because average wages in any given market may be skewed upward by particularly high earners. Moreover, QCEW average wages do not fully account for hours worked. Two markets may have the same average wage level, even the same number of workers, but the underlying wage rate would tell a vastly different story if, for example, the workers in one market worked twice as many hours. Taken together, QCEW average wage measures can only serve as a proxy for wages in this context.
The second and perhaps more serious limitation in the above analyses has to do with the endogeneity of HHI, the explanatory variable of interest. Previous research has established that characteristics of labor markets such as unemployment, productivity, and union membership affect both market concentration and worker compensation, but data regarding these factors are not readily available at the MSA-by-industry-group level. These omitted variables likely mean that the coefficients in the OLS models in tables 1 and 2 are biased, and certainly mean that the established measures are only correlational in nature.
Although it is difficult to correct the issue regarding average wages, it is possible to address the endogeneity issue through instrumental variables (IV) estimation. A local labor market’s own level of employment concentration may be instrumented using the average employment concentration of other localities within the same industry and year.31 That is, the concentration of related labor markets will be (strongly) correlated—something that is easily verifiable—but unrelated to the omitted variables—something that is unverifiable, yet highly plausible. For example, there is no reason to suspect that the concentration of meatpacking in Chicago, Fresno, and Atlanta affects the productivity of meatpackers in Green Bay. In effect, the HHI of related labor markets may serve as a less biased proxy for a given labor market’s own concentration level, and therefore as a better causal predictor of wages.
The equation to derive the instrument for any industry-group d across metropolitan areas i in any given year t may be expressed as
where i is an index of all MSAs with employment in industry d and year t, except for the given market’s MSA (a). The instrument is weighted by employment to account for size differences across the related labor markets.
With that calculated, the IV model may be specified as
where 
is the log of the annual average wage in a market constituted by metropolitan area a and industry-group d in year t, 
is the log of the weighted average employment HHI across related markets in the same industry and year (i.e., the log of the instrument), is a constant, and are year and market fixed effects, and is the error term.
In the baseline model with the IV specification (see model 9 in table 4), the magnitude of the coefficient on concentration more than triples versus the OLS specification (compare with model 1 in table 1). In the baseline IV model, a 10.0-percent increase in concentration is associated with a 1.0-percent decrease in wages.32 Practically speaking, this wage elasticity implies that a shift in employer concentration from the upper-bound threshold for unconcentrated markets (HHI = 1,500) to the lower-bound threshold for highly concentrated markets (HHI = 2,500)—a 66.7-percent increase in concentration—is associated with a 6.8-percent decrease in average wages. Stated differently, average wages in highly concentrated markets are expected to be at least 6.8-percent lower than average wages in unconcentrated markets, all else being equal. The effect size is slightly moderated after removing monopsony markets (HHI = 10,000) from consideration, but the relationship remains negative and highly statistically significant (see model 10 in table 4).
Table 4. Effect of labor market concentration on wages, instrumental variables regression, 2002–23| Variable | Model 9 | Model 10 |
|---|
Stage 1 (dependent variable: log employment HHI) |
|---|
Log employment HHI (related labor markets) | 0.2570[1] | 0.2481[1] |
|---|
| (0.0120) | (0.0122) |
Observations | 2,007,284 | 1,759,161 |
|---|
R-squared | 0.9370 | 0.9286 |
|---|
Stage 2 (dependent variable: log average annual wage) |
|---|
Log employment HHI | -0.1018[1] | -0.0897[1] |
|---|
| (0.0091) | (0.0089) |
Observations | 2,007,284 | 1,759,161 |
|---|
R-squared | 0.8423 | 0.8920 |
|---|
Mergers could significantly impact market power in the majority of U.S. labor markets
The 2010 merger guidelines from the DOJ and FTC state that “mergers resulting in highly concentrated markets that involve an increase in the HHI of more than 200 points will be presumed to be likely to enhance market power.”33 In essence, mergers that meet these criteria may receive extra scrutiny by antitrust authorities.
Using the QCEW HHI data, one can estimate how common it would be for a merger to warrant extra scrutiny in the labor market context. To do this, one simply needs to compare a labor market’s observed HHI with a counterfactual HHI calculated after simulating a merger between two (or more) employers. If the counterfactual HHI is above 2,500 and if the difference between the counterfactual HHI and the observed HHI is more than 200, then if that merger were to be proposed it would meet the criteria.
When simulating mergers between the top two firms in each MSA-by-industry-group labor market that had two or more employers in 2023 (N = 82,256), the condition for the enhanced market power presumption was fulfilled 60 percent of the time. Essentially, employer power could be meaningfully and significantly enhanced, according to antitrust review guidelines, in 3 out of every 5 labor markets (that are not already monopsonies) if the top two employers in those markets were to merge. More than 25.5 million Americans worked in these markets in 2023, accounting for nearly 1 in 5 private sector jobholders across the country.
Among MSA-by-industry-group labor markets with at least two employers and 100 workers in 2023 (N = 52,557), the condition for the enhanced market power presumption was fulfilled 44 percent of the time, and among markets with at least two employers and 1,000 workers (N = 16,426), the presumption was fulfilled 26 percent of the time. Therefore, even in thousands of relatively large labor markets, employer power could be meaningfully and substantially enhanced if the top two employers were to merge.
It is also possible to model the expected impact of a merger on wages. To do so, one simply needs to interpret some wage elasticity of concentration, 
, in response to the percentage change in concentration due to a merger. The formula for this is as follows:
where y is the expected percent change in average wages in a market constituted by area a and industry d in year t, and HHIC and HHIO are the counterfactual (after the simulated merger) and observed (before the simulated merger) HHIs for that market, respectively.
Going a step further, it is possible to model the average expected percent change in wages across a set of markets, 
, given merger-induced changes in concentration in each market:
where w is the employment level in a market and is used to weight the average calculation to account for differences in market size.
In summary, the regressions model how labor market concentration affects wages, and the merger simulation models how mergers could affect labor market concentration. In combination, then, these analyses can show how mergers may affect wages. With the wage elasticity of concentration () from the baseline OLS regression (model 1 in table 1), −0.0293, the employment-weighted average expected wage growth () across all MSA-by-industry-group markets with at least two employers in 2023 was −1.6. Thus, a merger of the top two employers in a local-level labor market is expected to reduce wages by 1.6 percent, per worker, on average. The average annual wage across these markets was around $61,300 in 2023, which means the average merger between the top two firms in the average of these markets could be expected to reduce workers’ annual pay by around $1,008 each ($63,000×−0.016). Across these markets, the average number of workers was 1,377. Therefore, a crude estimate of total annual wage loss resulting from a merger of the top two firms for the average of these markets is $1.4 million ($1,008×1,377), equivalent to the average annual pay of around 22 workers ($1,400,000/$63,000).
Conclusion
This article documents the development of a new set of statistics based on the QCEW that measure labor market concentration in the United States. These data show that the average local-level labor market in the United States is highly concentrated according to antitrust merger guidelines and that tens of millions of Americans work in uncompetitive markets. Moreover, these new data show that wages are negatively and significantly associated with employer concentration.
This article also simulates firm mergers to estimate the potential wage effects of further consolidation. This predictive approach finds that monopsony power could be meaningfully and substantially enhanced by mergers, according to antitrust review guidelines, in the majority of local-level labor markets. However, future work should take a more descriptive approach and examine the effects of actual (past) mergers. Given the aforementioned findings, it is expected that wages would grow negatively or at least more slowly in labor markets that have undergone recent mergers. By exploiting the structural discontinuities arising from such consolidation shocks, it may be possible to tease out better causal evidence of monopsony power.
ACKNOWLEDGMENTS: The author thanks Imani Drayton-Hill, Dave Hiles, Matthew Dey, Maribeth Rucker-Dong, and Kathryn Evans, all currently or formerly of the Office of Employment and Unemployment Statistics, for their valuable comments and guidance during the preparation of this article.
United States Department of Labor
