Effi Benmelech Northwestern Kellogg and NBER Nittai K Bergman Tel Aviv University Hyunseob Kim Cornell Johnson 72218 1 Disclaimer Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent the views of the US Census Bureau All res ID: 735332
Download Presentation The PPT/PDF document "Strong Employers and Weak Employees: How..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Strong Employers and Weak Employees: How Does Employer Concentration Affect Wages?
Effi Benmelech, Northwestern Kellogg and NBERNittai K. Bergman, Tel Aviv UniversityHyunseob Kim, Cornell, Johnson
7/22/18
1Slide2
Disclaimer
Any opinions and conclusions expressed herein are those of the authors 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.
2Slide3
What We Do
Use micro-level data from the U.S. Census Bureau to analyze how monopsony power in local labor markets affects wage behaviorFocus on labor market concentration using relatively localized geographic areas.Because the relevant market associated with job search is local (Moretti 2001).Labor mobility in the U.S. has declined significantly and job switches often occur between positions in the same area (Molloy and Wozniak 2014).
Construct county-industry-year level HHI of firm employment and relate these measures to wage behavior
3Slide4
What We Find
We analyze the effect of local-level labor market concentration on wages and find:Substantial variation in employer concentration both in cross-section and time-series
Negative relation between employer concentration and wagesEffect of concentration on wages is stronger when unionization rates are low
Sensitivity of wages to productivity growth reduced when employer concentration higher: less rent-sharing
Exposure to import concentration from China leads to more concentrated local labor markets
4Slide5
Data
Plant-level data from the Census of Manufacturers (CFM) and the Annual Survey of Manufacturers (ASM)Longitudinal Business Database (LBD) is used to construct measures of concentration of firms’ employment within a county-industry.Use the LBD to construct measures of labor market concentration as it tracks nearly the entire population of establishments at a yearly frequency.Impose additional filters on the data – to enable estimation of average wages and labor productivity
Require that firm-years have at least two plants under their ownership so that we can implement our identification strategy.
Data comprised of 650,000 plant-years over the 1977-2009 period
5Slide6
Figure I: Trends in Average Local-Level Employment Concentration, 1977–2009
6Slide7
Table I: Summary Statistics on Plant Observations from the CMF and ASM Sample
7
Mean
STD
Total value of shipment ($m)
95.95
379.45
Total wage ($m)
16.51
50.63
Total employees
348.56
816.68
Total labor hours (000)
817.25
3955.94
HHI (SIC3-county-year)
0.545
0.350
HHI (SIC4-county-year)
0.682
0.334
HHI (SIC3-county-year) = 1
0.227
0.419
HHI (SIC4-county-year) = 1
0.379
0.485
HHI (SIC3-year)
0.022
0.029
HHI (SIC4-year)
0.044
0.048
Log labor productivity
4.61
0.94
Average wage ($000)
41.84
14.24
Log average wage ($000)
3.67
0.35
Average wage ($000), production
37.61
14.33
log(employment, SIC3-county-year)
6.24
1.68
log(employment, SIC4-county-year)
5.66
1.76
Plants per segment (SIC3)
9.85
15.20
Plants per firm
31.89
39.14
Plant age
17.09
8.99
Unionization rate
0.228
0.129
Observations (plant-years)
656,000
—Slide8
Empirical Model
We begin with the following reduced-form baseline regression:log(avg. wages)pfjt = β0 + β
1HHIjct−1+ β
2
X
pfjt
+ β
3
Z
jct−1
+
δjt
+
µ
ft
+
ε
pfjt
X
pijt
is a vector of plant-level controls (log of labor productivity, log of the number of plants per segment within the firm, log of the number of plants per firm, and plant age).Zjct−1 is one-year lagged log of aggregate employment at the county-industry level. δjt is a vector of either industry or industry-by-year fixed effects. µft is a vector of firm or firm-by-year fixed effects. All standard errors are clustered at the county level.
8Slide9
Our Identification Strategy
Our specifications include either firm or firm-by-year fixed-effects.Thus our identification is achieved using within firm-year variation, comparing multiple establishments belonging to the same firm but located in areas or belonging to industries with varying levels of labor market concentration. We control directly for productivity in our specifications. Most threats to our identification strategy work through a productivity channel.Our battery of controls also removes any common cross-industry variation within firms alleviating cross-industry heterogeneity concerns.
We further show that our results continue to hold in a subsample of firms that operate multiple plants in only one industry segment.
9Slide10
Table II: Local Employer Concentration and Wages
10Panel A: 3-digit SIC Industries
(1)
(2)
(3)
(4)
(5)
(6)
Dep. Var.
Log avg. wages
HHI (SIC3-county-year)
−0.025
−0.044
−0.037
−0.028
−0.023
−0.049
−2.87
−5.41
−5.29
−5.64
−4.31
−9.41
log(employment, SIC3-county-year)
0.042
0.039
0.036
0.026
0.027
0.026
19.08
17.16
21.62
26.99
26.74
27.28
log(labor productivity)
0.156
0.155
0.102
0.095
0.094
0.065
47.19
46.31
55.77
49.09
44.94
36.78
log(plant per segment)
—
−0.042
−0.015
−0.017
−0.020
−0.008
—
−19.20
−10.68
−11.26
−11.42
−5.08
log(plant per firm)
—
0.041
0.021
−0.016
—
—
—
26.63
19.28
−8.96
—
—
Plant age (/100)
—
0.422
0.425
0.405
0.412
0.358
—
17.61
24.27
23.23
23.31
21.94
Year fixed effects
Y
Y
Y
Y
Industry fixed effects
Y
Industry-year fixed effects
Y
Firm fixed effects
Y
Firm-year fixed effects
Y
Y
Observations
656,000
656,000
656,000
656,000
656,000
656,000
R
2
20.45%
22.56%
43.90%
54.99%
62.55%
67.18%Slide11
Table II: Local Employer Concentration and Wages
11
(1)
(2)
(3)
(4)
(5)
(6)
Dep. Var.
Log avg. wages
HHI (SIC4-county-year)
−0.042
−0.063
−0.054
−0.043
−0.038
−0.055
−4.21
−6.51
−8.24
−8.34
−6.90
−10.73
log(employment, SIC4-county-year)
0.034
0.031
0.026
0.020
0.021
0.019
18.93
17.36
27.84
25.2524.8228.54log(labor productivity)0.1520.1510.0900.0940.0930.06043.2443.5153.4747.2743.4035.08log(plant per segment)—−0.036−0.013−0.015−0.017−0.007—−18.51−11.16−11.01−11.13−5.01log(plant per firm)—0.0340.018−0.019———24.8118.03−11.21——Plant age (/100)—0.4230.4300.4140.4210.374—17.3026.3123.6723.6023.32Year fixed effectsYYYYIndustry fixed effectsYYFirm fixed effectsYFirm-year fixed effectsYYObservations656,000656000656,000656000656,000656,000R219.44%21.28%48.37%54.67%62.23%69.22%
Panel B: 4-digit SIC IndustriesSlide12
Single Industry Firms
Baseline results continue to hold in the subsample of firms that operate multiple plants in only one industry segment.Using this subsample in conjunction with firm-by-year fixed effects, we largely sidestep cross-industry heterogeneity as an alternative explanation.
12Slide13
Table IV: Employer Concentration and Wages by Five-Year Time Period
13Panel A: 3-digit SIC Industries
(1)
(2)
(3)
(4)
(5)
(6)
Subsample period:
1977–1981
1982–1986
1987–1991
1992–1996
1997–2001
2002–2009
Dep. Var.:
Log avg. wages
HHI (SIC3-county-year)
−0.018
−0.012
−0.015
−0.031
−0.039
−0.029
−2.75
−1.78
−1.95
−4.59
−5.31
−4.18
log(employment, SIC3-county-year)
0.029
0.031
0.0300.0250.0210.02324.3523.0920.3816.4510.9811.75log(labor productivity)0.1220.1090.0950.0890.0720.08040.0936.5630.1329.3023.0229.07log(plant per segment)−0.009−0.014−0.024−0.028−0.020−0.025−3.87−5.32−10.36−11.70−8.37−10.21Plant age (/100)1.6810.9710.5910.5140.4730.26716.2416.3112.8518.3718.7114.48Firm-year fixed effectsYYYYYYObservations114,00087,000101,000110,000102,000143,000R266.02%65.65%62.84%62.30%56.36%60.87%Slide14
Sample Periods
Negative relation between employer concentration and wages approximately doubles in magnitude over the sample period.Consistent with a secular decline in worker bargaining power over time, as would be predicted byreduction in labor mobility in the United States (constraining the choice-set of workers as they search for employment and negotiate over compensation)drop in unionization rates within the United States beginning in the 1970s (Card 1992)
14Slide15
Table V: Local Employer Concentration, Unions, and Wages
15Panel A: 3-digit SIC Industries
(1)
(2)
(3)
(4)
(5)
(6)
Dep. Var.
Log avg. wages
HHI (SIC3-county-year)
−0.135
−0.153
−0.070
−0.094
−0.087
−0.079
−8.05
−9.40
−6.41
−11.62
−10.39
−9.86
log(employment, SIC3-county-year)
0.039
0.036
0.036
0.025
0.025
0.026
17.68
15.92
21.86
25.3125.0927.63log(labor productivity)0.1480.1480.1020.0920.0920.06548.7348.2155.7749.0944.9036.81Union0.1080.1090.1360.0970.1090.1692.142.154.964.474.781.45HHI x Union0.4070.4020.1450.2370.2280.1317.297.204.098.998.595.59log(plant per segment)—−0.044−0.015−0.018−0.020−0.008—−19.97−10.82−11.98−11.98−5.14log(plant per firm)—0.0400.021−0.016———26.8519.28−8.79——Plant age (/100)—0.4080.4270.3970.4050.358—17.4524.5822.8822.8621.88Year fixed effectsYYYYIndustry fixed effects
Y
Industry-year fixed effects
Y
Firm fixed effects
Y
Firm-year fixed effects
Y
Y
Observations
656,000
656,000
656,000
656,000
656,000
656,000
R
2
21.89%
23.98%
43.98%
55.37%
62.88%
67.20%Slide16
Employer Concentration, Unions, and Wages
Negative relation between the HHI labor market concentration measure and wages is significantly weaker amongst plants in industries with high unionization rates.In industries with unionization rates near zero, a one standard deviation increase in local-level employer concentration is associated with a decline in wages between 2.5% and 5.3%.In contrast, at the average unionization rate, the sensitivity between local level labor market concentration and wages declines by more than a third.
16Slide17
Employer Concentration and Sensitivity of Wage Changes to Productivity Changes
Hypothesize that labor market employer concentration impedes the translation of productivity growth to wage increasesEmployers use monopsony power to avoid wage increases, thereby capturing rents from increased productivityPositive relation between wage growth and productivity growth, measured at plant level (Consistent with Stansbury and Summers 2017 which uses aggregate data)A one standard deviation decrease in HHI from its mean increases the elasticity of production worker wages to productivity by approximately 9.5% (Column 5)
Alternatively: Moving from HHI employer concentration measure of zero to an HHI concentration measure equal to one (representing 22.7 percent of the sample), reduces the elasticity of wages to productivity by approximately 25%
17Slide18
Table IX: Local Employer Concentration and Wages Controlling for Labor Value-Added
18Panel A: 3-digit SIC Industries
(1)
(2)
(3)
(4)
(5)
(6)
Dep. Var.
Log avg. wages
HHI (SIC3-county-year)
−0.025
−0.044
−0.038
−0.028
−0.023
−0.049
−2.91
−5.45
−5.37
−5.61
−4.29
−9.47
log(employment, SIC3-county-year)
0.041
0.038
0.036
0.027
0.027
0.026
17.95
16.16
21.30
26.9926.7427.10log(labor productivity)0.1440.1440.0980.0970.0960.06143.5943.7048.8745.7341.6330.01log(labor VA)0.0160.0150.005−0.002−0.0010.0049.869.145.26−2.65−1.495.35log(plant per segment)—−0.041−0.014−0.017−0.020−0.007—−19.01−10.57−11.32−11.46−5.03log(plant per firm)—0.0410.021−0.016———26.6519.27−8.91——Plant age (/100)—0.4240.4240.4050.4130.357—17.6424.2323.2523.3421.93Year fixed effectsYYYYIndustry fixed effectsYIndustry-year fixed effectsYFirm fixed effectsYFirm-year fixed effects
Y
Y
Observations
656,000
656,000
656,000
656,000
656,000
656,000
R
2
20.67%
22.74%
43.92%
55.00%
62.55%
67.19%Slide19
China Import Penetration
andLocal Employer ConcentrationGrowing body of literature investigates the impact of increased trade with China on labor markets in the United States (Autor, Dorn, and Hanson (2013) and Autor
et al. (2014) Impact on wages, employment, voting patterns, etc.Investigate the effect of import penetration from China on local labor market
concentration.
Hypothesize
that by causing employers to shut down a fraction of their operations, increased import competition from China may have led to an increase in local labor market concentration.
Construct
industry-by-year measure of import penetration from China to
U.S. equal
to
industry
-level dollar value of imports scaled by total value of shipments in the
industry (plant
i
, industry
j
, year
t
):
19Slide20
Conclusion
Analyze effect of local level labor market concentration on wagesUsing manufacturing plant-level data from the U.S. Census from 1977 to 2009 provide five results:Wages significantly lower in local labor markets in which employers are more concentrated.Local-level employer concentration exhibits substantial cross-sectional and time-series variation and
increased over time
Negative
relation between labor market concentration and wages is stronger when unionization rates are low
Link
between productivity growth and wage growth is stronger when labor markets are less
concentrated
Greater exposure
to
“
China Shock
”
is associated with more concentrated labor markets
Argue that results are consistent with firms exploiting local-level monopsony power to reduce wages – particularly when labor bargaining power is weak.
20