Wage Inequality and Firm Growth By HOLGER M. MUELLER, PAIGE P. OUIMET, AND ELENA SIMINTZI*
* Mueller: New York University, Stern School of Business, 44
inequality, CEO pay has risen significantly
West Fourth Street, New York, NY 10012, NBER, CEPR, and ECGI (e-mail:
[email protected]); Ouimet: University of North Carolina at Chapel Hill, Kenan-Flagler Business School, Campus Box 3490,
McColl
Building,
Chapel
Hill,
NC
27599
over the past decades. Terviö (2008) and Gabaix and Landier (2008) link the rise in
(e-mail:
[email protected]); Simintzi: University of British Columbia, Sauder School of Business, 2053 Main Mall, Vancouver, BC, V6T
CEO pay to changes in the size of the largest firms in the economy. As the authors show,
1Z2, Canada (e-mail:
[email protected]). The authors are
the sixfold increase in CEO pay between 1980
grateful to Nick Bloom (discussant) for helpful comments.
While the dramatic rise in wage inequality over the past decades has been the topic of much research, its causes are still not perfectly understood. 1 In this and our companion paper (Mueller, Ouimet, and Simintzi 2016), we focus on the role of firms. After all, wages are paid by firms. Accordingly, understanding wage inequality at the firm level may shed light on its causes and ultimately may help us understand better trends in aggregate wage
attributed to the sixfold increase in the size of the largest companies during that period. In addition, variation in CEO pay closely tracks variation in firm size in the cross-section as well as across countries. Terviö (2008) and Gabaix and Landier (2008) both argue that the link between CEO pay and firm size is consistent with the efficient assignment of CEOs in market equilibrium. Building on Rosen’s (1981)
inequality. Mainly due to data availability, much of the empirical literature on pay practices at the firm level has focused on pay at the very top of the firm’s hierarchy: CEO compensation.
and 2003 in the United States can be fully
2
Along with the rise in aggregate wage
1 See Acemoglu and Autor (2011) and Atkinson, Piketty, and Saez (2011) for reviews of the literature. For recent empirical examinations of wage inequality in the United States and elsewhere, see, e.g., Card, Heining, and Kline (2013), Barth et al. (2016), Alvarez et al. (2016), and Song et al. (2016). 2 Friedman and Jenter (2010) and Edmans and Gabaix (2016) provide detailed reviews of the CEO pay literature.
economics of superstars, the authors suggest that “the economic impact of a manager’s decisions depends on the amount of resources under his control” (Terviö 2008, p. 642). 3 Accordingly, efficient matching posits that more talented CEOs should match with larger
3
See also Rosen (1982, p. 311): “Assigning persons of superior talent to top positions increases productivity by more than the increments of their abilities because greater talent filters through the entire firm by a recursive chain of command technology. These multiplicative effects support enormous rewards for top level management in large organizations.”
firms, implying that CEOs’ equilibrium pay is
changes in firm size can potentially explain
increasing with firm size.
trends in aggregate wage inequality, akin to
Although the rise in CEO pay over the past
the arguments made by Terviö (2008) and
decades has been spectacular, it can hardly
Gabaix and Landier (2008) in the context of
explain the rise in aggregate wage inequality.
CEO compensation.
Indeed,
commonly
used
measures
of
aggregate wage inequality, such as the 90/10
I. More Pay Inequality at Larger Firms
log wage differential, are unlikely to be much
In our companion paper (Mueller, Ouimet,
affected by CEO pay given that the income of
and Simintzi 2016), we explore the relation
th
most CEOs lies comfortably above the 90
between within-firm pay inequality and firm
percentile of the aggregate wage distribution.
size using firm-level survey data on employee
However, this does not mean that the link
pay for a broad cross-section of private and
between firm size, managerial talent, and pay
public firms in the United Kingdom for the
proposed by Terviö (2008) and Gabaix and
period from 2004 to 2013. Our final sample
Landier (2008) cannot be useful in explaining
consists of 880 firms. The survey data are
wage inequality more broadly. Many jobs
provided by Income Data Services (IDS), an
within a firm involve managerial skills and
independent research and publishing company
responsibilities. If more talented managers
specializing in the field of employment.
match with larger firms, then we should see
Important for our purposes, employers are
pay increasing with firm size not only at the
asked to group job titles into broad hierarchy
very top of the firm’s hierarchy but also in
levels based on required skills and tasks,
other hierarchy levels where managerial talent
including managerial responsibilities. Hence,
is important. By contrast, the pay of lower-
if a particular job title has different meanings
level employees should be largely invariant
at different firms (e.g., different managerial
with respect to firm size given that the actions
responsibilities), then it will be assigned to
of these employees are less scalable across the
different hierarchy levels.
firm. Taken together, these arguments imply
To give some examples, hierarchy level 1,
that pay differentials between top- and
our lowest hierarchy level, includes work that
bottom-level jobs within a firm should be
“requires basic literacy and numeracy skills
increasing with firm size. Lastly, if pay
and
inequality increases with firm size, then
straightforward
the
ability and
to
perform
short-term
a tasks
few to
instructions under immediate supervision.”
when lower hierarchy levels are compared to
Typical job titles are cleaner, laborer, and
one another, an increase in firm size has no
unskilled worker. By contrast, hierarchy level
significant
9, our highest hierarchy level, includes “very
inequality.
effect
on
within-firm
pay
substantial
Also, whenever the coefficient on firm size
experience in, and leadership of, a specialist
is significant, it is monotonically increasing in
function,
the
the pay ratio. For instance, moving from the
organisation’s overall strategy.” Typical job
25th to the 75th percentile of the firm-size
titles are HR director, finance director, and
distribution raises the pay associated with
head of legal.
hierarchy level 9 by 280.1% relative to the pay
senior
executive
roles
including
with
some
input
to
To obtain measures of within-firm pay
associated
with
hierarchy
level
1.
By
inequality, we compute for all (9×8)/2 = 36
comparison, the pay associated with hierarchy
hierarchy-level pairs the ratio of average
level 6 increases (only) by 59.7%, that
wages within a given firm and year (“pay
associated with hierarchy level 7 increases by
ratio”). For example, pay ratio 19 compares
138.2%, and that associated with hierarchy
the average pay of top-level executives, such
level 8 increases by 253.3%—all relative to
as finance and HR directors, with that of
the pay associated with hierarchy level 1.
employees at the bottom of the firm’s
Hence, a given increase in firm size has a
hierarchy, such as cleaners and unskilled
roughly 4.7 times bigger impact on pay ratio
workers, within the same firm and year.
19 than it has on pay ratio 16. (The complete
In order to examine the relation between within-firm pay inequality and firm size, we
set of results is described in Mueller, Ouimet, and Simintzi 2016.)
regress each pay ratio on the number of firm-
In sum, our results show that larger firms
level employees. Thus, we run 36 separate
exhibit significantly more pay inequality:
regressions. In spite of the large number of
wage differentials between top- and bottom-
regressions, our results reveal a remarkably
level
clear pattern. When higher hierarchy levels (6
between different top-level jobs—are all
to 9) are compared to either one another or
increasing with firm size. By contrast, wage
lower
pay
differentials comparing lower hierarchy levels
differentials between different hierarchy levels
to one another are invariant with respect to
are increasing with firm size. By contrast,
firm size. Hence, an HR director’s pay (level
hierarchy
levels
(1
to
5),
jobs—but
also
wage
differentials
9) increases relative to the pay of an unskilled
some of the rise in aggregate wage inequality
worker (level 1) as firm size increases.
over time. To address this question, we have
However,
ordinary
gathered wage data for a large number of
HR/Personnel officer (level 4) does not
developed countries. 4 The data are provided
increase relative to the pay of an unskilled
by LIS, formerly known as The Luxembourg
worker. Our results are consistent with
Income Study. LIS data are particularly well
theories emphasizing the efficient assignment
suited for our purposes as they use official
of managerial talent: while pay inequality is
data collected from individual countries’
increasing with firm size, the result is entirely
statistical offices. The data include labor
driven by hierarchy levels where managerial
income for a broad cross-section of employees
skills and responsibilities are important.
in a given country and year. We limit our
the
pay
of
an
Can our analysis say something about the
sample to full-time employees by excluding
possible causes of rising wage inequality over
employees identified as part-time and those
time? To address this issue, we focus on
that report working less than 35 hours per
within-firm variation by including firm fixed
week.
effects. That is, we examine the relation
employees in a given country and year, we
between changes in pay inequality at the firm
estimate the 10th and 90th percentiles of the
level and firm growth. Consistent with our
aggregate wage distribution. Our measure of
cross-sectional results, we find that wage
aggregate wage inequality is the commonly
differentials between top- and bottom-level
used 90/10 log wage differential.
Using
the
sample
of
full-time
jobs—but also wage differentials between
We source firm-size data from Thomson
different top-level jobs—become significantly
Reuters’ Worldscope. Worldscope provides
larger as firms grow over time. By contrast,
data on firm fundamentals for publicly listed
wage differentials comparing lower hierarchy
firms in many countries. Similar to the
levels to one another are unrelated to firm
analyses in Terviö (2008) and Gabaix and
growth.
Landier (2008), we focus on the largest firms in a given country. (This is also consistent
II. Wage Inequality and Firm Growth in Developed Countries The within-firm analysis raises the question of whether firm growth can possibly explain
4
The cross-country analysis described in this section is not part of our companion paper (Mueller, Ouimet, and Simintzi 2016). In our companion paper, we focus entirely on UK firms. In addition to showing that high-inequality firms are larger, we also show that they exhibit better operating performance, higher valuations, and larger equity returns, consistent with the notion that higher pay inequality is a reflection of better managerial talent.
with our firm-level analysis in Section I: the
Table 2 examines the relationship between
average firm in our UK sample has 10,014
aggregate wage inequality, expressed through
employees.) Specifically, we calculate the
the 90/10 log wage differential, and the
average number of employees for either the 50
average number of employees (in logs) of the
or 100 largest firms in a given country and
50 (100) largest firms in a country. The
year. LIS data are not available for every year
regressions in columns (1) and (4) include
and, for most countries, there is a gap of
country and year fixed effects. As is shown,
several years between surveys. On average,
there is a positive and significant association
we have wage data for six different years for
between changes in wage inequality and
the countries in our sample. Our final sample
employment growth by the largest firms in a
consists of all countries for which we have
country. The regressions in columns (2), (3),
both wage data and firm-size data: Australia,
(5), and (6) include country fixed effects but
Austria, Belgium, Canada, Denmark, Finland,
no year fixed effects. Instead, they include a
France, Germany, Greece, Italy, Netherlands,
linear time trend defined as the given year
Spain, Sweden, United Kingdom, and United
minus 1999. In columns (2) and (5), the
States.
positive coefficient on the time trend is reflective of the rise in aggregate wage [ Insert Table 1 Here ]
inequality
during
the
sample
period.
Table 1 shows for each country the sample
Importantly, in columns (3) and (6), adding
period, number of country-year observations,
firm growth to the regression reduces the
and change in average employment among the
coefficient on the time trend by 36.1% and
50 (100) largest firms during the sample
39.8%, respectively. Consequently, part of
period. As can be seen, firm growth is
what may be perceived as a global trend
pervasive among large publicly listed firms
toward more wage inequality may be driven
during the sample period. With the exception
by an increase in the size of the largest firms
of
in the economy.
Denmark,
the
change
in
average
employment among the top 50 (100) firms is
III. Conclusions
positive in all countries. We discuss firm-level evidence based on [ Insert Table 2 Here ]
UK data showing that within-firm pay inequality increases with firm size. Moreover,
as firms grow larger over time, within-firm
the Rise of West German Wage Inequality.”
pay inequality rises. Lastly, using wage data
Quarterly Journal of Economics 128 (3):
from 15 developed countries, we document a
967—1015.
positive association between aggregate wage
Edmans, Alex, and Xavier Gabaix. 2016.
inequality at the country level and growth by
“Executive
Compensation:
A
Modern
the largest firms in the economy.
Primer.” Journal of Economic Literature 54 (4): 1232—87.
REFERENCES
Frydman, Carola, and Dirk Jenter. 2010.
Acemoglu, Daron, and David Autor. 2011. “Skills,
Tasks,
Implications
and
for
Earnings.”
In
Economics,
Vol.
Technologies:
Employment
Handbook 4,
of
edited
by
Gabaix, Xavier, and Augustin Landier.
Labor
2008. “Why Has CEO Pay Increased So
Orley
Much?” Quarterly Journal of Economics
Alvarez, Jorge, Felipe Benguria, Niklas Engbom, and Christian Moser. 2016. the
123 (1): 49—100. Mueller, Holger M., Paige P. Ouimet, and
Amsterdam: Elsevier.
and
Financial Economics 2: 75—102.
and
Ashenfelter and David Card, 1043—1171.
“Firms
“CEO Compensation.” Annual Review of
Decline in
Earnings
Elena Simintzi. 2016. “Within-Firm Pay Inequality.” Unpublished. Rosen, Sherwin. 1981. “The Economics of Superstars.” American Economic Review 71
Inequality in Brazil.” Unpublished. Atkinson, Anthony B., Thomas Piketty, and
(5): 845—58.
Emmanuel Saez. 2011. “Income Inequality
Rosen, Sherwin. 1982. “Authority, Control,
in the Long Run.” Journal of Economic
and the Distribution of Earnings,” Bell
Literature 49 (1): 3—71.
Journal of Economics 13 (2): 311—23.
Barth, Erling, Alex Bryson, James C. Davis,
Song, Jae, David J. Price, Fatih Guvenen,
and Richard Freeman. 2016. “It’s Where
Nicholas Bloom, and Till von Wachter.
You Work: Increases in the Dispersion of
2016.
Earnings
Unpublished.
across
Establishments
and
Individuals in the United States.” Journal of Labor Economics 34 (2): S67—S97. Card, David, Jörg Heining, and Patrick Kline. 2013. “Workplace Heterogeneity and
“Firming
Up
Inequality.”
Terviö, Marko. 2008. “The Difference That CEOs
Make:
An
Assignment
Model
Approach.” American Economic Review 98 (3): 642—68.
TABLE 1— FIRM GROWTH IN DEVELOPED COUNTRIES Top 50 Firms
Top 100 Firms
Country
Sample Period
Obs.
Change in Firm Size
Sample Period
Obs.
Change in Firm Size
Australia
1985 - 2001
4
37.1%
1995 - 2001
2
16.2%
Austria
1994 - 2004
4
82.8%
1997 - 2000
2
18.3%
Belgium
1988 - 2000
5
112.2%
1992 - 2000
4
35.4%
Canada
1981 - 2010
10
73.1%
1981 - 2010
10
80.7%
Denmark
1995 - 2010
5
-2.1%
1995 - 2010
5
-4.3%
Finland
1987 - 2010
7
58.6%
1991 - 2010
5
46.7%
France
1994 - 2005
3
48.3%
1994 - 2005
3
40.3%
Germany
1984 - 2010
7
91.0%
1984 - 2010
7
87.3%
Greece
1995 - 2010
5
192.6%
1995 - 2010
5
201.7%
Italy
1987 - 2010
10
31.5%
1987 - 2010
10
30.3%
Netherlands
1983 - 2010
8
107.9%
1987 - 2010
7
87.1%
Spain
1995 - 2010
5
200.3%
1995 - 2010
5
185.9%
Sweden
1987 - 1995
3
13.6%
1987 - 1995
3
15.5%
United Kingdom
1986 - 2010
8
51.3%
1986 - 2010
8
43.5%
United States
1986 - 2010
8
55.8%
1986 - 2010
8
53.0%
TABLE 2— AGGREGATE WAGE INEQUALITY AND FIRM GROWTH
(1) Firm Size
Top 50 Firms (2)
0.211*** (0.0739)
Time Trend
Observations R-squared
(3)
(4)
0.145**
0.206***
(0.0671)
(0.0609)
Top 100 Firms (5)
(6) 0.183*** (0.0562)
0.0104***
0.00656***
0.0111***
0.00668***
(0.00139)
(0.00193)
(0.00127)
(0.00172)
92
92
92
84
84
84
0.892
0.857
0.863
0.944
0.914
0.923
Notes: The dependent variable is the 90/10 log wage differential. Firm size is the average number of employees (in logs) of the 50 or 100 largest firms in the country. If there are fewer than 100 firms with available employment data in a given country and year, the country-year observation is dropped from the top 100 sample. Time trend is the given year minus 1999. All regressions include country fixed effects. Those in columns (1) and (4) additionally include year fixed effects. Robust standard errors are in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level.