The effect of education on the minimum wage

The effect of education on the minimum wage
Christos Pargianas
University of Scranton, Scranton, PA, USA
ABSTRACT
This research shows for the first time that the level of education has a causal, negative effect on
the minimum wage. I use 2SLS, with historical educational data as an instrument for the level of
education in 2010, and I find that across the US states a one percentage point greater proportion
of college graduates is associated with a real minimum wage that is lower by 1.5%–1.6%. Also, in
order to control for state-level omitted variables, I regress the change in the minimum wage on
the change in education and I find again a negative, and significantly at the 1% level, effect.
Minimum wage is a policy that is chosen by governments according to voters’ preferences. The
results of this research imply that when the level of education increases voters prefer a lower
minimum wage.
KEYWORDS
Education; minimum wage;
2SLS regressions; voter
preferences
JEL CLASSIFICATION
H00; I28; H27
I. Introduction
This research is related to political economy papers
that examine how education affects the economy
through institutions and policies.1 Glaeser and Saks
(2005) show that an increase in education reduces
corruption, and Glaeser, Ponzetto, and Shleifer
(2007) argue that an increase in education leads to
more democracy. This research shows that an
increase in education reduces the minimum wage.
Pargianas (2015; Forthcoming) explain theoretically
how education affects institutions and policies and,
therefore, they provide theoretical support for
Glaeser and Saks (2005) and this research. More
specifically, they argue that college graduates and
unskilled individuals prefer different policies and
governments choose the policies that benefit the
group with the greater political power. An increase
in the proportion of college graduates increases their
political power and, therefore, policies change in
their favour. College graduates do not earn the minimum wage and this means that they do not lose
when the minimum wage decreases. On the other
hand, they are either entrepreneurs who employ
unskilled workers and pay to them the minimum
wage or employees who could receive a pay rise if
their employer’s profit increases after a decrease in
the minimum wage that reduces the unskilled labour
cost. As a result, a decrease in the minimum wage is
a policy they prefer.
Several papers have examined how a change in
the minimum wage affects wage inequality. More
specifically, DiNardo, Fortin, and Lemieux (1996)
and Lee (1999) find that during the 1980s, the
decline in the real value of the minimum wage
explains 20%–50% of the rise in wage dispersion,
80% of the growth in within-group wage inequality
and about 15% of the change in the return to schooling. This evidence suggests that the minimum wage
significantly affects economic outcomes and so the
question ‘what determines the minimum wage’ is of
great importance. This research shows for the first
time that the level of education has a causal negative
effect on the minimum wage.
II. Empirical evidence
Data description
The analysis in this section uses minimum wage data
from the Bureau of Labor Statistics Monthly Labor
Review, January Issues. In every state, the minimum
wage value that is used is the largest between the
federal minimum wage and the state minimum wage
CONTACT Christos Pargianas [email protected] 1
Earlier research has argued that education affects the economy, first, because human capital is a factor in the production of final output, and, second,
because it increases innovation and imitation rates. See, for example, Mankiw, Romer, and Weil (1992). In addition, Galor (2005) argues that education
affects people’s ability to adapt and remain productive in an economy that changes rapidly because of technological progress.
APPLIED ECONOMICS LETTERS, 2016
VOL. 23, NO. 11, 765–767
http://dx.doi.org/10.1080/13504851.2015.1105917
© 2015 Taylor & Francis
Downloaded by [EBSCO Publishing Distribution 2010], [Paige Riordan] at 21:08 20 May 2016
in the January of each year. For the construction of
the real minimum wage I follow Lee (1999), and I
divide the minimum wage by the median income for
each state. The source for the median income, the
Gini index, the percentage of blacks and Latinos, the
composition of the four US regions, and the education data across states, is the US Census Bureau. The
measure that is used for education is the fraction of
individuals 25 or older with at least a bachelor’s
degree.
Results
Table 1 presents the results of a simple OLS regression of the 2010 log real minimum wage on the
proportion of college graduates and other controls
in 2000.2 The controls I use are, first, the dummies
of the US regions, which control for region specific
characteristics, second, the proportion of blacks and
Latinos, which capture racial dissimilarity and, third,
the Gini index, which captures economic dissimilarity. In all cases, we can see that states with higher
proportion of college graduates are associated with
lower real minimum wage and that this effect is large
and always statistically significant at the 1% level.
The coefficients in the four specifications I use are
–.011 and –.012, and imply that a one percentage
point increase in the proportion of college graduates
decreases the real value of the minimum wage by
1.1%–1.2%. None of the control variables is statistically significant.
The problem with OLS is, first, that there might
be omitted variables that affect both education and
the minimum wage and, second, that there might be
reverse causality. The identification strategy I use in
order to address these issues is the same as the one
used by Glaeser and Saks (2005) and Barro and Lee
(2010). More specifically, I use the 2SLS method
with historical educational data as an instrument.3
The only potential issue that remains is that the
instrument may depend on state-specific omitted
variables which do not change over time and which
are the true determinants of the minimum wage. I
capture most of them, but not all, by including
region dummies in the regressions. The best way to
avoid all the bias from omitted state-level variables is
to estimate regressions with state fixed effects which
asks whether changes in education predict changes
in the minimum wage. I report the results of these
regressions in Table 2. Both the 2SLS and the fixedTable 1. OLS and 2SLS regressions of the real minimum wage on the proportion of college graduates.
OLS 2SLS
Dependent variable: log real minimum wage in 2010 (1) (2) (3) (4) (5) (6) (7) (8)
Proportion of college graduates in 2000 −.012*** −.012*** −.012*** −.011*** −.016*** −.016*** −.016*** −.015***
(.0019) (.0021) (.0018) (.0019) (.0020) (.0022) (.0018) (.0022)
Northeast .020 .021 .010 −.005 .040* .041** .032 .021
(.0213) (.0217) (.0254) (.0276) (.0236) (.0235) (.0268) (.0297)
South .018 .018 .003 .002 .010 .005 −.004 −.005
(.0128) (.0162) (.0183) (.0179) (.0097) (.0155) (.0145) (.0159)
West .034 .037 .026 .031 .040* .040 .033 .036
(.0205) (.0240) (.0216) (.0220) (.0218) (.0249) (.0232) (.0234)
Blacks 2000 .000 −.001 .000 −.001
(.0008) (.0010) (.0007) (.0010)
Latinos 2000 .000 −.002 .000 −.001
(.0009) (.0012) (.0008) (.0011)
Gini 2000 .402 1.017** .382 .717
(.3411) (.4597) (.3171) (.4539)
Constant −3.57*** −3.57*** −3.73*** −3.98*** −3.47*** −3.47*** −3.61*** −3.76***
(.0428) (.0450) (.1369) (.1910) (.0422) (.0445) (.1346) (.2009)
R2 .61 .61 .62 .65 .55 .55 .56 .60
1st stage F 29.73 20.07 27.27 18.09
# Obs. 50 50 50 50 50 50 50 50
Note: Robust standard errors appear in brackets. *, ** and *** denote statistical significance at the 10%, 5% and 1% level, respectively. In 2SLS regressions,
the instrument is the proportion of college graduates in 1960.
2
I use year 2000 independent variables because according to the theory (see the Introduction section to this paper), education affects the minimum wage
through the political process which is slow. Thus, it takes time to see the political effect of a change in the level of education. The results do not change if I
use year 2010 independent variables. 3
I use education data of year 1960 because is the earliest year with available data for all the states. The earliest data are from year 1940 but in 1940’s data
two values are missing (Alaska and Hawaii). The results do not change if I use year 1940 education as an instrument but in this case the instrument turns
out to be weak in some specifications.
766 C. PARGIANAS
Downloaded by [EBSCO Publishing Distribution 2010], [Paige Riordan] at 21:08 20 May 2016
effects methods confirm the significant negative
effect of education on the minimum wage.
More specifically, Table 1 also presents the
results of the 2SLS regressions. I use the same
specifications as in the OLS regressions and, again,
states with higher proportion of college graduates
are associated with lower real minimum wage. This
effect is larger in the 2SLS than in the OLS regressions, the coefficients of the 2SLS regressions are
between –.015 and –.016, and always statistically
significant at the 1% level. The variable ‘northeast’
is significant in two of the four specifications and
the variable ‘west’ in one. The variables that capture
economic and racial dissimilarity are always insignificant. Also, the instrument is not weak as we can
see from the first stage F-values, which are always
sufficiently high. Finally, in Table 2, I present the
results of the regressions of the difference in the
minimum wage on the difference in the proportion
of college graduates and the difference in other
controls, in order to avoid the bias from statespecific omitted variables. The main coefficient is
a little smaller, in absolute value, than in the OLS
and 2SLS regressions but always significant at the
1% level. Also, the coefficient of the change in the
Gini index is positive and significant.
III. Conclusions
This research shows that education has a causal
effect on the minimum wage in the USA. More
specifically, a one percentage point increase in the
proportion of college graduates decreases the
minimum wage by 0.7%–1.6%. A very interesting
topic for future research would be to examine the
effect of education on other economic policies like
the level of taxation, government spending, labour
market regulations, etc. Also, it would be interesting
to examine whether there are other variables that
predict the minimum wage and other policies. In
order to identify these variables, we need to have in
mind that policies change when the political power
of the different groups changes. As a result, variables
that affect the balance of the political power are
probably the best candidates.
Disclosure statement
No potential conflict of interest was reported by the author.
References
Barro, R. J., and J. Lee. 2010. “A New Data Set of Educational
Attainment in the World, 1950-2010.” NBER Working
Paper No. 15902, doi:10.3386/w15902.
DiNardo, J., N. M. Fortin, and T. Lemieux. 1996. “Labor
Market Institutions and the Distribution of Wages,
1973-1992: A Semiparametric Approach.” Econometrica
64: 1001–1044. doi:10.2307/2171954.
Galor, O. 2005. “From Stagnation to Growth: Unified
Growth Theory.” In Handbook of Economic Growth, edited by Philippe Aghion and Steven Durlauf, 1: 171–293.
doi:10.1016/S1574-0684(05)01004-X.
Glaeser, E. L., G. A. M. Ponzetto, and A. Shleifer. 2007. “Why
Does Democracy Need Education?” Journal of
Economic Growth 12: 77–99. doi:10.1007/s10887-007-
9015-1.
Glaeser, E. L., and R. E. Saks. 2005. “Corruption in America.”
Journal of Public Economics 90: 1053–1072. doi:10.1016/j.
jpubeco.2005.08.007.
Lee, D. S. 1999. “Wage Inequality in the United States
During the 1980s: Rising Dispersion or Falling Minimum
Wage?” The Quarterly Journal of Economics 114: 977–
1023. doi:10.1162/003355399556197.
Mankiw, N. G., D. Romer, and D. N. Weil. 1992. “A
Contribution to the Empirics of Economic Growth.” The
Quarterly Journal of Economics 107: 407–437. doi:10.2307/
2118477.
Pargianas, C. 2015. “Endogenous Economic Institutions and
Persistent Income Differences among High Income
Countries.” Open Economies Review 1–24. doi:10.1007/
s11079-015-9363-y. May 2015.
Pargianas, C. Forthcoming. “Endogenous Political
Institutions, Wage Inequality, and Economic Growth.”
Macroeconomic Dynamics. Advance online publication.
doi:10.1017/S1365100515000449.
Table 2. OLS regressions of the change in the real minimum
wage on the change in the proportion of college graduates.
Dependent variable:
difference in log real
minimum wage 1980–2010
OLS
(1) (2) (3) (4)
Difference in the
proportion of college
graduates in 1970–2000
−.007*** −.007*** −.009*** −.009***
(.0016) (.0018) (.0016) (.0017)
Difference in Share Black
1970–2000
.002 −.001
(.0037) (.0026)
Difference in Share Latino
1970–2000
.001 −.002
(.0010) (.0013)
Difference in Gini
1970–2000
.745*** .984***
(.2470) (.2998)
constant −.024 −.022 −.034* −.036*
(.0202) (.0229) (.0193) (.0212)
R2 .21 .22 .36 .38
# Obs. 50 50 50 50
Note: Robust standard errors appear in brackets. *, ** and *** denote
statistical significance at the 10%, 5% and 1% level, respectively.
APPLIED ECONOMICS LETTERS 767
Downloaded by [EBSCO Publishing Distribution 2010], [Paige Riordan] at 21:08 20 May 2016
Copyright of Applied Economics Letters is the property of Routledge and its content may not
be copied or emailed to multiple sites or posted to a listserv without the copyright holder’s
express written permission. However, users may print, download, or email articles for
individual use.