What Do Right-to-Work Laws Do? A Case Study Analysis Using
Synthetic Control Method
Ozkan Eren
Louisiana State University
Serkan Ozbekliky
Claremont McKenna College
Using the case study of Oklahoma and a recently developed econometric technique, we examine the impact of right-to-work laws on state-level labor market outcomes. Having eliminated the potential time
variant and invariant confounding e¤ects, our results show that the passage of right-to-work laws in Oklahoma decreased the private sector unionization rates. Several other state outcomes, on the other hand,
were not a¤ected by right-to-work laws. The …ndings for the private sector generally carry over to the
manufacturing sector.
JEL: J51, J58, J21
Keywords: Average Wages, Manufacturing Employment, Right-to-Work Laws, Synthetic Control Method,
Unionization.
Corresponding author. E.J. Ourso College of Business. Department of Economics, Louisiana State University, Baton
Rouge, LA, 70803. Tel: (225)-578-3900. Fax: (225)-578-3807. Email: [email protected].
y
Robert Day School of Economics and Finance, Claremont McKenna College, Claremont, CA 91711. Tel: (909)-607-0721.
Fax: (909)-621-8249. E-mail: [email protected].
1
Introduction
Recent enactments of right-to-work (RTW) laws in Indiana and Michigan have rekindled the debate over
the role of RTW laws on state economies and labor organizations. Opponents of the laws assert that
RTW laws lead to lower wages, lower safety and health standards that protect workers on the job, and
cause workers to receive union representation without incurring the cost. In his December 2012 address
to union workers in Michigan, President Obama stated: “....these so-called right-to-work laws, they don’t
have to do with economics, they have everything to do with politics” (USA Today, December 11, 2012).
Advocates contend that these laws create jobs, lead to higher wages, improve union accountability and
are morally right because they do not compel individuals to support a cause in which they do not believe.
For example, an editorial in Wall Street Journal associates RTW laws with the massive job growth gap
observed in the 1990s between Ohio and Texas (Wall Street Journal, March 3, 2008).1
There were some non-negligible di¤erences between RTW and non-RTW states in terms of economic
progress in the second half of the century. Relative to non-RTW states, RTW states on average grew by 1.5
percentage points faster. Focusing on more recent trends, we see that per capita income growth was higher
in the RTW states in the last three decades (Table 1). Employment growth rate also exhibited a faster
pace in the RTW states during the same time period. We observe similar favorable statistics in terms of
total unemployment and manufacturing employment growth as well. These remarkable di¤erences across
states have led researchers to investigate the economic e¤ects of RTW laws.
From a research standpoint, the evidence regarding the role of RTW laws on state economies is mixed
at best. Some studies …nd signi…cant e¤ects of RTW laws on various state outcomes, while others …nd
no e¤ect (see for example, Hirsch 1980, Holmes 1998, Farber 2005, Lafer and Allegretto 2011). The lack
of consensus among previous studies is likely to be driven by challenges in identifying the e¤ects of RTW
laws. Even in the absence of a causal relation, it is possible to observe signi…cant correlations between
1
In 1990s, Texas added more than one and a half million new jobs, while Ohio lost more than ten thousand jobs. In
addition to RTW laws, the editorial attributes the job growth gap to NAFTA and the absence of state income tax in Texas.
1
RTW laws and state outcomes due to systematic di¤erences across states. For example, it is a well-known
fact that some states have more hostile attitudes towards unions and, among many other factors, these
state-speci…c attitudes may have facilitated the passage of RTW laws in these states. Omission of these
attitudes in the estimation of the impact of RTW laws, say, on private sector unionization is then likely
to yield misleading results. Apart from this, the timing of RTW law enactments adds further challenges.
After the …rst wave between the mid-1940s and mid-1950s, only a few states adopted RTW laws with
long intervals between each instance. In this respect, the usual regression based approach may not be
the best …t in the current context.
The purpose of this paper is to examine the impact of RTW laws on state-level labor market outcomes in Oklahoma, which passed an RTW amendment to the state constitution in 2001. The passage
of RTW laws in Indiana and Michigan in 2012 set aside, Oklahoma is the most recent RTW state with
comprehensive pre- and post-treatment data on our outcomes of interest. To address the aforementioned identi…cation problems, we use the synthetic control method, which was introduced in Abadie and
Gardeazabal (2003) and later extended in Abadie et al. (2010). The synthetic control method is an appealing data-driven procedure to examine the e¤ects of policy interventions, in particular for case studies
where only one (or very few) unit(s) undergoes a treatment in a given sample period. The basic idea is
to construct a weighted combination of control units based on pre-intervention characteristics, which is
expected to provide a better counterfactual than a single control unit (or a simple average of all control
units) for the treated unit(s). The synthetic control method generalizes the usual di¤erence-in-di¤erence
(DD) approach by allowing for time-variant unobserved confounders. In this respect, the estimates of the
policy intervention obtained through the synthetic control method are robust not only to time invariant
unobservables, but also to unobservable confounders that vary over time as long as these confounders
vary in a similar way after the intervention.
Using the synthetic control method and aggregate level data from 1983 to 2007, we …nd that the
passage of RTW laws in Oklahoma signi…cantly decreased private sector unionization rates. On the
2
other hand, at least in the short run, we do not …nd evidence that RTW laws had any impact on total
unemployment rate or on average private sector wages. Further examination of the impact of RTW laws
in the manufacturing sector alone do not alter our results.
2
Background
2.1
Oklahoma and the Right-to-Work Laws
In 1935, with the passage of the Wagner Act, Congress granted organized labor statutory sanction to
…re employees for refusal to join a union. The law gave rise to a movement to curb the additional power
bestowed upon unions at the state level. The 1947 Taft-Hartley amendments to the 1935 Wagner Act
granted states the power to ban the union shop (the so-called Right-to-Work laws), a contract provision
that requires all employees to join and pay dues to the union. By the time of the Taft-Hartley Act, …ve
states had already passed such laws (Arkansas and Florida in 1944, and Arizona, Nebraska and South
Dakota in 1946). Since then and until the most recent cases of Indiana and Michigan in the year 2012,
twenty-two states have passed RTW laws. Figure 1 shows these states and the years the laws were
enacted.
Although the passage of RTW laws in Oklahoma is relatively recent, the debate has gone on for
decades, with the …rst referendum occurring in 1964. An important milestone in the history of the RTW
laws in Oklahoma was the re-election of Governor Francis Keating in 1998. In his inaugural address,
Keating, who was a strong supporter of RTW laws, set the passage of RTW laws as one of the central
goals of his second term. After Republicans gained many seats in the House and Senate in 2000, both
houses voted to submit RTW laws to a popular vote. In September 2001, Oklahomans passed a rightto-work amendment to the state constitution by a margin of 54 percent to 46 percent in a referendum,
which made Oklahoma the 22nd RTW state. Almost immediately, several unions …led a federal lawsuit
seeking to prevent the amendment’s implementation. These attempts, however, did not overturn the
3
amendment.
At the time Oklahoma passed the RTW laws, the state’s economy was in relatively good shape. The
unemployment rate in 2000 stood at 3% and it was not only well below the national average but also
lower than 18 of the then 21 RTW states. Over the decade prior to RTW laws, partly owing to the state’s
concentration in the oil and gas industries, overall employment grew by 22%, which was also ahead of
the national average. During the same time period, per capita income grew by 15%, compared to 16%
growth for the country as a whole.
2.2
Previous Research
As noted, obtaining convincing estimates for the e¤ects of RTW laws on state outcomes is a daunting
task. The major obstacle stems from challenges in distinguishing the e¤ect of RTW laws from other
observable and unobservable state characteristics. The …ndings in the literature on various outcomes are
mixed at best. Using aggregate level data and treating RTW laws as exogenously determined, Hirsch
(1980), Warren and Strauss (1979) obtain negative e¤ects of RTW laws on private sector unionization
rates. Studies that control for unobserved factors, on the other hand, usually …nd no signi…cant e¤ect
of RTW laws on state level private sector unionization (see for example, Farber 1984 and Lumsden and
Peterson 1975). Turning to wage and employment e¤ects, some studies …nd positive and signi…cant
e¤ects, while others …nd no to negative e¤ects of RTW laws (see, for example, Moore and Newman 1985
for a review, Farber 2005, and Kalenkoski and Lacombe 2006).
Among many others, one other notable study is Holmes (1998), who examines the location decision
of manufacturing entrepreneurs. The author did not speci…cally explore the isolated e¤ect of RTW laws,
but instead used it to proxy for a state’s business climate. As such, a state is described as pro-business if
it is an RTW one and anti-business otherwise. In order to circumvent the potential confounding e¤ects
arising from systematic di¤erences across states, Holmes examines the change in manufacturing activity
between the borders of adjacent states with di¤erent RTW laws. The intuition behind this estimation
4
strategy is that the unobserved factors (i.e., attitudes towards unions, fertility of the soil) will be very
similar within a given distance from the border. Holmes (1998) …nds that when one crosses the border
into a pro-business state, the manufacturing employment on average increases roughly by one-third.
However, it is important to note that this increase re‡ects several pro-business policies (i.e., low taxes,
lax environmental regulations) adopted by RTW states and not just the law itself.
3
Empirical Methodology
In its simplest form, the idea of the synthetic control method is to construct a weighted combination
of unexposed (control) units, which is expected to provide a better counterfactual for the (treatment)
unit that is exposed to an intervention. To begin, consider that there are S + 1 states and suppose that
only one state is continuously exposed to an intervention of interest after a time t. Let YitN denote the
outcome that would be observed for any state i (i = 1:::::S + 1) at time t (t = 1::::T ) in the absence
of the intervention and T0 be the number of pre-intervention periods such that 1
T0 < T . Let YitE
denote the outcome of i after being exposed to the intervention of interest in periods T0 to T . This setup
implies that the intervention has no e¤ect before the implementation period; for t 2 f1; :::::T0 g and all
i 2 f1; :::::S + 1g; we have YitN = YitE :
Prior to going further, it is crucial to note that the model relies on the assumption of no interference
between units; the outcomes of the unexposed states are not a¤ected by the intervention. That being
said, the e¤ect of the intervention for state i at time t can be expressed by
it
= YitE
YitN and, borrowing
from the common practice in the treatment literature, the observed outcome can be written as
Yit = YitN +
it Dit ;
where Dit is an indicator that takes the value of one if state i at time t is exposed to the intervention
and zero otherwise. Let us de…ne the only state that is continuously exposed to the intervention after T0
5
(1
T0 < T ) as being the …rst state; then we have Dit = 1 if i = 1 and t > T0 and zero otherwise. In
this setup, we aim to estimate (
1T0 +1 ; :::::;
1T )
and the e¤ect of the intervention for the …rst state for
t > T0 is
1t
= Y1tE
Since Y1tE is observed, in order to estimate
Y1tN = Y1t
Y1tN :
we need to estimate the counterfactual Y1tN : Following
1t ,
Abadie et al. (2010), suppose that Y1tN is given by a factor model
Y1tN =
where
t
t
+
t Zi
+ 't ui +
it ;
(1)
is an unknown common factor with constant factor loadings, Zi is a vector of observed covariates
with the corresponding
t
vector of unknown parameters, 't is a vector of unobserved common factors,
ui is a vector of unknown factor loadings and, …nally,
it
are unobserved transitory shocks at the state
level with zero mean. Now, consider a vector of weights W = (! 2 ; :::::; ! S+1 ) such that ! S
0 for
s = 2; :::::; S + 1 and ! 2 + ::: + ! S+1 = 1: Each value of the vector W represents a potential synthetic
control, which is nothing but a weighted average of the control states. The value of the outcome variable
for each synthetic control indexed by W is
S+1
P
! s Yst =
t
+
t
s=2
S+1
P
s=2
! s Zi + 't
S+1
P
! s ui +
s=2
S+1
P
!s
it :
s=2
Suppose that there are (! 2 ; :::::; ! S+1 ) such that
S+1
P
s=2
! s Ys1 = Y11 ; :::;
S+1
P
s=2
! s YsT0 = Y1T0 ; and
S+1
P
s=2
! s Zs = Z1 :
(2)
As long as the number of pre-intervention periods is large relative to the scale of transitory shocks and
6
under the standard conditions, Abadie et al. (2010) show that
Y1tN
S+1
P
s=2
! s Yst
0:
(3)
Hence, we can obtain an estimate of the e¤ect of the intervention by
^ 1t = Y1t
S+1
P
s=2
! s Yst
for t 2 fT0 + 1; :::::T g:
It is important to note that equation (1) extends the usual DD approach as the model does not impose
't to be constant over time. As is well known, the traditional di¤erence-in-di¤erence method allows for
the presence of time-invariant unobserved confounders, and taking the time di¤erences eliminates these
unobservables. The synthetic control method, on the other hand, allows for the unobserved confounders
to vary. As shown in Abadie et al. (2010), equation (1) implies that a synthetic control can …t Z1 and
a long set of pre-intervention outcomes Y11 ; :::Y1T0 only as long as it …ts Z1 and u1 ; thus,
and
S+1
P
s=2
! s us = u1 holds.
S+1
P
s=2
! s Zs = Z1
The optimal weight vector W to construct the synthetic control is chosen to minimize the distance
between the pre-intervention characteristics of the …rst state and the remaining control states.2 To
determine the signi…cance of the estimated impact using the synthetic control method, Abadie et al.
(2010) suggest the use of “placebo”or “falsi…cation”tests. Similar to the classical permutation tests, the
idea of the placebo test is to apply the synthetic control method to each control unit as if it were exposed
to the intervention and compare the actual estimated e¤ect with that of each control unit. Under the
hypothesis that the intervention has an impact, the actual estimate is expected to be large relative to the
distribution of the placebo estimates.
2
See Abadie et al. (2010) for detailed discussion of the optimal weight vector selection.
7
4
Data
The data for this study is at the state level and comes from several di¤erent sources covering a time
span of 25 years from 1983 to 2007. Appendix A provides detailed information regarding these sources.
We choose 1983 as the initial year because it is the …rst year that the Current Population Survey (CPS)
began collecting monthly information for some of the important variables used in the analysis (e.g., union
status).3 To circumvent any potential confounding e¤ects of the Great Recession, we restrict our analysis
to prior to 2008.4 Oklahoma adopted RTW laws in September 2001, leaving us with a six-year of postintervention data. This period seems like a reasonable time span to observe the impact of the RTW laws
on state level outcomes, at least in the short run.
Our state level outcomes of interest are private and manufacturing sector unionization rates, total
unemployment rate, manufacturing employment rate, and average hourly private and manufacturing
sector wages.5 The set of observed predictors (Z) includes the log of the state’s population and total
land square miles, log of per capita income, and the percentage of the state’s labor force that are white,
male, college graduates and those residing in a metropolitan area. The observed predictors are averaged
over the entire pre-intervention period; moreover, for all outcomes of interest, we augment the relevant
lagged values biennially over the pre-intervention period. The average values of the covariates along with
the lagged outcomes constitute our set of pre-intervention predictors for Oklahoma and all other states.
Figure 2 plots the raw data trends from 1983 to 2007 for the outcome variables. Overall, the trends in
state outcome variables for Oklahoma largely tracks those of the rest of the nation. Even though economic
variables are trending in the same direction, there are some important di¤erences. For instance, private
sector unionization rates have been consistently lower in Oklahoma (Panel A). In Panel D, we see that
the downward trend in manufacturing employment observed in the last three decades is much ‡atter in
3
Beginning in 1983, union status questions were added to the earnings supplement of the monthly CPS outgoing rotation
group …les. The monthly earnings supplement, absent the union status questions, began in January 1979.
4
Several other papers employed a similar approach to minimize the confounding e¤ects that may occur because of the
Great Recession (see, for example, Neumark and Muz 2014).
5
Wages are de‡ated by the year’s consumer price index (2000=100).
8
Oklahoma. Beginning with the early 1990s, we also observe average manufacturing hourly wages falling
behind the national average (Panel F). The …rst two columns of Table 2 also provide the mean of the
selected pre-intervention predictors for Oklahoma and the rest of the states.
5
Empirical Results
5.1
Unionization
Prior to the enactment of the RTW laws, the synthetic control that resembles Oklahoma the most with
respect to the predictors of private sector unionization is constructed using nine states. The rest of the
states are assigned zero weights. Table B1 in Appendix B presents these states. The …rst four states
with the largest weights are North Dakota (22%), followed by South Carolina (19%), Colorado (16%),
and Wyoming (14%). The third column of Table 2 displays the mean of the selected pre-intervention
characteristics of synthetic Oklahoma. The average gap between the characteristics of Oklahoma and its
synthetic counterpart is substantially smaller than it is between Oklahoma and control states (Columns
1-3 of Table 2).
The left column of Panel A of Figure 3 shows the private sector unionization rates from 1983 to 2007
for the actual and the synthetic Oklahoma and highlights several important points. First, consistent with
the overall trend observed in the U.S. in the last three decades, private sector unionization rates have
been declining in Oklahoma. Next, even though it is not perfect, unionization rates for the synthetic
Oklahoma track the trajectory closely for almost the entire pre-intervention period with a root mean
squared prediction error (RMSPE) of 0.54.6 Finally, right after the passage of the law, we observe a
noticeable divergence between the two lines. This …nding provides tentative evidence of a negative impact
of the RTW laws on private sector unionization. The average post-intervention gap is slightly above 0.92
percentage points. That is, if the state of Oklahoma had not adopted the RTW laws, private sector
6
The pre-intervention MSPE is the average of the squared discrepancies for the outcome variable (e.g., private sector
unionization rate) between Oklahoma and its synthetic counterpart during 1983-2001.
9
unionization rates would be 0.92 ( ) percentage points larger. Taking Oklahoma’s union membership
rate of 4.4% (YOK,2000 ) in the year 2000 as our benchmark, the 0.92 percentage point gap corresponds to
a roughly
YO K ,2 0 0 0
=20.6% reduction in private sector unionization rates.
Of course, a natural and an important question to ask at this point regards the precision of this
estimate. To determine this, we conduct a series of placebo studies by iteratively applying the synthetic
control method to all control states. In each iteration, we reassign the passage of the RTW laws to one
of the 49 control states (shifting Oklahoma to the control pool) as if one of the control units had the
intervention in 2001. We then estimate the e¤ect of the RTW laws for each placebo study by taking the
ratio between the post-intervention gap and the state’s unionization rate in the year 2000 and we do this
for each post-intervention year (i.e.,
Yi;2000 ; i
it
= 1:::::S + 1 and t = T0 + 1:::::T ). The right column of Panel
A of Figure 3 presents the results. The grey lines represent the e¤ect of RTW laws on private sector
unionization for each of the 49 states, while the superimposed black line denotes the e¤ect for Oklahoma.
As it is visible from the …gure, the e¤ect line for Oklahoma is unusually large relative to the distribution
of the lines for the control states, which is likely to indicate a “nonrandom” reduction in private sector
unionization rates in Oklahoma after the adoption of RTW laws.
Turning to the manufacturing sector unionization, the synthetic Oklahoma is constructed using eight
states with Utah (38%), Louisiana (15%), Montana (14%) and Wisconsin (10%) being the …rst four states
with the largest weights. Table B2 in Appendix B present the states with positive weights. The RMSPE
between Oklahoma and its synthetic counterfactual is equal to 0.79. Looking at the left column of Panel
B of Figure 3, we see that the trends from private sector unionization carry over here. Speci…cally, the
passage of RTW laws in Oklahoma indicate a roughly
YO K ,
2000
=19.6% reduction in manufacturing sector
unionization rates.
Compared with placebo studies, Oklahoma e¤ect line seems to plunge right after the RTW laws
enactment (right column of Panel B of Figure 3). This sudden (and largest among all lines) drop in the
e¤ect line, however, does not appear to sustain its downward trend. As such, we observe a slight bounce
10
back of the line to its level in 2003.
5.2
Employment
The next set of results pertains to state level employment outcomes. Focusing …rst on total unemployment
rate, the counterfactual that most resembles Oklahoma is built using a weighted combination of Colorado
(32%), South Dakota (19%), Louisiana (17%), Michigan (16%) and Maine (15%) with a pre-intervention
RMSPE of 0.34. The rest of the states are assigned zero weights (Table B3 in the Appendix B). Column
5 of Table 2 shows the selected mean of the pre-intervention characteristics of synthetic Oklahoma. We
observe much more a¢ nity between Oklahoma and its synthetic counterpart relative to Oklahoma and
all other states (Columns 1, 2 and 5 of Table 2). Looking at the right column of Panel C of Figure 3, the
Oklahoma e¤ect line ( YO K ,2 0 0 0 ) is not unusually large relative to placebo lines, which suggests that the
passage of RTW laws had no e¤ect on Oklahoma’s unemployment rate.
As for manufacturing employment, the synthetic Oklahoma is constructed using nine states with a
pre-intervention RMSPE of 0.09. Table B4 in Appendix B present these states and Column 6 of Table
2 provides the selected pre-intervention characteristics for synthetic Oklahoma. Looking at the right
column of Panel D of Figure 3, we do not observe any discernible pattern for the e¤ect line of Oklahoma
from the rest of the placebo lines. This indicates no e¤ect of RTW laws on manufacturing employment
in Oklahoma.
5.3
Wages
Our …nal set of evidence pertains to log average hourly wages. The synthetic control that resembles
Oklahoma the most with respect to the predictors of private sector average wages is constructed using
six states with the …rst four states with the largest weights being Wyoming (54%), Arizona (17%),
Alaska (11%) and Maine (7%). Table B5 in Appendix B presents the rest of the states with positive
weights. The left column of Panel E of Figure 3 plots the log of average hourly wages for the entire
11
period for the actual and synthetic Oklahoma; the pre-intervention RMSPE is 0.01. The two lines almost
perfectly overlap prior to adoption of RTW laws and they begin to diverge with the law. After an initial
divergence, however, the gap narrows up and the lines cross roughly …ve years after the implementation
of the laws. Looking at the Oklahoma e¤ect line, we once again do not observe any unusual divergence
from the placebo lines. This evidence is likely to rule out any e¤ect of the RTW laws on average wages
in Oklahoma.7
Similar to private sector wages, we do not …nd any impact of RTW laws on average manufacturing
wages in Oklahoma (Panel F of Figure 3).8
5.4
Robustness Checks
We undertake additional sensitivity checks to examine the validity of our synthetic control estimates.
First, we restrict our control pool to include only non-RTW states, instead of using all states. The
synthetic control method estimates using only non-RTW states as our control pool produce results that
are similar to those presented in the paper.9 Second, with the exception of wages, we reported the
e¤ect estimates with respect to their values in the year 2000. As an alternative, we calculated the e¤ect
estimates using the ratio between the post-intervention gap and the average of the relevant outcome over
the entire pre-intervention period (i.e.,
it
Yi;pre
intervention
; i = 1:::::S +1 and t = T0 +1:::::T ). The results are
qualitatively similar to those presented in the paper. Third, we extended our set of predictors to include
variables such as states’minimum wage, per capita corporate tax, total road mileage and gross product.
We also dropped all additional predictors and run the speci…cations with only lagged outcome values. The
results remain intact. Next, we implemented a year-by-year matching to construct the synthetic controls.
7
Since wages are expressed in log form, we do not need to take the ratio of the post-intervention gap relative to a
benchmark period.
8
Beginning with 2004, we observe a noticeable divergence in the manufacturing wage trends of Oklahoma and its synthetic
counterpart. To examine this divergence further, we extended our data to cover more recent years. Speci…cally, we use data
from 1983 to 2010. The two lines continue to drift apart during the Great Recession years, reaching a gap of roughly 20% in
2010. However, it is not clear whether this further divergence is solely attributable to RTW laws. These results are available
upon request.
9
Note that the states with positive weights change when we restrict the sample to include only non-RTW states.
12
Speci…cally, rather than augmenting the relevant lagged outcomes biennially over the pre-intervention
period into each synthetic control speci…cation, we added all lagged outcomes from 1983 to 2001. The
results are virtually identical to those presented in the paper. All these additional results are available
upon request.
Finally, as noted in Section 3, the synthetic control method relies on the assumption of no interference
between units. This implies no e¤ect of Oklahoma’s RTW laws on control states’outcomes. There are
not many routes by which this assumption can be violated. One potential scenario is that the passage
of RTW laws may have increased the anti-union sentiment in other states or national unions fearing
a potential epidemic e¤ect may have reacted to the adoption of RTW laws by diverting resources to
Oklahoma from other states. In both cases, interference is likely to lead to lower unionization rates in
the control states and this would imply an attenuation bias in our unionization estimates. The passage of
RTW laws may have also attracted investment to Oklahoma from other states. This would then suggest
that the unemployment (manufacturing employment) gap was upward (downward) biased, while wage
gaps can also be downward biased because of a lower labor demand in the control states. These potential
biases, however, are unlikely to rule out no e¤ect of RTW laws.
6
Conclusion
Since the …rst wave of enactments in the mid-1940s, RTW laws have been at the center of policy discussions. Recent passage of the laws in Indiana and Michigan has also rekindled the debate over the role
of RTW laws in state economies. There is, however, still no clear consensus among policy makers and
researchers. This lack of consensus among researchers predominantly stems from challenges in distinguishing the e¤ect of the laws from other observable and unobservable state characteristics. Even in the
absence of a causal relation, it is possible to obtain signi…cant correlations between RTW laws and state
outcomes.
Using the case study of Oklahoma and a recently developed econometric technique, this paper ex13
amines the impact of RTW laws on various state-level labor market outcomes. Having eliminated the
potential time variant and invariant confounding e¤ects, our results indicate that the passage of RTW
laws in Oklahoma signi…cantly decreased private sector unionization rates. State outcomes, including
total unemployment rate and private sector average wages, on the other hand, were not a¤ected by RTW
laws, at least in the short run. The …ndings for the private sector generally extend to the manufacturing
sector.
Before concluding, there is at least one caveat to keep in mind. Oklahoma is a relatively small state
and even before the passage of RTW laws, private sector unionization was low, compared to the rest
of the nation. Considering other larger non-RTW states with high private sector union densities such
as Wisconsin, where the debate over RTW laws is currently intense, the adoption of the laws may have
di¤erent e¤ects. Nevertheless, the …ndings of the paper may still shed light on the e¤ects of RTW policies
on local labor market outcomes.
14
References
[1] Abadie, Alberto and Javier Gardeazabal. 2003. “The Economic Costs of Con‡ict: A Case Study of
the Basque Country.” American Economic Review 93(1): 113-132.
[2] Abadie, Alberto, Alexis Diamond, and Jens Hainmueller. 2010 “Synthetic Control Methods for
Comparative Case Studies: Estimating the E¤ect of California’s Tobacco Control Program.”Journal
of the American Statistical Association 105: 493-505.
[3] Farber, Henry S. 1984. “Right-to-Work Laws and the Extent of Unionism.” Journal of Labor Economics 2: 319-352.
[4] Farber, Henry S. 2005. “Non-Union Wage Rates and the Threat of Unionization.” Industrial and
Labor Relations Review 58 (3): 335-52.
[5] Hirsch, Barry T. 1980. “The Determinants of Unionization: An Analysis of Interarea Di¤erences.”
Industrial and Labor Relations Review 33: 147-161.
[6] Hirsch, Barry T. and David A. Macpherson. 2003. “Union Membership and Coverage Database from
the Current Population Survey: Note.” Industrial and Labor Relations Review 56 (2): 349-54.
[7] Holmes, Thomas J. 1998. “The E¤ects of State Policies on the Location of Industry: Evidence from
State Borders.” Journal of Political Economy 106(4): 667-705.
[8] Kalenkoski Charlene and Donald Lacombe. 2006. “Right-to-Work Laws and Manufacturing Employment: The Importance of Spatial Dependence.” Southern Economic Journal 73: 402-418.
[9] Lafer, Gordon and Sylvia Allegretto. 2011. “Does Right-to-Work Create Jobs: Answers from Oklahoma.” Economic Policy Insitute Brie…ng Paper 300. Washington, D.C.: EPI.
[10] Lumsden, Keith and Craig Peterson. 1975. “The E¤ect of Right-to-Work Laws on Unionization in
the United States.” Journal of Political Economy 83: 1237-1248.
15
[11] Moore, William J. and Robert J. Newman. 1985. “The E¤ects of Right-to-Work Laws: A Review of
the Literature.” Industrial and Labor Relations Review 38: 571-585.
[12] Neumark David and Jennifer Muz. 2014. “The “Business Climate”and Economic Inequality.”NBER
Working Paper Series. No. 20260.
[13] USA Today. 2012. “Obama: Right-to-Work Laws are Politics.” December 11.
[14] Wall Street Journal. 2008. “Texas v. Ohio.” March 3.
[15] Warren , Ronald S. and Robert P. Strauss. 1979. “ A Mixed Logit Model of the Relationship between
Unionization and Right-to-Work Legislation.” Journal of Political Economy 87: 648-654.
16
Table 1: Average Trends in Various State-Level Outcomes
RTW States
Non-RTW States
(1)
(2)
Per Capita Income Growth Rate
1.82
1.67
Employment Growth Rate
2.57
1.87
Unemployment Rate
5.39
5.73
Manufacturing Employment Growth Rate
-2.26
-3.12
NOTES: Average trends in state-level outcomes from 1980 to 2007.
17
18
4.40
7.20
8.50
11.60
17.50
25.00
3.00
5.70
7.00
11.39
13.26
14.07
2.65
2.75
2.96
2.58
2.64
2.77
15.00
11.14
9.99
87.39
54.13
20.77
59.17
(1)
8.28
10.35
13.77
13.92
18.13
23.43
3.92
6.79
7.34
12.45
15.96
19.20
2.77
2.81
2.91
2.64
2.62
2.70
14.97
10.65
10.11
88.06
53.59
22.40
64.78
(2)
All Control States
4.72
6.86
8.67
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
14.26
11.12
10.03
89.36
53.80
20.75
52.96
(3)
Private
Sector
Unionization
13.09
17.18
24.94
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
14.81
11.05
10.00
91.69
54.19
20.52
69.95
…..
…..
(4)
Manufacturing
Sector
Unionization
…..
…..
…..
…..
…..
…..
3.38
6.42
7.14
…..
…..
…..
…..
…..
…..
…..
…..
…..
14.08
11.13
9.98
92.43
53.85
20.74
38.15
(5)
Unemployment
…..
…..
…..
…..
…..
…..
…..
…..
…..
11.30
13.40
14.04
…..
…..
…..
…..
…..
…..
14.84
11.07
10.10
90.46
53.81
23.21
65.03
(6)
Manufacturing
Employment
Synthetic Control
NOTES: The variables are only a subset of those used in the analysis. The remainder are excluded in the interest of brevity. The full set of sample statistics are available upon request.
Private Sector Unionization Rate (2000)
Private Sector Unionization Rate (1992)
Private Sector Unionization Rate (1984)
Manufacturing Sector Unionization Rate (2000)
Manufacturing Sector Unionization Rate (1992)
Manufacturing Sector Unionization Rate (1984)
Total Unemployment Rate (2000)
Total Unemployment Rate (1992)
Total Unemployment Rate (1984)
Manufacturing Employent Rate (2000)
Manufacturing Employent Rate (1992)
Manufacturing Employent Rate (1984)
Log of Average Private Wages (2000)
Log of Average Private Wages (1992)
Log of Average Private Wages (1984)
Log of Average Manufacturing Wages (2000)
Log of Average Manufacturing Wages (1992)
Log of Average Manufacturing Wages (1984)
Log of Population
Log of Total Land Square Miles
Log of Per Capita Income
White (%)
Male (%)
College Graduate (%)
Living in Metropoitan Area (%)
Synthetic Control Predictors
Oklahoma
Table 2: Summary Statistics: Selected Lagged Outcome Variables and Pre-Intervention Characteristics
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
2.64
2.74
2.94
…..
…..
…..
13.57
11.43
10.07
92.59
54.44
20.36
49.5
(7)
Log of Average
Private
Wages
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
…..
2.59
2.62
2.77
15.18
12.02
10.10
90.78
54.93
20.8
59.52
(8)
Log of Average
Manufacturing
Wages
1947
1985
1946
1963
1947
1951
1946
1955
1947
1958
1947
1947
2001
1946
1954
1944
1954
1953
1947
1947
1976
1954
Ο Right to Work States
Ο No Right-to-Work States
Figure 1: US States by RTW Laws and Enactment Years-Recent enactments (Indiana and Michigan) in 2012 are
excluded.
19
Panel B: Manufacturing Unionization
5
0
Private Sector Unionization Rate
5
10
Manufacturing Sector Unionization Rate
10
15
20
25
15
Panel A: Private Sector Unionization
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Control States
Oklahoma
2
Private Sector Unemployment Rate
4
6
8
Manufacturing Sector Employment Rate
10
12
14
16
18
20
Panel D: Manufacturing Employment
10
Panel C: Unemployment
Control States
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Control States
Oklahoma
Panel F: Manufacturing Average Wages
Log of Average Private Sector Wages
2.6
2.7
2.8
2.9
3
Log of Average Manufacturing Sector Wages
2.5
2.6
2.7
2.8
Panel E: Private Sector Average Wages
Control States
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Control States
Oklahoma
Control States
Figure 2: Trends in State Outcomes. The panels contain the raw data trends (1983-2007) for Oklahoma and
the average of all other states. Wages are de‡ated by the year’s consumer price index (2000=100).
20
2
-.5
Relative Gap in Private Sector
Unionization Rate
0
.5
Private Sector Unionization Rate
4
6
8
1
10
Panel A: Private Sector Unionization
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Synthetic Oklahoma
Oklahoma
Control States
5
-2
Relative Gap in Manufacturing Sector
Unionization Rate
-1
0
1
2
Manufactruing Sector Unionization Rate
10
15
20
25
Panel B: Manufacturing Unionization
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Synthetic Oklahoma
Oklahoma
Control States
2
-.5
Relative Gap in Private Sector
Unemployment Rate
0
.5
Private Sector Unemployment Rate
4
6
8
1
10
Panel C: Unemployment
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
year
Synthetic Oklahoma
Oklahoma
21
Control States
2010
9
-.4
Relative Gap in Manufacturing Sector
Employment Rates
-.2
0
.2
Manufacturing Sector Employment Rate
10
11
12
13
14
.4
Panel D: Manufacturing Employment
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Synthetic Oklahoma
Oklahoma
Control States
-.2
Percentage Gap in Private Sector
Average Wages
-.1
0
.1
.2
Log of Average Private Sector Wages
2.6
2.7
2.8
2.9
3
.3
Panel E: Private Sector Average Wages
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Synthetic Oklahoma
Oklahoma
Control States
-.2
2.5
Percentage Gap in Manufacturing Sector
Average Wages
-.1
0
.1
.2
Log of Average Manufacturing Wages
2.6
2.7
2.8
Panel F: Manufacturing Average Wages
1980
1990
2000
2010
1980
1990
year
Oklahoma
2000
2010
year
Synthetic Oklahoma
Oklahoma
Control States
Figure 3: Trends in Various State Outcomes. Oklahoma vs. Synthetic Oklahoma. The left panels contain
trends (1983-2007) for Oklahoma and synthetic Oklahoma while the right panels indicate the e¤ects of RTW
laws in Oklahoma and the placebo e¤ects in all states. Panels A, B, C and D express the e¤ects relative to
the corresponding outcome rates in the year 2000. The preintervention period is from 1983 to 2001. Wages
are de‡ated by the year’s consumer price index (2000=100).
22
Appendix A: Data Sources
Unionization Rates
Union membership as a percentage of the private sector civilian nonfarm labor force. Manufacturing
unionization rate is de…ned similarly. Source: Union Membership and Coverage Database available at
www.unionstats.com. See Hirsch and Macpherson (2003) for a general description of the database.
Unemployment Rate
Total unemployment as a percentage of the civilian nonfarm labor force. Source: Bureau of Labor
Statistics, Local Area Unemployment Statistics.
Manufacturing Employment Rate
Manufacturing employment as a percentage of the private sector civilian nonfarm labor force. Source:
Bureau of Economic Analysis, U.S. Department of Commerce, Regional Economic Measurement Division,
Regional Economic Information System CD.
Private Sector Average Wage Rates
All values are de‡ated by the year’s average consumer price index (2000=100). Source: Union Membership and Earnings Data Book, Bureau of National A¤airs. The database is constructed by Barry Hirsch and
David Macpherson.
Manufacturing Sector Average Wage Rates
All values are de‡ated by the year’s average consumer price index (2000=100). Source: Bureau of Labor
Statistics, Current Employment Statistics- State and Metro Area Estimates.
State Population
Source: Bureau of Census, Bureau of Economic Analysis.
23
Land Square Miles
Source: Bureau of Census, State and County Quick Facts.
Per Capita State Income Data
All values are de‡ated by the year’s average consumer price index (2000=100). Source: Bureau of
Economic Analysis, U.S. Department of Commerce, Regional Economic Measurement Division, Regional
Economic Information System CD.
Gender, Racial, Residential and Educational Statistics
Percentage of the civilian labor force that are male, white, college graduate and those residing in metropolitan area. Source: Current Population March Supplement.
24
Appendix B: State Weights
Table B1: State Weights in the Synthetic Oklahoma-Private Sector Unionization
State
Alaska
Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Weight
State
0.00
0.03
0.00
0.00
0.00
0.16
0.00
0.00
0.00
0.00
0.05
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Weight
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
NOTES: The synthetic Oklahoma is constructed using the states with positive weights.
The set of predictors include lagged values of the outcome variable, log of the state's
population and total land square mileage, log of per capita income and the percentage
of the state's labor force that are white, male, college graduates and those residing in
metropolitan area. Lagged values of the outcome variable are added biennially over the
entire preintervention period. All other predictors are averaged over the same period.
25
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.22
0.00
0.00
0.00
0.00
0.19
0.09
0.00
0.07
0.00
0.00
0.00
0.00
0.00
0.00
0.14
Table B2: State Weights in the Synthetic Oklahoma-Manufacturing Unionization
State
Alaska
Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Weight
0.00
0.00
0.00
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.08
0.00
0.00
0.00
0.15
0.00
0.00
0.00
0.00
0.00
0.00
0.00
State
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Weight
0.14
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.38
0.00
0.00
0.04
0.00
0.10
0.06
NOTES: The synthetic Oklahoma is constructed using the states with positive weights.
The set of predictors include lagged values of the outcome variable, log of the state's
population and total land square mileage, log of per capita income and the percentage
of the state's labor force that are white, male, college graduates and those residing in
metropolitan area. Lagged values of the outcome variable are added biennially over the
entire preintervention period. All other predictors are averaged over the same period.
26
Table B3: State Weights in the Synthetic Oklahoma-Unemployment
State
Alaska
Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Weight
0.00
0.00
0.00
0.00
0.00
0.33
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.17
0.15
0.00
0.00
0.16
0.00
0.00
0.00
State
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
0
Weight
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.19
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
NOTES: The synthetic Oklahoma is constructed using the states with positive weights.
The set of predictors include lagged values of the outcome variable, log of the state's
population and total land square mileage, log of per capita income and the percentage
of the state's labor force that are white, male, college graduates and those residing in
metropolitan area. Lagged values of the outcome variable are added biennially over the
entire preintervention period. All other predictors are averaged over the same period.
27
Table B4: State Weights in the Synthetic Oklahoma-Manufacturing Employment
State
Alaska
Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Weight
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.24
0.00
0.00
0.00
0.21
0.00
0.02
0.00
0.00
0.00
0.00
0.00
0.09
State
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Weight
0.00
0.00
0.00
0.00
0.00
0.04
0.00
0.00
0.32
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.01
0.00
0.00
0.00
0.06
0.00
0.00
NOTES: The synthetic Oklahoma is constructed using the states with positive weights.
The set of predictors include lagged values of the outcome variable, log of the state's
population and total land square mileage, log of per capita income and the percentage
of the state's labor force that are white, male, college graduates and those residing in
metropolitan area. Lagged values of the outcome variable are added biennially over the
entire preintervention period. All other predictors are averaged over the same period.
28
Table B5: State Weights in the Synthetic Oklahoma-Private Sector Average Wages
State
Alaska
Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Weight
0.11
0.00
0.17
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.06
0.07
0.00
0.00
0.00
0.00
0.00
0.00
State
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Weight
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.54
NOTES: The synthetic Oklahoma is constructed using the states with positive weights.
The set of predictors include lagged values of the outcome variable, log of the state's
population and total land square mileage, log of per capita income and the percentage
of the state's labor force that are white, male, college graduates and those residing in
metropolitan area. Lagged values of the outcome variable are added biennially over the
entire preintervention period. All other predictors are averaged over the same period.
29
Table B6: State Weights in the Synthetic Oklahoma-Manufacturing Average Wages
State
Alaska
Alabama
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Weight
0.11
0.00
0.01
0.00
0.00
0.20
0.00
0.00
0.00
0.00
0.00
0.27
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
State
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
Weight
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.03
0.00
0.36
0.00
0.00
0.00
0.00
0.00
0.00
0.00
NOTES: The synthetic Oklahoma is constructed using the states with positive weights.
The set of predictors include lagged values of the outcome variable, log of the state's
population and total land square mileage, log of per capita income and the percentage
of the state's labor force that are white, male, college graduates and those residing in
metropolitan area. Lagged values of the outcome variable are added biennially over the
entire preintervention period. All other predictors are averaged over the same period.
30