New evidence on the effect of right-to-work laws on productivity and population growth.
Hicks, Michael J. ; LaFaive, Michael ; Devaraj, Srikant 等
In his study of right-to-work (RTW) laws, Richard Vedder (2010)
outlined the classical-liberal argument regarding workplace liberty and
offered evidence of the effects of RTW legislation on employment and
output in individual states. He found that RTW laws have a positive
impact on both jobs and output as firms and workers move to states with
greater economic freedom.
This article extends Vedder s work by examining the impact of RTW
laws on productivity and population growth. We begin with a review of
the literature on both issues. Second, we reprise the exposition of
labor demand theory offered by Hicks and LaFaive (2013), directly tying
their work to estimates of total factor productivity (TFP), the Solow
residual, and labor productivity across RTW and non-RTW states. Third,
we re-evaluate earlier estimates of the effect of RTW laws on population
growth offered by Hicks (2012) and Hicks and LaFaive (2013). In doing
so, we incorporate an identification strategy introduced by Hicks (2012)
designed to adjust for population growth unrelated to RTW legislation.
The Impact of Right-to-Work Legislation: A Review of the Literature
The National Labor Relations Act of 1935 (the Wagner Act) did not
permit states to restrict union contracts from mandating universal union
membership within a union-represented establishment. In 1947, the
Taft-Hartley Act allowed states to opt out of this requirement and
allowed employees to avoid union membership and payment of union dues.
The affirming vote of states to permit this union opt-out has become
known as RTW legislation.
Some state legislatures actually voted for RTW laws prior to
passage of the Taft-Hartley Act. The states that first passed RTW laws
were heavily concentrated in the South, Soudiwest, and Great Plains.
Those regions were not heavily industrialized at the time and did not
possess large transportation sectors. Furthermore, there was
considerable cultural opposition in the South toward the labor movement
(Black and Black 1989). Since the Taft-Hartley Act, 25 states have
passed RTW legislation, with Indiana doing so twice. The two most recent
adopters are in tire heavily unionized manufacturing states of Indiana
and Michigan.
A large body of analysis has been performed regarding the impact of
RTW legislation. The role of RTW legislation on unionization levels and
rates has a long empirical history (Dickens and Leonard 1985, Freeman
1985, Farber 1987, Lazear 1988, Reder 1988, Jarley and Fiorito 1990,
Moore 1998) and generally concludes that RTW laws reduce private-sector
union participation, although some later works (e.g., Koeller 1994) find
no impact.
Newman (1983) focused on RTW laws in the South from the 1940s
through 1970s and found that RTW legislation was a significant
contributor to population growth as labor-intensive manufacturing firms
moved to RTW states.
Holmes (1998) extended Newman's work by using RTW as an
identification tool to siphon out the impact of other business-friendly
policies on firm location at the county level. His study was especially
important in that it included a broad range of business-friendly
policies in a carefully executed study of counties in different states
but with contiguous borders. Holmes (1998) reports a very large increase
in manufacturing employment in places with business-friendly policies
and no unusual geographic complications. For example, he notes that
while Louisiana and Mississippi are both RTW states, their border shows
stark differences in manufacturing location because Louisiana has a long
history of unfriendly policies toward business.
More recently, Stevans (2009) introduced an endogeneity correction
in the adoption of a RTW law. Since it is possible that local factors
(such as strong unions) prevent the passage of RTW laws, any test of RTW
versus non-RTW states is not a natural experiment. Stevans found that
after applying an econometric correction for the endogeneity problem,
there were no wage or employment effects of RTW legislation. Vedder,
Denhart, and Robe (2011) conducted a study of RTW using a pooled OLS
(ordinary least squares) model from 1977 through 2007. They found a
roughly 1 percent increase in the growth rate of per capita personal
income for states passing RTW legislation. The strength of their work is
its parameterization of several contributing factors and the length of
time analyzed, but they do not correct for the endogeneity problem.
Hicks (2012) estimated the impact of RTW laws on manufacturing
employment, manufacturing incomes, and the share of manufacturing income
in states from 1929 through 2005. He examined the actual effect of RTW7
legislation using an identification strategy that isolated old Southern
states and 1947 manufacturing employment to account for political
factors affecting the passage of RTW laws. Hicks reported no impacts on
manufacturing employment, share, or incomes from the full sample.
However, there were statistically significant contributions of RTW laws
to manufacturing income growth in the vast majority of states that had
adopted legislation since 1950. Criticism of the empirical design of the
study prompted a follow-up article (Hicks and LaFaive 2013) that looked
at shorter time periods. The new study found that prior to 1970, RTW
legislation had only a small effect on population, income, and
employment. However, later periods (1971-90 and 1991-2011) saw a large
and statistically significant influence of RTW.
Vedder (2010) extends the RTW literature by identifying two likely
effects of RTW legislation--namely, higher population growth and levels,
and higher labor productivity. His work offers its greatest contribution
in explaining the transmission of policy to the action of individuals
and firms. Vedder (2010: 178) suggests that his empirical work "is
likely not the last word," and it is in that spirit we proceed to
refine his empirical framework.
Economic Theory and Right-to-Work Legislation
Minimal formal modeling of the impact of RTW legislation exists in
the economic literature. The work outlined above is either empirical or
offers a descriptive theory of effects. Reed (2003) provides a review of
the complexity of economic theory surrounding RTW laws and develops
several arguments. The first of these is that the presence of RTW laws
may permit nonunion workers to free ride, eroding the strength of unions
to bargain and thereby reducing the wage premium for workers. Second,
the increased requirement for effective unions under an RTW environment
may motivate unions to demonstrate their effectiveness by securing
higher wage and benefit gains. These two arguments, both of which are
plausible, may occur simultaneously.
The impact of RTW laws on firm location decisions is of interest
both as a research and policy question. The effects of RTW legislation
during different periods of regional growth offer some evidence of the
overall effect of RTW. Among the questions that can be asked of RTW are
whether or not unionization leads to differences in firm productivity,
and whether wages and benefits vary across regions with different levels
of unionization. Moreover, insofar as wages and benefits are not the
primary cost differential between union and nonunionized firms, other
matters may play a bigger role in firm location decisions. For example,
negotiating with unions may be costly, and much of the cost-increasing
effects of unions are embedded in work rules and decreased flexibility
in hiring and firing, not pay. Earlier researchers have offered a formal
model of a production function (Hicks and LaFaive 2013), which we
reprise here.
We model a simple production technology [theta](N) that is solely
dependent on labor, N. As described above, suppose that the level of
unionization affects productivity, then [theta](N[u]), but the direction
of the effect is unclear so d[theta](N[u])/du >/< 0. The wage
effect of unionization w(u) is such that RTW could increase, decrease,
or leave wages unchanged (dw[u]/du >/< 0). From this, we construct
a familiar labor demand function:
(1) [phi] = p[theta](N[u]) - w(u)[??],
where profit, [pi], for a firm is comprised of the multiplicative
product of the price, p, and a labor-only production function
[theta](N). From this is subtracted the wage rate, w, times employed
labor units N. The first-order condition with respect to unionization is
(2) [partial derivative][pi]/[partial derivative]u = p[theta]'
(N[u])N'(u) - w'[u] [??].
If we assume that [theta]'(N[u]) > 0, then Hicks and
LaFaive's (2013) model yields some straightforward results: if
[partial derivative][theta](N[u])/[partial derivative]u [greater than or
equal to] 0 and [partial](w[u])/[partial derivative]u [less than or
equal to] 0, then [partial derivative][pi]/[partial derivative]u
[greater than or equal to] 0. More simply, profits could be higher with
unions if labor productivity benefits from unionization. However, in the
opposite case, if [partial derivative][theta](N[u])/[partial
derivative]u [less than or equal to] 0 and [partial
derivative](w[u])/[partial derivative]u [greater than or equal to] 0,
then [partial derivative] [phi]/[partial derivative] [less than or equal
to] 0. These alternative findings imply that unionization may either
increase or decrease firm profitability depending on the impact of
unions on the productivity of labor.
Reed (2003) explains the uncertainty surrounding the direction of
the impact of unionization that makes formal modeling unclear. He argues
that productivity and wage effects of unions vary by industry and time.
So, the conditions outlined above provide strict relationships, which
may vary either through aggregation or across time. Thus, in the
preceding model, the effect of RTW legislation transmits to the
aggregate economy through an uncertain pathway. This uncertainty leaves
the effect of RTW legislation largely an empirical question to be
explored in a labor productivity model and reliant on careful treatment
of the data.
For example, RTW legislation may well have been influenced by
initial union conditions or local preferences. Strong unions in
industrialized states may have blocked die legislation, while less
industrialized states would be more likely to endorse RTW legislation.
Moreover, heavily industrialized states may enjoy manufacturing
clusters, such as automobile production, that continued to attract new
firms seeking the benefits of agglomerations. This feature may lead to
an observed [partial derivative][theta](N[u])/[partial]u [greater than
or equal to] 0 that is unrelated to RTW legislation. Also, during
periods of rapid employment growth in heavily unionized sectors, unions
may have served as employee screening tools for employers and so boosted
profitability. Later, as employment declined, unions may have aided in
the retention of low-productivity workers thus reducing labor
productivity.
Conversely, the convergence of state-level industrial structure in
the past half century would tend to push increased levels of more
unionized industries (primarily manufacturing and transportation because
mining, a heavily unionized industry, is not particularly mobile) in
states that had historically low levels of manufacturing. In other
words, states became more similar over the last half century, and this
necessarily meant more manufacturing in the South. This result could
have occurred without any consideration of RTW legislation. Moreover, it
is not clear that in the absence of weakened union influence the optimal
firm decision would be to hire more workers. Consequently, what is most
helpful in understanding the empirics is in the derivation of TFP and
the Solow residual in RTW versus non-RTW states, which accounts for the
growth in productivity not accounted for by the growth of inputs. It is
to that matter we now turn our attention.
Productivity Effects of Right-to-Work Laws
In this section, we estimate total factor productivity in the
context of RTW legislation. Vedder (2010) argues that higher output
elasticity of RTW states will boost aggregate output, an observation
that is confirmed by his empirics. However, what is critical is the
determination of the transmission mechanism of this growth and its
decomposition across RTW and non-RTW states.
We begin with a Cobb-Douglas production function. Constructing a
time series from the Annual Survey of Manufacturers, we estimate this
model at die state level for manufacturing firms from 2004-11 (Hicks
2013). This is a relatively straightforward model, where we seek to
extract the TFP across the dimension of RTW. Summary statistics are
reported in Table 1.
These data have a limited time period, beginning with annual
surveys in 2004. However, this suits our purpose since we will approach
the problem of evaluating the relative influence of RTW laws across two
samples during a period in which no RTW laws were changed. Indiana
passed RTW legislation in 2012 and Michigan in 2013, so we limit our
analysis to 2011. Oklahoma passed RTW legislation in 2001; thus it is
likely a full movement to equilibrium would not have occurred. We
address this in the results.
Our model is the familiar Cobb-Douglas production function:
[Y.sub.i,t] = A[[K.sup.[alpha].sub.i,t] [N.sup.[beta].sub.i,t], from
which we wish to recover empirical estimates of TFP, A. We also compute
the Solow residual across samples of RTW and non-RTW states. TFP is the
growth in output attributable to technological change in the capital and
labor basis model. The Solow residual is an expansion upon the TFP
estimate since it accounts for changes output not explained by the
growth of inputs. In this sense, it is the combined growth in the
changes in marginal product across inputs and the total technological
change across time. TFP is derived from an estimation of the
Cobb-Douglas production function, while the Solow residual is derived
from growth accounting. Our calculation of the Solow residual is drawn
from Hulten (2001) and takes the form: R = dY/Y - [s.sub.k] dK/K -
[s.sub.n] dN/N where s is the share of each input. The results from our
Cobb-Douglas model are reported in Table 2.
These results are distinctly similar to the canonical estimates of
the Cobb-Douglas production function as constant returns to scale across
aggregate sectors. At interest across these samples are the estimates of
TFP. In RTW states, our estimates of TFP are 2.022, while for non-RTW
states the estimate is 1.16. (1) Moreover, when we combine the sample
and include an RTW variable, we find it is positive and statistically
significant, and includes the value of other coefficients, suggesting
that RTW does matter. These results hold when we omit Oklahoma from the
analysis due to its 2001 adoption of RTW legislation.
Together these results strongly suggest that the presence of RTW
legislation increased total factor productivity of manufacturing firms
from 2004-11. However, the estimate of total factor productivity, A, has
a critical weakness in that it provides an estimate across the average
input mix. The Solow residual addresses that weakness by providing a
single technology estimate while accounting for changes in input share
of capital and labor. Table 3 reports these results.
In both the derived TFP from our Cobb-Douglas production function,
and in our growth accounting of the Solow residual, we find much higher
levels of productivity growth in RTW states than in non-RTW states.
Consistent with received theory, output declines are occurring in
non-RTW states. This is consistent with an interpretation of [partial
derivative][theta](N[u])/[partial derivative]u [less than or equal to] 0
from the derivative of die labor demand function.
This is also consistent with the findings from Vedder (2010). This
model has some inherent weaknesses, not least of which is an estimation
across the business cycle, which included significant changes to
manufacturing (see Hicks 2013 for a summary). To address this problem,
we launch a second empirical strategy to test productivity.
We use data on manufacturing firms from the 2007 Survey of Business
Owners (SBO) collected by the U.S. Census Bureau from a random sample of
businesses in the United States. The data used in our analysis are based
on the Public Use Microdata Sample (PUMS) released by the U.S. Census
Bureau in August 2012. The sample includes all businesses from the U.S.
nonagricultural sector that were in existence during 2007, filed tax
returns with the Internal Revenue Service, and had revenues of more than
$1,000. The Census Bureau identified these firms using IRS Form 1040,
Schedule C; Form 1065, U.S. Return of Partnership Income; Form 1120,
U.S. Corporate Income Tax Return; Form 941, Employer's Quarterly
Federal Tax Return; and Form 944, Employer's Annual Federal Tax
Return. Summary statistics appear in Table 4.
To test these data, we posit a very simple model of firm
productivity:
(3) log ([R.sub.i]/[N.sub.i]) = [alpha] + [[beta].sub.1]
[RTW.sub.i] + [[beta].sub.2] log([w.sub.i]) + [[beta].sub.3][T.sub.i] +
[[theta].sub.S] + [[alpha].sub.i] + e,
where the receipts per employee are a function of a fixed
intercept, location in an RTW state, average wages per employee in the
firm, w, tenure T, a vector of firm size categories, S, and a dummy
variable for each state (Alabama is the omitted state). Results, with
state-clustered standard errors, appear in Table 5.
These results point to a direct impact of location in an RTW state
on productivity, as measured by firm-level output per worker, for a
random sample of almost 50,000 manufacturing firms in 2007. The effect
is similar to the Cobb-Douglas and Solow residual results, statistically
significant, and supportive of results reported by Vedder (2010).
Our exploration of industry- and firm-level productivity suggests
that the effect of unionization, through the absence of RTW legislation,
is negative and significant, and also affects firms' capital
structures across labor markets, as evidenced by the estimates of total
factor productivity from the Cobb-Douglas production function and the
Solow residual. In order to more fully explore this, we turn our
attention to population growth in RTW states.
Modeling the Impact of Right-to-Work Laws
In examining the role of RTW laws in fostering migration, Vedder
(2010) acknowledged that factors other than RTW legislation influence
migration patterns. The problem is that there is little expectation that
RTW laws devolve upon states in a random fashion. Thus we adapt the
endogeneity treatment from Hicks and LaFaive (2013) to address this
concern.
Hicks and LaFaive (2013) observed that places that were relatively
poor in the middle of the 20th century also possessed a latent
anti-union sentiment, which led to early passage of RTW legislation. The
ensuing half century has seen many of these places grow faster than the
nation as a whole, for reasons as diverse as expanded political freedom
for minority groups to the widespread adoption of air conditioning.
Consequentially, a model that treats the introduction of RTW legislation
as a random event would bias any estimate of its impact. For that
reason, we must suspect endogeneity within the RTW legislation and
measures of economic performance such as population growth.
To correct for this problem, we employ an identification strategy
for the adoption of an RTW law, with an eye toward isolating RTW and
other unobserved variables that may affect our economic variables of
interest. Here we posit that adoption of an RTW law would be influenced
by the importance of manufacturing within a state at the time the 1947
Taft-Hartley Act was adopted and the political environment surrounding
unions at that time. To represent these variables, we use manufacturing
income in 1947 and a binary variable representing the old Southern
states (i.e., those states that seceded from the union). The identifying
equation for RTW is:
(4) E([R.sub.i,t]|[M.sub.i][S.sub.i]) = [alpha] + [[beta].sub.1](M)
+ [[beta].sub.1](S) + [u.sub.i,t],
where dM/dt = 0, and ds/dt = 0. The resulting estimate
[[??].sub.i,t] is conditioned on two variables that do not vary with
time. This equation offers two consequences regarding the endogeneity
and concomitant policy concerns above. We believe the endogeneity
concern is addressed through the identification of factors that would
contribute to a decision to adopt RTW legislation in states. The time
invariant nature of the regressors in this first-stage estimate
introduces a first-stage, fixed-effects estimate of [[??].sub.i,t],
using a technique introduced by Fernandez-Val and Vella (2011).
This approach captures any time invariant heterogeneity from which
concomitant policy variables would have their greatest source. To
correct for time-varying heterogeneity (unequal variances), we employ a
feasible generalized least squares (FGLS) estimate, because each of die
subestimates are for short periods that potentially suffer from small
sample-related problems, as well as from period-specific heterogeneity
(Wooldridge 2002). These two steps provide a safeguard against die
incidental variable concern.
For our estimation, we examine the conterminous 48 states and die
District of Columbia from 1947 through 2011. Summary statistics appear
in Table 6.
We construct a very basic treatment model to estimate the impact of
RTW legislation:
(5) log([dP.sub.i,t]/dt) = [alpha] + [beta]([[??].sub.i,t]) +
[[??].sub.j,t]) + [delta][??][Y.sub.j,t] + [theta][Y.sub.i,t - n] +
[[epsilon].sub.i,t],
where the dependent variable P is population in state i, in year t.
Population growth is estimated as a function of a common intercept,
[alpha] a presence variable for RTW legislation, in state i, in year t,
and tire weighted average of that variable in contiguous states,
weighted with a first-order contiguity matrix, [??]. This formulation is
designed to account for cross-border effects of RTW legislation in
adjacent states.
These two elements are corrected with the expected value of RTW
from the endogeneity equation (4) above, which is designed to identify
the adoption of an RTW legislation. The regression includes a first
order spatial contiguity element to correct for spatial autocorrelation
(8WYj t), a temporal autoregressive element ([theta][[gamma].sub.i,t -
n]) with optimal lag lengths selected through an informational criterion
recommended by Bozdogan (2000). We include an error term,
[[epsilon].sub.i,t], iid, [right arrow] (0, [[sigma].sup.2]). All
variables employed in the analysis pass individual and common unit root
tests and so are assumed stationary.
There are some econometric considerations in the estimation
process. The FGLS are estimated with White's (1980)
heteroskedasticity invariate, variance-covariance matrix. The estimate
of E([R.sub.i,t]) does not appear to suffer from weak instrumentation
concerns, with and F-statistic of 511.5, and both instrumental variables
enjoying statistical significance far better than 0.01 percent. We offer
an alternative specification to deal with spatial autocorrelation,
employing a method proposed by Pesaran (2006). We report and interpret
both results, which we call Model 1 and Model 2, respectively.
We estimate the relationship between RTW legislation and population
economic variables from 1947 to the present over three distinct time
periods: from 1947 through 1970, 1970 through 1990, and 1990 through
2011. The purpose of this approach is to evaluate both the impact of RTW
on population and whether or not effects varied across time periods. The
full time period estimates are reported in Table 7.
Table 8 reports the selected results (RTW coefficient only) from
two different specifications (Model 1 and Model 2) across the three
different time periods and the Wald test results from the comparison of
growth rates between time periods. Our analysis assumes that growth
rates for population are measures of overall economic well-being and
that RTW legislation affects them through a labor demand function. This
labor demand function yields conflicting theoretical possibilities as to
the impact of unions, which has been the challenge to existing research
in this area for some time (Reed 2003). Results above suggest that RTW
would be productivity enhancing and so promote population growth. We
also assume that the results above permit us to interpret the RTW
legislation dummy variable as clean, in the sense that it does not
capture other policy variables that are not perfectly coincident. While
the estimation process leads to this assumption in our interpretations,
the relaxation of this assumption simply alters the interpretation from
a strict RTW effect, compared to that of a combined suite of policies of
the type offered by Holmes (1998).
These results indicate that RTW legislation has a positive and
statistically significant influence on population growth during the
length of the observed period (the first column of results). The effect
is not discernable from 1947-70 but is in the later periods.
Furthermore, these impacts are relatively large, with growth rates
boosted just over 1 percent for population. We believe these results are
insensitive to alternative specifications that address the concomitant
variable problem (see Hicks and LaFaive 2013 for a fuller treatment of
this issue). These findings are suggestive of Vedder, Denhart, and Robe
(2011), and especially Vedder (2010).
The estimates in each of these categories tell a similar story.
From 1947 through 1970, the presence of RTVV legislation played no role
in population growth. A Wald test confirmed that for population growth,
the 1947-70 period was lower than either the later period (1971-90) or
the overall time. Moreover, RTW laws in adjacent states had no
measurable effect during this period. Whatever the cause, it is clear
that RTW legislation did not affect population growth during the more
than two decades after Taft-Hartley passed, a time of brisk increases in
manufacturing employment (Figure 1).
The period of nearly static employment growth in the most heavily
unionized sectors, from 1971-90, experienced a very different effect of
RTW, having a strong impact on population growth of 1.5 percent. In all
three cases, a Wald coefficient test found statistically different
coefficient values for this period when compared to values in the
earlier period (1947-70).
[FIGURE 1 OMITTED]
By the final period, from 1991-2013, the effect of RTW on these
three measures had lessened from the 1971-90 period but remained both
statistically significant in each case and important in terms of the
size of the impact (all roughly 0.8 percent higher growth in states with
an RTW law). The adjacent RTW variable was neither economically nor
statistically significant in any of our estimates.
Our research suggests that in the early days following
Taft-Hartley, RTW legislation had no meaningful impact on aggregate
economic growth measures in states in which it had passed. During the
beginning of the manufacturing employment stagnation (1971-90), that
changed, with RTW laws exerting a significant impact on growth of all
three measures. In the period 1991-2013, the impacts of RTW on growth
slowed modestly, but remained large enough that they should command
economic policy attention.
Conclusion
Richard Vedder (2010) offered an important addition to the
literature on RTW legislation with his description of the influence
individual choice plays in both population growth and labor productivity
in states where RTW legislation has passed. This article has focused on
the theory and empirics of the matter, extending both the argument from
Vedder into a labor demand function, and the empirics of industry and
firm productivity and population growth.
We estimate a Cobb-Douglas production function for manufacturing
industries at the state level and find that total factor productivity in
non-RTW states was about 57 percent of the level in RTW states. Our
derivation of the Solow residual suggests that non-RTW manufacturing
productivity was roughly 64 percent of the RTW states. Furthermore, our
firm-level analysis from the 2007 Survey of Business Owners found that
RTW states achieved higher productivity (sales per employee) than firms
in non-RTW states. These results extend Vedder's (2010) examination
of productivity of RTW laws across three different estimation
strategies.
Our second empirical strategy examined the impact of RTW
legislation on population growth from 1947 to 2013. We employ an
identification strategy offered by Hicks (2012) that includes 1947
manufacturing employment and the geography of the old South to isolate
union disposition among voters. Our findings suggest that from 1947-70,
RTW legislation had no effect on aggregate measures of economic activity
between states. However, that outcome changed for the 1971-90 and
1991-2013 periods, when the presence of an RTW law boosted state
population growth by 1.1 percent to 1.5 percent--results that support
Vedder's (2010) work. Thus, our study extends the literature by
carefully and more fully examining the effect of RTW legislation in
promoting both population growth and productivity growth.
References
Black, E., and Black, M. (1989) Politics and Society in the South.
Cambridge, Mass.: Harvard University Press.
Bozdogan, H. (2000) "Akaike's Information Criterion and
Recent Developments in Information Complexity." Journal of
Mathematical Psychology 44(1):62-91.
Bureau of Economic Analysis (2015) "Regional Economic
Accounts." Washington: U.S. Department of Commerce.
Dickens, W., and Leonard, J. S. (1985) "Accounting for the
Decline in Union Membership, 1950-1980." Industrial and Labor
Relations Review 38(3):323-34.
Farber, H. S. (1987) "The Recent Decline of Unionization in
the United States." Science 238 (13 November): 915-20.
Fernandez-Val, I., and Vella, F. (2011) "Bias Corrections for
Two-Step Fixed Effects Panel Data Estimators." Journal of
Econometrics 163(2):144-62.
Freeman, R. B. (1985) "Why Are Unions Faring Poorly in NLRB
Representation Elections?" In T. A. Kochan (ed.) Challenges and
Choices Facing American Unions, 45-64. Cambridge, Mass.: MIT Press.
Hicks, M. J. (2012) "Right-to-Work Legislation and the
Manufacturing Sector." Center for Business and Economic Research,
Ball State University.
--(2013) "Manufacturing Productivity through the Great
Recession: What Does it Mean for the Future?" Center for Business
and Economic Research, Ball State University.
Hicks, M. J., and LaFaive, M. (2013) "Economic Growth and
Right-to-Work Laws." Mackinac Center for Public Policy.
Holmes, T. J. (1998) "The Effect of State Policies on the
Location of Industry: Evidence from State Borders." Journal of
Political Economy 106(4):667-704.
Hulten, C. (2001) "Total Factor Productivity: A Short
Biography." In C. R. Hulten, E. R. Dean, and M. Harper (eds.) New
Developments in Productivity Analysis, 1-54. Chicago: University of
Chicago Press.
Jarley, P., and Fiorito, J. (1990) "Associate Membership:
Unionism or Consumerism?" Industrial and Labor Relations Review 43
(2): 209-24.
Koeller, C. T. (1994) "Union Activity and the Decline in
American Trade Union Membership." Journal of Labor Research
15(1):19-32.
Lazear, E. P. (1988) "Employment at Will, Job Security, and
Work Incentives." In Proceedings of the Conference on Employment,
Unemployment, and Hours of Work, Science Center Berlin, September 17-19,
1986. London: Allen and Unwin.
Moore, W. J. (1998) "The Determinants and Effects of
Right-to-Work Laws: A Review of the Recent Literature." Journal of
Labor Research 19(3):445-69.
Newman, R. J. (1983) "Industry Migration and Growth in the
South." Review of Economics and Statistics 65(1):76-86.
Pesaran, M. H. (2006) "Estimation and Inference in Large
Heterogeneous Panels with a Multifactor Error Structure."
Econometrica 74(4):967-1012.
Reder, M. W. (1988) "The Rise and Fall of Unions: the Public
Sector and the Private." Journal of Economic Perspectives
2(2):89-110.
Reed, W. R. (2003) "How Right-to-Work Laws Affect Wages."
Journal of Labor Research 24(4):1-18.
Stevans, L. K. (2009) "The Effect of Endogenous Right-to-Work
Laws on Business and Economic Conditions in the United States: A
Multivariate Approach." Review of Law and Economics 5(1): 595-614.
U.S. Census Bureau (2004-11) Annual Survey of Manufacturers.
Washington: U.S. Census Bureau.
--(2007) "Survey of Business Owners." Washington: U.S.
Census Bureau.
Vedder, R. (2010) "Right-to-Work Laws: Liberty, Prosperity,
and Quality of Life." Cato Journal 30(1):171-80.
Vedder, R.; Denhart, M.; and Robe, J. (2011) "Right to Work
and Indiana's Economic Future." Indiana Chamber of Commerce.
Available at http://share.indianachamber.com/media/docs/Indiana Right To
Work-1-27-11.pdf.
White, H. (1980) "A Heteroskedasticity-Consistent Covariance
Matrix Estimator and a Direct Test for Heteroskedasticity."
Econonietrica 48(4):817-38.
Wooldridge, J. M. (2002) Econometric Analysis of Cross-Section and
Panel Data. Cambridge, Mass.: MIT Press.
(1) A Wald test rejects the equality of these coefficients:
t-statistic = -8.44, for II0: [[beta].sub.RTW] = [[beta].sub.non-RTW]
Michael J. Hicks is the George and Frances Ball Distinguished
Professor of Economics and Director of the Center for Business and
Economic Research at Ball State University. Michael LaFaive is the
Director of the Morey Fiscal Policy Initiative at the Mackinac Center
for Public Policy. Srikant Devaraj is Research Assistant Professor at
the Center for Business and Economic Research at Ball State University.
TABLE 1
CAPITAL AND LABOR LEVELS AND EXPENDITURES
Mean Median Std. Dev.
Capital Stock ($1,000s) 20,562,598 13,407,505 22,112,637
Capital Expenditure ($1,000s) 2,884,433 2,143,272 3,089,931
Employment 251,765 177,486 249,326
Payroll ($1,000s) 11,824,359 8,550,733 12,245,945
Output ($ 1,000s) 102,000,000 75,053,127 110,000,000
SOURCE: U.S. Census Bureau (2004-11).
TABLE 2
COBB-DOUGLAS ESTIMATES
All RTW All Non-RTW Full Sample
States States Full Sample w/RTW
Total Factor
Productivity 2.022 *** 1.166 *** 1.439 *** 1.371 ***
(6.61) (5.60) (9.77) (9.45)
Capital 0.401 *** 0.386 *** 0.403 *** 0.393 ***
(5.24) (6.86) (10.79) (10.73)
Labor 0.557 *** 0.632 *** 0.598 *** 0.609 ***
(6.55) (10.23) (14.92) (15.50)
RTW Dummy -- -- -- 0.126 ***
(4.23)
Adj-[R.sup.2] 0.90 0.96 0.94 0.94
NOTES: N = 370, OLS, estimates. *** denotes statistically
significant at the 0.01 level, ** at the 0.05 level, * at the
0.10 level. Each estimate has been treated with White's (1980)
heteroskedasticity corrections.
TABLE 3
SOLOW RESIDUAL, TFP, AND OUTPUT GROWTH
U.S. MANUFACTURING, 2004-11
R (Solow Residual) TFP d(log[Y])
RTW 3.99 2.022 0.006
Non-RTW 2.58 1.166 20.004
TABLE 4
PRODUCTIVITY AND RIGHT-TO-WORK LAWS
Mean Median Std. Dev.
Receipts per Employee ($1,000s) 206.12 135.71 435.55
Right-to-Work States 0.36 0 0.48
Pay per Employee ($1,000s) 36.70 34 24.07
Established in
1980-89 0.209 0 0.407
1990-99 0.189 0 0.391
2000-02 0.054 0 0.225
2003 0.019 0 0.137
2004 0.022 0 0.147
2005 0.018 0 0.134
2006 0.017 0 0.129
2007 0.010 0 0.101
Employment Size
5-9 0.134 0 0.341
10-19 0.165 0 0.371
20-49 0.205 0 0.404
50-99 0.148 0 0.355
100-249 0.112 0 0.315
250-499 0.027 0 0.163
500-999 0.008 0 0.089
1,000+ 0.003 0 0.053
SOURCE: U.S. Census Bureau (2007).
TABLE 5
PRODUCTIVITY ESTIMATION RESULTS
Variable Coefficient
[alpha] 1.96
(61.41)
RTW 0.0748 ***
(32.95)
Wages 0.8561 ***
(98.09)
Tenure Category Yes
Size Category Yes
State Fixed Effects Yes
N 49,814
[R.sup.2] 0.42
NOTES: *** denotes statistically significant at the 0.01 level,
** at the 0.05 level, * at the 0.10 level; t-statistics in
parentheses, for standard errors clustered by state.
TABLE 6
SUMMARY STATISTICS
Mean Median Std. Dev.
Population 4,195,693 2,787,000 4,747,510
Right to Work 0.297 0 0.457
Right-to-Work Adjacency 0.316 0.25 0.335
Real Personal Income 173,000,000 103,000,000 215,000,000
Total Employment 2,287,311 1,539,370 2,459,790
Real Wages 23,015 1,556 26,475
SOURCES: Bureau of Economic Analysis (2015); Hicks
and LaFaive (2013).
TABLE 7
STATE POPULATION GROWTH RATE, 1947-2013
Model 1 Model 2
Intercept 0.0058 *** 0.001
(6.84) (-1.47)
Right to Work 0.009 *** 0.0007 ***
(4.52) (2.71)
Adjacent Right to Work -0.01074 *** -0.002
(-3.59) (-0.74)
Spatial Autocorrelation 0,59 *** --
(2.23)
AR(1) 0,56 *** 0.457225 ***
(15.25) (4.77)
Adjusted R-squared 0.53 0.48
F-statistic 928.2 994.3
Durbin-Watson stat 1.74 1.82
NOTES: N = 2,303; *** denotes statistically significance at die
0.01 level, ** at the 0.05 level, * at the 0.10 level. All
estimates in pooled OLS.
TABLE 8
STATE POPULATION GROWTH RATE, RTW COEFFICIENTS, AND COEFFICIENT
RESTRICTION TESTS
Wald Test
Model 1 RTW HO = 1947-2013 HO = 1947-70
Coefficient value value
1947-70 0.000252 -- n/a
1971-90 0.15 ** 2.83 *** 2.95 ***
1991-2013 0.01 *** 2.49 *** 2.65 ***
Wald Test
Model 1 RTW HO = 1947-2013 HO = 1947-70
Coefficient value value
1947-70 -0.0016 -- n/a
1971-90 0.013 ([dagger]) 1.53 ([dagger]) 1.81 *
1991-2013 0.011 ** 2.17 ** 2.65 ***
Wald Test
Model 1 HO = 1971-90 HO = 1990-2013
value value
1947-70 -- --
1971-90 n/a 0.93
1991-2013 -33.2 *** n/a
Wald Test
Model 1 HO = 1971-90 HO = 1990-2013
value value
1947-70 -- --
1971-90 n/a 0.29
1991-2013 -0.36 n/a
NOTES: *** denotes statistically significant at the 0.01 level,
** at the 0.05 level, * at the 0.10 level, and ([dagger]) at the
0.15 level.
