plcy_691H_Full_Draft.docx

39 

Full text

(1)

What were the Macroeconomic Effects of Wisconsin’s 2015 Passage of

a Right to Work Law?

By

Lorcan Farrell

Senior Honors Thesis

Department of Public Policy

University of North Carolina at Chapel Hill

April 15, 2020

Approved:

________________________

Dr. Jeremy Moulton

(2)

Abstract

Unions and right to work laws have been a contentious subject since the passage of the

Taft-Hartley Act. While the debate spans many decades most of the research on the topic has focused primarily on individual level effects. Using the synthetic control method, and the state of

Wisconsin as a case study, this paper analyzes the state level economic effects of a RTW law.

The results show no effect across the Housing Price Index, GDP, manufacturing wages, or income per capita. This calls into question the purpose and worth of RTW laws.

Introduction

In the past year, high profile labor disputes have thrown unions back into the spotlight.

In 2018, the West Virginia branches of the American Federation of Teachers and the National Education Association went on strike to protest low pay and benefits. That sparked a wave of

similar movements in states such as Oklahoma, Kentucky, Arizona, and North Carolina (Rios, 2018). Teachers flexed their collective bargaining power in order to negotiate for increased wages, halts to benefit cuts, and better working conditions. In September of 2019 49,000 workers

at General Motors walked off the job after GM and the United Auto Workers failed to negotiate a new contract. Sticking points between the two sides included wages, healthcare, and job

security. These actions would not be possible without the respective unions acting as an organizing force and representing the workers in negotiation. While the workers obviously are happy with the unions enabling them to have their demands met, the school districts and GM see

(3)

the state level, looking at the macroeconomic effects of a law passed to reduce the strength of unions in Wisconsin.

Prior to 1932 labor relations were mostly unregulated. On a case-by-case basis, courts would decide to grant companies injunctions in order to force workers to stop strikes. A large portion of strikes failed because of these injunctions (National Labor Relations Board, n.d.). In

1932, Congress passed the Norris – La Guardia Act. This significantly strengthened unions. Courts were no longer allowed to issue injunctions against non-violent labor disputes. Employers

could no longer make agreeing not to join a union a condition of employment. They also could no longer interfere in the formation of unions within their company. The pendulum swung in the opposite direction towards the workers and strikes became commonplace. In 1934,

Congress passed the Wagner Act. It created the National Labor Relations Board and formalized the collective bargaining rights of employees. The framework the Wagner Act established is still

the national labor policy today. Workers have the right to join unions and employers are obligated to negotiate with those unions (National Labor Relations Board, n.d.). However, just because workers have the right to form unions does not mean that states cannot pass laws to

weaken those unions.

Pro-business advocates see unions as a limit on business potential. They stifle growth by

increasing costs, and work stoppages or other union actions can result in millions of lost

revenue. These sentiments lead to the 1947 Taft-Hartley Act. The act was an amendment to the Wagner Act that established some restrictions on unions. These included requirements to

negotiate in good faith, restrictions on dues, and restrictions on actions taken against employers. The focus of this thesis, however, is the section that allowed states to pass

(4)

A right-to-work law is a law that prohibits employers from making union membership a condition of employment. The goal is to reduce union strength by reducing their membership

and resources. A union’s bargaining power comes from the number of members it has. A threatened strike is much more effective the more workers that will be walking off the job. Similarly, the functional ability of a union is higher the more due paying members it has. RTW

laws increase the free-rider problem that many unions suffer from. If membership drops while coverage remains steady, then fewer resources per worker are available.

While RTW laws directly affect unions, often times the intent of the law is more focused on signaling to businesses that may be looking to expand in the state. A RTW law serves as a way for states to appear as friendly and open towards businesses. The hope is to sway expansion

decisions toward the state, in an attempt to boost economic growth. When signing Wisconsin’s 2015 RTW law, Governor Scott Walker explained that the reason for pushing for the law was

because RTW legislation is on the checklist of every site selector exploring potential growth sites (Opoien, 2015).

Similar to the topic of unions as a whole, RTW laws are divisive. The National Right to

Work Committee is one of the major advocates of right-to-work laws. They refer to states without RTW laws as “compulsory-unionism states” or “forced-dues states.” Their

arguments center on the claim that making union dues a condition of employment results in a net loss of income for workers. Claims that RTW states are more affordable, have greater economic growth, and are home to higher-paying jobs are the rallying cries for those seeking to increase

the spread of RTW laws (National Right to Work Committee, n.d.). On the other side of the argument is the AFL-CIO. The largest collection of American unions sees RTW laws as a threat.

(5)

states have more discrimination and higher pay gaps. They also make an opposing claim to the National Right to Work Committee, arguing that RTW laws result in lower wages for

workers (AFL-CIO, n.d.).

There is a large body of research exploring the effects of RTW laws on workers. For the most part, it is inconclusive. Both sides of the argument cite numerous studies to support their

view. However, there is significantly less research on how RTW laws affect states as a whole, which is the more important question for policymakers. State legislatures should be taking action

to improve their state. A small positive effect for one group that leads to a larger negative effect for another group would overall be hurting the state as a whole. In order to truly understand the merits of a RTW law those voting on it need to understand how the state as a whole will be

affected, not just individual groups. It does not matter if employees see a wage benefit if

employers face a larger cost, which negatively affects the state’s economy. 27 states have passed

RTW laws, though West Virginia’s is currently being challenged in the state supreme court. The purpose of this thesis is to estimate the effects of the RTW law on four different statewide outcomes in one state, Wisconsin. Wisconsin’s private sector RTW law came into effect in

March of 2015. This paper will compare macroeconomic and labor market outcomes in the state before and after the law came into effect seeking to answer the question: do RTW laws

have state-level benefits. I chose to focus on Wisconsin’s RTW law because it is the most recently passed bill for which there is significant post data. It also has the added benefit of coming as a surprise when the Wisconsin legislature passed it in 2015. Then Governor Scott

Walker had dismissed the idea of a RTW bill as a distraction in the few months prior. Then in February the state legislature fast tracked the bill during a special session, getting it passed and

(6)

time to prepare, reducing the likelihood that the treatment effect would begin to happen prior to the law taking affect. A potential disruption to this theory is the fact that in 2011 Scott Walker

pushed through an end to collective bargaining for public sector employees. This could have resulted in spillover effects reducing private sector unionization rates, which would contaminate the pre-treatment data for this analysis. However, analysis shows that while there was a drop in

private union membership and coverage rates in 2011, it was only temporary, and trends continued as normal the next year (Bono-Lunn 2020). Kentucky and West Virginia have both

passed such laws more recently, but as discussed earlier West Virginia’s has not gone into effect and Kentucky’s has been in effect for just over one year. Michigan and Indiana also adopted RTW laws this decade effective in 2013 and 2012 respectively. I considered those too close to

the Great Recession to be able to make meaningful comparisons to the data from before the law took effect.

The next chapter of this thesis will explore the existing literature and how this paper contributes to it. Chapter 3 is a discussion of how the synthetic control method functions and what data I used for the analysis. The final two chapters cover the findings from the analysis and

how they can be applied to actionable public policy.

Literature Review

The literature on the effect of RTW laws is extensive. Since the passage of Taft-Hartley each side of the debate has sought to find proof that their argument was correct. This constant

back and forth has interfered with the creation of generally accepted conclusions.

Common outcome measures for analyzing the effect of RTW laws on workers are wages

(7)

regressions. The controls accounted for the differences in worker characteristics between states with RTW laws and those without. These included age, sex, education, and hourly versus salary

work. The regression with the full suite of controls found that states with RTW laws had lower wages. This effect was not limited to union workers, rather it was across all workers in the state. Shierholz and Gould defined compensation outside of wage as employee sponsored healthcare

and pensions. They ran the regression with the same controls on the probability that workers would have access to employee sponsored healthcare and pensions. The results matched those

from the wage regression. Employees in a state with a RTW law are less likely to have access to either employee sponsored healthcare or pensions.

Roberts and Habans (2015) used a similar multivariate regression on wages but added

fixed effects to take into account that workers cluster and the assumption of observational independence cannot be guaranteed. On top of that, Roberts and Habans performed a separate

analysis using propensity score matching in order to pair workers that are as similar as possible across treated and non-treated states. This is yet another step towards attempting to isolate the effect of a RTW law from a large number of potential confounders. The authors also modified

their first multivariate regression by including an interaction between the existence of a RTW law and the social, demographic, or occupational group membership of a worker. This

interaction allowed the authors to analyze the effect of RTW laws on wage inequalities across the groups. Across both of their methods, Roberts and Habans find that states with RTW laws have lower average hourly wages than those without. The magnitude of the reduction varies over

the two methods, and in both it is less than the effect found by Shierholz and Gould. However, there is still a statistically significant confirmation of the hypothesis that RTW laws reduced

(8)

A common theme across the debating literature is that prior research failed to properly control for any number of confounding variables. Reed (2003) argued that historic economic

conditions were an, as of then, overlooked potential confounder of the outcomes of past research. The argument behind the claim is that not all states are created equal, specifically that RTW states tend to be poorer and have lower wages even before the passage of a RTW law. Thus, it is

likely that any difference in wages after the passage of a RTW law is primarily due to the preexisting economic conditions. Reed controlled for the economic condition prior to passing a

RTW law with a variable for the log of a state’s personal per capita income in 1945. Reed chose 1945 because all RTW laws except for Florida were passed after 1945. Wyoming, Louisiana, and Idaho were dropped due to passing their laws significantly after 1945; Oklahoma was counted as

non-RTW state because their law came into effect outside of the window of analysis. Reed’s final regression also included an interaction between the RTW indicator and 1945 personal per

capita income, along with a measure of farming’s share of total state earnings in 1945. The analysis showed that on average states with a RTW law saw higher wages than those without. In the 18 RTW states that were part of the analysis, 16 are estimated to have higher wages due to

their passage of a RTW law. This broad confirmation of the average effect shows that the average was not driven by a few outliers. There are questions about whether 1945 personal per

capita income is a good measure of preexisting economic conditions. Also Reed included very few controls for any other potential confounding variables. However, his analysis and conclusion are a good example of how there is evidence for both sides of the RTW argument.

A debate similar to the one over worker compensation surrounds the effect of RTW laws on union strength and membership. Most of the literature claims that RTW laws have a negative

(9)

significant effects on unions. Moore and Newman (1985) showed that the difference arises from whether researchers treat the existence of a RTW law as exogenous or endogenous. If RTW laws

are thought to just be indicators of the tastes of the residents of a state, then the presence of a RTW law will have little effect on unionization rates (Lumsden and Petersen 1975). However, a synthetic control analysis of Wisconsin after the passage of their RTW law in 2015 found a sharp

drop in union membership rates (Bono-Lunn 2020).

While both sides of the debate on RTW law’s effects on workers have been thoroughly

researched, there is significantly less literature on the law’s effect on states. Kalenkoski and Lacombe (2006) examined how RTW laws affect the industry distribution of employment in a state, finding that states with RTW laws have a higher proportion of manufacturing employment.

This follows the claim that RTW laws make states more attractive to manufacturers. However, the outcome was not the main variable, rather the paper focused on how most RTW research

fails to control for geographically correlated factors. Similar to Roberts and Habans (2015), Kalenkoski and Lacombe showed that adding those controls reduce the previously shown effects.

One of the few papers to look at multiple state level outcomes is Eren and Ozbeklik’s

(2015) synthetic control study of Oklahoma. The use of a synthetic control follows the trend of recent RTW research attempting to control for more and more potential confounders. The

variables of interest were a series of labor market outcomes including the standard wages and union membership rates, along with the employment-population ratio and the level of

manufacturing employment. The only variable that was significantly affected was the private

(10)

This paper builds off the work of Eren and Ozbeklik (2015). By performing a similar analysis on Wisconsin, it can help confirm whether the results are externally valid. It also

considers additional state level economic outcomes in an attempt to broaden the research into the higher-level effects of RTW laws.

Methods and Data

Method

This thesis makes use of the synthetic control method. In terms of econometric research

designs, the synthetic control is a relatively new one. Abadie and Gardeazabal first used it in their paper exploring the economic costs of conflict through a study of the Basque

Country (Abadie & Gardeazabal, 2003). The most prominent example of the method is Abadie et

al.’s paper on the impact of California’s tobacco control program (Abadie, Diamond, & Hainmueller, 2010).

In its simplest form, the synthetic control method is a program effect case study. It looks at one treated unit in an attempt to distill the treatment effect. In this paper, the treated unit is Wisconsin. A common method for situations like this where the treatment has already occurred

without a clear option for a control unit is a difference in difference test. However, neighboring states make imperfect controls when looking on a statewide level. When the comparison is

occurring at a regional level then the border region between two states is a good control because often times there is little change between the two sides besides state policies. However, changes become much more significant the greater the area of analysis. This is due to different

demographics, industries and even geographies. These differences can cause different trends in the outcome variable making it hard to determine treatment effects. Figure 1 and Figure 2 below

(11)

neighboring states. It is clear that none of the four states are great candidates to be used as comparison points for a difference in difference test. They either have completely different

trends across the period of analysis, or there is no clear best counterfactual to choose.

(12)

Figure 2: Comparison of the GDP trends in Wisconsin and neighboring states

Rather than select an existing untreated unit to use as a comparison, the synthetic control

method creates its own. This control is created by using a weighted average of the pool of untreated units. The aim is to match the pre-treatment trends of the control and the treated unit as closely as possible. Thus, the differences between the control and treated unit after the treatment

can be attributed to the treatment itself.

Data

The method uses control variables to perform the pre-treatment matching. The weights that provide the smallest error across the controls are used to predict the outcome variable. This thesis also borrows from Eren and Ozbeklik (2015) in the selection of control variables for

matching. The controls fall into a series of categories. The first is union membership information. In addition to the membership-rate outcome variables, this includes both

(13)

coverage rate across public and private sectors. These are important controls to match on because RTW laws are passed to weaken unions. Therefore, any treatment effect would have presumably

been driven by the change in union membership. All union related data came from the Union Membership and Coverage Database constructed by Barry Hirsch and David Macpherson. They build the database using monthly Current Population Survey (CPS) data. The next category is

state demographic data from the CPS accessed through Minnesota Population Center’s Integrated Public Use Microdata Series (IPUMS). These variables are the percentage of the

population that is white, the percent that is college educated, the percent that lives in a metro area, and the percent that are male. These demographics tend to be predictors of union membership or coverage rates (Moore and Newman 1985). The last category are some

miscellaneous state level variables that are used to more closely match the characteristics of the synthetic Wisconsin to the actual Wisconsin. These controls are the population and land

area. This data came from the Census Bureau.

The primary outcome variables for this paper are quarterly gross domestic product, and the quarterly Federal Housing Finance Agency housing price index (HPI). These are

both indicators for the macroeconomic health of the state. GDP as a measure of the total economic output of the state, and housing prices as an indicator of consumer spending and

confidence. The two other outcome variables are income per capita and manufacturing wages. Income per capita is another way of checking to see if people across the state are seeing benefits from the law. Manufacturing wages is the most specific of the outcome variables. While it is not

a measure of the statewide economy, RTW laws are primarily directed to affect the

manufacturing industry so it is a good way to measure if the law influences those closest to it. It

(14)

affects are seen in Wisconsin as in Oklahoma, the likelihood of external validity is higher. GDP, HPI, manufacturing wages, and income per capita are used as matching variables in the analyses

where they are not the outcome of interest.

All the data are quarterly if available or annual if not. The treatment went into effect March 11th 2015 (Mautz, n.d.). For the purpose of the analysis, the treatment period is the

2nd quarter of 2015. The pre-treatment data ranges back to the 1st quarter of 2010.

This was selected in order to avoid contamination by the Great Recession in 2008. As mentioned

in the introduction, Wisconsin’s 2011 ban on public sector collective bargaining could have contaminated the data as well. However, the evidence suggests any influence from the change was temporary (Bono-Lunn 2020). The analysis goes until the 4th quarter of 2018, which is the

last quarter for which there is complete data. Most of the variables used as controls are observed as averages across the pre-treatment period of analysis. Some are observed at specific dates to

better improve trend matching. These are the first date of the analysis, Q1 2010, the last date before the treatment goes into effect, Q1 2015, and the midpoint between them, Q2 2014. Each analysis also includes three lagged observations of the variable of analysis at the fourth quarter

of 2010, second quarter of 2012, and first quarter of 2015. The full list of variables used to match in each synthetic control can be found in Tables 1-4 in the Appendix.

Empirical Results

Housing Price Index

(15)

percent), Louisiana (16.2 percent), Alaska (10.6 percent), and Montana (6.5 percent). Table 5 contains the values of the pre-treatment matching variables for Wisconsin, the synthetic control,

and the unweighted average of all other states. On average, the difference between Wisconsin and the control is smaller than the difference between Wisconsin and the average of all other states.

Figure 3 shows the quarterly HPI for Wisconsin, and the synthetic control, from the first quarter of 2010 until the last quarter of 2018. It illustrates a few key points. First, the synthetic control closely matches Wisconsin for most of the pretreatment period. The RMPSE was 2.252. Even though the HPI used was seasonally adjusted there is still some evidence of seasonality in

the trend, which the control was able to match. Next, after the introduction of the RTW law in 2015 the two trend lines begin to diverge. The average post-treatment difference between the

(16)

In order to determine the significance of this difference and check whether it is just a result of random noise I ran a series of placebo tests. This was done by running the exact same synthetic control with each of the control states taking the place of Wisconsin as the treated unit.

The quarterly difference between the treated units and the control for each state was plotted below in Figure 4. The light grey lines are the differences for each of the control states, the

green highlighted line is Wisconsin. It is clear to see that while Wisconsin’s differences appear to be on the higher end, they are not significantly different from those of other states. The placebo figure excludes states that were extreme outliers due to poor matching in the synthetic control in

(17)

GDP

The next variable of interest is Wisconsin’s state level GDP, which for this analysis is measured in millions of 2019 USD. In this instance, the synthetic Wisconsin is made up of the 5

different states in Table 2. For GDP the 5 states, by weight, are Delaware (50.5 percent), Maryland (20.2 percent), Illinois (18.4 percent), Maine (9.1 percent), and California (1.8

percent). Table 6 contains the values of the pre-treatment matching variables for Wisconsin, the GDP based synthetic Wisconsin, and the unweighted average of all other states. The gap between Wisconsin and its synthetic counterpart trends smaller than the difference between Wisconsin

(18)

The GDP for Wisconsin and the control are plotted in Figure 5. Once again, the synthetic control matched very well to actual Wisconsin’s pretreatment trend. The RMPSE is 1088.362.

While this is much higher than the RMPSE for the HPI, it is important to remember that RMPSE varies based on the scale of the variable. In this case, the scale is hundreds of thousands, so a RMSPE of 1088.362 still indicates good matching in the pre-treatment period. In the

post-treatment period, there is very little difference between the control and the actual trend line. The mean difference is 1866.518. With a baseline GDP of 300,793 this difference is less than 1%.

Even if it was statistically significant, it is not practically significant. Still in order to confirm the results I ran the same placebo test as with the HPI analysis. As Figure 6 shows the green line that is Wisconsin is no different from any of the other states. The placebo figure excludes states

(19)
(20)

The third set of results pertains to the RTW law’s effect on Manufacturing Wages. This synthetic control consists of 7 states found in Table 3. The top five by weight are Delaware (70.0

percent), Ohio (19.6 percent), Colorado (4.7 percent), Alaska (2.3 percent), and Washington (2.1 percent). Table 7 contains the values of the matching variables for Wisconsin, the manufacturing wages synthetic control, and the unweighted average of all other states. It shows that the

synthetic control matches the actual Wisconsin values better than taking the average across all the control states.

Figure 7 plots the average manufacturing wages for Wisconsin and the synthetic

Wisconsin. The RMPSE is 0.3146 and while it is not as closely matched as the other controls, it does a good job of following the general trends in the pre-treatment period. In the post-treatment

period, the average difference between the control and actual Wisconsin is 0.1614. With a baseline of 23.92, the average treatment effect is a less than 1% change in manufacturing wages.

Figure 8 shows the results of the placebo test. At no point in the post-treatment period is the difference between Wisconsin and the synthetic Wisconsin any different from that of other states. The placebo figure excludes states that were extreme outliers due to poor matching in the

(21)
(22)

The final analysis pertains to income per capita in Wisconsin from 2010 to 2018. The synthetic control that most closely matches pre-treatment Wisconsin is made up of 9 states which

can be found in Table 4. The top five by weight are Montana (33.1 percent), Maine (22.2 percent), Vermont (17.5 percent), Rhode Island (8.9 percent), and Minnesota (4.7 percent). Table 8 shows the values of the control variables for Wisconsin, the synthetic control, and the unweighted average of all other states. On average, the difference between Wisconsin and the control is smaller than the difference between Wisconsin and the average of all other states.

Figure 9 shows the income per capita for both Wisconsin and the synthetic control. For most of the pre-treatment period, the control follows the actual Wisconsin nearly exactly. The

RMPSE is 152.1444 The average post-treatment difference is -88.01. The 2015 baseline income per capita was 45,937. The estimated treatment effect is effectively zero. When compared to all

(23)
(24)

A concern associated with the synthetic control method is that the results can be polluted by one state playing an outsized role in the creation of the control unit. In order to account for

this, a leave-one-out robustness check is performed. This check iteratively drops the states with the largest weights from the panel of control states and reruns the analysis. The resulting synthetic control is then plotted versus the main synthetic control. If there are no major

differences, then the results of the analysis can be considered valid. Three of the four synthetic controls in this thesis passed the robustness check. The manufacturing wages synthetic control

gives Delaware a weight of 70%, when Delaware is dropped the control trend changes drastically. However, because there was no significant treatment effect measured this is not a major concern. The graphs for the test can be found in the appendix.

Conclusion

Since the Taft-Hartley act enabled their creation, RTW laws have become one of the

primary tools for policy makers seeking to curtail the strength of unions. Whether or not

weakened unions should be a goal of policy makers has been the subject of debate since at least

the passage of the Taft-Hartley act. It was a debate that had died down as unions lost ground across the nation; however, the recent rise in inequality and collective action has seen discussion rekindled. However, even as unions lost ground RTW laws still served a purpose for policy

makers. The passage of a RTW law was used to signal to businesses that a state was seeking to provide favorable conditions for businesses looking to expand.

(25)

control method, various state-level economic indicators were examined, pre and post the law going into effect. The results indicated that across all four variables of interest the addition of the

RTW law had no significant effect. While manufacturing wages, and the housing price index appeared to show some treatment effect, there was no way to distinguish said effect from statistical noise. GDP and income per capita in Wisconsin showed no change at all in the

quarters after the law went into effect.

This is a surprising result. Studies have definitively shown that RTW laws are effective in

reducing the strength of unions. And, while the exact effects are unknown, there is literature showing that weakened unions have an effect on workers. That effect not showing up in the statewide economy warrants further study. It is possible that the effect takes more than four years

to become noticeable, or that it shows up in other indicators not studied in this paper. Potentially any negative effect is being balanced out by a matching positive effect elsewhere in the

economy. Those possibilities all assume the hypothesis that a RTW law would have any effect on the statewide economy. A final possibility is that unions are such a small part of the economy that any changes to them will not result in a statewide effect. Union coverage rates barely rise

above 10%, meaning 90% of the workforce is not directly affected by any weakening of unions. The lack of any results also calls into question the validity and strength of signaling that a state

has a business-friendly environment. If this signaling was successful, one would expect an increase in business operations within a state to increase the statewide GDP.

Recommendations

(26)

divisive topic. Given that, at the moment, there is no clear evidence that a state is improved or worsened by a RTW that divisive debate would be a waste of political capital and goodwill.

Until further evidence is gathered, it is best to table the RTW debate.

For further research, there are multiple paths to take. The first would be to repeat this study with different states as the units of analysis. This would allow insight into how

generalizable the effects of RTW laws are. Because each state has its own composition of industries, and workforce, and regulations, it is entirely possible that RTW laws affect each state

in a unique way. The second path of further research is a look into why Wisconsin’s statewide economy did not appear to be affected. This could be done through simply waiting to allow more time to pass, taking a broader approach with more variables of interest, or exploring anything

like a potential wealth transfer that balanced out positive and negative effects. The third path would be to explore whether there is a significant effect that arises from using a RTW law to

(27)

Appendix

Synthetic Control State Weights

Table 1: HPI Synthetic Control States

State Weight Pennsylvania 35.3% Delaware 27.1% Maine 16.2% Alaska 10.6% Montana 6.5% Colorado 4.2%

Table 2: GDP Synthetic Control States

State Weight Delaware 50.5% Maryland 20.2% Illinois 18.4% Maine 9.1% California 1.8%

Table 3: Manufacturing Wages Synthetic Control States

State Weight Delaware 70% Ohio 19.6% Colorado 4.7% Alaska 2.3% Washington 2.1% Maine .8% Maryland .5%

Table 4: Income per Capita Wages Synthetic Control States

State Weight

Montana 33.1%

Maine 22.2%

Vermont 17.5%

Rhode Island 8.9%

Minnesota 4.7%

(28)

Illinois 3.9%

Maryland 3.0%

California 2.7%

Synthetic Control Matching Variables

Table 5: Comparison of matching variables between Wisconsin, HPI Synthetic Wisconsin, and Average of all States except Wisconsin

Wisconsin HPI Synthetic Wisconsin All States except Wisconsin

Manufacturing Wages 22.30429 22.45134 25.47844

Manufacturing

Wages(2010q1) 21.55 21.66414 23.90667

Manufacturing

Wages(2012q2) 21.72 22.07214 24.26905

Manufacturing

Wages(2015q1) 23.92 22.81014 25.24143

Manufacturing Union

Membership(2014q1) 12.5 11.5859 9.928571

Private Union

Coverage(2014q1) 7.1 7.3622 8.542857

Manufacturing Union

Coverage(2011q1) 9.85 11.9051 11.22857

Manufacturing Union

Coverage(2014q1) 12.85 12.2436 10.55238

Total Union

Membership(2014q1) 9.15 12.5622 14.67619

Proportion of Employment that is Manufacturing(2015q1)

8.149579 6.078105 6.250272

GDP(2010q1) 249054.9 249031.4 368797.2

GDP(2015q1) 300793.7 297546.2 452600.4

Population 3148484 5252053 7168120

Housing Price Index

-Adjusted 202.721 202.7195 226.3021

Housing Price Index

-Adjusted(2015q1) 212.44 211.9381 222.8938

Housing Price Index

-Adjusted(2010q4) 204.39 201.0295 202.3043

(29)

Adjusted(2012q2)

Table 6: Comparison of matching variables between Wisconsin, GDP Synthetic Wisconsin, and Average of all States except Wisconsin

Wisconsin GDP Synthetic Wisconsin All States Except Wisconsin Private Union

Membership 6.245238 6.229967 8.033862

Private Union Coverage 6.77381 6.944343 8.824339

Manufacturing Union

Membership 10.24762 9.853752 9.859788

Total Land Area 54310 18804.28 71251.05

Manufacturing Wages 22.30429 22.83973 25.47844

Manufacturing

Wages(2010q1) 21.55 22.25299 23.90667

Manufacturing

Wages(2012q2) 21.72 21.89989 24.26905

Manufacturing

Wages(2015q1) 23.92 23.94113 25.24143

Manufacturing Union

Coverage(2011q1) 9.85 9.8673 11.22857

Manufacturing Union

Coverage(2014q1) 12.85 10.306 10.55238

Proportion of Employment that is

Manufacturing(2015q1) 8.149579 5.336214 6.250272

Population 3148484 4682753 7168120

GDP 274987 275495.8 441230.4

GDP(2015q1) 300793.7 302847.1 452600.4

GDP(2010q4) 259625 259770.2 380519.7

GDP(2012q2) 274601.1 274695.8 402890.7

Table 7: Comparison of matching variables between Wisconsin, Manufacturing Wages Synthetic Wisconsin, and Average of all States except Wisconsin

Wisconsin Manufacturing Wages Synthetic Wisconsin

All Other States Except Wisconsin Manufacturing Union Membership 10.24762 9.977519 9.859788 Manufacturing Union Coverage(2010q1

(30)

Private Union Membership 6.245238 6.118376 8.033862

Private Union Coverage 6.77381 6.918181 8.824339

Manufacturing Union

Coverage(2011q1) 9.85 10.0967 11.22857

Manufacturing Union

Coverage(2014q1) 12.85 11.2338 10.55238

Proportion of Employment that is

Manufacturing 8.074103 6.046047 6.281974

GDP 274987 175450.5 441230.4

Population 3148484 3158054 7168120

Population(2010q1 - 2012q2) 3135397 3133654 7069178 Population(2012q2 - 2015q1) 3161532 3182039 7216770

Manufacturing Wages 22.30429 22.36265 25.47844

Manufacturing Wages(2015q1) 23.92 23.36616 25.24143 Manufacturing Wages(2010q4) 22.05 21.76021 24.46286

Manufacturing Wages(2012q2) 21.72 21.7078 24.26905

Table 8: Comparison of matching variables between Wisconsin, Income per Capita Synthetic Wisconsin, and Average of all States except Wisconsin

Wisconsin Synthetic WisconsinIncome per Capita All States ExceptWisconsin Percentage that is White 90.55476 89.62256 78.53545

Percentage that is Male 50.11428 49.24236 49.09365 Manufacturing Employment 239289.9 134064.7 277720.8

Total Employment 1953567 1893831 4282025

Private Union Membership 6.245238 6.032348 8.033862

Private Union Coverage 6.77381 6.933157 8.824339

Total Land Area 54310 68751.36 71251.05

Manufacturing Wages 22.30429 22.34204 25.47844

Manufacturing Wages(2010q1) 21.55 21.74112 23.90667 Manufacturing Wages(2012q2) 21.72 21.89946 24.26905 Manufacturing Wages(2015q1) 23.92 23.14528 25.24143

Manufacturing Union

Coverage(2011q1) 9.85 11.7894 11.22857

Population 3148484 3239993 7168120

(31)

Income per Capita(2015q1) 45937 45962.41 52145.19

Income per Capita(2010q4) 39802 39665.44 44925.9

Income per Capita(2012q2) 42808 42757.74 48133.57

(32)
(33)
(34)

Figure 13: Wisconsin Manufacturing Wages Placebo with all states shown

Figure 14: Wisconsin Income per Capita Placebo with all states shown

(35)
(36)
(37)
(38)
(39)

Bibliography

Abadie, A., Diamond, A., & Hainmueller, J. (2010). Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control Program. Journal of the

American Statistical Association, 105(490), 493–505. JSTOR.

Abadie, A., & Gardeazabal, J. (2003). The Economic Costs of Conflict: A Case Study of the Basque Country. The American Economic Review, 93(1), 113–132.

AFL-CIO. (n.d.). Right to Work | AFL-CIO. AFL-CIO. Retrieved October 3, 2019, from

https://aflcio.org/issues/right-work

Bono-Lunn, D. (2020). The Work or the Free Ride? The Impacts of U.S. Right-to-Work Laws on Wages, Free Riding, and Unionization. University of North Carolina at Chapel Hill.

Bureau, U. C. (n.d.). State Area Measurements and Internal Point Coordinates. The United States Census Bureau. Retrieved March 27, 2020, from

(40)

Economic Analysis, B. (n.d.). GDP and Personal Income Interactive Data. BEA.Gov. Retrieved March 27, 2020, from https://apps.bea.gov/iTable/index_regional.cfm

Eren, O., & Ozbeklik, I. S. (2011). Right-to-Work Laws and State-Level Economic Outcomes: Evidence from the Case Studies of Idaho and Oklahoma Using Synthetic Control Method (1101 Classification-JEL J51, J58, J21; Working Papers). University of Nevada, Las Vegas ,

Department of Economics. https://ideas.repec.org/p/nlv/wpaper/1101.html

Eren, O., & Ozbeklik, S. (2015). What Do Right-to-Work Laws Do? Evidence from a Synthetic

Control Method Analysis. Journal of Policy Analysis and Management, 35(1), 173–194.

https://doi.org/10.1002/pam.21861

Hirsch, B., & Macpherson, D. (n.d.). Union Membership and Coverage Database. Retrieved March

27, 2020, from http://unionstats.com/

Kalenkoski, C. M., & Lacombe, D. J. (2006). Right-to-Work Laws and Manufacturing Employment:

The Importance of Spatial Dependence. Southern Economic Journal, 73(2), 402–418. JSTOR.

https://doi.org/10.2307/20111898

Krisher, T. (2019, September 16). Union votes to strike at General Motors’ US plants. AP NEWS.

https://apnews.com/2b221852192a412d8bc409b87ce0f949

Lumsden, K., & Petersen, C. (1975). The Effect of Right-to-Work Laws on Unionization in the

United States. Journal of Political Economy, 83(6), 1237–1248. JSTOR.

Mautz, K. (n.d.). Wisconsin’s “Right-to-Work” Law. WISCONSIN LEGISLATIVE COUNCIL. Moore, W. J., & Newman, R. J. (1985). The Effects of Right-to-Work Laws: A Review of the

(41)

National Labor Relations Board. (n.d.-a). 1947 Taft-Hartley Substantive Provisions | NLRB | Public Website. National Labor Relations Board. Retrieved October 3, 2019, from

/about-nlrb/who-we-are/our-history/1947-taft-hartley-substantive-provisions

National Labor Relations Board. (n.d.-b). Pre-Wagner Act labor relations | NLRB | Public Website. National Labor Relations Board. Retrieved October 8, 2019, from

/about-nlrb/who-we-are/our-history/pre-wagner-act-labor-relations

National Labor Relations Board. (n.d.-c). The 1935 passage of the Wagner Act | NLRB | Public

Website. National Labor Relations Board. Retrieved October 8, 2019, from /about-nlrb/who-we-are/our-history/1935-passage-wagner-act

National Right to Work Committee. (n.d.). Benefit Enjoyed in States with Right To Work Laws

Archives. National Right To Work Committee. Retrieved October 3, 2019, from

https://nrtwc.org/category/benefits-of-right-to-work/

Norris-La Guardia Act, United States Congress, Title 29. Labor U.S. Code (1932).

https://www.govinfo.gov/content/pkg/USCODE-2010-title29/html/USCODE-2010-title29-chap6.htm

Opoien, J. (2015, February 20). Wisconsin Republicans announce plans to fast-track right-to-work, Walker will sign if passed. The Capital Times. https://madison.com/news/local/govt-and-politics/

wisconsin-republicans-announce-plans-to-fast-track-right-to-work/article_e9fad3e7-3e65-5e99-8bf0-00a790a9dcce.html

Reed, W. R. (2003). How right-to-work laws affect wages. Journal of Labor Research, 24(4), 713–

730. https://doi.org/10.1007/s12122-003-1022-1

Rios, E. (2018, March 23). Educators across the US are using the West Virginia teachers’ strike to

(42)

https://www.motherjones.com/politics/2018/03/educators-across-the-us-are-using-west-virginias-teachers-strike-to-inspire-their-own-battle-plans/

Roberts, A. J., & Habans, R. A. (2015). Exploring the Effects of Right-to-Work Laws on Private Wages. https://escholarship.org/uc/item/5n091465

Ruggles, S., Flood, S., Goeken, R., Grover, J., Meyer, E., Pacas, J., & Sobek, M. (2020). IPUMS

USA: Version 10.0 [dataset]. https://doi.org/10.18128/D010.V10.0

Shierholz, H., & Gould, E. (n.d.). The compensation penalty of “right-to-work” laws. Economic

Policy Institute. Retrieved October 2, 2019, from https://www.epi.org/publication/bp299/

Wessels, W. (1981). Economic Effects of Right to Work Laws. Journal of Labor Research, 2, 55–75.

Figure

Updating...

References

Updating...

Related subjects :