Abstract
While considerable efforts have been made by legislators, business associations, and political organizations to pass right-to-work (RTW) laws in states across the country, the empirical evidence on the effect of adopting an RTW law on labor market outcomes and state budgets is both varied and mixed. This article provides a forecast on the effect of RTW laws on important labor market outcomes—including earnings, employment, unionization, and inequality. It also investigates RTW’s impacts on two particularly affected industries (manufacturing and construction) and three demographic groups (African-American, Latino/a, and female workers). The findings are subsequently applied to the state of Illinois to project the potential law’s impact on Illinois workers and on the state’s tax revenues. By and large, as a policy prescription, RTW would generate harmful effects to Illinois’ economy, lower its capacity to provide essential public services and degrade the quality and condition of the state’s labor force.
Introduction
The nine million–job shortfall induced by the late-2007 economic recession and the election of Republican governors, and legislative majorities since 2010, has caused a number of states to rethink policies on employment and income growth. One proposed policy change in many struggling states is the implementation of a right-to-work (RTW) law, which limits the ability of labor unions to collect “fair share” fees from the workers they represent and influence the conditions of employment for a workplace (Devinatz 2011). In 2012, Indiana became the first state in America to adopt an RTW law since Oklahoma in 2001. Michigan followed in early 2013, and in March of 2015, Wisconsin became the twenty-fifth state to adopt an RTW provision.
Since the 2010 elections, initiatives to include RTW laws in a state’s legislative agenda have begun in at least ten other collective bargaining (CB) states, including Illinois. Since 2011, Republican lawmakers in each of the following states have introduced RTW legislation: Rhode Island (Spetrini 2013), New Jersey (New Jersey State AFL-CIO 2011), Illinois (Hyde Park Johnny 2013), Kentucky (Flack 2013), Pennsylvania (Ryan 2013.), Maine (Stone 2013), Ohio (Vardon 2013), and New Hampshire (Leubsdorf 2013). The latter three states defeated attempts to pass RTW legislation either by popular vote or in their state legislatures. In Minnesota, a constitutional amendment establishing the state as an RTW state was introduced in 2012 (Ragsdale and Brooks 2012). Finally, in 2015 the Republican dominated Missouri House narrowly failed to override Democratic governor Nixon’s veto of an RTW bill (Hancock 2015). In early 2015, Illinois’ newly elected Republican governor proposed county-by-county “union free zones,” modeled on the passage of such a measure in Kentucky. 1 Later in the spring, the state’s General Assembly voted 72-0 against an RTW bill. 2
While considerable efforts have been made by legislators, business associations, and political organizations to pass RTW laws in states across the country, the empirical evidence on the effect of adopting an RTW law on labor market outcomes and state budgets is both varied and mixed. This paper provides a forecast on the effect of RTW laws on important labor market outcomes—including earnings, employment, unionization, and inequality. It also investigates RTW’s impacts on two particularly affected industries (manufacturing and construction) and three demographic groups (African-American, Latino/a, and female workers). The manufacturing industry has been a focus of RTW study because it is subject to both low-wage foreign competition and capital-labor substitution effects. Economic theory has postulated that industries, which require less than a college degree, are vulnerable to relocation or technological redundancy. Consequently, these industries need to reduce labor cost (i.e., via RTW) in order to protect domestic jobs. The construction industry is also a typical case study because it has relatively high paying occupations and is often heavily unionized. It is also an industry dependent on and subject to government contracts and regulations. Advocates contend that RTW would reduce construction costs, thereby extending the value of construction expenditures. The national findings are subsequently applied to the state of Illinois to project the potential law’s impact on Illinois workers and on the state’s tax revenues.
What Is an RTW Law?
The term right to work is a misnomer that has little to do with the right of a person to seek and accept gainful employment. Rather, RTW is a state-level policy that prohibits a labor union from negotiating union security clauses with an employer. In essence, it is a restriction on the liberty of the two parties to form a contract, with the intent to discourage union activities. Union security clauses are the legal mechanisms by which unions overcome the collective action, or “free-rider,” problem (Olson 1965). The free-rider problem arises because individuals who stand to enjoy the fruits of the collective also have an incentive to avoid making any contribution to the cause, especially if they believe the group will succeed without their support. When a significant number of individuals elect to “free ride,” the collective becomes resource-starved, causing it to underperform or fail altogether. To avoid this fate, rules, which limit the ability of an individual to shirk their obligation to the group, are necessary.
Union security clauses are not automatic. They are negotiated contract provisions that regulate the collection of union dues. In CB states, the parties are free to negotiate a range of union security options. Labor unions typically prefer “union shop” terms, requiring every person who benefits from union representation to pay either union dues or fees. In RTW states, the parties are barred from negotiating union security clauses, defaulting instead to “open shop” terms, whereby the payment of dues or fees is optional for workers represented by the union. Between these two policy poles are arrangements that require represented persons to pay a proportion of full dues or even to allow union objectors to contribute dues to charity. Such arrangements are, however, prohibited by most RTW legislation.
In an RTW state, objectors to unions still receive all the benefits of unionization if they are covered by a union contract. It is important to understand that a labor union does not control the composition of the bargaining unit. Thus, coverage is not decided by the union, but rather by an administrative agency (i.e., a federal or state labor board) that determines if a particular worker shares a “community of interests” with the unionized group. By both law and the practicalities of the workplace, however, unions must represent all persons covered by their organization. In RTW states, persons covered can refuse to pay dues or fees and yet receive all the benefits that are earned from the contributions of others, including the union wage premium, health insurance, pension benefits, and representation through the grievance system. 3
In a CB state, on the other hand, a labor union and employer may (but are not mandated to) agree to a union security clause that requires all covered persons to pay dues or fees to finance CB activities. In such situations, someone seeking to avoid paying dues or fees to the union has three options: (1) voluntarily separate from their job, (2) convince union leadership to negotiate an open shop, or (3) persuade fellow workers to decertify the union. Given that the last two outcomes are difficult to achieve, the most viable option for dissenters is to work elsewhere.
Review of Previous RTW Studies
A recent study released by the Congressional Research Service finds that “existing empirical research is inconclusive” and that “even the most sophisticated studies are unable to fully isolate the effects of varied union security policies” from the effects of RTW laws (Collins 2012, p. 14). Table 1 briefly summarizes the highest regarded research on the topic in the past few years. The intent of this article is to add to the extant studies on this issue by providing a more conclusive analysis.
Concise Review of Previous RTW Studies, 1980-2011.
RTW = right to work; CB = collective bargaining.
Previous studies on the impact that RTW has on earnings are varied and mixed. At one extreme, one paper found a 7.9 percent increase in wages due to the law (Reed 2003). The study suffers from serious statistical deficiencies, including a small sample size, inadequately controlling for other factors, which may increase wages over time, and assigning wage growth that would have occurred in the absence of an RTW law. 4 On the other end of the spectrum, two researchers found that RTW lowers the price of labor by as much as 18.3 percent. If the “price of labor” is closely aligned with a worker’s actual wage, as economic theory purports, then RTW could impact average wages by as much as 18.3 percent (Garofalo and Malhotra 1992). In the middle are studies which show that RTW has no statistically discernible effect and those which show a moderate decrease in wages as a consequence of RTW laws (Eren and Ozbeklik 2011; Gould and Shierholz 2011; Hogler 2011; Lafer 2011; Moore 1980; Stevans 2009). Due to flawed methodologies in the studies, which find increases in wages in the law, a full literature review suggests that RTW by itself lowers wages by between 0 and 5 percent on average. It should be noted that one study found that RTW, while lowering worker wages, is associated with a 1.9 percent increase in proprietor incomes, indicating that RTW is a transfer of income from employees to owners, with little trickle-down effect (Stevans 2009; Table 1).
Studies that analyze employment levels also range in their results. Those studies that report total employment impacts generally find no statistically significant association between RTW and total employment in a state, with a 2011 report actually finding a 1.0 to 3.0 percent decline in employment (Lafer and Allegretto 2011). Early research on RTW’s impact on manufacturing employment tended to find a moderate (2.1-6.6 percentage point) increase in manufacturing’s share of a state’s economy associated with having an RTW law; however, recent research that found no effect on manufacturing employment calls into question the statistical approach and conclusions of previous works (Eren and Ozbeklik 2011). In general, RTW by itself appears to have no or a very small effect on a state’s employment level (0.0% plus or minus about 3.0%). While some special interests may find these conclusions surprising, in reality these common estimates are quite reasonable, given that surveys of large firm location decisions report that RTW laws are not regarded as important factors influencing where a firm does business. Instead, labor skills and costs, state and local tax incentives, highway accessibility, energy availability and costs, and proximity to major markets are far more important (Lafer 2011).
One goal of labor unions is to protect worker safety and health. A 2011 study found, after accounting for characteristics specific to individuals and states, which may affect health insurance spending, that living in an RTW state was associated with a 2.6 percent reduction in employer-sponsored health insurance benefit spending on average (Gould and Shierholz 2011). To the extent that health benefits translate to healthier, more productive, and safer employees, RTW laws appear to have a negative impact. RTW laws also weaken the effectiveness of unions to protect member safety, with higher construction fatality rates in RTW states (0.14-0.18 per thousand workers) compared with CB states with high union density (0.11 per thousand workers; Zullo 2011).
One area where there is a general consensus among researchers is on the negative effect that RTW laws have on union membership and union power. The accumulation of studies indicates that RTW laws decrease unionization by between 5 and 8 percentage points (Moore 1980). A 2004 study focused on union density and concluded that RTW laws “exert an independent and strongly negative effect,” reducing union density by 8.8 percentage points on average after holding other factors constant (Hogler, Shulman, and Weiler 2004).
That RTW lowers the level of unionization in a state is intuitive. The work that unions perform is not cost-free. In addition to organizing members, negotiating and administering contracts, and dealing with political affairs, labor unions in RTW states must expend resources on activities aimed at discouraging members from defecting even while getting by with less financial resources at their disposal. By hampering the collection of dues, advocates for RTW understand that the policy puts organized labor in a weaker position and tends to result in lower union membership rates. This, paired with the inconclusive effect of RTW on wages and employment, has led one researcher to assert that the true intent of RTW laws is based on “hidden objectives” that are more ideological than pragmatic: “less influence for unions, less bargaining power for workers, more wealth for the wealthy, and more misery for the immiserated” (Hogler 2011, p. 303).
Methodological Approach of Analysis
This study predominately uses data from two sources: the Current Population Survey Outgoing Rotation Groups (CPS-ORG) and the Covered Employment and Wages program (commonly called ES-202 data). CPS-ORG data—which are collected, analyzed, and released by the U.S. Department of Labor Bureau of Labor Statistics (BLS)—report individual-level information on twenty-five thousand respondents nationwide each month. The records allow for analysis of such outcomes as wages, union wage effects, and wage inequality while controlling for demographic, geographic, educational attainment, and other work variables. The CPS-ORG dataset is also remarkably large; capturing data on 3,207,587 individuals aged sixteen to eighty-five in the United States over the ten-year period from 2003 to 2012. Weights are provided by the BLS to match the sample to the actual total U.S. population that is sixteen years of age or greater for each year. These weights adjust the influence of an individual respondent’s answers on a particular outcome to compensate for demographic groups that are either underrepresented or overrepresented compared with the actual population. In 2012, for example, there are 315,164 observations; the weighted number of observations is 243,284,338. The data were extracted from the user-friendly Center for Economic and Policy Research (CEPR) Uniform Data Extracts (Center for Economic Policy Research 2012) and use the preferred real wage variable, which converts all worker incomes into uniform hourly wages and amend them to 2012 dollars using the Consumer Price Index Research Series Using Current Methods (BLS 2012) inflation adjustment.
The ES-202 program provides the most complete database of monthly employment and quarterly wage information in America. ES-202 data are collected, analyzed, and released by the BLS. The records include data for workers covered by either state unemployment insurance laws or by the Unemployment Compensation for Federal Employees system, comprising over 96 percent of America’s workforce. ES-202 data are released as county-level estimates and are based on the location of the job rather than on a worker’s residence. The data include information for eighty-seven private-sector industries in 3,174 U.S. counties over the eleven-year period from 2001 to 2011. Consequently, even after bounding the analysis to average incomes between $10,000 and $200,000 to remove unusual outliers, the ES-202 program yields a large sample of 1,729,662 observations.
There are, of course, limitations to both datasets. For ES-202 data, many self-employed workers, domestic workers, nonprofit employees, and Armed Forces members stationed in America are excluded from the dataset. ES-202 also does not capture data on hours worked, and cannot be broken down by full-time and part-time workers. Finally, the BLS provides estimates of average employment and average total wages by industrial classification. Many of the following estimates using ES-202 data therefore present an average effect of RTW on averages, rather than at an individual level. CPS-ORG data, contrary to ES-202 data, report a worker’s state of residence rather than state of employment, so the results may be biased by workers who live in RTW states but work in CB states (e.g., living in Iowa but working in Illinois) and vice versa. CPS-ORG data, moreover, are based on household survey responses rather than on administrative payroll reports. Nevertheless, employing both datasets provides for a level of robustness in the findings as both tend to reach similar conclusions.
Lastly, to analyze the simple correlations between RTW and both employee benefits and workplace fatality rates in the construction industry, data were compiled from two additional sources. The Economic Census, jointly administered by the U.S. Census Bureau and U.S. Department of Commerce, provides state-level data on the dollar amounts spent by employers on both legally required benefits and fringe benefits. The most recent year in which the Economic Census was carried out and for which information is available is 2007. The information on fatality rates in construction is from the BLS Census of Fatal Occupational Injuries (CFOI), which provides data from 2008 to 2010.
The analysis conducted employs four statistical approaches. Each method aims to account for unmeasured characteristics, separating out an accurate estimate of the causal effect that RTW laws have—or do not have—on labor market outcomes.
For the CPS-ORG data, two strategies are used. In one empirical design, an ordinary least squares (OLS) regression model is used to determine wage and wage inequality estimates. The model is run three times: once evaluating the correlation between RTW and economic outcomes on the whole population sixteen years of age or older; a second evaluating the impact of RTW on outcomes, which narrows the analysis to “working-age” individuals who are sixteen to sixty-four years of age but adds in controls for demographic, educational, industry, and work factors; and a third time to include state-level effects that capture other state policies or variables that may influence outcomes. Note that incorporating the state-level effects provides very conservative estimates. In each run, the model is weighted so that the sample matches the U.S. population.
The second strategy involves a probit regression model to analyze the effect of RTW on whether an individual is employed as a union member. This approach allows for estimates of the probability of being employed and the probability of being a union member. Similar to the OLS regression strategy, this model is run three times: once including the whole sample sixteen years of age or older, a second time with working-age individuals and with a host of control variables, and finally a fully controlled run with other state-level factors.
The third empirical design incorporates the ES-202 data and uses a population mixed regression model with random intercepts, which allows for a proper estimate of the effect of RTW laws. Essentially, this method assigns a value to each county in the United States, which simplifies and quantifies county characteristics and policies apart from whether or not it has an RTW law.
The final strategy, which uses ES-202 data, combines advanced statistics and the case study strategy to compare contiguous counties where RTW is the law of the land on one side and CB is the state law across the border (Dube et al. 2010). In 2011, the contiguous border counties included 195 RTW counties and 180 CB counties, including Washington, D.C. 5 Since “the counties border each other, differences due to geographic and locational factors should be minimized,” with the county economies likely to look similar and be interconnected so that RTW is one of the only differences from one county to its neighboring county across the border (Holcombe and Lacombe 2004, p. 412). In this border-county regression model, a collection of ratios based on the spatial distance of each CB county’s borderline that “touches” an RTW county are used as weights and applied.
RTW Impacts on Labor Market Outcomes
Table 2 provides descriptive statistics of the employed workers in 2012, sorted by residence in either a CB state or an RTW state. The statistics, from the CPS-ORG program, are itemized into three general categories—work and industry, demographics, and education—and illustrate some noticeable differences between CB and RTW states.
Descriptive Statistics of Employed Workers, CB versus RTW, 2012.
Source. Center for Economic Policy Research. (2012). “CPS Uniform Data Extracts.” Current Population Survey Outgoing Rotation Groups.
Statistics are adjusted by the outgoing rotation group earnings weight to match the total population sixteen years of age or older. CB = collective bargaining; RTW = right to work; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research.
Workers earn more in CB states. Without adjusting for any factors, the average hourly wage of a worker is $2.36 higher in CB states than in RTW states. Notably, while the unemployment rate is 1.1 percentage points higher in CB states, much of this can be attributed to the fact that more people participate in the labor force in CB states. In CB states, 64.1 percent of individuals sixteen years or older were either employed or were looking for a job in 2012 compared with 63.2 percent in RTW states. In fact, CB states had a slightly higher aggregated employment rate among this group than RTW states in 2012 (+0.2 percentage points). Thus, while RTW states as a whole do have a lower unemployment rate compared with CB states as a whole, they also have a lower employment rate.
It could be that workers in CB states earn more because of a number of other work factors, including unionization, being a full-time worker (working 35 hours per week or more), or the industry in which one works. In 2012, 15.2 percent of all employed workers in CB states were members of a labor union, nearly triple the 5.7 percent union membership rate of RTW states. Employees in CB states worked somewhat shorter workweeks than their RTW counterparts (−1.3 hours) and were actually less likely to be employed by the public sector (−0.6 percentage points).
A number of demographic variables may also influence a worker’s hourly wage. The employed labor force in CB states had more females (+0.6 percentage points) and white non-Latinos (+4.1 percentage points) than the RTW workforce. CB states also tended to have more immigrants (+3.5 percentage points), fewer married individuals (−1.8 percentage points), and fewer military veterans (−1.8 percentage points) than RTW states. A higher share of individuals in CB states were also both employed and in school than in RTW states (+0.5 percentage points). This last finding is not unsurprising, as workers who reside in CB states also tend to be more highly educated than their RTW counterparts. For example, 36.1 percent of CB workers have attained a bachelor’s degree or more. By contrast, 31.3 percent of the RTW workforce has achieved that level of education. Given that education is very strongly correlated with labor market outcomes, it is critically important to control for this difference in outcomes to understand the actual effect of RTW laws on wages, unionization, employment, and inequality.
Across all models, the data strongly indicate that RTW lowers worker earnings. The following section presents results from all models, beginning with individual-level estimates of the effect of RTW on the real hourly wages of all employed workers. County-level estimates of RTW’s impact on private-sector workers are then presented before turning to effects on the manufacturing and construction industries.
Effect on the Real Wage of All Workers: An Individual-Level Analysis
A simple correlation between RTW and average hourly wage (in constant 2012 dollars) shows that RTW laws tend to lower a worker’s wage by $2.08 per hour. At the same time, union membership is associated with a $3.74 per hour increase in a worker’s hourly wage (Table 3). These results do not control for a host of other factors that may influence a worker’s hourly wage but are statistically significant and do suggest that RTW laws and labor unions have noteworthy impacts on a worker’s income.
RTW Effects on Real Hourly Wage, All Employed Workers.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
Full dataset includes 1,730,415 observations; working-age observations include 1,638,946 respondents. Analysis based on simple OLS regression model. Observations weighted to match U.S. population. RTW = right to work; CB = collective bargaining; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research; OLS = ordinary least squares.
Based on the average yearly growth in wages over ten years.
To investigate the true causal effect of RTW laws on a worker’s hourly wage, the second run of the model—which focuses on the working-age population eighteen to sixty-four years old—accounts for demographic variables, educational attainment, industry, and time-specific effects. In this context, the impact of RTW on a worker’s wage is actually more negative, at $2.15 less per hour. The real (i.e., inflation-adjusted) wages of workers in RTW states, however, have been growing by $0.09 per hour per year compared with a $0.20 per hour per year decline in CB states. The small $0.29 hourly advantage per year due to RTW, however, begins to diminish after 7.5 years, and after fifteen years, CB wages actually again grow faster than RTW wages. Union membership, on the other hand, is statistically associated with a $1.97 per hour increase in a worker’s real wage on average. In addition, unions shield workers in RTW states against the drop in wages associated with RTW laws, by $1.09 on average in those states.
Over the eighteen recessionary months from December 2007 to June 2009, the real wage of an employed worker increased by $0.44 per hour on average in CB states and by $0.22 per hour on average in RTW states. This increase in the wages of those who have a job during the Great Recession is likely the result of wage stickiness, of employers laying off lower skilled workers, and of increased worker protection in CB states by unions. Since the recession, real wages have rebounded, increasing by $0.71 per hour on average in CB states and just $0.41 per hour in RTW states. The analysis indicates that, as a labor market policy, RTW has hurt workers during and since the recession, lowering wages from where they would otherwise be in a CB setting.
Contrarily, union membership protected worker wages during the recession and has helped increase them in the postrecession period. During the recession, being a union member was statistically associated with per hour real wages that were $0.19 higher than nonunion workers in addition to the union wage premium of $1.97. Since the recession, union wages have grown by $0.28 per hour higher than nonunion wages in addition to the union wage premium, for a $2.25 real per hour wage premium.
Table 4 combines all RTW and union variables over time to understand RTW’s standalone effects from 2003 to 2012. Compared with a worker in a CB state, a worker in an RTW state earned $1.58 less per hour in real terms over the period. At a normalized 2,080 hours worked per year, RTW reduced a worker’s total wages by $32,871 over the ten-year period in constant 2012 dollars. Compared with nonunion workers in RTW states, union workers in CB states performing the same work in the same industry with the same level of education earned $3.68 more per hour on average over the period just by both being a union member and living in a CB state. Over ten years, assuming a normalized 2,080-hour work year, a worker in an RTW state who is not a union member would, on average, earn $76,523 less than a comparable worker in a CB state who is a union member (Table 4).
The Effect of RTW on Hourly Wages, 2003-2012.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
Full dataset includes 1,730,415 observations; working-age observations include 1,638,946 respondents. Analysis based on simple OLS regression model. Observations weighted to match U.S. population. RTW = right to work; CB = collective bargaining; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research; OLS = ordinary least squares.
Based on the average yearly growth in wages over ten years.
Referring back to the far right column in Table 3, the results once again show a strong union wage premium at $1.80 per hour for CB states and $1.09 per hour for RTW states. The outputs also indicate that real wages for employed workers actually increased slightly during the Great Recession ($0.33 per hour), that wages have recovered faster in CB states than RTW states, and that unions have helped lift worker wages in the postrecession years. However, this run suggests that RTW has no discernible effect on a worker’s wages, except that RTW caused wages to grow by $0.12 per hour on average. These results could indicate that other state-level policies and factors are actually far more important in determining a worker’s wage than is RTW.
Another method to estimate the effect of RTW laws on earnings is to “normalize the data” and analyze the results in terms of percentages. Table 5 shows estimates based on this type of analysis. A simple correlation test projects that real wages are reduced by 8.2 percent due to RTW laws and that being a union member increases a worker’s real wage by 23.4 percent on average. These estimates, however, do not control for other important factors, such as demographics, work and industry characteristics, and education.
RTW Effects on Natural Log of Real Hourly Wage, All Employed Workers.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
Full dataset includes 1,730,415 observations; working-age observations include 1,638,946 respondents. Analysis based on simple OLS regression model. Observations weighted to match U.S. population. RTW = right to work; CB = collective bargaining; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research; OLS = ordinary least squares.
Based on the average yearly growth in wages over ten years.
After adding in those other important factors in order to really parse out RTW’s effects, RTW is found to lower worker wages by 7.3 percent on average in comparison with CB states. RTW, however, is statistically associated with a small 0.3 percent annual increase in wages for the employed population. Some of these gains, however, were lost in the postrecession years, which have seen a 0.9 percent decrease in wages in RTW states. When controlling for additional variables, unionization raises a worker’s wages by 12.9 percent (compared with 10.3 percent in RTW states). Nationwide, unionization has also grown real wages by 0.8 percent in the postrecession years beyond the 2.8 percent national rebound in hourly wages (Table 5).
Upon capturing other state-level factors, it becomes clear that RTW laws have a negative impact on worker wages now and over time. The final run on the model reports that RTW laws are strongly associated with a 2.1 percent drop in real wages, have no effect on wage growth over time, and reduced wages by 1.0 percent further since the recession than they otherwise would have been in a CB setting. In this run, union membership is found to raise a worker’s hourly wage by 11.9 percent in CB states and 9.9 percent in RTW states. Moreover, since the recession, a worker who is a member of a labor union has, on average, experienced an additional 0.6 percent bounce back in wages above all other workers (Table 5).
Effect on the Total Wages and Salaries of Private-Sector Workers: A County-Level Analysis
Limiting the analysis exclusively to private-sector industries produces remarkably similar results (Table 6). In the first population mixed regression model, the effect of RTW is a $2,030 drop in wage and salary income on average. Similarly, a second approach which allows for RTW to have a positive or negative yearly effect on wages suggests that the law reduces worker wages by $2,338 on average, with wages subsequently rebounding slightly over the next twenty years but never surpassing those of CB states. That is, wages in RTW states inched closer to CB states each year by $75.84. Extrapolating this finding, wages in RTW states will catch up to CB states by $1,517 after twenty years, but this projected increase will not be enough to compensate for the initial $2,338 annual wage gap between RTW and CB states. 6 Furthermore, after twenty years, wages in CB states grow faster each year than in RTW states. Therefore, any wage gains spurred by RTW over time never make up for the initial downward pressure of RTW on wages.
RTW Effects on Wages and Salaries of Private-Sector Workers.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 1,730,415 private industry-county groupings; border-county observations include 4,080 pairs. Analysis based on multilevel mixed regression model. RTW = right to work; BLS = Bureau of Labor Statistics.
Based on the average yearly growth in wages over ten years.
If Illinois adopted an RTW law next year, over ten years, a private-sector worker who currently earns $35,000 in wage and salary income will earn $24,393 less on average (Table 7). Finally, results from studying only industries in RTW counties and CB counties, which border one another, again show a strong negative impact of RTW on worker wages (−$2,222; Table 6).
Projected RTW Effect on an Illinois Private-Sector Worker Earning $35,000 per Year.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 1,730,415 private industry-county groupings; border-county observations include 4,080 pairs. Analysis based on multilevel mixed regression model. RTW = right to work.
Once again, “normalizing the data” helps to reveal the effect of RTW laws on earnings in terms of percentages and generate telling results. Table 8 shows that average industry wages in RTW counties are between 5.7 and 7.5 percent lower than in CB counties. Conducting the analysis using only the border counties approximates that wages grow by 0.41 percent each year in RTW counties due to the law, but the effect may diminish over time. One result from the border-county approach is that RTW laws seem to have positive impacts on earnings (Table 8). Given the otherwise uniform finding that RTW laws lower private industry wages, this outlying result may simply be an anomaly—perhaps a separate, unrelated feature of the border counties tends to favor the private sector on the RTW side, such as a more “pro-business” environment (Holmes 1998).
RTW Effects on the Growth in Wage and Salaries of Private-Sector Workers.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 1,730,415 private industry-county groupings; border-county observations include 4,080 pairs; wage growth results include 1,598,763 observations. Analysis based on multilevel mixed regression model. RTW = right to work; CB = collective bargaining; BLS = Bureau of Labor Statistics.
Based on the average yearly growth in wages over ten years.
In the first mixed regression approach, the RTW indicator is statistically associated with a 0.33 percent increase in year-to-year wage growth on average (Table 8). From 2001 to 2011, wages in CB counties grew by a nominal (i.e., not adjusted for inflation) annual average of 2.67 percent compared with 3.00 percent for RTW counties. Pairing this finding with results from Table 5 implies that implementing an RTW law would be associated with an initial $2,030 decrease in the average private-sector worker’s wages, and it would take twenty-four years for a worker’s income to catch up to the value of what his or her earnings would have been if he or she would have continued to work in a CB state.
The aforementioned approach, however, assumes that RTW’s effect on wage growth is constant over time, but this may not be the case. It could be true that RTW has positive effects on wage growth in some years and negative impacts in others. The second column of Table 8 provides additional insights on RTW’s impact on wage growth. From 2001 to 2011, this run on the model reports that incomes in RTW industry counties grew by 2.96 percent compared with 2.72 percent in the CB counties. But the approach also indicates that industries in RTW counties are beginning at a lower base wage growth rate (−0.41 percent), suggesting that the RTW effect could partially be due to those economies simply playing “catch-up” to their CB counterparts (Holmes 1998, 14; Solow 1957). More importantly, the small yearly growth premium from RTW laws (+0.33 percent) only appears to last for 5.9 years before beginning to recede back to CB wage growth levels. Indeed, after 10.3 years, wages in CB states again grow faster than those in RTW states. These estimates are also projected for Illinois in Table 9. Accordingly, average industry wages in RTW counties never catch up to those in CB counties due to RTW legislation. Instead, wages make slight short-term gains before once again losing ground and ultimately hurting worker wages in the long run (Table 9).
Projected RTW Effect on Illinois Private-Sector Worker Wage Growth.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 1,598,763 private industry-county groupings. Analysis based on multilevel mixed regression model. RTW = right to work; CB = collective bargaining; BLS = Bureau of Labor Statistics.
Effect on Manufacturing and Construction Industry Wages
RTW laws are strongly statistically associated with a $2,815 total wage penalty for manufacturing workers on average, all else constant. Manufacturing employees in RTW counties suffer an annual wage loss (−$72.75 each year) due to the law compared with their CB counterparts. Applying this difference to Illinois, where there are over seven hundred forty thousand manufacturing employees and the industry is the second largest employer in the state, predicts that, over ten years, a manufacturing worker who earns $35,000 this year would bring home $31,252 less in total wages on average as a result of Illinois becoming an RTW state in 2014 (Table 10).
Projected RTW Effect on an Illinois Manufacturing Worker Earning $35,000 per Year.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 373,394 manufacturing industry-county groupings; border-county observations include 3,952 pairs. Analysis based on multilevel mixed regression model. RTW = right to work; CB = collective bargaining; BLS = Bureau of Labor Statistics.
Estimates on the percentage change in earnings corroborates this finding, as RTW laws are associated with manufacturing wages that are 8.6 percent lower than in CB counties on average. Moreover, an analysis of wage growth reveals—similar to the results for all private-sector industries—that average RTW manufacturing wage growth begins at a lower base rate (−0.61 percent), grows faster than CB manufacturing wages (+0.34 percent per year), but declines after a short time (6.6 years; Figure 1). Once again, RTW as a policy prescription does not by itself raise average wages above what they would have otherwise been in a CB state. In terms of earnings, manufacturing workers are better off in CB states.
Average manufacturing wage growth, CB states versus RTW states.
For construction workers, RTW laws have extremely negative earnings consequences. By all metrics, there is no discernible added benefit annually from working in the construction industry in an RTW county as opposed to a CB county. In fact, in terms of total wages, RTW laws reduce construction industry wages by $6,062 on average, without making any gain over time relative to CB states. Over ten years, a construction worker who earns $35,000 this year would therefore bring home $60,620 less in total wages on average as a result of Illinois becoming an RTW state (Table 11). While the industry only represents 5 percent of the state’s workforce, there are more than 311,000 construction employees in the state; therefore, the negative impact of RTW would be substantial.
Projected RTW Effect on an Illinois Construction Worker Earning $35,000 per Year.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 76,202 construction industry-county groupings. Analysis based on multilevel mixed regression model. RTW = right to work; CB = collective bargaining; BLS = Bureau of Labor Statistics.
The normalized approach substantiates this result by finding that construction wages are far lower in RTW counties, with workers taking a 22.2 percent pay reduction compared with their counterparts in CB counties. There is no statistically significant bonus for construction employees to working in an RTW county. Instead, because mean construction wages are lower in RTW counties and grow on average by the same percentage as in CB counties, simple compounding means that RTW legislation would serve as a considerable net negative for the earnings of employed construction workers.
Recapping RTW’s Effect on Earnings
Common conclusions can be drawn from both the individual-level analysis on the hourly wages of all workers and the county-level evaluation of the annual wage and salary incomes of private-sector workers. First, RTW has a negative effect on worker wages, holding all else constant. RTW laws lower a worker’s hourly wage by about $2 per hour. RTW is also found to be responsible for between a 2.1 and an 8.2 percent plunge in a worker’s hourly wage, with a middle estimate of a 7.3 loss in real hourly wages. Similarly, RTW is statistically associated with a 5.7 to 7.5 percent slump in the yearly earnings of private-sector workers. Moreover, RTW is strongly correlated with a drop in private-sector incomes ranging from $2,030 to $2,338. For the manufacturing and construction industries, RTW’s effect on annual total wage and salary income is respectively −$2,815 (−8.6 percent) and −$6,062 (−22.2 percent). In sum, fourteen out of sixteen evaluations of RTW’s impact on earnings report a negative impact, one produces a negative but statistically insignificant estimate, and one outlying analysis results in a positive impact.
Second, the impact of RTW on wage growth is unclear. In general, RTW appears to have a small positive effect on wage growth over time. The effect, however, ranges from 0 to 0.83 percent per year, with some results reporting declining effects over time. In total, two out of six models with controls find statistical evidence that wage growth due to RTW diminishes over time, three report suggestive evidence that wages fall, and one reports no effect of RTW on raising worker wages at all. Thus, while RTW may have a very minimal positive impact on real wage growth, it is unclear whether the increase would ever be enough to compensate for the policy’s prominent downward effect on wages.
Third, RTW laws had no positive effect on worker wages during the recession and have had an unambiguously negative effect on real wages in the postrecession years. The individual-level hourly wage approach provides suggestive evidence that RTW laws had a negative effect on worker wages during the Great Recession. In addition, there is statistically significant evidence that RTW has reduced real wages by 1.0 percent more than they are in CB states since the recession, resulting in hourly wages that are $0.30 to $0.40 lower.
Finally, in terms of real hourly wages, unions have fared far better at raising and protecting worker earnings than RTW laws. Union membership is correlated with a $3.74 and 23.4 percent increase in the hourly wage of an employed person. Holding all else equal, union membership has an unambiguously positive impact on a worker’s wages. When controlling for other factors, including demographics, work and industry variables, and educational attainment, the union hourly wage premium is still $1.80 to $1.97 (12-13 percent) per hour. While RTW limits a union’s ability to improve the wages of its members, the union wage effect is still quite robust in RTW states: the models report about a $1.10 and a 10 percent increase in per hour earnings from union membership in RTW states compared with nonunion workers in those states. While labor unions did not have a negative impact on worker wages during the recession, there is only suggestive evidence that they provided a positive boost to the wages of the employed. However, labor union membership has contributed to a $0.15 to $0.30 increase in worker wages (0.6-0.8 percent) above the rest of the employed workforce since the recession officially ended.
Effect on the Probability of Being Employed: An Individual-Level Analysis
If an RTW law significantly influences business location decisions, there should be noticeable effects on employment levels in RTW states compared with CB states. This section looks at the effect of RTW and other factors on the probability of an individual being employed. The model is a probit regression model, which reports the direction of the effect and its statistical significance (i.e., whether a factor is associated positively or negatively with an outcome, and whether it is likely to actually be a determinant of an outcome). To get at the magnitude of statistically significant factors, average marginal effects are generated. Essentially, this approach allows researchers to determine the average impact of a policy on the probability of an event occurring for the whole population. In this case, the event is having a job. Note that the actual share of the U.S. population sixteen years and older that did have a job in 2012 was 58.6 percent. Thus, the actual probability that any person sixteen years or older was employed was 58.6 percent in 2012.
Table 12 reports results from the probit regression model (which of course includes all persons in the dataset rather than solely the employed), beginning with a simple correlation before controlling for a host of demographic and educational factors, and then adding in a state-level variable to capture other state characteristics (e.g., a state’s unemployment insurance policy, a state’s weather, or a disproportionate share of a state’s youth attending college and choosing not to be employed), which provide a very conservative estimate. Initially, RTW seems to have a positive impact on an individual’s probability of being employed. Without controlling for any factors, RTW is associated with a 0.4 percentage point increase in the chance of an individual being employed. After accounting for factors such as age, gender, race/ethnicity, immigration status, veteran status, marital status, and level or education, RTW laws are associated with a 1.4 percent increase in the probability of being employed. However, since the recession, RTW has been linked with a smaller 0.4 percentage point bump in the likelihood of having a job. By comparison, during the Great Recession, the probability of having a job fell by 1.9 percentage points for all American residents. Since the recession ended in June of 2009, the chances that one has a job actually fell by 2.4 percentage points more leading to a 4.3 percentage point dip in the probability of being employed (Table 12).
RTW Effects on the Probability of Being Employed, Population Sixteen Years and Older.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for all respondents from 2003 to 2012.
Full dataset includes 3,194,414 observations; working-age observations include 2,529,880 respondents. Analysis based on probit regression model, average marginal effects. Observations not weighted to match U.S. population. RTW = right to work; CB = collective bargaining; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research.
Based on the average yearly change in the likelihood of being employed over ten years.
When other state-level policies and unmeasured factors are accounted for (e.g., a state’s unemployment insurance policy, a state’s weather, or a disproportionate share of a state’s youth attending college and choosing not to be employed), RTW has no statistically significant effect on the probability of being employed. In fact, there is suggestive evidence of a negative impact on employment. Since the recession, from 2010 to 2012, RTW laws lowered an individual’s likelihood of having a job by 1.2 percentage points. While RTW may have had a small positive effect prior to the Great Recession, the estimates in Table 12 indicate that it has hurt workers since the financial crisis.
Importantly, many other factors have a much larger effect on the probability of being employed than RTW laws. A white non-Latino individual is between 6.2 and 8.2 percentage points more likely to have a job than an African-American. Higher levels of education are very strongly statistically associated with having a job: those with just a bachelor’s degree are 10.5 percentage points more likely to have a job than those with just a high school degree, and at the extreme, individuals with a doctorate degree are 33.0 percentage points more likely to be employed than persons without a high school degree or equivalent.
Effect on All Private Industry Jobs: A County-Level Analysis
With regard to employment, Table 13 provides estimates of the effect of RTW laws on both the average change in the number of employed persons per industry and the percentage change per industry (since the size of industries is different across counties). Three-digit North American Industry Classification System (NAICS) groups categorize the private-sector industries. 7 In the assessments conducted on per person changes in industry employment, the estimated effects of RTW are approximately the same. Holding all else constant, an RTW law is statistically associated with a small 4.18 to 4.50 per industry increase in jobs per county compared with CB counties. Inclusion of the RTW and year interaction terms, however, finds that the boost in change in employment that is correlated with RTW diminishes entirely after 8.9 years. In measuring the percentage change in employment, RTW laws are associated with a minor 0.42 percent increase in private-sector employment each year. But once again the effect diminishes over time. While it appears that having an RTW law augments private-sector county employment by a total of 6.5 percent after ten years, just 3.6 years later, CB counties make up the job deficit entirely. That is, while it may appear that RTW has positive impact on employment over the first few years, over time the effect diminishes and employment returns to the level at which it would have been in the absence of an RTW law. Since RTW is also associated with slower wage growth compared with CB counties after 10.3 years, one compelling hypothesis for this trend is that increased consumer demand by CB workers incentivizes new businesses to open or relocate in CB states over time, raising employment in CB counties.
RTW Effects on Growth in Employment for Private-Sector Workers.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 1,598,763 private industry-county groupings. Analysis based on multilevel mixed regression model. RTW = right to work; BLS = Bureau of Labor Statistics.
Based on the average yearly growth in employment over ten years.
Effect on Manufacturing and Construction Industry Employment
The evidence on the impact that RTW laws have on manufacturing and construction industry employment is mixed. Table 14 chronicles RTW’s effects on manufacturing job growth and share of the private-sector economy. From left to right, the table reports estimates on RTW’s effects on the change in manufacturing employment, growth in manufacturing employment, and the ratio of manufacturing to total private-sector employment, including all counties in the third column and solely the border counties in the final column. Each model but for one reports similar results: the positive effects of RTW laws are transitory and become negative over time. While RTW laws are statistically associated with a 17.55 job increase in manufacturing per county, each year this gain is reduced by 1.98 jobs. While RTW provides some initial protection against manufacturing job losses, employment decline was −2.42 percent per year on average in RTW counties from 2001 to 2011 and was actually more negative than CB counties in the last two years of analysis (2010 and 2011).
RTW Effects on Manufacturing Employment.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 345,657 manufacturing industry-county groupings; border-county observations include 3,952 pairs; manufacturing share evaluations based on 34,475 observations. Analysis based on multilevel mixed regression model. RTW = right to work; BLS = Bureau of Labor Statistics.
Based on the average yearly growth over ten years.
The estimates of the effect that RTW has on manufacturing’s share of the economy—the ratio of manufacturing employment to total private-sector employment—partially conflict with the work of other researchers (Holmes 1998; Kalenkoski and Lacombe 2006). For all U.S. counties, holding all else constant, there is no statistical evidence that RTW laws have any impact on manufacturing’s share of private employment. The border-county analysis, on the other hand, provides results akin to extant research, reporting that RTW laws are statistically associated with a 3.43 percentage point increase in manufacturing’s share of the economy. In summary, there is inconsistent evidence across models and methods; RTW laws may increase manufacturing’s share of the economy but they may also have no effect.
RTW laws are causally related to slight per year increases in construction employment (+12.85 jobs and +2.00 percent), but once again, those advantages disappear in the long run. After thirteen years, the average yearly construction employment growth in RTW counties is less than in CB states. Table 15 summarizes the impact of RTW into average yearly effects over ten years.
RTW Effects on Construction Employment.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012) for industry-county pairs from 2001 to 2011.
Full dataset includes 42,680 construction industry-county groupings. Analysis based on multilevel mixed regression model. RTW = right to work; BLS = Bureau of Labor Statistics
Based on the average yearly growth in employment over ten years.
Recapping RTW’s Effect on Employment
The general conclusion is that RTW’s impact on job gains is mixed. In terms of the probability of having a job, the effect of RTW ranges from a 1.2 percentage point decrease to a 1.4 percentage point increase, with a middle estimate of a positive 0.4 percentage point rise in the likelihood of being employed in any job. For the county-level private industry analysis, RTW laws are associated with a 0.42 percent increase in private-sector employment each year, but the effect diminishes over time. The number of construction jobs expands by 2.0 percent in RTW states, but this growth slowly recedes. Over the eleven years of the analysis, the average growth in jobs is just 0.55 percent due to RTW, while wages and salary income was −$6,062 less per year. Thus, the law’s effect on employment ranges from (1) being positive, to (2) being positive but less effective over the course of just a few years, to (3) being negative. A more conclusive claim is that demographics and educational attainment are the largest measurable factors determining whether or not an individual has a job.
RTW Impact on Union Membership
Analogous to the probit regression model approach to determining RTW’s impact on the probability of having a job, this section analyzes the probability of an employed worker being a union member. First, a simple correlation detects a statistically significant negative correlation between RTW laws and union membership. RTW laws are associated with a 9.7 percentage point decline in the probability of being a union member (Table 16). A reduction in probability of being in a union is especially significant given that just 12.2 percent of all workers were union members from 2003 to 2012. 8 For the working-age population eighteen to sixty-four years of age, the 2003 to 2012 union membership rate was 12.5 percent.
RTW Effects on the Probability of Being a Union Member, Population Sixteen Years and Older.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
Full dataset includes 1,738,315 observations; working-age observations include 1,645,675 respondents. Analysis based on probit regression model, average marginal effects. Observations not weighted to match U.S. population. RTW = right to work; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research.
Inserting controls for demographic, industry, work, and education variables and limiting the research to the working-age employed population yields similar results. RTW laws lower a worker’s probability of being a union member by 9.9 percentage points on average. During the Great Recession, the union membership rate increased by 0.6 percentage points among employed workers. As the employed worker pool shrunk, unions protected the jobs of their members while those not represented by a labor union were more likely to lose their job. During the recession, RTW, on the other hand, was statistically associated with an additional 0.4 percentage point decline in the probability of being a union member, beyond the 9.9 percentage point average reduction. After adding in state-level factors, RTW is statistically associated with a 1.5 percentage point decrease in the probability of being a union member. This run on the model suggests that other state-level factors may also be important determinants of whether a worker is a union member (e.g., political or ideological predispositions of the state; Table 16).
However, outside of the industry of employment, almost no factor is as significant as RTW laws in determining a worker’s probability of being a union member. Being a manufacturing worker raises a worker’s likelihood of being a union member by 14.0 percentage points, and being a government employee at any level increases the probability by between 14.1 and 22.9 percent. Other characteristics have only a fraction of the impact. An individual is most likely to be a member at age fifty. Being fifty years old is associated with a 5.1 percentage point increase in the likelihood that one is a union member compared, for instance, with being twenty-five or fifty-five years old. Additionally, all else equal, men are 4.2 percentage points more likely to be a union member than are women, African-American and Latino/a workers are respectively 2.6 percentage points and 0.8 percentage points more likely to be in a union than white non-Latino workers, immigrant workers are 1.4 percentage points less likely to be union members than native-born workers, and veterans are 0.7 percent more likely to be union members on average (Table 16).
Conclusions from this section are consistent. The industry in which a worker is employed has the largest impact on the probability of being a union worker, non-white workers tend to be more likely to be union members than white workers, and individuals are most likely to be in a union at age fifty. Overall, RTW has a clear negative impact on union membership, but the effect may be as little as a 1.5 percentage point reduction in union membership and as large as a 9.9 percentage point fall.
RTW Effects on Inequality Measures
Given that RTW laws have nearly unambiguous negative effects on average wages and the probability of being a union member but have blurred effects on the probability of being employed, it could be true that the policy has an effect on inequality and the shrinking American middle class. 9 To get a sense of RTW’s effect on earnings inequality, a “90/50 ratio” in a simple regression model is used. The 90/50 ratio is simply the real hourly wage of the poorest person in the richest 10 percent of earners divided by the real hourly wage of the median individual of a given population over a period of time. In this analysis, the 90/50 ratio of each state in each month from 2003 to 2012 (120 months) is determined and assigned to individual workers who lived in the state in that month.
One individual typically does not have a drastic impact on inequality, so these results should generally be understood as correlations showing the sign of the effect rather than actual cause-effect estimates. Nevertheless, this section provides insights, which add to the literature on inequality and on RTW. Across the entire sample, the average 90/50 ratio (or inequality index) was 2.274. That is, on average, a worker in the Top 10 percent in terms of hourly wage earned at least 227.4 percent as much as the median earner in each American state from 2003 to 2012. 10 The respective lowest and highest 90/50 ratios in the dataset were 1.53 and 3.60.
Under the basic correlational approach, RTW laws are associated with a −.006 effect on an individual’s state-level inequality index (Table 17). The negative correlation implies that RTW laws slightly reduce wage inequality by compressing wages downward. After adding in the control variables, RTW laws are again correlated with a negative (−.012) impact on a worker’s state-level inequality index. But during the recession, RTW was associated with a .022-point increase in a worker’s state-level 90/50 ratio on average (.014 in RTW states compared with −.008 in CB states). RTW was further associated with an additional .037-point total increase in the postrecession era, suggesting that RTW might be contributing to increasing inequality over time. In total, over the study years, the independent RTW effect was associated with a slight annual increase in wage inequality. During the recession, inequality also rose by a marginal amount compared with what it otherwise would have been for workers in RTW states, indicating that RTW harmed workers during the recession.
RTW Effects on Wage Inequality, 90/50 Ratio.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
Full dataset includes 1,730,415 observations; working-age observations include 1,638,946 respondents. Analysis based on simple OLS regression model. Observations weighted to match U.S. population. 90/50 ratios for each state in each month are computed and assigned to each observation. RTW = right to work; CB = collective bargaining; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research.
Based on the average yearly growth in wages over ten years.
Perhaps the most interesting findings in Table 17 are those pertaining to union membership and the national trend over time. Being a union member in an RTW state is statistically associated with a reduction in inequality. In RTW states, unions promote a fairer distribution of wages. The effect of unionization in CB states is unclear, as the signs diverge. Perhaps a bit ominously, the model indicates that nationally, not only is wage inequality rising, but also it is increasing exponentially over time. From 2003 to 2012, the inequality index rose by between .050 and .060 uniformly across the nation. Since the real median wage of all workers in 2012 was $17.00 per hour, this finding suggests that the ninetieth percentile worker earns at least $0.85 more per hour than he or she earned in 2003. Assuming a normalized 2,080 hours worked per year, national inequality growth has increased the 90th percentile worker’s annual earnings by at least $1,770 relative to what they would have been in 2003, with an even greater redistribution of earnings for those in the rest of the top 10 percent. Without the growth in inequality, this $1,770 would have been distributed more evenly across the economy.
Effects on Race and Gender Wage Outcomes
Another measure of inequality is to analyze RTW’s effects on the real wages of African-American, Latino and Latina, and female workers. Table 18 presents the average real hourly wages of all workers by race or ethnicity in RTW states and CB states from 2003 to 2012. The results are broken down by native-born and immigrant workers but are not adjusted for other important characteristics such as education or industry. The results show the negative effect of RTW on all major racial/ethnic identifications. RTW has particularly negative effects on the wages of native-born minority groups: per hour work incomes are at least $2.49 lower in RTW states for native-born African-American, Latino/a, and Asian workers compared with their respective CB counterparts. By contrast, wages are $1.82 per hour lower in RTW states than in CB states for whites born in the United States. For foreign-born workers, RTW has a negative effect, but generally less negative than for those born in America. Only the wages of immigrant white workers fall by more on average than their native-born equivalents due to RTW; immigrant white workers experience a $2.20 hourly wage penalty from RTW. Among immigrant workers, Asians suffer the largest drop in hourly wages associated with RTW (−$2.65), while African-Americans (−$1.76) and Latinos and Latinas (−$0.80) still see a wage penalty. Although Latino/a workers experience the smallest RTW penalty, it should be noted that their average hourly wages are the lowest of all racial/ethnic groups in America. The wages of all other groups have further to fall.
RTW Effects on Working-Age African-American Wages.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
Full dataset includes 1,730,415 observations of workers of all ages. Observations weighted to match U.S. population. RTW = right to work; CB = collective bargaining; CPS-ORG = Current Population Survey Outgoing Rotation Groups; CEPR = Center for Economic and Policy Research.
Table 19 reports RTW effects on African-American, Latino/a, and female wages after controlling for all other important variables. All else equal, RTW laws have consistently negative impacts on the real hourly wages of employed African-Americans (Table 19). RTW laws are statistically associated with a $0.00 to $0.26 decrease in African-American real hourly wages on average, in addition to the wage penalty faced nationwide by African-American workers of between $2.21 and $2.61 less per hour and the uniform negative effect of RTW laws (−$2.10 per hour). In dollar figures, RTW laws are found to lower an African-American worker’s wage by a total of $2.36 per hour compared with what it would be in a CB state. Moreover, RTW is statistically associated with between a 1.9 to 8.5 percent wage penalty for working-age African-American workers.
RTW Effects on Working-Age African-American, Latino/a, and Female Wages.
Source. Authors’ analysis of CPS-ORG, CEPR Uniform Data Extracts (2012) for employed workers from 2003 to 2012.
For Latino/a workers, the results are more mixed but largely negative (Table 19). The findings show that RTW may lower the wages of Latino and Latina workers by $2.10 per hour (or 8.3 percent). After adding in the state-level controls, RTW is found to raise wages by $0.90 per hour (0.8 percent). When added to the estimates from Table 18, the consensus in the models is that at worst RTW lowers the wages of Latino/a workers and at best marginally raises them.
RTW’s effect on the wages of female workers is also somewhat unclear but largely negative (Table 19). Generally the findings clearly illustrate that women earn significantly less than men. Simply by being a female, a woman’s per hour wage is reduced by about $4.70, or about 17.5 percent. RTW has a negative impact on women, though the impact is less negative than it is for men. Female employees earn $1.90 less per hour than they would in a CB state. However, when state-level effects are included in the analysis, RTW laws yields a smaller $0.10 per hour decline on the real wages of female workers. The normalized model, though, shows that RTW laws lower the average female worker’s wages by 1.9 to 7.6 percent compared with their counterparts in CB states who perform the same work in the same industry with the same degree.
For African-Americans and women, RTW tends to have negative impacts on real hourly wages—RTW lowers African-American wages by between $0.19 and $2.39 per hour and by between 4.7 and 9.8 percent according to the four models. For women, RTW reduces hourly wages by between $0.00 and $2.15 and by between 2.2 and 9.2 percent. The impact of RTW on Latino/a workers ranged from an 8 percent loss in earnings to less than a 1 percent increase.
Relationship with Benefits and Workplace Fatalities: Construction Industry Case Study
Given that one goal of labor unions is to protect worker health and safety, and that RTW laws reduce unionization, this section explores the association of RTW laws with employer-provided benefits and workplace fatalities in the construction industry. The construction industry is selected as a case study for three reasons: (1) the earnings and employment effects of RTW previously reported are so negative that the industry warrants further analysis, (2) it is an industry that is particularly susceptible to workplace fatality, and (3) construction data for workplace fatalities and benefits packages is easily accessible. While the data do not allow for advanced analytics, the results are nonetheless suggestive of the effect that RTW might have on these important health and wellness outcomes.
Benefits provided to workers in addition to their salaries can be split into two main categories: legally required benefits and fringe benefits. Legally required benefits include expenditures made by employers for Social Security and Medicare contributions, unemployment insurance, worker’s compensation, and state temporary disability payments. In contrast, fringe benefits are voluntary expenditures made by employers for items such as life insurance premiums, pension plans, medical insurance premiums, welfare plans, and other union negotiated benefits.
The Economic Census provides state-level data on the dollar amounts spent by employers on both legally required benefits and fringe benefits in the construction industry. In 2007, the most recent year in which an Economic Census was conducted, employers in construction sectors spent roughly $5,050 per employee on legally required benefits in RTW states. In comparison, CB states spent over $1,200 per employee on legally required benefits during this same time period. The difference in spending is even more pronounced when comparing dollars spent on fringe benefits or voluntary expenditures by employers in construction. Firms spent an average of $4,613 per employee on fringe benefits in states with RTW laws in 2007. In this same year, construction employers spent approximately $2,918 more per worker on fringe benefits in states with CB rights. The disparity in these experiences is displayed in Figure 2.
Dollars spent per employee in construction sectors, CB states versus RTW states, 2007.
The BLS Survey of Occupational Injuries and Illnesses (SOII) provide state-level data for nonfatal cases of work-related injuries and illnesses that are recorded by employers under the Occupational Safety and Health Administration’s (OSHA) recordkeeping guidelines. A 2009 report conducted for Congress by the Government Accountability Office (GAO) found that many employers did not report workplace injuries and illnesses because they did not want to increase workers’ compensation costs and negatively impact their chances of winning contracts. Workers also did not report job-related injuries often out of fear of being disciplined or fired. In total, the GAO found that data from OSHA did not include up to two-thirds of all workplace injuries and illnesses. In addition, 53 percent of health practitioners reported experiencing pressure from companies to downplay injuries or illnesses, and 47 percent reported experiencing this pressure from workers.
The underreporting of occupational injuries and illnesses suggests that a comparison of the experiences of RTW states and CB states would be unreliable. Instead, fatality rates provide a more accurate assessment of comparative experiences on the state level. Simply stated, deaths of workers on the job are difficult to conceal. Fatal injury rates depict the risk of incurring a fatal occupational injury and can be used to compare risk among different worker groups. Data from BLS CFOI report that construction fatality rates in RTW states averaged 13.1 deaths per 100,000 workers from 2008 to 2010. In CB states, on the other hand, fatality rates are much lower at 9.6 deaths per 100,000 workers. For construction workers in Illinois, the fatality rate was lower at 9.4 deaths per 100,000. Figure 3 graphically represents these findings.
Incidence rates of fatal injuries in construction sectors in CB states, RTW states, and Illinois, 2008-2010.
Although this section did not incorporate any advanced statistical analysis, construction workers in RTW states experienced worse health and wellness outcomes than workers in CB states. Workers in RTW states on average received benefits packages that were $4,126 lower, including $1,208 less in the legally required benefits and $2,918 less in fringe benefits. Additionally, construction workplace fatalities in CB states are lower than in RTW states by 3.5 deaths per 100,000 workers.
Predicted RTW Impacts on Illinois
A forecast of anticipated labor market impacts, should Illinois adopt an RTW law, can be estimated by applying the average national effects of RTW to the state. For the following estimates, it is important to understand that the effects will be gradual and will not trigger instant changes to the labor market. 11
If Illinois adopted a statewide RTW law, earnings would fall by around 5.7 to 7.3 percent over time. For manufacturing and construction workers, RTW would reduce their wage and salary incomes by 8.6 and 22.2 percent, respectively. The effect of RTW on wage growth is unclear. RTW would have a small positive effect on wage growth that would diminish over time: a wage growth rate of between 0 and 0.4 percent per hour is predicted, with estimates closer to the former as Illinois earnings begins to align with those in RTW states. The union wage premium in Illinois would also fall by about 2.0 percentage points.
It is unclear whether RTW would increase employment in Illinois. The conservative, middle-of-the-road estimates purport that RTW would spur between a 0.4 and a 0.55 percent increase in jobs. Accordingly, the state’s unemployment rate would initially fall from 9.1 to between 8.3 and 8.5 percent. There is evidence that this small reduction in unemployment, however, may cease over a few years. Additionally, RTW on a local level would have particularly negative impacts on workers in the Chicago economic region, but the results are quite similar across each of the other regions (Figure 4). While the total employment impacts of local RTW zones range from a loss of 180 jobs in the Champaign-Urbana region to a 342-job gain in the Rockford area, the total employee compensation would decrease in all local markets. The Champaign-Urbana, Quad Cities, Rockford, and Springfield-Decatur communities would all experience worker earnings declines of around $40 to $60 million. A conservative estimate is that labor income would consequently decline by $16 million in the Peoria-Bloomington region and $104 million in the St. Louis economy. Tax loss would also occur in varying degrees based on the size of the regional economy. Thus, in the communities where employment slightly increases, local RTW zones create low-wage jobs and eliminate middle-class occupations.
Predicted economic impacts (isolated) of local RTW zones by integrated region.
While a degree of uncertainty characterizes employment effects, other labor market outcomes are much clearer. The unionization rate for Illinois would unambiguously decline from 14.6 to between 4.7 and 13.1 percent over time. RTW would reduce the hourly wages of African-American workers by between 1.9 and 8.5 percent and between 1.9 and 7.6 percent for women. For Latino/a workers, RTW appears to reduce hourly wages by as much as 8.3 percent but may raise them by a minimal 0.8 percent. RTW would also reduce the annual benefits packages offered to Illinois construction industry workers by as much as $4,126.
As previously noted, construction workplace fatalities are 3.5 deaths per 100,000 workers lower in CB states than in RTW states. A forecast of anticipated work-related fatalities for Illinois construction workers, should RTW be enacted, can be estimated by comparing Illinois fatality rates and fatality rates in RTW states. The average incidence rate of fatal injuries from 2008 to 2010 for construction workers in Illinois was 9.4 deaths per 100,000 workers. In Illinois in 2008, thirty-two construction workers were killed on the job while twenty-seven workers died in 2009 and another twenty-seven were killed in 2010. If RTW were to be passed in Illinois, it could be estimated that an additional 10.67 Illinois construction workers would lose their lives on an annual basis. This estimate assumes that construction industry production would be similar to levels experienced from 2008 to 2010. Extrapolated over the span of a decade, approximately 107 additional Illinois workers would suffer fatal work-related injuries in construction sectors due to enacting an RTW law. Since this assumes long-term production similar to that seen in the Great Recession when output recessed, this may in fact also be a conservative estimate of the increase in fatalities.
To determine tax impacts, findings generated from both the CPS-ORG data and ES-202 data were employed, providing for a degree of sensitivity analysis. The CPS-ORG findings show that RTW lowers real worker wages by 2.1 to 7.3 percent. They also show that the probability of being employed increases by 0.4 to 1.4 percent from RTW, although since the recession RTW laws have decreased the probability by 1.2 percent. These findings allow for the creation of a three by two matrix to determine the effect that implementing an RTW law would have on Illinois.
In 2012, the CPS-ORG data report that there were 5,470,769 residents older than sixteen years of age in Illinois. Of those, 60.1 percent were employed, higher than the share for all CB (58.6 percent) and RTW states (58.5 percent), for a total employed population of 3,286,695 workers. Often overlooked is that the proportion of Illinois’ population sixteen years or older that is in the labor force is 66.1 percent, greater than the proportion for all CB states (64.1 percent) and for all RTW states (63.2 percent). Consequently, the state’s 2012 unemployment rate of 9.1 percent was actually higher than the rate for all CB (8.6 percent) and RTW states (7.4 percent). If either more unemployed individuals in Illinois had decided to drop out of the labor force or more people in RTW states had opted to enter the labor force and look for a job, the 2012 unemployment numbers would look very different. If Illinois’ labor force participation rate matched that of RTW states, the state’s unemployment rate would have been 4.9 percent in 2012, and if RTW states’ labor force participation rate matched that of Illinois, their unemployment rate would have been 11.5 percent in 2012. Claims that RTW laws lower the unemployment rate should be subject to serious analysis.
CPS-ORG data also report that the average wage in Illinois matched the mean wage for all CB states in 2012, at $22.95 per hour and that the union membership rate for Illinois workers was 14.6 percent, slightly lower than the rate of 15.2 percent for all CB states.
Table 20 displays the matrix of estimates of the impact of adopting an RTW law in Illinois. The fall in real hourly wages by either 2.1 percent or 7.3 percent adjusts downward the average real wage to $22.46 or $21.27 per hour. At the time of this analysis RTW could cause the state’s unemployment rate to either fall to 7.0 percent or 8.5 percent or, if recent RTW trends were upheld instead, increase to 10.9 percent. In only one scenario are the effects of implementing an RTW law positive on labor income and state income tax revenues in Illinois. In the five other scenarios, both total worker earnings and state income tax revenues would be projected to decline. The middle-of-the-road estimates of the effect that RTW would have on labor income indicate a $1.9 billion to $8.9 billion annual reduction in labor income across the state, with resultant yearly tax revenue declines of between $76.9 million and $355.0 million. Note that, due to “wage stickiness” and the fact that many CB agreements would remain valid and binding for a few years after Illinois became an RTW state, these estimates are likely more appropriate after the third year of the law’s adoption.
Estimated Impacts of Adopting an RTW Law in Illinois, Individual-Level Model.
Source. CPS-ORG (2012).
RTW = right to work; CPS-ORG = Current Population Survey Outgoing Rotation Groups.
The labor income effect is the average hourly wage (adjusted by the wage effect) multiplied by the average usual weekly hours worked by Illinois residents (33.86 hours) multiplied by fifty-two weeks multiplied by the number of employed persons (adjusted by the employment effect).
The state tax effect is the change in labor income multiplied by a uniform average tax rate of 4 percent. Since 2011, the Illinois state income tax is a flat rate of 5 percent of net income. Given standard exemptions/deductions and that Manzo and Bruno (2013) found the average state income tax rate for Illinois residents was 4.7 percent, the assumed 4 percent rate provides a conservative estimate.
The findings from the ES-202 data show that RTW lowers worker wages by 5.72 percent and stimulates a one-time 0.32 percent growth in the employment base (Table 21). Comparable estimates from a second run on the model report that RTW laws reduce wages by 7.54 percent initially with subsequent 0.41 percent gains in earnings each year and that employment grows by 0.42 percent per year with declining effects after seven years. Table 21 displays five-year estimates of the impact of adopting an RTW law in Illinois. The fall in real hourly wages reduces the average hourly wage to $21.63 and $21.31 and grows the employment base by 10,517 and 12,489 workers in the first year. While RTW would help lower Illinois’ unemployment rate by 0.64 percentage points by the third year of adoption, the model estimates that by the fifth year all employment gains would vanish. In total, by year 5, Illinois would suffer a loss in labor income between $35.97 billion and $39.66 billion. Accordingly, this drop in income would result in a five-year reduction of between $1.44 billion and $1.59 billion in income tax revenues to Illinois. These ranges of estimated impacts do not factor in changes in productivity or consumer demand or the possibility that some of the reduced labor income retained by the employer would be spent within the state generating some additional corporate taxes. Nonetheless, the analysis generally submits that enacting an RTW law would have negative consequences for the Illinois labor force, economy, and state budget.
Estimated Impact of Adopting an RTW Law in Illinois, Each Year Compared with Base, County-Level Model.
Source. Authors’ analysis of ES-202 Covered Employment and Wages, U.S. Department of Labor BLS (2012).
Due to “wage stickiness” and the fact that many CB agreements would remain valid and binding for a few years after Illinois became an RTW state, these estimates are most appropriate after the third year of the law’s adoption. RTW = right to work; BLS = Bureau of Labor Statistics; CB = collective bargaining.
The labor income effect is the average hourly wage (adjusted by the wage effect) multiplied by the average usual weekly hours worked by Illinois residents (33.86 hours) multiplied by fifty-two weeks multiplied by the number of employed persons (adjusted by the employment effect).
The state tax effect is the change in labor income multiplied by a uniform average tax rate of 4 percent. Since 2011, the Illinois state income tax is a flat rate of 5 percent of net income. Given standard exemptions/deductions and that Manzo and Bruno (2013) found the average state income tax rate for Illinois residents was 4.7 percent, the assumed 4 percent rate provides a conservative estimate.
Conclusion
While considerable efforts have been made by legislators and political organizations to pass RTW laws in states across the country, the empirical evidence on the effect of adopting an RTW law does not support prescribing it as an economic policy tool. As noted, “right to work” is a misnomer that has little to do with the right of a person to seek and accept gainful employment. Instead, RTW really means: the right to work in a compromised union setting but to nonetheless receive the benefits of collective representation without having to contribute toward the cost of obtaining those benefits.
By every measure this study demonstrates that workers earn more in CB states. In 2012, without adjusting for any factors, the average hourly wage of a worker was higher in CB states than in RTW states. Even after controlling for other factors, fourteen out of sixteen models report a negative impact of RTW on earnings, and another produced a statistically insignificant effect. RTW laws also have strong negative effects on manufacturing and construction industry wages. Additionally, RTW laws had no positive effect on worker wages during the recession and have had an unambiguously negative effect on real wages in the postrecession years. If Illinois were to pass an RTW law, workers would experience a reduction in their hourly work income and, consequently, a reduced consumption capacity.
The effect of RTW on earnings growth is unclear. In general, RTW appears to be associated with a very small positive effect on wage growth, with the possibility that the effect diminishes over time. In total, two out of six models with controls find statistical evidence that wage growth due to RTW diminishes over time, three report suggestive evidence that it shrinks, and one reports no effect or decline on wage growth at all. Thus, if Illinois were to pass an RTW law, workers may for a brief time experience very minimal wage growth, but the increase would never be enough to compensate for the policy’s prominent downward effect on wages. Furthermore, any income gain would be far less than what would have been earned if the workers had continued to be employed in a CB state. In other words, RTW would have the effect of diminishing the benefits of CB, thereby lowering wages for all workers, union or nonunion. Simply put, RTW is a recipe for lower lifetime earnings.
In terms of real hourly wages, unions have fared far better at raising and protecting worker earnings than RTW laws. Union membership has an unambiguously positive impact on a worker’s wages. While RTW limits a union’s ability to improve the wages of its members, the union wage effect is still quite robust in RTW states. Additionally, union membership has contributed to a small increase in wages above the rest of the employed workforce since the recession officially ended. If Illinois were to enact RTW, it would strip workers of one device (i.e., CB) that has proven to boost incomes since the 2007-2009 recession.
While the current unemployment rate is slightly higher in CB states, much of this could be attributed to the fact that more people participate in the labor force in those states. While RTW states as a whole do have a lower unemployment rate compared with CB states as a whole, they also have a lower aggregated employment rate. Digging deeper, RTW’s impact on employment is mixed. In terms of the probability of having a job, the effect of RTW ranges from a small decrease to a small increase. For the private sector, RTW laws are associated with a 0.4 percent increase in private-sector employment each year, but the effect diminishes over time. The law’s ultimate effect on Illinois’ employment may at best be slightly positive but temporary, and at worst negative. When isolated and measured against other job creation variables, RTW has no proven record of stimulating meaningful economic growth. By comparison, Illinois would reap far greater economic benefits by increasing its investment in education and skilled-based training.
RTW has an obvious negative impact on union membership. While RTW contributes to the exorbitantly rising wage inequality in America, union membership is found to reduce wage inequality in RTW states. Additionally, for female, African-American, and Latino/a workers, RTW tends to have negative impacts on real hourly wages. RTW in Illinois would not only reduce union membership but also remove structures that reduce racial and gender inequality.
Finally, RTW would substantially reduce labor income in Illinois, resulting in a dramatic decline in state income tax revenues. As Illinois struggles with difficult choices about how to pay for investments in hard (e.g., bridges) and soft (e.g., education) infrastructure, RTW would leave both the state’s current citizens more vulnerable and future constituents more likely to inherit a diminished capacity to prosper.
RTW laws have overall negative impacts for American workers. RTW laws should also not be viewed as buffers, which improved labor market outcomes during the Great Recession. Indeed, the conclusive negative consequences of RTW laws (lower earnings, lower union membership rates, negative effects on both female and African-American workers, and negative to neutral impacts on Latino/a worker wages) outweigh the inconclusive benefits. RTW does not appear to be an effective policy mechanism to advance a state’s economic recovery. By and large, as a policy prescription, RTW would generate harmful effects to Illinois’ economy, lower its capacity to provide essential public services, and degrade the quality and condition of the state’s labor force.
Footnotes
Acknowledgements
The authors thank the anonymous reviewers and special editor Victor Devinatz for comments and suggestions, which greatly helped to improve this article.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.