Right to Work or Right to Vote? Labor Policy and American Democracy

Abstract

There is growing attention to the role of organized labor in maintaining and expanding democratic institutions in the United States. In this article, we investigate the effect of right-to-work laws on electoral democracy in the states. We theorize a series of mechanisms by which labor unions contribute to the maintenance and expansion of democratic institutions, including contributing money to campaigns and influencing the electorate. Right-to-work laws, by limiting labor unions’ ability to raise funds, reduce the strength of these mechanisms and send signals to political elites about the organizational balance of power in their states. Using recent advances in difference-in-differences analysis, we find that right-to-work laws had a substantial negative effect on state-level electoral democracy in recent decades, even net of Republican control of government. Although the difficulty of causal identification in this context warrants caution, the findings speak to the importance of organized labor in shaping democratic institutions.

In the last two decades, Republican state legislatures in many current battleground states have sought to pass—and frequently succeeded in passing—right-to-work (RTW) legislation seeking to curb labor union power. RTW laws prohibit unions from collecting dues from non-unionized workers in otherwise unionized workplaces. Denying unions the ability to collect these dues is economically and institutionally important for unions because they rely on the collective action and bargaining leverage of large groups of workers. RTW laws increase the incentives for free riding by employees who, through withholding dues while benefiting from union workplace gains, weaken the scope of collective action and limit unions’ financial viability (Olson 1965). These laws are also of significant political consequence; among other things, they can deprive unions—many of which have been an important coalitional ally of the Democratic Party—of the funds to support candidates, mobilize voters, influence elections, and otherwise exert political influence (Fiegenbaum, Hertel-Fernandez, and Williamson, 2018).

Democrats, Republicans, unions, and businesses are all aware of this reality and have fought for their respective positions accordingly. RTW laws have long been a staple policy mandate for conservatives looking to gain a political advantage in state politics, and many of the contests over RTW legislation in the 2010s were in battleground states such as Wisconsin, Michigan, Ohio, and Missouri, with voters in the latter two states ultimately approving ballot initiatives that overturned RTW legislation. In 2022, Tennessee followed Alabama in entrenching RTW in its state constitution, while in 2023, the Michigan legislature—having switched to Democratic Party control—overturned its RTW law. The issue remains alive within numerous states, and efforts to expand RTW laws have been aided further by the Supreme Court’s decision in Janus v. AFSCME (2018) that effectively extended right-to-work provisions to all public employees, whether federal, state, or local.

There is extensive scholarly examination of the impact that RTW laws have for unions, workers, and the economy, particularly within the disciplines of economics and sociology, which have debated the impact that these laws have for unions, for wages, for rates of free riding, and for inequality (citations follow this section). Our focus here is on the political consequences that RTW laws have on the nation’s democratic vitality, particularly at a time when democratic procedures are being threatened by polarized politics, voting rights retrenchment, and threats of electoral subversion (Mettler et al. 2022). We examine the impact that RTW laws have for the functioning and health of democratic electoral institutions in states that have implemented these laws. Although RTW laws have arguably played an important role in the economic and political landscape since the 1940s, we limit our analysis to recent decades during which time democratic institutions have polarized across the states—with some states lowering the costs of voting and implementing more partisan-balanced legislative district maps, and other states restricting voter registration and ballot access and drawing gerrymandered district maps designed to thwart state-level electoral competition (Grumbach 2022).

Using recent advances in difference-in-differences methods, we estimate the effect of RTW laws on a state-level measure of electoral democracy. The results suggest that the effect of RTW is substantial: states that adopt RTW laws reduce their electoral democracy scores by about 0.5 standard deviations relative to states that do not. In supplemental analyses, we show that some of this effect occurs through union membership decline. However, only Republican state governments have passed RTW laws in recent decades, and studies have found that Republican control of state government itself reduces state electoral democracy (Grumbach 2022). The relationship between RTW laws and democracy could be confounded as a result. Nonetheless, we find a similarly large effect of RTW laws even holding constant party control of state government.

Drawing on a large secondary literature, we suggest that RTW laws reduce unions’ capacity to influence worker partisanship, racial attitudes, perceptions of economic security, and access to political information that better enable workers to monitor and hold elected officials accountable; that RTW laws also reduce unions’ ability to increase workers’ expression of those preferences through electoral and non-electoral participation; and, finally, that RTW laws reduce labor unions’ capacity to engage in campaign spending and elite lobbying in “organizational combat” with employers and other political opponents which, at times, have supported policies that weaken democratic institutions. Because electoral institutions are mostly governed by state legislatures, this argument supports our finding that RTW laws reduce electoral democracy at the state level.

We do want to emphasize certain limits to our findings. RTW laws are part of complex political feedback processes (Hertel-Fernandez 2018), and the laws are likely to be both a cause of state-level democratic backsliding, as well a consequence of the ascendance of coalitions in state politics motivated to entrench their political power by restricting access to others. With over two dozen states having passed RTW laws over the course of more than eight decades, we should also be cautious about lumping the impact of all RTW states. The current wave of RTW laws has important similarities with past waves, but also important differences; previous eras occurred in largely southern agricultural states with low union density whereas the recent laws have occurred in midwestern industrial states with historically strong unions. Laws governing campaign finance were different, partisan politics—by current measures—was less polarized, and important federal laws, such as the Voting Rights Act of 1965, were not in place. Moreover, not all RTW laws are identical, with some state statutes including qualifications fought for by influential, if not legislatively victorious, union lobbying.

In recent years, there has been increasing interest among political scientists in both labor politics (e.g., Ahlquist 2017) and the nation’s democratic vitality (e.g., Levitsky and Ziblatt 2023; Mettler et al. 2022), but rarely have these two topics been examined together, especially in the quantitative literature. This absence is further notable given the wealth of studies in comparative politics that center labor movements with democratization (e.g., Collier 1999; Rueschemeyer, Stephens, and Stephens 1992; Ziblatt 2006), as well as an abundance of scholarship in American political development that maintains the importance of the labor movement in the United States in establishing and expanding critical democratic laws and institutions (e.g., Frymer 2008; Orren 1995; Schickler 2016). The results of our study affirm these literatures and support long standing advocacy among political theorists for the necessity of a democratic workplace to the health of a democratic society (e.g., Anderson 2017; Dahl 1985; Pateman 1970).

Labor and American Democracy

Recent research on both domestic and international threats to democratic stability has typically given only scant attention to the importance of labor unions. But labor unions are significant political institutions, importantly involved in local and national politics with clear implications for democracy. Internationally, there is an extensive literature documenting the influence of labor movements and working-class organizations on state and democratic formation, whether in Europe (Rueschemeyer. Stephens, and Stephens 1992; Boix 2003), Latin America (Collier 1999), or Asia (Chu 1998; Neureiter 2013).

In the United States, by contrast, the labor movement rose after the establishment of formal democratic institutions in the Constitution and early expansions of voting rights. However, labor played an important—and often underemphasized—role in later expansions to American democracy, particularly in the twentieth century. Labor unions lobbied for an expansive welfare state, the establishment of Progressive and New Deal-era administrative agencies focused on worker rights and labor protections, and political reforms to fight corruption and to ensure easier access to the ballot box. By the middle of the twentieth century, the American Federation of Labor (AFL) and Congress of Industrial Organizations (CIO), both separately and together, had become prominent supporters and financial benefactors of leading civil rights, social welfare, and voting rights legislation.

This was not without important complications and caveats. W.E.B. Du Bois (1935) famously chastised white workers for not acting in solidarity with Black co-workers in the mass reaction to Reconstruction. The AFL’s rise coincided with their quite active support for anti-Chinese labor laws in the late nineteenth century, and both the AFL and CIO would struggle with racial and gender discrimination within their union ranks and locals, as well as having sometimes tenuous and fractured relationships with civil-rights and immigrant-rights movements well into the last decades of the twentieth century (Frymer 2008; Ngai 2004).

But unions were not merely critical voters to the Democratic Party’s electoral coalition, they were institutionally embedded in organizational legislative and campaign drives (Greenstone 1969; Schlozman 2016). Furthermore, many unions, and particularly the CIO’s attention to a broader class of workers that had previously been excluded by the AFL, saw their role as incorporating workers and broader populations who were left out of conventional politics (Fine and Tichenor 2009; Goldfield 2020). The CIO’s mobilizing efforts had broader feedback effects as well, such as helping spur a realignment among Northern liberals towards greater support for civil rights (Schickler 2016). With both the CIO’s rise and the migration of millions of Black workers to Northern industrial states with more heavily unionized workforces, civil rights groups became more invested in the labor movement, and vice versa (Korstad and Lichtenstein 1988). By the early 1960s, the AFL–CIO, led by progressive leaders such as Walter Reuther of the United Auto Workers, had become important financial backers of the Civil Rights Movement, participating in and helping fund the March on Washington and sending funds to bail Martin Luther King, Jr. and other civil rights activists out of jail in Birmingham, Alabama (Frymer 2008; Lee 2008). National unions publicly endorsed and mobilized around the Civil Rights Act of 1964 and Voting Rights Act of 1965 as well as the unsuccessful Equal Rights Amendment of the 1970s. And, in 2000, the AFL–CIO reversed its longstanding opposition to immigration, calling for undocumented immigrants to be granted citizenship. Today, the union movement has become the nation’s largest mass-membership organization of people of color (Bronfenbrenner and Warren 2007).

Unions are, in an important sense, micro-democracies, a place where workers vote directly for representatives who help govern their workplace, negotiating through collective bargaining agreements everything from hours and wages and retirement benefits, to political and legal rights while on the job, to standards of care and community (Dahl 1985, Frymer 2005). Like all democracies, these micro-union democracies can and have had their shortcomings, including discriminatory behavior, corruption, and stagnant leadership. At the same time, unions have frequently been effective at promoting collectivity, solidarity, and equality, all of which provide ideological and instrumental reasons for unions to support democracy both as a principle and a substantive reality in the United States. We return to the specific mechanisms by which labor unions might maintain or expand democracy in the theory section. First, however, we provide background on right-to-work laws, which affect labor union strength in the United States.

Right-to-Work Laws and Labor Union Strength

Right-to-work laws are among the most consequential labor policies in the United States. They largely derive from the Taft-Hartley Act of 1947, passed with a Republican majority aided by Southern Democratic support in Congress over a veto by President Harry Truman. The legislation, adopted amidst a postwar moment of unprecedented strikes and labor stoppages across the country, was designed to scale back certain union rights established in the National Labor Relations Act (NLRA) of 1935. That law had provided workers the right to organize and bargain collectively over wages, hours, and conditions of employment, and prohibited an array of “unfair labor practices” frequently used by employers to defeat union drives. The NLRA also enabled collective bargaining agreements to require all workers covered by the contract to pay union dues.

Taft-Hartley pushed back on many of these rights. Among other things, it expanded unfair labor practices to include more activities by unions, prohibited unions from participating in secondary boycotts, made it easier for employers to hire non-union workers in union settings, and required that “union shops”—in which new employees are required to join the union immediately after being hired—be permitted only where a majority of workers voted for it and state law allowed it. Most importantly for our purposes, Section 14b of Taft-Hartley allowed states to modify federal policy on union security clauses in collective bargaining agreements, so long as the modification further limited such agreements. This enabled states to enact so-called “right-to-work” laws that prohibited unions from collecting otherwise mandatory fees from nonconsenting workers in otherwise unionized workspaces.

There are currently 26 right-to-work (RTW) states. Until the 2000s, most RTW laws were passed in Southern, Southwest, and Great Plains states. Four states—Arizona, Arkansas, Florida, and Nebraska—established RTW laws in the years just before Taft-Hartley, while fourteen more states enacted such laws in the decade after the Act’s passage. Conservative and business lobbyists, notably the National Right to Work Committee and business leaders like Fred Koch (the father of the prominent conservative donors, the Koch Brothers), mobilized behind these laws with the intent of scaling back economic regulations and weakening both unions and their allies in the Democratic Party (Dixon 2007; Lee 2014; Newman and Skocpol 2023). But by and large, the states that passed RTW during this time already had low rates of union membership. During the 2000s, a new wave of state legislatures passed RTW laws, this time including Midwest swing states such as Michigan and Wisconsin that had higher rates of union membership—and more competitive electorates. Two other midwestern states, Ohio and Missouri, passed laws that were later overturned by voter referendums.

The actual impact of RTW laws on union strength has been greatly debated. In theory, RTW laws create a serious collective action problem for unions by allowing free riders to avoid paying dues while at the same time benefiting from a non-excludable good. As Mancur Olson (1965) has contended, a rational employee motivated solely by economic self-interest will “free ride” by withholding union dues if they can do so without incurring countervailing costs, even if they benefit from and support the union, because paying dues is unnecessary to receive the benefits. Accordingly, RTW laws ought to lower union membership and reduce an existing union’s ability to collect money from workers, and this should in turn curtail the union’s ability to spend money on behalf of broader workplace issues in public policy and politics (Olson 1965, 2). Indeed, the Janus decision, which effectively extended RTW principles to public employees, was vociferously opposed by unions and enthusiastically supported by businesses on exactly these grounds (Brief of Economists and Law and Economics 2018; Fisk and Malin 2019).

Descriptively, unionization rates in RTW states are about three times lower than in non-RTW states (Fortin, Lemieux, and Lloyd 2022). But because RTW states are meaningfully different from non-RTW states in many respects, determining the causal impact of these laws on unionization has often proved challenging. The nation’s first RTW laws were passed in states that already had weak labor movements and were more reliant on agriculture and newly developing Sunbelt economies rather than industrial sectors (see, e.g., Farber 1984; Jones 2000); many of these early RTW states also maintained undemocratic Jim Crow regimes. By contrast, except for a brief period of a limited RTW law in Indiana in the 1950s, no state with large industries and historically strong labor movements had established such laws until the twenty-first century. Moreover, labor union membership post-Taft Hartley has declined nationally, in RTW states and in most non-RTW states alike, suggesting that RTW laws are not the only reason for labor’s struggle to maintain membership. Several other important factors contribute to declining union membership, including Supreme Court interpretations of national labor policy, conservative appointments to the NLRB, the increased aggressiveness of employers to commit unfair labor practices in union campaigns, and trade agreements like NAFTA that accelerate broader developments and pressures of an increasingly globalized economy (Bronfenbrenner 1994; Stiglitz 2017; Choi et al 2021).

At the same time, numerous studies have found RTW laws to have a negative effect on union membership and union activities and are not simply a preexisting product of anti-union attitudes among workers or anti-union politics in the states (e.g., Moore and Newman 1985; Ellwood and Fine 1987; Feigenbaum, Hertel-Fernandez, and Williamson 2018; Hogler, Shulman, and Weiler 2004; Chun 2023; VanHeuvelen 2023; Zullo 2008). Difference-in-differences analyses of the recent wave of RTW laws implemented in the 2010s consistently show robust and significant negative effects of RTW on union membership (Fortin, Lemieux, and Lloyd 2022; Murphy 2023; see too Eren and Ozbeklik (2016) who find a decline in private union membership in Oklahoma after the passage of its RTW law, and Brief of Economists and Law and Economics (2018) who point to the success of decertification drives in Iowa driven by RTW laws). ¹ Fortin, Lemieux, and Lloyd (2022) estimate that RTW laws in recent decades reduced union membership by 1.9 to 3.4 percentage points relative to states that did not implement RTW (refer to figure A1 in the online appendix for event study estimates). Within highly unionized industries, RTW reduced union membership by nearly 5 percentage points.

There is also a small but important literature on the political impacts of RTW laws. Although certain state legislative efforts to pass RTW laws have had notable bipartisan support (most recently Oklahoma in the early 2000s), RTW laws have by and large been passed by opponents of unionization with the intent of hurting not only their capacity for collective bargaining, but also their political power (e.g., Farhang and Katznelson 2005; Lee 2014; Phillips-Fein 2011)—a case of designing policy to weaken political opponents by using “policy feedback as a political weapon” (Hertel-Fernandez 2018, 364). This literature provides a helpful starting point for exploring the effect of RTW laws—not just for the political balance of power between labor and anti-labor coalitions, but for democracy overall.

How Right-to-Work Laws Can Impact Democracy in the States

In the previous sections, we discussed long-term relationships between organized labor and democratic reform and described how RTW laws reduce the strength of labor unions. In this section, we outline mechanisms by which RTW laws, by reducing the strength of labor, could reduce the quality of electoral democracy in the American states. State governments retain authority over many of the procedural mechanisms involved in running elections and representing populations, including legislative districting, policies surrounding voter registration, casting ballots, and election certification. This state-level authority over the election process has only increased in the aftermath of the Supreme Court’s decision in Shelby County in which a critical component of federal oversight via the Voting Rights Act was disabled. RTW laws also vary at the state level, making the effect of RTW laws on electoral democracy in the states a relatively tractable empirical question.

While a primary function of labor unions is to organize around workplace issues, many unions conduct a wider range of political activities that plausibly impact democratic outcomes. These include collecting campaign contributions, organizing and supporting political protests, endorsing candidates and policies, and mobilizing people on election day, both union and non-union, to vote (Ahlquist 2017). We theorize two sets of mechanisms by which RTW laws might reduce electoral democracy in the states: those centered on workers, and those centered on elites. Some of these mechanisms are direct, such as unions’ electoral mobilization and campaign contributions; others are indirect, such as signals that policy victories over labor send to political elites about the balance of power in state politics. All these mechanisms share a common root, which is that RTW laws can serve to demobilize unions by depriving them of financial resources and weakening their membership base that they rely on to mobilize around democratic activities.

Importantly, through these mechanisms, weakening labor has the potential to weaken democratic institutions both by actively contributing to democratic backsliding and by preempting expansions to democratic institutions. The initiation of authoritarian Jim Crow regimes after Reconstruction is a story of backsliding, and over subsequent decades, southern states diverged from northern states that were actively expanding access to voting. Similarly (but not to a comparable degree as Jim Crow), in recent decades some non-RTW states expanded access to voting through policies such as same-day registration (Grumbach and Hill 2022) and creating independent commissions to draw balanced legislative district maps. We argue that RTW not only contributed to the ability of state governments to actively weaken democratic institutions through policy and administrative action, but also staved off threats of active expansions to electoral democracy.

Worker-Based Mechanisms

By weakening union strength, RTW laws reduce unions’ capacity to provide information and political socialization to American workers which in turn encourages and maintains civic norms through well-constructed social and political networks (Ahlquist and Levi 2013; Iversen and Soskice 2015; Macdonald 2021; Mosimann and Pontusson 2017). Unions directly provide information to their members, not only about collective bargaining but about politics, such as the policy positions of electoral candidates and evidence about how policies affect workers’ lives (Stevens and Greer 2005; Western and Rosenfeld 2011). Unions disseminate this political information via regular emails, newsletters, flyers, and meetings of workers, as well as election-specific communications such as issue briefs, candidate endorsements, and get-out-the-vote information (Macdonald 2019). Unions and union members have historically been more likely than other organized interest groups to discuss public policy issues (Schlozman, Verba, and Brady 2013). Through these efforts, union membership shapes individuals’ political knowledge (Macdonald 2019), policy preferences (Ahlquist et al. 2014; Ahlquist and Levi 2013; Kim and Margalit 2017; Mosimann and Pontusson 2016), and voting behavior (Francia and Orr 2014; Leighley and Nagler 2007).

This political socialization through union membership can have profound impacts not just on preferences, but on how workers view their own group identities and relationships to other identity groups. Unions provide a shared space for individuals to socialize and, often through informal conversation, build a politicized “working class” group identity (Macdonald 2019; Newman and Skocpol 2023). Recent research has found that individuals who belong to unions are more likely to identify as working class, all else equal, and that this identification predicts support for redistribution and the welfare state (Franko and Witko 2023).

While working class identity affects economic policy preferences, it is more important for this study that the development of working-class identity may also reduce the salience of other countervailing identities, or increase solidarity toward workers with different racial, gender, religious, or national identities. Surveys of the American public in recent years have consistently shown that “ethnic antagonism” is central to antidemocratic attitudes in the mass public (e.g., Bartels 2020). U.S. labor unions in recent years have explicitly linked considerations of racial justice and democracy in platforms and endorsements in recent years. A 2022 AFL–CIO resolution titled “The Urgent Fight for Racial Justice” endorsed congressional legislation to expand democratic institutions and the “revers[al] [of] state voter suppression laws passed in the wake of attempts to undermine the legitimacy of the 2020 presidential election.” ² Recent work has found that union membership decreased racial resentment among white workers (Frymer and Grumbach 2021), and research in comparative politics has found that unions, by encouraging solidarity between workers, can mute the impact of far-right organizations on worker preferences (Arndt and Rennwald 2016; Mosimann, Rennwald, and Zimmerman 2019; Rennwald and Pontusson 2021). Thus, it is plausible that union membership could increase support for a more inclusive democracy in the mass public by reducing the racial and ethnic conflict that often forms the foundation for calls to reduce the size of the electorate.

Unions, as central organizational members of the Democratic Party’s extended network (Bucci and Reuning 2021), increase voting for Democrats and decrease voting for Republicans. This mechanism of increasing Democratic and reducing Republican voting is partly direct, with most unions endorsing and contributing to Democratic candidates. It is also partly indirect, as union membership increases workers’ support for redistribution (Macdonald 2019). This influence on economic attitudes, in turn, should decrease voting for the party relatively more opposed to redistribution, the Republican Party.

Most specifically, Feigenbaum, Hertel-Fernandez, and Williamson (2018) find that RTW laws significantly decreased Democratic Party vote shares, and increased Republican control of state governments, in recent decades. The authors point to a number of mechanisms that likely undergird this overall effect: RTW reduced voter turnout; RTW reduced union voter mobilization capacity in RTW states such that Democratic voters were less likely to be contacted to vote; and RTW reduced the share of campaign spending by labor unions relative to control states. This finding suggests that RTW laws might reduce democracy through the mechanism of increasing Republican control of state governments—and by reducing the challenges Republicans might face in attempts to weaken democratic institutions.

So far, the worker-centric mechanisms we have discussed are direct and informational, typically involving union leaders, organizers, and rank-and-file members sharing political information and appeals. We also argue that there are indirect effects of unions on democratic attitudes through an economic security mechanism. RTW laws hurt unions, which reduces workers’ job security and the reach of the union wage premium. Broadly, historical and comparative scholars point to economic decline and precarity—and particularly the scapegoating of minority groups for economic problems—as central to the rise of authoritarianism in the twentieth century (Payne 1996). More specifically, recent behavioral research on the United States shows that job losses increased white workers’ support for the candidacy of Donald Trump (Baccini and Weymouth 2021), as well as for Trump-associated policies such as mass deportations of undocumented immigrants (Hopkins, Margalit, and Solodoch 2023).

Union membership affects workers’ attitudes—and it also affects the frequency and intensity with which workers express those attitudes through political participation. This is accomplished by both influencing workers’ beliefs about the efficacy and social benefits of participation, as well as through subsidies to participate such as transportation to protest actions or information about how to register to vote. Social scientists have long considered labor unions to be important forces in mobilizing voter turnout (Delaney, Masters, and Schwochau 1988; Rosenfeld 2010; Zullo 2008). Leighley and Nagler (2007) find that the decline of labor unions in the late twentieth century reduced voter turnout of workers and upwardly skewed the class distribution of the U.S. electorate. There is also evidence that unions increase voter turnout especially among workers of color (Kim 2016).

Non-electoral participation is also central to union strategy. Unions are a leading source of mobilization for public protests, not simply and formally worker strikes, but broader public boycotts, rallies, and petition drives. The labor movement’s ability to send large numbers of bodies to public events should not be underestimated; unions have the pre-existing organization to mobilize quickly and impactfully in ways that are unique among organizations in American politics. They are arguably equally prominent as a mobilizer and backbone for many adjacent movements, ³ from providing financial and organizational support for numerous civil rights groups in the 1950s and 60s to the anti-Iraq war movement in the 2000s to the recent mobilization around voting rights in conjunction with civil rights organizations. ⁴

In summary, RTW laws weaken democratic institutions by reducing unions’ capacity to provide workers with information and political socialization, which affects their economic attitudes, racial attitudes, partisanship, and participation. However, we note that existing research suggests that RTW only reduces a state’s union membership by a few percentage points. While decreasing union membership by approximately 3 percentage points is meaningful, it is not clear that a decline in union membership of this size would have a substantial downstream effect on democratic institutions. If RTW affects democratic institutions, it is likely due to additional mechanisms beyond RTW’s direct effect on union membership, such as unions’ political spending and signals to elites about the balance of organizational power in a state’s political environment.

Elite Mechanisms

Thus far, we have explored ways in which unions influence workers in ways that might bolster democracy. But unions also promote democracy through elite-centered mechanisms, such as by financially supporting pro-democracy candidates and elected officials and by lobbying for public-interest policies. By making it harder to recruit and retain members, RTW laws limit unions’ ability to engage in these activities at the same scale and with the same effectiveness.

Political donations and lobbying are not commonly considered democracy-promoting activities. They tend to be dominated by an “unheavenly chorus” of highly informed and well-resourced individuals and interest groups that are distinctly unrepresentative of the broader American public (Schattschneider 1960). This presents a challenge for effective democratic governance, which is premised on the “equal consideration of the interests and preferences of all citizens” (Schlozman, Verba, and Brady 2013, 96). Unions have helped rectify this imbalance—if only somewhat—by repeatedly serving as a primary organizing force for the diffuse interests of American workers, counterbalancing the immense political influence of concentrated economic interests (e.g., Hacker and Pierson 2010), and by increasing the descriptive representation of the working class in elected office (Sojourner 2013). Overwhelmingly, labor contributions in the United States support candidates from the Democratic Party; in the 2016 general election, for instance, nearly 90% of the labor sector’s $217 million in contributions supported Democratic candidates (Muller 2018). But, as Feigenbaum, Hertel-Fernandez, and Williamson (2018) find, RTW laws can serve to significantly decrease campaign spending by unions in affected areas. By reducing the financial resources at unions’ disposal, RTW laws have the potential to limit the ability of unions to serve this critical organizing role on behalf of low- and middle-class workers. Even if unions did not serve the long-term interests of democracy by representing the interests of the (vastly underrepresented) working class, this skewing of contributions toward the relatively more pro-democracy major party in contemporary U.S. politics would support the notion that present-day union contributions contribute to maintaining democracy.

RTW laws also signal to the political elite that unions are losing power. This may lead their legislative and electoral allies to assign less weight to union leaders’ lobbying efforts or feel that supporting labor’s policy agenda is less conducive to their own electoral fortunes. More significantly, as we describe with case study evidence in the next section, RTW laws likely embolden and energize those actors seeking to entrench their own interests in part by scaling back democratic access to the voting booth because they perceive the marginalizing power of their once most powerful opposition.

This precise dynamic has been on display in recent decades. As business organizations and their coalition partners advocated for unpopular policies that were far to the right of constituency opinion in the states (Hertel-Fernandez 2019), RTW laws repeatedly limited unions’ ability to act as a countervailing force to challenge the political power of business organizations. More broadly, as the economic policy agenda of conservative organizations and the Republican Party grew increasingly unpopular, their incentive to weaken countervailing organizations like unions—and democratic institutions, themselves—increased (Hacker and Pierson 2020; for this general theory in the context of comparative democratization, see Ziblatt 2017). In the next section, we illustrate these mechanisms through qualitative analysis of RTW states, with a brief case study of Wisconsin.

RTW Laws on the Ground

The 2010s saw both a wave of anti-labor laws and a wave of democratic backsliding in the U.S. states through partisan gerrymandering and policies that increased the cost of voting. Were these two transformations simply the result of a common cause, conservative ascendence in the states? Or do anti-labor laws like RTW also have an independent effect on the possibilities for democratic backlash through the mechanisms that we outlined in the previous sections? Evidence from the states during this period suggests that political coalitions understood policies like RTW as affecting the balance of power in their states—and as allowing the Republican Party to take more extreme positions than they otherwise would have without fear of electoral or organizational sanction. Indeed, the Republican Party over this period moved farther away from the center in terms of their legislative voting (Shor and McCarty 2022) and the public policies that they passed (Caughey and Warshaw 2022). The GOP’s stance on state-level democratic institutions also moved farther away from the center (Grumbach 2023). To accomplish this, elite Republicans and allies pushed to reduce labor power, especially in critical midwestern states where unions provide critical operational support to the maintenance of Democratic Party majorities (Hertel-Fernandez 2018).

The 2010 midterm elections resulted in significant gains for Republicans, enabling them unified governmental control of eleven new states, including key presidential battleground states like Michigan, Ohio, Pennsylvania, and Wisconsin. Republican state legislatures, some newly in the majority, moved actively in their political agenda, and many included an effort to retrench worker rights, both those in unions and those without (Lafer 2013). Numerous state legislatures quickly passed laws narrowing public sector collective bargaining rights, and three passed outright RTW laws (although Ohio’s legislative act was subsequently overturned by public referenda). After implementing RTW, some states passed legislation that increased the cost of voting for citizens or drew gerrymandered district maps. However, most RTW states made voter registration and casting a ballot more inaccessible through executive and administrative action. Importantly, both explicit legislation and observed democratic institutional outcomes, such as average wait times for in-person voting or the percent of adult citizens who are registered to vote, are contained in the State Democracy Index measure we use.

After RTW, Indiana, for example, moved to reduce early voting sites, especially in Marion County, which contains the more Democratic-leaning city of Indianapolis. In the mid-2010s, the state also purged 28.7% of voter registrations. ⁵ Indiana later moved the deadline for submitting absentee ballot applications to 12 days from 8, restricted efforts to extend polling place hours, and began to require citizens to provide identification when requesting an absentee ballot. In Oklahoma, the percent of eligible citizens registered to vote declined by 5 percentage points after RTW.

In other RTW states, such as Michigan, the implications for democratic institutions stemmed from blocking or delaying reforms that would increase voter turnout and blocking challenges to its gerrymandered state legislative and U.S. House district maps. After the passage of Michigan’s RTW law, the policy’s champions convened to celebrate the negative effect that the law would have on labor unions’ ability to run get-out-the-vote initiatives (Hertel-Fernandez 2019, 147). Other states were implementing major election reforms during this period, such as same day voter registration policies that are understood to increase turnout, especially among young people (Grumbach and Hill 2022). Michigan’s architects of RTW believed that, by reducing unions’ ability to turn out voters, they would preempt ballot initiatives to make voting more accessible. Ultimately, these coalition partners asserted that RTW had won the state of Michigan for Donald Trump in 2016 (Hertel-Fernandez 2019, 196).

Due to a greater amount of journalistic attention, the mechanisms behind the relationship between RTW and democracy are especially clear in Wisconsin. Wisconsin Governor Scott Walker signed Act 10 in 2011, a “budget repair bill” that largely eliminated collective bargaining rights for the state’s 175,000 public employees. Wisconsin had been the first state in the nation to provide collective bargaining for public employees, notably empowering those workers in state politics for a half century. As in some other states, Republican legislators exempted police and firefighters to avoid criticism from more conservative constituencies. In the following months, buoyed by their success with Act 10, the Republican legislature passed both a new law mandating voter ID in primary and local elections as well as a quite aggressive partisan gerrymander, further entrenching their position. In 2015, Republicans passed a broader private sector RTW law prohibiting employers and unions from mandating dues from non-unionized workers, and shortly after, passed yet another voter ID bill (Kaufman 2018). As in other states, the changes to democratic institutions can also be seen in administrative behavior; in the election cycles following RTW, Wisconsin went from nineteenth among states in wait times for in-person voting to thirty-first.

The rhetoric of political leaders in Wisconsin on the potential for policy feedback from anti-labor laws was clear. Wisconsin Senate Majority Leader Scott Fitzgerald explained that “if we win this battle, and the money is not there under the auspices of the unions, certainly what you’re going to find is President Obama is going to have a … much more difficult time getting elected and winning the state of Wisconsin” (quoted in Wis. Educ. Ass’n Council v. Walker, 824 F. Supp. 2d 856, 876 n.17 (W.D. Wis. 2012)). In response to a question as to whether Wisconsin will “ever get to a completely red state,” Governor Scott Walker replied, “Well, we’re going to start in a couple of weeks with our budget adjustment bill [containing a RTW law]….the first step we’re going to deal with collective bargaining for all public employee unions, because you use divide and conquer” (quoted in Lafer 2013). As the CEO of the State Policy Network Tracie Sharp said on the eve of the 2016 election, the goal is to “defund and defang one of our freedom movement’s most powerful opponents, the government unions,” with the longer-term goal to “deal a major blow to the left’s ability to control government at the state and national levels. I’m talking about permanently depriving the left from access to millions of dollars in dues extracted from unwilling union members every election cycle” (Pilkington 2017, emphasis original). She added after the election, “when you chip away at one of the power sources that also does a lot of the get-out-the-vote … I think that helps for sure” (Hertel-Fernandez 2019).

With a diminished organizational counterweight from labor unions, well-resourced groups in the Republican Party’s extended network pulled the party’s candidates rightward during this period, especially on economic issues. Policies to reduce the power of labor in politics were central to these groups’ strategies. The Koch network and its flagship advocacy organization, Americans for Prosperity, were an important agenda-setter on RTW laws. As Skocpol and Hertel-Fernandez (2016, 694) found, the existence of a paid Americans for Prosperity director increased the probability that a state would restrict public sector collective bargaining by 30%. We provide descriptive analysis of the “Conservative Troika” as an organizational force in setting the agenda for RTW laws during this period. We use data from Hertel-Fernandez (2019, ch. 6), which measures the organizational strength of Americans for Prosperity, the American Legislative Executive Council (ALEC), and the State Policy Network across states as of 2011. This allows us to estimate the correlation between state Conservative Troika strength and the likelihood of implementing RTW in the future, (see figure 2). ⁶

Figure 1 Right-to-work enactment in the States

Figure 2 Conservative troika strength and future RTW implementation

In summary, RTW laws helped elect Republicans to state government and emboldened Republicans to take more extreme policy positions in recent decades—including on issues of voting rights and redistricting. RTW laws also helped preempt the emergence of election law reforms that would expand the electorate, such as automatic and same day voter registration, that were diffusing across many other states during this time period. In the next section, we describe our data and difference-in-differences strategy to estimate the effect of RTW laws on electoral democracy in the states.

Data and Estimation Strategy

We collect data on RTW laws and on state-level electoral democracy for all fifty states (Frymer, Grumbach, and Hill. 2024). We use data on the year of states’ RTW implementation from Feigenbaum, Hertel-Fernandez, and Williamson (2018). For this treatment variable, we code a state as treated beginning in the year its RTW statute or constitutional amendment is implemented.

Our measure of state-level electoral democracy comes from the State Democracy Index (SDI) from Grumbach (2022). The SDI draws on fifty-one variables related to who in a state is eligible to vote (e.g., felon disenfranchisement policies), the cost of voting for the state’s average eligible voter (e.g., no-fault absentee voting or the average wait-time to vote in person), the partisan balance of legislative districting (e.g., the partisan efficiency gap; see Stephanopoulos and Warshaw 2020), the security and integrity of state election administration (e.g., post-election audits), and the correspondence between a state’s public opinion and policy outcomes (see Caughey and Warshaw 2018). A Bayesian factor analysis model then uses those fifty-one variables to generate State Democracy Index scores for each state-year that best predict the values of those fifty-one variables.

We use a difference-in-differences design to estimate the causal effect of RTW laws on electoral democracy in the states. Difference-in-differences protects against time-invariant confounders. More substantively, this means that we control for all long-term differences—both observed and unobserved—between states in terms of characteristics such as state political culture, geography, racial and ethnic demographics, and the legacy of Jim Crow. ⁷

Importantly, in some specifications we also adjust for party control of state government. This is a major potential time-variant confounder; research suggests that Republican control of state government reduced democratic performance during this era (Grumbach 2022), so it could be the case that there is no causal relationship between RTW laws and democratic institutions. However, if difference-in-differences estimates show a relationship between RTW laws and democracy even when holding constant party control of state government, it would be strong evidence for our theory of an independent causal effect of RTW on democratic institutions.

Our main difference-in-differences specifications are event studies and estimators from Callaway and Sant’Anna (2021). Event studies interact state treatment status with leading and lagging indicators of the state’s treatment status, which allows us to estimate dynamic treatment effects of RTW laws over time as well as assess pre-trends (treatment effects before RTW laws take effect). In the following event-study equation, D represents RTW treatment status for state i in year t, α represents fixed effects for state i, and λ represents fixed effects for year t. Temporal leads and lags are represented by L.

$$ {y}_{it}={\alpha}_i+{\lambda}_t+\sum \limits_{l=-K}^{-2}{\beta}_l{D}_{it}^l+\sum \limits_{l=0}^L{\beta}_l{D}_{it}^l+{\varepsilon}_{it} $$

However, in staggered difference-in-differences contexts, two-way fixed effects models (of which traditional event studies are one variety) have recently been discovered to face potential problems in aggregating treatment effects. Specifically, treatment-effect estimates from two-way fixed effects models are a weighted average of all possible two-period difference-in-differences comparisons in the data, and if treatment effects vary over time, some of those weights can even be negative (Goodman-Bacon 2021). We therefore use alternative treatment-effect aggregation estimators from Callaway and Sant’Anna (2021), which compare treatment effects to “clean” control units and avoid the weighting problem. We use four aggregators of the difference-in-differences treatment effects, which aggregate by calendar time (“Calendar”), treatment duration (“Dynamic”), group (“Group”), and group-time (“Group-Time”) (Callaway and Sant’Anna 2021).

These models compare never-RTW states with states that adopted RTW laws in the 2000-2018 period. They exclude “always-treated” cases—states that have always had RTW laws throughout the 2000 to 2018 period—from the control group, given that RTW laws could have effects on democracy long into the future. As a robustness check, we also include two-way fixed effects models in online appendix table A1, which do include always-RTW states in the control group.

Overall, our difference-in-differences design protects against the main threats to causal inference, but we must satisfy the parallel trends assumption. Substantively, this means worrying about time-variant confounders that might be correlated with the timing of RTW implementation, or endogenous processes where declines in electoral democracy cause RTW laws rather than the other way around.

We take this threat seriously. It could be the case that RTW laws are passed and implemented by coalitions that are otherwise using their political authority to weaken democratic institutions simultaneously. In other words, we worry that RTW laws are not actually causing changes to democratic institutions in the states but are instead just a policy that tends to occur when a coalition promoting restrictions on the democratic process takes power. In such a situation, trends in democratic institutions in the states would have progressed as they did even absent the rise of RTW laws. As mentioned, however, some of our analyses control for party control of state government, which protects against a potential time-variant confounder related to coalitions.

Results

We first present the descriptive relationship between RTW laws and electoral democracy in the states. Figure 3 shows states’ electoral democracy scores from 2000 through 2018, with colors distinguishing states that never implemented RTW (light gray), implemented RTW prior to 2000 (dark gray), or implemented RTW between 2000 and 2018 (with blue representing pre-RTW years and red representing post-RTW years). While it is difficult to visually infer the effects of RTW, figure 3 shows a consistent gradual decline in democracy scores after Oklahoma implements RTW in 2001 and Wisconsin implements RTW in 2015. Kentucky and West Virginia show little change in democracy scores after their years of RTW implementation (2017 and 2016, respectively). We also see that Indiana and Michigan experience precipitous declines in their state electoral democracy scores between 2011 and their RTW implementation year, 2012. These declines in democracy stem in large part from gerrymandering. Partisan gerrymandering is important for electoral democracy, since it affects how citizens’ votes translate into legislative seats, and, ultimately, policy outcomes. States, especially under Republican control, in the 2010s drew district maps with what are likely the most extreme levels of partisan gerrymandering in U.S. history (Grumbach 2022). However, Indiana and Michigan implemented RTW and developed their legislative maps with similar timing. Thus, later on, we check whether our difference-in-differences results are driven by outliers that experience a rapid decline in electoral democracy (often due to implementing new gerrymandered district maps) around the time of implementing RTW, like Indiana and Michigan. To foreshadow the results, we find that RTW laws significantly reduce state-level democracy even when excluding the difficult cases of Indiana and Michigan. ⁸

Figure 3 Descriptive results

While visualizing descriptive state-level trends is informative, our difference-in-differences estimates better summarize the average treatment effects of RTW. Figure 4 shows event-study estimates of the effect of RTW. The figure shows a substantial negative effect of RTW on state-level electoral democracy beginning the year of RTW implementation (about -0.5 standard deviations) and remaining relatively consistent in the post-treatment period.

In contrast to the plot on the left, the plot on the right adjusts for party control of government. Recent literature points to the importance of Republican control of government as a cause of state-level democratic backsliding, with, for example, states that came newly under Republican control after the 2010 elections moving swiftly to enact gerrymandered district maps (Grumbach 2022). Considering this, the similarity in RTW treatment-effect estimates in both plots is striking. The treatment-effect estimates in the plot on the right are smaller than those on the leftmost plot, but only slightly. These results are consistent with theories that RTW laws affect state-level electoral democracy through mechanisms above and beyond RTW laws’ effect on Republican electoral victories in the states.

The event study plots also present pre-trends. If there is a clearly observed “effect” of RTW prior to implementing RTW, this could be evidence that states that go on to implement RTW are already trending differently in terms of electoral democracy from states that will not implement RTW—a violation of the parallel trends assumption. Figure 4 shows that these pre-trends are statistically insignificant. Despite their statistical insignificance, the leftmost panel (“Full RTW Effect”) shows decreasing pre-trends in the years -3 through -1 prior to implementing RTW, which could reflect endogeneity (Freyaldenhoven, Hansen, and Shapiro 2019). The rightmost panel, which adjusts for Republican control of state government, does not show the same pattern in the pre-trends. Taken together, these results suggest that states under GOP control were already trending slightly downward democratically prior to implementing RTW, and that adjusting for GOP control in the model removes this trend.

In figure 4, we present our main treatment-effect estimates using recent advances in difference-in-differences estimators. These new estimators circumvent important weighting problems in two-way fixed-effects estimators in situations where treatment timing is staggered across units (Goodman-Bacon 2021), as is the case in RTW implementation in the states. We use four treatment-effect estimators from Callaway and Sant’Anna (2021), which aggregate difference-in-differences treatment-effect estimates by calendar time (“Calendar”), by time to treatment (“Dynamic”), by treatment cohort (“Group”), and by treatment cohort and calendar time (“Group-Time”). ⁹

Figure 4 Event study estimates of right to work effect

The results shown in figure 5 are clear and consistent: RTW laws reduced state-level electoral democracy over recent decades. The treatment-effects range from -0.55 standard deviations with the Calendar estimator to -0.74 with the Dynamic estimator. For comparison, Grumbach (2022, 12) finds that unified Republican control of state government had a treatment effect of about -1 standard deviations over this time period. Taken together, these results suggest that RTW laws were a smaller but still important contributor to democratic backsliding in the U.S. states.

Figure 5 Right to work effect with alternative difference-in-differences estimators

To further investigate the mechanisms behind this effect, we estimate the relationship between state union membership and democracy scores in table 1 using a two-way fixed-effects estimator. Importantly, because union membership is causally downstream from RTW laws, we do not include both RTW laws and union membership in these models. (In addition, we use two-way fixed effects here because union membership is a continuous rather than binary variable.) The results suggest that a decline in union membership of 1 percentage point, one mediating mechanism between RTW laws and democracy, reduces state-level democracy by between 0.066 and 0.083 standard deviations (p<0.1 in all models). Based on recent literature that finds that RTW laws reduce union membership by about 3.4 percentage points, and our finding that RTW reduces democracy scores by between 0.55 and 0.74 (figure 5), we estimate that between 30% and 54% of the effect of RTW goes through the mechanism of decreased union membership

Table 1 Granger causality test

Robustness and Causality

One concern is that one or two outlier states might be driving the RTW results. In particular, states that implement RTW around the time they implement a gerrymandered district map, such as Michigan and Indiana, could bias our estimates in a negative direction. To address this concern, we run a “leave-one-out” cross-validation exercise that addresses the potential role of outliers (Broderick, Giordano, and Meager 2020). We re-run the difference-in-differences models 50 times, each time leaving out one state from the analysis. ¹⁰ We present the results in figure 6. Given the descriptive declines in democracy scores in new RTW states in Indiana and Michigan shown earlier in figure 3, we were surprised at the robustness of these leave-one-out results. Furthermore, excluding both Indiana and Michigan from an analysis yields somewhat attenuated by still highly robust difference-in-differences estimates, with coefficients ranging from -0.30 to -0.54 standard deviations on the State Democracy Index (and each coefficient remains statistically significant at the p < 0.01 level). By contrast, excluding recent RTW states like West Virginia and Wisconsin does not affect the results much at all.

Figure 6 Leave one-out analysis

An additional concern relates to reverse causality. Changes to political institutions like legislative districts or election administration often affect downstream political terrain. It might be the case that democratic backsliding causes RTW laws rather than the other way around. To investigate this possibility, we provide a Granger causality test in table 1. ¹¹ The Granger causality test looks at whether past changes in state democracy scores affect the likelihood of implementing a RTW law in the future. Table 2 shows the results for democracy scores lagged by a number of years ranging from one to four. Across the models, the effect of democracy scores on subsequent RTW law implementation is small and statistically insignificant; specifically, the coefficients in table 2 are about an order of magnitude smaller than those from the main results of the effect of RTW laws on state democracy scores. In substantive terms, this finding means that it is highly unlikely that RTW laws occur because of earlier declines in democracy.

Table 2 Relationship between union membership and state democracy

Conclusion

Scholars offer different explanations for the expanding variation in democratic institutions across states in recent decades, pointing to changes to racial and demographic threat (e.g., Bartels 2020), money in politics (Fordham 2022), and the interaction of geography and minoritarian institutions (Levitsky and Ziblatt 2023), among others. In this article, we highlight the role of organized labor, and specifically its decline, in the American political economy as an additional important force. Despite increased attention from political media, a rise in public support for unions, and a notable increase in strike activity and labor activism in the post-COVID-19 era, union membership continues to decline (U.S. Bureau of Labor Statistics 2022). We argue that this decline, in part stemming from RTW laws, adds an important dimension to understanding the current battlegrounds and polarization over democratic institutions and voting rights in the United States.

We have argued that organized labor, through a number of mechanisms—including the influence unions have over workers’ partisanship and voting, workers’ public and racial attitudes, the importance of unions to an individual’s pay scale and economic security, and the critical role that unions’ play as lobbyists and providing campaign contributions—importantly shapes the historical development and maintenance of democratic institutions. A recent wave of RTW laws that were designed by legislators to reduce labor union capacity and influence within government have further served to weaken the effectiveness of these mechanisms which, in turn, have opened opportunities for legislators hoping to constrict access to the voting booth and insulate their positions and interests from more electorally representative democratic institutions in the states.

Results from our difference-in-differences analysis are consistent with this argument. We find that RTW laws since 2000 had a negative effect on democratic institutions in the states of about one-half of a standard deviation in state-level electoral democracy. Importantly, we find that the effect of RTW laws occurs through channels outside of RTW laws’ effect on party control of state government.

Nonetheless, we believe that the quantitative results should be interpreted with some caution. The 2010s was a transformative decade in American politics, involving rapidly rising partisan polarization, technological and economic changes to the media environment, important Supreme Court rulings on campaign finance and the Voting Rights Act, and a legislative redistricting year coinciding with a Republican wave election in the states. In causal inference terms, these forces, combined with multidirectional feedbacks between labor policy and democratic institutions, open up possibilities for endogeneity and violations of the parallel trends assumption.

There remains considerable work to be done on the question of labor and democracy. On the comparative side, further research should investigate how labor policy and law, such as the function and capacity of the National Labor Relations Act in the United States compared to sectoral bargaining systems, moderate and mediate the relationship between labor and democracy. Within the study of the United States, labor politics and policy remain largely absent from studies of democratic health and decline. Among the critical questions is the impact of union decline for partisan and ideological polarization, the decline of the Democratic Party, the rise of the Republican Party, and whether the diversification of the working class and of labor union members in terms of race, gender, and immigration status affects the political influence of the working class in both national and state politics. Although it has been argued that the rise of economic inequality in the United States has influenced the rise of political polarization, political scientists have not done enough to examine the relationship between union decline and rising economic disparities (McCarty, Poole, and Rosenthal 2016; Rosenfeld 2014). We also know little about the variation in terms of unions’ effect on workers and democracy across sectors and workforces, such as the difference between public- and private-sector union membership, or the distinct politics of police and public safety unions as opposed to service and industrial economies.

But more broadly, our goal here is to join the new scholarly attention as discussed earlier to both the importance of the labor movement for the nation’s democratic vitality, particularly within quantitative social science, as well as increasing focus on the important ways that shifts in the nation’s economy in the last several decades directly impacts its politics (e.g., Hacker et al. 2021). Scholarly and journalistic inquiries into the threats to democracy, particularly those seeking to explain why a seemingly increasing number of voters support candidates promoting the scaling back of democratic rights and access to the voting booths, have debated the importance of class-related mechanisms like “economic anxiety” with that of racial resentment and xenophobia as explanations. By contrast, our study and an emerging literature have focused on the feedback between labor policy and conflict over race, citizenship, and democracy. Whereas a focus on the “economic anxiety” of the American worker serves to individualize and de-politicize the phenomena, focusing on the work of political activists, lobbyists, and legislators makes clear how central and concerted the efforts to establish Right to Work policies have been at destabilizing and reorganizing many of the foundations of politics today.