Workplace Domination and Labor Unions

Abstract

Recent decades have seen a precipitous decline in union membership, with deleterious consequences for the working class. Yet political theorists have devoted little attention to unions. In this article, I argue that unions help solve collective action problems that otherwise stymie workers’ attempts to take industrial action and compel employers to take better account of their interests. Unions thereby afford workers a form of partial protection from domination. This has three important implications. First, insofar as workers have claims against subjection to domination, they must be afforded opportunities to unionize. Second, insofar as the ability of unions to afford workers such protection depends on their being able to credibly threaten industrial action, it is essential that unions and their members have the right to take such action. Finally, workers have duties to join the union that represents them and contribute to its legitimate activities. This lends support to various forms of compulsory unionization.

Recent decades have witnessed a precipitous decline in union membership. In Australia, for instance, in 1976 union density—the percentage of all employees belonging to a union—was 51%. By 2016 this had fallen to 14% (APL 2018–19). In the United Kingdom, union density was 54% in 1980 (Disney, Gosling, and Machin 1993) compared to 23.4% in 2018 (Department for Business 2019). And in the United States, in 1964 union density was 29.3% (excluding agricultural workers; Hirsch, Macpherson, and Vroman 2001). By 2020, it had declined to 10.8% (US Bureau of Labor Statistics 2021b). Similar developments have taken place in virtually all other developed countries.

Some welcome these trends. In 2007, for instance, Australian workplace relations minister Joe Hockey declared that the role of unions in Australian economic development was “essentially over” and claimed that “Australians are choosing not to join the unions because they see them as irrelevant to their lives” (ABC 2007). Nevertheless, evidence suggests that declining union density has had deleterious consequences for workers. Unions, for instance, help ensure higher wages, particularly for lower-status workers. Declining union density, then, has contributed to the sharp increase in economic inequality witnessed in recent decades. Western and Rosenfeld (2011, 528), for instance, find that declining union density explains 20 to 30% of the growth in US private sector wage inequality since 1973. Unions have also historically played a major role in politics: mobilizing voters, lobbying in defense of their interests, and supporting candidates sympathetic to the worse-off. As such, declining union density has also undermined the political power of the working class (cf. O’Neill and White 2018; Rosenfeld 2014, chap. 7).

Against this backdrop, it is surprising that although unions have been extensively studied by social scientists, ¹ political theorists have devoted little attention to them. ² As such, the justification—if there be such— of the rights of workers to unionize, take industrial action, and so on, remains somewhat obscure. Extant accounts typically appeal to the effect of labor unions and industrial action on the general welfare (see, e.g., Mill 1848). Yet such arguments are hostage to the empirical evidence concerning the overall effects of unionization. And though I tend strongly toward a pro-union interpretation, the matter is highly contested. ³ My aim in this article, then, is to develop an alternative account. Unions, I argue, help solve collective action problems that otherwise stymie workers’ attempts to take industrial action and compel employers to take better account of their interests. In this way, unions afford workers protection from domination. Because individuals have claims against being subjected to domination, workers must be afforded substantive rights to unionize and take industrial action—even if the exercise of those rights leads to suboptimal socioeconomic outcomes.

My account has at least two further implications. First, experts generally agree that declining union density has been at least partially driven by the sustained attack on union organizing, industrial action, and collective bargaining spearheaded by conservative governments in recent decades (see, e.g., Brady 2007, 89). Corporations have also played a prominent role. The New York Times reports that Amazon, for instance, has in recent times threatened workers who were leading union drives with dismissal, established websites claiming that workers would have to skip meals to pay union dues, and even arranged for traffic signals to be modified to make it harder for union canvassers to approach workers in parking lots (Streitfeld 2021). My account explains why such developments and practices are problematic.

Second, several theorists have recently defended various forms of compulsory unionization. Reiff (2020, 11), for instance, argues that every workplace should have a union that all workers should be required to join and support as a condition of their employment. The standard objection to such arrangements is that they abridge individual liberty. ⁴ Indeed, critics often maintain that the power of unions rests on their ability to coerce workers into becoming members and participating in industrial action. F.A. Hayek (1960, chap. 18) went so far as to claim that for this reason “the whole basis of our free society is gravely threatened by the powers arrogated by the unions.” My account offers a powerful response to this charge. Protection from domination, I argue, constitutes a collective good for all workers, regardless of whether they are union members. Those who decline to join the unions that represent them thus unfairly free-ride on union members. This entails that workers have duties of fairness to unionize. Thus, liberty-based complaints of this kind have little force. Compulsory unionization merely requires employees to discharge their duties of fairness to their fellow workers, just as compulsory taxation requires individuals to discharge such duties to their fellow citizens in virtue of the collective goods produced by the state. ⁵

The structure of the article is as follows. The first section explicates the idea of workplace domination. The second section introduces the problems of collective action that attend industrial action, shows how unions help solve them, and argues that this affords workers a form of protection against domination. In the next section, I argue that the considerations advanced in the first two sections entail that workers have duties of fairness to join and participate in the activities of the unions that represent them. The final section concludes by considering a range of related issues.

Workplace Domination

Any discussion of workplace domination must begin with an account of domination more generally. The argument that follows could be phrased in terms of any reasonable view, of which there are several on offer in the literature. Because it has been by far the most influential, I adopt the conception articulated by Pettit (1997, 52–58). On this view, an agent A will subject some other agent or agents to domination where A has the power to interfere in some or all the choices of those others on an arbitrary basis. This occurs where A can exercise their power of interference in a manner such that they are “not forced to track what the interests of those others require” (55). A will, in turn, be forced to track others’ interests if and only if it is either impossible or very difficult for A to fail to take appropriate account of those others’ interests in decision-making, or A will be liable to suffer some non-trivial penalty if they fail to do so (58). Slavery, for instance, is the paradigm case of domination insofar as slaveholders (irrespective of whether they choose to do so) do not typically face any nontrivial barriers in, nor penalties for, visiting interference upon their slaves without any regard for their interests. Slaves are thereby subject to the arbitrary will of slaveholders and thus subject to domination.

The fundamental assertion of this article is that individuals have rights against subjection to domination. There are at least three reasons why this might be thought to be the case. First, where others have the power to interfere in our affairs on an arbitrary basis, they will often be liable to exercise that power for their own purposes, to our cost. Domination, then, constitutes a serious threat to our other interests. Secondly, many neo-republicans have argued at length that liberty requires the absence of domination (see esp. Pettit 1997). A slave whose master is not disposed to intervene in their choices is nevertheless unfree, given the master’s power to interfere at will. Insofar as we have claims to individual liberty, then, we have claims against subjection to domination. Finally, others have argued that domination constitutes an objectionable form of hierarchy and, as such, social inequality (e.g., Kolodny 2019). Insofar as we also have claims to be treated as equals, then, we have claims against subjection to domination.

Whichever of these accounts we favor, it is a familiar observation that certain structural features of the labor market in typical liberal democracies give rise to the domination of workers by employers (see, e.g., Anderson 2017, chap. 2; González-Ricoy 2014; Gourevitch 2016, 315–18; Hsieh 2005; Pettit 1997, 142–43; and Reiff 2020, 95–102). As Coase (1937) observes, the fundamental purpose of commercial firms is to overcome the transaction costs (e.g., negotiation time) that attend interactions between independent actors specializing in different aspects of the production process. Firms mitigate these costs by employing workers with a variety of skillsets and coordinating their activities under the authority of managers. The scope of managerial authority is defined by the terms of the contractual relationships between employers and employees (mediated by employment law). Such contracts typically afford managers very broad discretion to direct workers’ activities (see esp. Grossman and Hart 1986). After all, if it were necessary to renegotiate the terms of each worker’s employment every time changes to the production process were called for, this would reintroduce many of the transaction costs that commercial firms are supposed to avoid. Managerial authority, then, is typically very expansive.

Moreover, managerial decisions can subject workers to significant and pervasive interference. Under typical regimes of industrial law, managers can alter the tasks assigned to workers. They can affect the remuneration workers will receive and the volume of work they are required to do. They can alter the times at which employees must arrive and leave their places of work, as well as the persons with whom they will be in close association in the workplace. They can regulate the times at which workers may choose to consume meals and how they dress. In the United States, employers may subject employees to lengthy security checks (without paying them for the time taken), punish or dismiss them for their Facebook posts, and compel them to work overtime. In some industries, managers routinely regulate workers’ bathroom breaks and, in some instances, deny workers such breaks altogether. ⁶ And so on.

Managers, then, enjoy a significant power of interference over workers. This need not amount to domination, provided workers can compel managers to track their interests in the exercise of that power. However, unlike the democratic state wherein regular elections and contestation mechanisms (e.g., courts) compel politicians to track the interests of their citizens (albeit very imperfectly), very few workers have the power to elect their managers or formally contest managerial decisions. Yet the matter is not straightforward. Arneson (1993, 139; emphasis in original), for instance, argues that, with respect to their working conditions, individuals enjoy a “considerable degree of control in the form of exit rights.” This might be of relevance in two ways. First, because it is costly for businesses to replace staff, the ability of workers to threaten to resign in response to decisions injurious to their interests might compel managers to track the interests of their workforce. Alternatively, it might be argued that, in failing to resign, workers give tacit consent and thereby waive whatever claims they might otherwise have had against subjection to domination. ⁷

Yet, for exit rights to play either of these roles, workers must surely be able to exercise them without unacceptable cost. Employers are unlikely to take seriously threats on the part of workers to resign if the costs of their doing so would be prohibitive. And we should hardly be willing to say that failing to resign constitutes tacit consent if resignation would be excessively costly. By analogy, a woman subject to the whims of a violent partner plainly does not give consent to the treatment to which she is subjected by failing to exit that relationship if doing so would cut her off from her children.

Unfortunately, for the vast majority of workers, resignation is very costly indeed. ⁸ Workers are often bound by noncompete clauses, which prohibit them from taking jobs with similar firms, greatly limiting their opportunities for alternative employment. As of 2021, almost 40% of the US labor force had been subject to some such agreement, with many binding employees for longer than 12 months (Starr, Prescott, and Bishara 2021, 60–68). Moreover, the US Federal Reserve (Board of Governors 2020, 16) reports that in 2019, the median American adult had only around $5,300 in savings. In the same year, the median duration of unemployment for workers was around 9.2 weeks (US Bureau of Labor Statistics 2021a). Many workers, then, simply cannot afford to resign in response to adverse managerial decisions.

Joblessness also typically entails a range of psychological costs such as stress, anxiety, and depression, the effects of which can persist long after individuals find new employment (cf. Paul and Moser 2009). Over time, workers also often develop firm-specific human capital: skills and knowledge primarily of value to the firm in which they are employed. This, then, renders workers less competitive for roles in other firms and liable to command less remuneration (on this point see González-Ricoy 2014, 240, and Hsieh 2005, 128). Finally, as González-Ricoy (2014, 240) points out, giving up one’s job often means surrendering an important source of self-esteem and severing valuable relationships with others in one’s workplace. Taken together, these costs are very significant.

As such, I hold that very many workers are dominated by their employers. This, of course, is not absolute. Industrial regulations and the need to retain staff surely do force employers to take some account of workers’ interests. My point is simply that the scope of arbitrary managerial authority is typically very expansive, such considerations notwithstanding. Insofar as we have good reason to object to domination—whether it is partial or total—we have good reason to hold that this state of affairs is problematic.

Labor Unions and the Worker’s Dilemma

In my view, the principal value of unions is that they afford workers a form of partial protection against domination by ensuring that employers face credible threats of industrial action. We can define industrial action as any form of collective action on the part of workers that aims to pressure employers into taking better account of their interests. Industrial action can impose significant costs on employers. In late 2019, for example, a 40-day strike by 48,000 workers at General Motors cost the company an estimated $4 billion (Wayland 2019). Where employers face credible threats of industrial action, then, they are compelled to track the interests of their workers more closely in decision-making to negate such threats. Thus, the ability of workers to issue credible threats of industrial action constitutes an important form of protection against domination. There is significant evidence that the ability of workers to issue and execute such threats is highly effective in this respect. More frequent strike action has often been associated with higher wages, for instance (see, e.g., Rubin 1986).

These observations are familiar (see, e.g., Gourevitch 2016, 317). However, extant discussions of industrial action (at least among political theorists) almost entirely neglect the role of unions. That is a critical omission. In the absence of unions, employers can largely discount the possibility of industrial action. This is because industrial action presents workers with a collective action problem.

Consider, first, that in most cases the costs of participation in industrial action are significant. All forms of industrial action involve confrontation between workers and managers that most would strongly prefer to avoid for its own sake and for its potentially adverse professional consequences. There are also often material costs. Strikers, for example, forego wages they would otherwise have earned, which most workers can ill afford. More drastically, in some jurisdictions such as the United States, employers are legally permitted to hire replacements to permanently take the jobs of striking workers (see NLRB v. Mackay Radio & Telegraph Co., 304 U.S. 333 [1938]). For it to be rational for any individual worker to participate, then, the expected benefits of doing so would need to be very substantial.

The benefits of successful industrial action can be very significant. However, these are not identical to the benefits of participation in industrial action. Industrial action sometimes fails to produce benefits of any kind. And even when industrial action does succeed, the benefits typically accrue to both participants and nonparticipants. Many workplaces are governed in accordance with policies that regulate pay and conditions for all workers. Industrial action is often taken to protest the failure of employers to conform to these policies or to seek improvements to their terms. Either way, the benefits—if the action succeeds—flow to all workers, irrespective of whether they participate. The benefits of participation for each individual worker, then, will be a function of the benefits they can expect to receive from successful industrial action weighted by the marginal difference they can expect to make to the probability of that action’s success by participating.

Unfortunately, each individual worker’s participation will typically make very little difference in this respect. Consider strikes, for instance. In any large organization, any individual’s refusal to work is unlikely to make any significant difference to the costs of strike action to their employer, and thereby to management’s incentives in negotiating with their workforce. It is hardly as though any single worker’s agreeing to return to the assembly line would have made any nontrivial difference to General Motors’ incentives to address workers’ grievances in 2019, for instance. Such considerations will surely generalize to other forms of industrial action in the vast majority of instances. Thus, because most workers can do very little to affect industrial action’s prospects of success and will usually receive the benefits whether they participate or not, it will almost invariably be rational for workers to decline to participate and thereby avoid the associated costs. As such, industrial action will generally fail to occur—even in cases where the benefits for workers of successful industrial action would significantly outweigh the costs to each worker of participating. Let us call this the worker’s dilemma.

It is a well-established finding that workers in industries with higher union density are much more likely to take industrial action (see, e.g., Cornfield 1991, 35). Unions, then, help resolve the worker’s dilemma. Following Offe and Wiesenthal (1980, 78–79), I hold that the fundamental way in which unions are able to achieve this is by partially redefining workers’ interests, thereby changing the “standards according to which [the costs of engaging in industrial action] are subjectively estimated” in ways that favor collective action.

There are several ways in which unions seek to redefine workers’ interests. First, unions typically seek to encourage workers to identify as part of a collective and thereby to “think in terms of group interests and group gains and losses” (Kelly 1998, 34; emphasis in original). In this way, workers might treat potential gains to their fellows’ interests as giving them additional reasons to participate in industrial action on top of those afforded by their own interests. The popular union slogan “an injury to one is an injury to all” is surely an expression of this idea. Relatedly, unions also aim to encourage workers to accept the idea that membership “is of value in itself” and that the costs of membership and collective action “have to be accepted as necessary sacrifices” for the advancement of all workers’s collective interests (Offe and Wiesenthal 1980, 78–79). The idea, in effect, is that because unions encourage workers to identify closely with their fellows, workers will come to favor participating in industrial action for its own sake (perhaps as an expression of their collective identity), even if their individual contribution is very unlikely to make any difference.

Finally, and in my view most significantly, unions seek to inculcate norms of solidarity that require collective action in the face of workplace injustice. Many individuals have strong preferences to act cooperatively with others, conditional on their having a sufficient degree of assurance that others will also cooperate (Fehr and Schurtenberger 2018). This is often understood under the rubric of norms: widely shared rules of behavior with which individuals prefer to conform, conditional on their believing that a sufficient number of others in their reference group do or will also conform. A central function of such norms is to help solve collective action problems (see, e.g., Ostrom 1998).

Unions, then, act as “norm entrepreneurs,” promoting norms according to which workers ought to act collectively to protect and promote their interests. ⁹ The very existence of unions, whose purpose is to advance workers’ interests via collective action, plausibly helps inculcate and reinforce attitudes of this kind. Beyond this, workers are often dissatisfied with various aspects of their employment situations. Yet studies suggest that they are unlikely to feel that it is appropriate to take collective action in response unless they perceive these aspects as matters of unfairness or injustice that ought to be resisted (see e.g. Buttigieg, Deery, and Iverson 2008; Klandermans 2015, 219–21). Unions, then, characteristically seek both to frame adverse features of individuals’ employment situations in these terms and (as already noted) encourage workers to identify as part of a collective, thereby promoting the idea that collective resistance is appropriate.

Yet such norms, given their conditional nature, will be of little help if workers lack sufficient assurance that others will also participate. Unions have an essential role to play in providing such assurance. The worker’s dilemma is clearly an instance of the more general prisoner’s dilemma. There is significant evidence that individuals are much more likely to cooperate in such cases where they can communicate in advance and precommit to doing so (Ostrom 1998, 6–7). Unions facilitate such communication, providing forums in which workers can debate the merits of taking industrial action and credibly commit to participating. Moreover, studies of protest movements have often found that leadership structures are essential to coordinate activities, mobilize participants, and disseminate information (for a seminal discussion, see Jenkins 1983). In the workplace context specifically, some empirical research suggests that even workers with strong solidaristic attitudes will typically fail to engage in industrial action without union leadership (see, e.g., Dixon, Roscigno, and Hodson 2004). Unions, then, help overcome potential uncertainties that might otherwise stymie workplace activism. Leaders, in consultation with members, can authoritatively determine the form industrial action should take, how long it should go on for, what its success conditions might be, and so on.

Thus, unions have a crucial role to play in overcoming the worker’s dilemma. They help ensure (though, of course, cannot guarantee) that employers face credible threats of industrial action, thereby compelling employers to track workers’ interests more closely in decision-making. Unions are often highly effective in this respect.

Consider collective bargaining. In Australia and the United States (among other places), the terms of many individuals’ employment are regulated by collective agreements negotiated between union representatives and management. In some European countries, by contrast, a system of “sectoral bargaining” prevails, in which union representatives negotiate industry-wide agreements with groups of employers. Agreements of both kinds tend to achieve outcomes that are significantly more favorable to workers’ interests. For instance, workers covered by collective agreements typically enjoy a significant “wage premium.” In the United States, Rosenfeld (2014, 45) finds that unionized private sector workers typically enjoy wages around 22% higher than relevantly similar non-union workers. Such workers are also around 17% more likely to be covered by employer-provided health insurance and about 20% more likely to have an employer-provided pension (62–64).

Unions also have long played a leading role in the introduction and enforcement of protective legislation and regulations. The US labor movement played a critical role in the passage of the Fair Labor Standards Act (1938), the Civil Rights Act (1964), and the Occupational Health and Safety Act (1970). There is a vast literature demonstrating that unions ensure greater consideration is given to workers’ interests by ensuring regulatory compliance. Reviewing decades of studies across a wide array of jurisdictions, Morantz (2017, 520–26) finds, inter alia, that unionized workplaces are significantly more likely to adhere to standards of worker health and safety, more likely to compensate individuals injured in the workplace, and typically exhibit lower rates of traumatic and fatal injuries. She also finds that workplaces with a significant union presence are more likely to adhere to regulations requiring the payment of overtime and exhibit lower rates of wage discrimination against women and people of color.

In these and many other ways, unions compel employers to track the interests of their workforce more closely than they would otherwise have done. This constitutes not only a form of protection against workplace domination but also a powerful affirmation of the ideal of collective self-rule in the workplace. An obvious implication of the claim that citizens have rights against domination is that those empowered to visit interference upon others must be compelled to exercise that power only in accordance with those others’ shared interests. ¹⁰ The only practical way in which this might be realized is to ensure that the former are subject to the control of the latter. In the case of the state, this requires a robust contestatory democracy, wherein citizens actively participate and in which government is appropriately responsive to citizens’ inputs (see esp. Pettit 1997). A similar rationale may be offered for unions in the workplace. Unions enable workers to demand that their employers track their interests more closely in decision making, and—by solving the worker’s dilemma—afford them the collective power necessary to lend force to such demands. They thereby enable workers to enjoy a form of collective self-rule in the workplace, just as the institutions of democracy enable this in the political case.

Collective self-rule, moreover, expresses a commitment to each worker’s legitimate interests in freedom and equality. Freedom, at least on the view presupposed in this article, just is nondomination, and nondomination in the workplace is constituted, in part, by the collective self-rule unions can realize for workers. Moreover, as Medearis (2020, 246–27) forcefully argues, industrial action—the means by which unions afford workers protection from domination—has an inherently egalitarian dimension. The power of industrial action, after all, lies in its collective character. Large numbers of workers act together, taking on costs in an attempt to force employers to address their joint demands. And where industrial action and threats thereof prove effective in affording workers protection against domination, each will enjoy a kind of equal status in the workplace, insofar as it will be a matter of common knowledge that no one has the power to interfere with any other person in a way that fails to track their interests. (I borrow this observation from Pettit 1997, 71.)

For these reasons, then, workers must be afforded the right to form and join unions. It is not enough, moreover, for this to be a mere abstract prospect. Unionization must be a real, practical possibility for workers. Corporate practices and government regulations that aim to render unionization inordinately difficult are, to that extent, unjust. Before April 2010, for instance, US airline and railway workers seeking to establish unions were required to hold a vote securing support from a majority of all workers whom the union might represent, rather than merely a majority of voters—an almost impossibly high threshold to clear (Rosenfeld 2014, 24–25). Arrangements of this kind are plainly indefensible, as are anti-union corporate practices of the sort that Amazon (among many others) is alleged to have engaged in.

Moreover, insofar as the power of unions to compel employers to track workers’ interests more closely rests on their ability to issue credible threats of industrial action, it is also essential that unions and their members have rights to take industrial action and that they enjoy the fair value of such rights. That is, regimes of industrial law must afford workers opportunities to take industrial action sufficient to provide an effective measure of protection against domination. Regimes that severely constrict the ability of workers to take industrial action are also, to that extent, unjust. For example, following reforms instituted in 2005, most Australian workers are only permitted to take industrial action following the expiration of a previous collective agreement and with permission from the Fair Work Commission, among other requirements (see FWC 2020). The commission, in turn, is obliged to order workers to terminate industrial action when it is “causing or threatening to cause, significant economic harm” to the employer, to any employee covered by the relevant collective agreement, or to the Australian economy as a whole. Most workers, then, have no right to take industrial action when employers threaten mass redundancies, endanger workers’ safety, or breach their obligations under industrial law or enterprise agreements. And even when they can take action, workers are prohibited from imposing significant costs on employers, which greatly weakens their bargaining position. Arrangements of this sort are deeply problematic.

However, unions are highly controversial. F. A. Hayek (1960, chap. 18), for instance, argued that they intensify economic inequality by increasing the wage differential between unionized and non-unionized workers and lead to price inflation by driving up wages. Vedder and Gallaway (2002, 128) hold that unions are associated with “lower rates of growth in income and jobs.” And Taylor (2013, 597) argues that unions increase the cost of labor to employers, disincentivizing hiring and thereby shrinking the range of alternate employment opportunities open to workers. This raises the costs of exit and therefore intensifies workplace domination.

These are consequentialist arguments, concerning which the evidential picture is mixed. For instance, there is some evidence that unions may indeed be a cause of inflation (see, e.g., Hung and Thompson 2016). On the other hand, insofar as unions tend to increase wages for lower-status workers, higher union density actually appears to be associated with decreased inequality (see, e.g., Western and Rosenfeld 2011). Yet even if the evidence was uniformly negative, it would still fail to undermine the argument developed, above. Individuals, I argue, have moral rights to unionize. Such rights are not contingent on their exercise failing to have any negative effects on the general welfare. There would be little point in any right if it could be violated whenever the general welfare would be better promoted by doing so. Surely, none of us can reasonably demand that other members of the community arrange their working lives to our maximum advantage. Still less can we reasonably demand that other workers tolerate domination in their workplaces so that we might enjoy less in our own—particularly when the option of affording ourselves protection against domination by unionizing is also open to us.

Let us consider, then, another family of objections. These seek to show that industrial action is necessarily rights-violating. If so, the fact that unions help workers issue credible threats of industrial action would argue against permitting workers to unionize, rather than (as I suggest) in favor.

I think it is undeniable that industrial action is sometimes rights-violating. Needful patients, for instance, plausibly have rights against doctors to provide lifesaving medical treatment. Strike action on the part of doctors where it entails the violation of such rights, then, is unjustifiable. Yet the charge at hand is that all industrial action is rights-violating. Several arguments might be made to this end, and we lack the space to discuss them all. Here, then, I shall simply consider what I take to be the three most promising suggestions. ¹¹

First, industrial action typically involves workers declining to discharge their contractual duties to their employer, thereby violating their employers’ rights under those contracts. Thus, one might conclude that industrial action is necessarily rights-violating. This argument fails. Employers do have presumptive rights against workers that they discharge their contractual duties. Industrial action may be unjustified where it involves the violation of these rights. Yet such rights may be undermined in a wide variety of ways. Dobos (2022, 256–62), for instance, points out that exploitative contracts do not plausibly give rise to any right on the part of employers to workers’s labor power. Industrial action on the part of workers whose pay or conditions are exploitative, then, does not involve a violation of any legitimate right. Nor will industrial action in any other case where the contractual rights of employers have been undermined for other reasons.

Second, one might point to the fact that strikers decline to work while maintaining that their jobs remain their own, insofar as such workers characteristically insist that employers may not hire replacements. Some have argued that this violates the rights of would-be replacements and employers to freely enter into contractual relations of their choosing (see, e.g., Shenfield 1986, 11). Yet this objection clearly depends on the claim that, by striking, workers forfeit their rights to their jobs. Otherwise, employers could have no right to attempt to hire replacements. Why should this be? The only plausible answer seems to be that workers’ claims to their jobs are conditional on their discharging their contractual duties to their employer. Insofar as industrial action typically involves workers declining to discharge some or all these duties, workers forfeit their claims to their jobs. Yet, as we have just seen, the contractual rights of employers (and, thus, the correlative duties of workers) may be undermined in a variety of ways. Where these rights have been undermined, workers who take industrial action in response do not forfeit their claims to their jobs, precisely because doing so fails to breach any duty owed to their employer (for further discussion, see Gourevitch 2016, 10).

Third, one might suggest that striking workers violate employers’ property rights. Insofar as employers own their business’s productive assets, strikers “effectively render the employer unfree to use his property as he sees fit” (Gourevitch 2016, 310). However, the property rights of employers are not unlimited. Such rights are artifacts of political institutions, established and sustained by the state (Murphy and Nagel 2002). Like everything else the state does, then, property rights ultimately depend on the cooperative efforts of the entire citizenry. For that reason, the privileges afforded to individuals in virtue of their property rights must admit of an appropriate justification to all citizens. We cannot simply use our property as we and we alone see fit. Thus, if employers’ property rights over productive assets are to be said to entail a right to hire replacements, this must admit of an appropriate justification to all citizens whom this will affect. It is not at all difficult to see that no such justification is available. Were employers permitted to hire replacements, this would profoundly undermine the power of the strike threat and thereby render the power of labor unions to protect workers from domination much weaker. Because workers have rights against subjection to workplace domination, such arrangements are plainly not justified. ¹²

We cannot rule out some more compelling argument being produced in the future. At least for the present, however, we seem to have little reason to accept that industrial action is necessarily rights-violating. Still, as we have seen, the right to take industrial action is not unlimited. For this reason, it is appropriate for the state to regulate industrial action. Any legal regime is bound to be imperfect. Still, the guiding ideal must be to afford workers the space to take justified—non-rights-violating—industrial action in response to reasonable grievances to limit the extent of employer domination. State intervention is appropriate only where necessary to protect the (legitimate) rights of employers and the wider community. That is a far more permissive stance than presently prevails in many contemporary liberal democracies.

A final, more moderate, challenge has it that, rather than mitigating workplace domination, unions merely redistribute it. Critics occasionally argue that unions also subject workers to domination. Shenfield (1986, 44), for instance, holds that individuals are “willy-nilly subject in large measure to the power of some union if [they wish] to earn a living. …The purpose of union leadership [is] …to dominate them.” At least with respect to workers’ interests in avoiding domination, then, unions do not make anyone better off, let alone express any sort of commitment to workers’ freedom or equality.

This objection highlights the fact that the internal organization of unions matters a great deal. Union officials do enjoy a significant amount of power over the workers they represent. To ensure that this power does not amount to domination, it is essential that workers be able to compel union officials to pursue a set of objectives acceptable to all those whom they represent in light of their interests in the workplace; that is, to track their shared interests. It seems obvious that, in general, the most effective means of ensuring this will be for unions to be organized democratically.

This means, in the first instance, that union officials must be elected by the membership. When union leaders’ positions are tied to their approval among members via elections, this gives rise to powerful incentives to track the interests of those members in decision-making. However, this alone will not be sufficient. Different proper subsets of workers may have distinctive interests, and it may be that candidates elected by a majority of workers have limited incentive to engage with the interests of smaller subsets of the workforce whom they are also supposed to represent. Unions, then, must also provide opportunities for members to exercise effective influence over day-to-day decision making via regular opportunities for consultation and deliberation. Members must take advantage of those opportunities. And union members must be robustly disposed to give appropriate weight to the interests of all members in policy formation and decision-making on the basis of those consultations. A union arranged along these lines would, I think, avoid subjecting members to domination insofar as union officials would be forced to track workers’ interests.

Of course, in the real world, unions often fall short in this respect. Participation in internal union decision-making processes is often very minimal, leaders with the best intentions may misperceive members’ interests, and so on. In many cases, then, unions will enjoy some measure of arbitrary power over their members. That is regrettable. However, workers are still likely to be far better off with respect to their interests in avoiding domination where they are represented by an effective union. The disvalue of the domination to which they may be subjected by their union is likely to be far outweighed by the value of the protection from employer domination their union affords them. And, even if this were not so, such considerations do not furnish us with anything like an objection to unions, all things considered. As discussed, the evidence overwhelmingly suggests that workers are better off with respect to their wider interests when represented by an effective union. If one must inevitably be dominated in the workplace, then, surely it is better to be partially dominated by one’s union rather than solely by one’s employer.

Labor Unions and Free-Riding

The idea that failing to join one’s union constitutes free-riding is common in union circles (e.g., Marin-Guzman 2019), and the idea that individuals have duties not to engage in free-riding behavior has been discussed extensively among political theorists. Rawls (1971, 342-–50), for instance, held that the parties behind the veil of ignorance would favor a principle of fairness, according to which individuals “are not to gain from the cooperative efforts of others without doing [their] fair share.” ¹³ Yet theorists have thus far devoted little attention to the specific case of unions. ¹⁴ I think that the arguments of the previous sections lend support to the claim that workers represented by a union who fail to join are typically free-riders. Insofar as free-riding is wrongful, workers represented by a union have pro tanto duties to join and contribute to its activities.

We should begin by defining free-riding as follows:

Free-riding: A’s ɸing amounts to free-riding if and only if, by ɸing, A consumes some collective good without making an appropriate contribution to the process or processes by which that good is produced.

Three points stand in need of explication. First, goods are collective in the relevant sense where they are the nonexcludable product of social cooperation. A good is nonexcludable if and only if providing that good to some member of a group entails providing that good to all members of that group. Consider radio stations. Broadcasting a radio signal to some persons in a given area entails broadcasting to all persons in that area. As to the matter of social cooperation, many nonexcludable goods do not depend on human activity, at all (e.g., gravity). Clearly, individuals cannot free-ride on the production of such goods.

Second, free-riding is wrongful chiefly because it is unfair. As Cullity (2008) forcefully argues, unfairness involves a failure to exhibit the form of impartiality appropriate to the context in question. With respect to the production of collective goods, the appropriate form of impartiality arises out of two elementary facts. First, there are goods whose benefits for the members of some group are such that that group has reason to produce them. Second, the production of such goods typically requires a sufficient proportion of the group in question to contribute to some cooperative enterprise. For many communities, for example, the benefits of systems of public transport are such that they ought, collectively, to establish such systems. Because no individual could establish or operate such a system alone, ensuring the availability of public transport requires a scheme of collective action.

Contributing to such collective endeavors, then, is simply “what must be done by individuals if the group is to do what it ought” (Cullity 2008, 10). To refuse to contribute to the production of collective goods from which one benefits, then, is to exhibit a form of objectionable partiality in favor of oneself-i.e., to arrogate to oneself the privilege of declining to contribute to the good in question, while simultaneously depending on others failing to arrogate to themselves that same privilege. Fare-evaders on public transport, for instance, afford themselves the preferential advantage of free travel while simultaneously depending on others declining to do so. The benefits of public transport would not exist if no one paid their fare, after all. It is in this respect that fare-evasion is unfair.

Finally, free-riding involves a failure to make appropriate contributions to collective goods. There will often be many ways in which one can contribute. Some will be sufficient to discharge one’s duties not to free-ride, and others will not. Paying one’s fare on a public bus is (typically) sufficient to discharge one’s duties in this respect. Sending lunch to the mechanics responsible for maintaining public buses is clearly not. How, then, to distinguish between appropriate and inappropriate contributions?

For an answer, we can appeal to the account of unfairness adumbrated earlier. Failing to contribute to the production of collective goods constitutes a failure on the part of individuals to do what must be done if the group to which they belong is to do as it ought (i.e., to produce the collective good in question). Some agent A’s Ψing, then, will constitute an appropriate contribution to the production of some collective good if and only if contributions of that type are necessary in the circumstances for the production of the good in question (even if A’s token act of Ψing is not itself necessary). Paying one’s fare on public transport obviously qualifies, in this sense, insofar as it is necessary for a sufficient number of persons to pay their fares for such services to be viable (even if the viability of the service does not depend on any particular individual’s fare). Free lunch for the bus mechanics is clearly not necessary in this sense.

Stated more fully, then, I hold that where some good, G, is nonexcludable, produced by social cooperation among the members of some group, and of sufficient benefit to the members of that group such that they have collective reason to produce it, those who benefit from G have duties of fairness to make contributions of the kind necessary for the group to successfully produce G. I think that these conditions are straightforwardly satisfied in the case of unions. We have already identified the central good that unions produce: protection from domination (itself clearly the product of collective action among workers). It also seems difficult to deny that workers have powerful reason to act collectively so as to ensure such protection. Insofar as domination strikes at our fundamental interests, individuals have rights against subjection to workplace domination. As we have seen, most workers can hope to enjoy protection from domination only by acting collectively under the auspices of a union.

Many different sorts of contributions might be required to ensure that unions afford workers protection from domination. Still, the power of unions to compel employers to track the interests of workers more closely in decision-making fundamentally derives from their ability to credibly threaten industrial action. Contributions appropriate to the production of that good, then, will be those necessary for unions to maintain the credibility of that threat. Three such contributions will be of particular importance. First, workers must join the union that represents them, because unions must command a membership of a size sufficient for industrial action to pose a nontrivial threat to the interests of employers. Second, workers must be disposed to participate in union-authorized industrial action, provided the action is justified (i.e., will not violate the legitimate rights of employers or the wider community). Otherwise, employers would rationally attach little weight to any threat to take industrial action. And third, workers must refrain from undermining industrial action where it takes place. Suppose, for instance, that Dave goes on strike and Lisa agrees to take on his duties in addition to her own, thereby neutralizing the effects of his absence. Actions of this sort limit the costs to employers of industrial action and threaten to render it ineffectual as a means of compelling management to track workers’ interests.

Finally, then, protection from domination in the workplace is plainly nonexcludable in virtue of the fact of workers’ shared interests. Suppose Dave is a union member whose employer is compelled to track his interests more closely than they would have done in the absence of the credible threat of industrial action that his union facilitates. Dave thus enjoys a form of protection against domination. Suppose, further, that Lisa does not belong to the union but has interests in common with Dave in the workplace, such as fair pay, a reasonable work schedule, health and safety protections, and so on. Thus, insofar as the activities of the union compel their common employer to track Dave’s interests more closely, they will de facto also compel their employer to track Lisa’s interests more closely. Lisa, thus, will also enjoy protection against domination in the workplace by virtue of her co-workers’ union activities.

Yet perhaps this is not enough to establish that Lisa has duties of fairness to join and contribute to the union’s activities. The most significant objection to the idea of duties of fairness, in general, is that the mere receipt of benefits is not sufficient to ground duties to contribute to the systems by which they are produced (see esp. Nozick 1974, 90–96). ¹⁵ If Geoffrey’s neighbors set up a public address system that plays music he occasionally enjoys, for instance, he is surely not thereby obliged to help maintain that system. Maybe, then, the mere fact that non-unionized workers receive the benefit of protection from domination in the workplace fails to establish that there is a duty to unionize.

This objection fails. Recall that individuals’ duties of fairness to contribute to the production of collective goods are fundamentally duties of appropriate impartiality to do “what must be done by individuals if the group is to do what it ought” (Cullity 2008, 10). Citizens ought to pay their fares on public transport, given that the community to which they belong has reason to produce that good given its benefits for members. Geoffrey has no duty to contribute to the maintenance of the public address system because the benefits of such systems are simply too trivial for the neighborhood to have reason to produce them (12–13). In contrast, were his neighbors digging a well to provide drinking water to the houses on the street where none would otherwise have been available, there would be nothing counterintuitive in the suggestion that Geoffrey would treat his neighbors unfairly by declining to contribute to their efforts. The benefits of drinking water are very significant, and neighborhoods clearly have reason to ensure its availability.

The issue, then, is whether protection from domination is a significant enough benefit for workers, collectively, to have reason to produce it. There can surely be little argument on this score. Subjection to the arbitrary will of others is very bad for our interests. Protection against such domination is, as such, surely a good that workers have powerful reason to want, and thus a good that groups of workers have reason to produce. Moreover, protection from workplace domination comes with a series of associated gains that accrue to both union members and nonmembers. As we have seen, workers’ pay and conditions are typically much better when they are the product of collective agreements between union representatives and employers. Such agreements standardly apply to all employees, not merely to union members. In Austria, Belgium, Finland, and France, more than 90% of workers are covered by such agreements, for instance. Yet, Austrian union density is presently around 26%, Belgian union density is around 49%, Finnish union density is around 59%, and French union density is around 11% (OECD 2021). Similarly, enhanced compliance with rules mandating health and safety protections, nondiscrimination, and the like is plainly of significant benefit to all workers.

Yet what about workers who do not benefit from their union? Some unions, perhaps due to corruption or inefficiency, fail to realize benefits sufficient to outweigh the costs of participation for workers. Although such unions are surely rare, I concede that the arguments of this section do not apply in such cases. A more interesting possibility concerns workers whose interests are set back by their labor union. Union-negotiated collective agreements typically specify a set of wages and conditions that apply to all workers of a given type. Workers are usually better off under such agreements. Yet there is surely a (very) small number of highly exceptional employees who would be able to secure still better wages and conditions were they able to negotiate one on one with their employer. As Shenfield (1986, 43) puts the point, “Workers of above-average quality would get higher wages were it not for uniformities imposed by union action.” Insofar as free-riding implies the receipt of benefits, such employees are not free-riders.

In response, recall that domination consists in others having the power to interfere with us without being forced to track our interests—regardless of whether they choose to actually exercise that power. Even if a small minority of workers might enjoy better conditions without union representation, then, this would be solely by virtue of the grace and favor of their employers, who would retain their ability to interfere with such workers arbitrarily. In the absence of a union, then, exceptional employees would suffer domination. The fact that unions provide some measure of protection from domination is a very significant benefit to be set against diminished pay and conditions. Thus, the group of employees who genuinely fail to benefit from union representation are those for whom the diminution in their pay and conditions is significant enough to outweigh the benefit of protection from domination afforded them by their union. Surely very few workers fall into this category.

And even in the case of such workers, just as groups sometimes ought to produce certain collective goods in virtue of the benefits of those goods, they also sometimes ought to produce such goods for moral reasons. Individuals’ duties of fairness with respect to the production of collective goods are duties of appropriate impartiality that require individuals to do what must be done by them for their group to do as it ought. Thus, just as individuals may have duties of fairness to contribute to collective goods from which they benefit, they may also have such duties to contribute to the production of goods where the group to which they belong has moral reason to produce them (Cullity 2008, 11; Umbers and Moss 2021, 90–97). Suppose that some political community’s disability care services were funded by a special tax. Citizens who declined to pay, irrespective of whether they personally benefited from such services, would be open to much the same sort of complaint as any free-rider—that they had arrogated to themselves the preferential advantage of declining to pay the tax while simultaneously depending on others failing to afford themselves that same advantage. If no one paid the tax, after all, the community would fail to provide care for the disabled and would therefore fail to act as it morally ought. As discussed earlier, individuals have moral rights against subjection to domination and therefore have moral reasons to ensure protection from domination in the workplace. Even if they themselves do not benefit, all things considered, from union activities, then, exceptional employees might nevertheless have duties of fairness to unionize. After all, if all workers declined to contribute, workers collectively would fail to act as they morally ought insofar as they would fail to produce the good of protection from domination.

As such, I hold that workers have duties of fairness to join the unions that represent them, to participate in justifiable union-authorized industrial action, and to refrain from undermining such action. Individuals who fail to do so treat their fellow workers unfairly, arrogating to themselves the benefits of partial protection from domination, while depending on others taking on the costs associated with the provision of such protection.

Conclusion

By way of a conclusion, I want to consider how the foregoing considerations connect to four related issues. First, it has long been recognized that unions, themselves, also present workers with a collective action problem given the collective character of the benefits they produce and the insignificance of each worker’s contribution. Olson (1965, 66–91) influentially suggests that the existence of unions must therefore be explained by either coercion or the provision of “selective incentives” (benefits accessible only to union members). My account suggests a further (though not mutually exclusive) hypothesis. The purpose of unions is to resolve the worker’s dilemma, itself a different though related collective action problem. Much recent research has shown that groups often recognize where they face collective action problems, find ways to resolve these, and thereby promote their interests (see esp. Ostrom 1998). The fact that unions help resolve the worker’s dilemma, then, might also help explain how the collective action problem posed by unions themselves is sometimes overcome.

Second, I have argued that certain structural features of typical labor markets give rise to the domination of workers by employers. However, a number of influential scholars have recently argued that, in addition to giving rise to interpersonal domination, such structures also constitute a form of domination, in themselves. ¹⁶ On the influential account developed by Gourevitch (2013), for instance, the highly unequal distribution of control over productive assets characteristic of typical liberal democracies constitutes a form of domination insofar as it means that workers are effectively forced to sell their labor to some employer or other, rendering them vulnerable to exploitation.

Unions might help mitigate this structural domination in at least two ways. First, and most obviously, unions help reduce economic inequality by ensuring higher wages for workers. Second, mitigating structural domination will require reforming these social structures. Many such reforms are possible. A sufficiently generous universal basic income, for example, would plausibly mean that individuals were not effectively compelled to work for capital-owners. All such proposals, however, face an important and underrecognized challenge. Insofar as their object is to increase the power of workers in the labor market, their implementation would seriously undercut the interests of socioeconomic elites and corporations who benefit enormously from the status quo. These actors would certainly resist any attempt on the part of the state to implement such reforms. Such resistance, moreover, is likely to be effective. Elites and corporations, after all, are highly influential in democratic politics—far more so than the typical citizens who would be the beneficiaries of such reforms (see, e.g., Gilens 2012). As Gourevitch and Stanczyk (2018) point out, then, the only realistic hope for the implementation of the sorts of reforms to the labor market necessary to mitigate structural domination lies in the ability of the working class to mount an effective response to this inevitable opposition. It is hard to see how this might be accomplished without unions.

Such resistance, after all, will require much political activism, which in turn involves collective action. Such collective action is extremely unlikely to occur spontaneously. And unions, given their organizational capacities, are uniquely well positioned to afford workers the necessary leadership and coordination. The prospects for alleviating structural domination, then, are in an important respect tied to the fortunes of unions. This gives us further reason to insist that workers must have substantive opportunities to unionize.

Third, as noted in the introduction, the free-riding argument lends important support to proposals that would render union membership a condition of employment by undercutting the charge that such arrangements violate individual liberty. A further response to this charge follows from the arguments developed in the first section, as well as the considerations just discussed. For if most workers have little effective choice but to work for some employer or other, they are effectively compelled to enter into relations of domination with employers. Rendering employment conditional on union membership, then, plausibly enhances workers’ liberty. Even if the structures of the labor market subject workers to domination by compelling individuals to work, those employment relations are liable to be far less dominating (and thereby less injurious to liberty) where workers are represented by an effective union that can coordinate industrial action and thereby compel the employer in question to track their interests more closely. ¹⁷

Finally, we have thus far focused exclusively on employees of firms. Yet recent years have witnessed an enormous rise in so-called gig work, a distinguishing feature of which is that workers are typically treated as independent contractors, rather than employees. ¹⁸ I think that such workers also suffer domination, though in a different manner. The domination of employees constitutively involves subjection to managerial authority on an ongoing basis. Yet, gig workers are not typically subject to the formal authority of managers (Rahman and Thelen 2019, 195–96). They are simply offered work that they may either accept or decline. However, many philosophers have emphasized that relations of asymmetric dependence also give rise to domination. And there can surely be little doubt that gig workers are often highly dependent on the platforms through which they are offered work. This, in turn, affords such platforms an enormous amount of arbitrary power over their “workforce,” with sometimes troubling consequences. Rideshare companies such as Uber and Lyft, for instance, do not disclose the details of a job (distance, fare, and so on) until after a driver accepts an offer to take it on. Such platforms reserve the right to refuse access to drivers who cancel disadvantageous rides (e.g., those that will take a lot of time but command little remuneration). That, clearly, is something many gig workers can ill afford, leaving them little choice but to comply.

In the United States, the Biden administration is presently seeking to implement legislation that would ensure the rights of gig workers to unionize and take industrial action. That would be a welcome development. Were gig workers able to threaten to impose costs on such companies in response to decisions injurious to their interests, this might provide them with some measure of protection against domination. Were gig workers to refuse en masse to take on jobs, for instance, that would impose serious costs on the platforms they work for. It is hard to believe that gig workers could credibly threaten to do so without a union.

Yet gig work also presents significant challenges for unions on top of those that arise in traditional workplaces. For instance, we have already noted that the ability of unions to overcome the worker’s dilemma rests partially on their inculcating and leveraging norms of solidarity. Such norms are effective in promoting collective action only under certain conditions. It is generally accepted, for instance, that such norms tend to command wide adherence only where defection can be easily observed and readily sanctioned (see, e.g‥ Olson 1965, 60–65). However, most gig workers have little contact with one another and therefore will not typically be able to observe or sanction one another’s conduct. Considerations of this kind are liable to weaken the ability of unions to credibly threaten industrial action, which in turn weakens their ability to protect gig workers from domination by attempting to issue such threats. It may be, then, that unions can best protect gig workers from domination in other ways. Rahman and Thelen (2019, 196–97), for instance, suggest that unions ought to seek to develop coalitions among workers and consumers that might challenge the power of such platforms and compel states to impose tighter regulations on their activities. Many other strategies are possible. The coordination of industrial action may be critical to the power of labor unions to protect workers from domination. Fortunately, however, it is not the only tool at their disposal.