‘I’m Outta Here’: Theorizing the Role of Exit in the Ideal of Non-Domination

Accounts of non-domination have tended to emphasise the role resources and other capacity and voice building mechanisms can play in giving people the power and the institutional means of living lives that are free of domination. Yet the role of exit - of institutionally protected means of withdrawing from relationships - has remained undertheorized in accounts of non-domination. Drawing on a range of public policy examples, this paper seeks to shed light on the ways in which, and under what conditions, institutionalised means of exit can contribute to realising the ideal of non-domination. It shows that while rights of exit and low exit-costs can play an essential role in protecting people from dependence on the arbitrary wills of others, it is only under certain conditions these can be said to contribute to the realisation of the ideal of non-domination in a broader sense. Understanding the relationship between exit and non-domination, it further argues, gives us a clearer (if more complicated) picture of the relationship between non-domination and sources of power such as monetary resources and voice.


Introduction
Non-domination is about the secure enjoyment of freedom from arbitrary power (Lovett 2018;Pettit 2014;Shapiro 2012). Because it looks at freedom from a relational standpoint, it directs our attention at the relations of power between individuals and between individuals and organisations. And it tells us to be concerned not merely with actual instances where people are mistreated or abused as a result of being in a vulnerable position vis-à-vis some other actor, but rather the structural relations of power between them. People may be said to enjoy freedom as non-domination to the extent that their well-being is not (too) dependent on the arbitrary good will of another person or organisation. A wife who would face financial destitution were she to leave her husband, or her husband leave her, is in a too vulnerable position to enjoy nondomination. The same is true of a worker who is too dependent on the arbitrary good will of their employer for their future financial well-being.
There are many reasons to be concerned with whether people are secure in their enjoyment of freedom from arbitrary power. Structural inequalities in power can facilitate and perpetuate abusive practices within families and in the relationship between employers and employees. It can also erode trust and thus act to undermine even relationships characterized by affection and love Trägårdh 2006, 2010) which perhaps explains why the literature on nondomination has yielded such a rich and interesting research program, exploring both questions of theory and public policy (Lovett and Pettit 2009).
In this paper, I seek to shed light on the public policy implications of non-domination by exploring the relationship between exitof the relative ease with which people can separate themselves from relationships with individuals or organisationsand non-domination.
Much of the literature on non-domination, as Taylor (2017) has noted, has focused primarily on developing democratic and rule of law approaches to non-domination. Indeed, this has been the primary focus of Pettit's (1997Pettit's ( , 2001Pettit's ( , 2014 work on republican liberty. Though it would be wrong to say that Pettit's work does not have anything to say about exit, the relationship between exit and non-domination has remained somewhat undertheorized. Pettit, along with Shapiro (1999, 2012 and Lovett (2009Lovett ( , 2010, have developed accounts of non-domination that highlight the central role of low exit-costs in certain contexts, particularly in terms of protecting people from domination within marriage (see also Costa 2013;Halldenius 1998Halldenius , 2015Hobson 1990). Warren (2011) has developed an account of how exit can help free people from domination and improve democratic governance. And Taylor (2013Taylor ( , 2017, perhaps more than anyone, has focused on exit-based policy mechanisms to highlight the role markets and competition can play in freeing people from relationships of dependence by reducing their exit-costs.
But the relationship between exit and non-domination is more complex than has been reflected in the research literature. Much of the existing literature assumes that the relationship between exit and non-domination is relatively simple and unidirectionalin the sense that lowering exit costs is generally assumed to further non-domination. While this assumption is valid in many contexts, for reasons I explain below the question of whether lowering exit costs furthers non-domination is highly context dependent and requires a more fact-specific inquiry. Indeed, under certain conditions, non-domination is best furthered by raising the exit-costs for certain people, or agents.
In the paper, I try to show that because freedom as non-domination is concerned with the relations of power between individuals and between individuals and various types of organisations, the precise relationship between exit and non-domination varies depending on the circumstances. While in many contexts the availability of a low-cost exit will serve to enhance relationships of non-domination, under certain circumstances the ideal of non-domination is better served by limiting or circumscribing the possibilities for exit. I also try to clarify what these circumstances are. Building on the implications of this, I look at the relationship between exit, voice and non-domination, and try to suggest that under certain circumstances, policymakers in effect have to choose between two models of how to realize the ideal of non-domination, where exit and voice are made to play different roles. I argue that when the relevant liberty interest depends on being able to change the status quo, exit-costs should be made low even when it comes at the expense of voice; however, when the liberty interest depends on being able to maintain the status quo, exit-costs should be set at a higher level and processes for exercising voice given primacy.
In the next section I attempt to clarify the meaning of non-domination and outline a few of the central disagreements within the literature on non-domination that have bearing on the questions explored in this paper. In the subsequent three sections I develop the central claims of the paper, before going on to pinpoint and clarify some of the implications of the arguments presented.

The Meaning of Non-domination
It is important to begin by clarifying what non-domination means and to outline the contours of the debate regarding some of the key interpretative disputes, as these carry policy implications that will be explored later in the paper.
In broad terms, non-domination can be defined, in Lovett's words, "as a sort of structural independenceas the condition of not being subject to the arbitrary or uncontrolled power of a master" (2018). As an ideal and a normative concept, non-domination has two major virtues: it helps us think productively about what freedom means and what it requires in terms of institutional design, and it helps us identify (some of the) reasons why inequality is something to be concerned about. With respect to the former, consider the paradigmatic example of the slave who has a benevolent master (Lovett 2018). Looked at through the lens of negative liberty which defines freedom as the absence of interference by others (Berlin 2002)a slave who has a benevolent master that allows him or her to pursue various activities without interference would need to be considered free, at least in some respects. Yet this seems wrong, and the concept of freedom as non-domination gives us an account of why. The reason a slave is not free (even under conditions where they are not interfered with in their pursuit of various ends) is because of the structural inequalities in power that exist between the slave and the master, which render the slave completely dependent on the arbitrary whims and benevolence of their master. At the master's discretion and whim, all the opportunities the slave has enjoyed as a consequence of the master's benevolence could be taken awaywithout the slave having any means of vindicating his or her interests. To really free the slave, in other words, one must attend to the structural relations of power (see Lovett 2010;Pettit 1997Pettit , 2001Pettit , 2014Shapiro 2012;Skinner 2008; see also Halldenius 2010). With respect to inequality, the concept of non-domination helps us pinpoint what it is about inequality that we often object to, which is that inequalities in power or resources can be used to illicitly dominate more vulnerable parties (Shapiro 2012).
Yet, what does it mean to be 'secure in the enjoyment of freedom from arbitrary power'? This question boils down, in part, to how we ought to think about and go about estimating how vulnerable a position somebody is in, given different institutional contextshow much does one have to be at risk of losing to be in a vulnerable position? Do people's most fundamental and basic interests have to be at risk, or is it sufficient if somebody is at risk of suffering a substantial drop in their standard of living (even if where they will land is not a state of destitution)? 1 Here the answers in the research literature range from the (slightly) more minimalist (Shapiro 2012) to more demanding accounts of basic interests (Lovett 2009(Lovett , 2010Pettit 1997Pettit , 2001Pettit , 2014. Lovett (2009Lovett ( , 2010, for instance, suggests that we ought to think of a person's level of dependency as reflecting their subjectively perceived cost of exiting a relationship or organisation. When it comes to the question of what non-arbitrariness means, Lovett (2018) distinguishes between two answers to this question: defining non-arbitrariness procedurally and defining it democratically. On the former view, "[p]ower is not arbitrary … to the extent that it is reliably controlled by effective rules, procedures, or goals that are common knowledge to all persons or groups concerned" (Lovett 2018). On the latter view, "power is not arbitrary … to the extent that it is directly or indirectly controlled by the concerned persons or groups themselves" (Lovett 2018). How one answers this question will shape, for obvious reasons, what one takes to be the institutional and policy implications of the ideal of non-dominationparticularly concerning the role the democratization of human relationships should play in freeing people from arbitrary power (for an interesting take on what domination means in republican political theory, see Thompson 2018).
The answers given to the more philosophical questions thus underpin the type of policy recommendations that different scholars have put forth to protect people from domination. In broad terms, the primary approaches to realizing the ideal of non-domination have revolved around: (1) enshrining the rule of law, (2) democratization, and (3) facilitating individual independence (see Taylor 2017;Lovett 2018). These are, of course, not mutually exclusive and most theorists that have engaged with the issue have put forth recommendations that could be construed as fitting within all three of these approaches. But, with that said, to echo a point Taylor (2017) has made, the primary focus in much of the scholarship has been on policies that either seek to harness democratic inclusion and participation or to enshrine judicial protections and the rule of law. Taylor himself has sought to rectify this slight imbalance by theorizing and shedding light on market and exit-based strategies for freeing people from arbitrary power. And it is on his effort to theorize the role exit-based policy mechanisms can play that I hope to build in this paper.

Theorizing the Relationship between Exit and Non-domination
Before saying more about the relationship between exit and non-domination, it is important to clarify what exit means in this context. For the purposes of this paper, I will use Taylor's definition of exit (Taylor 2017). Drawing on Hirschman's (1970) typology, he defines exit as "separation: the attempt not simply to leave a dysfunctional organization but also to escape its authority and the various obligations of membership in it" (2017, p. 18). As should be clear from this definition, the act of (fully) exiting a relationship or an organisation is quite far reaching in its implications, since it involves not merely leaving an organisation or relationship, but escaping the obligations that were attached to that membership. A person who renounces their membership in an organisation, stops attending the meetings, but who is still liable to pay some sort of continuous severance fee, or to donate a large portion of their estate to the organisation in the event of their passing, has not been able to fully exit the relationship with the organisation. The exit under those circumstances will only have been partial, not complete.
Something that should be immediately clear, when we consider what exit means, is how inseparable it seems from the concept of liberty in some broad sense. The freedom to do this or that, to make choices, to switch allegiances seems inseparable from what it means to be able to act as a free agent. This is particularly the case if we take freedom to mean negative liberty, the absence of interference by others. A person who is prevented from leaving a spouse, or an employer, or indeed their place or country of residence, would not enjoy negative liberty. For Nozick (2013), similarly, the availability of exit is central to a free society. A free society, he argues, should permit people to experiment with different forms of social and economic organisation, though the formation of various forms of voluntary associations-on the condition that these associations remain voluntary, which is to say, allow their members to exit.
That low exit-costs can enhance relationships of non-domination is also a central insight of the scholarship on republicanism, as has already been mentioned. When the cost of quitting one's job, or of being fired, is made lower than it otherwise would be through a generous unemployment insurance, one is made less dependent on the arbitrary good will of one's employer. The same logic applies in the marital context. If the financial costs of leaving a spouse are made low, a potential source of domination is removed.
Yet exit also has a connotation not of liberty but of licenseof stepping away from obligations, of disloyaltywhich perhaps explains the ambivalent attitudes it seems to elicit in some of the literature on non-domination: at once something close to a necessary condition for liberty, and simultaneously a threat to the civic bonds of the community (see Taylor's 2017 discussion of how exit is treated among republican thinkers). Thus, in the context of discussing the importance of women having the option and the means to leave dominating husbands, the availability of exit is treated as central to freedom from arbitrary power; whereas, in the context of discussing rich individuals moving to a different country to avoid paying their fair share of taxes, the availability of exit is seen as something that undermines democracy and therefore also freedom as non-domination (see for instance Pettit 2014).
The relationship between exit and non-domination is thus more complicated than it may appear at first glance. In the next three sections, I attempt to paint a more nuanced picture of the relationship between exit and non-domination. I make two central claims: (1) under certain circumstances (which I attempt to specify), the best means of bringing about relationships of non-domination will be to limit or circumscribe the possibilities for exit. (2) In some contexts, low-exit costs can undermine the strength of voice, which means, in effect, that policymakers have to choose between exit-based and voice-based means of furthering non-domination.

Exit in the Context of Power as a Relational Good
From an individual perspective, having the option and the means to exit a relationship (be it with a spouse or an employer) at a low cost to oneself seems to unequivocally enhance one's freedom from arbitrary power. It is harder for a spouse, an employer, or indeed a particular state to exercise domination over another individual if that person has the option and the means to remove themselves from the relationship at any point in time without suffering too much hardship. Indeed, the knowledge that a person has the option and the means to exit a relationship and yet chooses to stay gives us good reason for thinking that the person is there voluntarily and is not dominated. This is why, with good reason, lowering the cost of exitby enabling no-fault divorce, the provision of generous social insurance, etc.is generally seen as a way of empowering individuals and achieving the kind of structural independence the ideal of non-domination calls for.
However, power is a relational good, which means that the empowerment of one individual can undermine the relative power and independence of another (though who and in what way will depend on several different factors). In a context, for instance, where the opportunities for women to find gainful employment are low and there is no social insurance system to support women who get divorced and find themselves without a job, policies that make it easier to get a divorce and reduce or eliminate any legal obligation to financially support a former spouse after the dissolution of the marriage would have the effect of lowering the cost of exit for the husband and making the wife more dependent on their husband's arbitrary good will. Or consider another example: under conditions of high wealth inequality, policies that make it easier for individuals and companies to relocatebe it to another province or region or another countrywill have the effect of lowering their exit costs, rendering them more independent vis-à-vis the powers that be in the political community in which they reside, while causing the rest of the political community to become more dependent on their arbitrary good will (if we assume the cost to the community of them leaving, whether in the form of a loss of tax revenue or jobs, if we take the example of a company relocating, would be substantial enough).
This complicates the relationship between exit and the ideal of non-domination. And it suggests that under certain circumstances, the best way to realize the ideal of non-dominationif by that we mean to create conditions where as many people as possible are free from relationships of dominationmay be to limit or circumscribe the possibilities for exit.
Let us consider a few examples, beginning with the example of rich individuals using the threat of exitof emigrating from the country or of relocating a company they own, or something along those linesas leverage to push lawmakers to set tax rates and write regulations they find favourable. There are several policy directions one can go to try to alleviate this problem. One policy direction would be to try to create a greater level of equality with respect to the influence different members of the political community enjoy over policy decisions. This might entail, for instance, instituting policies that seek to achieve a more equitable distribution of wealth and resources. Alternatively, it might entail policies that have the effect of reducing the influence of lobbyists and money in politics (see Pettit 1997, 2014)such as imposing limits on campaign contributions, or regulating the amount of time devoted to each major candidate for office on national broadcasting networks. In broad terms, the primary vehicle through which Pettit and most other republican theorists envisage limiting the influence of lobbyists and money in politics is facilitating a more inclusive and deliberative democratic processby equalizing and strengthening the opportunities for voice, in other words.
The role of exit in this context is less straight forward. While Warren (2011) has shown how policies that facilitate low-cost exit from potential relationships of domination can also serve to enhance the quality of democracy, Pettit (1997, 2014) sees exit, particularly when it takes the form of wealthy individuals moving to another country to reduce their tax bill, as something that is potentially detrimental to the political community and its enjoyment of freedom as non-domination. As Hirschman showed (Hirschman 1970(Hirschman , 2013, one of the ways to strengthen voice, under certain conditions, is to make exit more difficult. And this is also precisely what Pettit suggests could be done to prevent wealthy individuals and corporations from relocating to another country to pay less taxes (2014). This could be achieved, for example, through a citizenship-based taxation system, as used by the US (though administrative obstacles would no doubt reduce its effectiveness), or through other policies that complicate or increase the cost of relocating one's assets to another country. The aim here would not be to make it illegal or impossible to move to another country or territoryaims that would be incompatible with the ideal of non-dominationbut rather to raise the cost of exit sufficiently to discourage wealthy individuals from using the threat of exit to gain political leverage. 2 By raising the cost of exit from the political community, individuals are in effect made more dependent on the quality of its political decisionsthus incentivising the use of voice. 3 This underscores that there are different ways of achieving the structural equality that underpins the ideal of non-domination. When exist-costs are made low for the different parties in a relationship, it increases their mutual independence and helps ensure that they are not dependent on the arbitrary good will of another for their wellbeing. In contrast, the path described in the example above involves increasing the mutual dependence of the different parties in the relationship on the rules that govern the collective that they together constitute. In the first scenario, exit costs are made low and voice is given a peripheral role; in the latter scenario, exit costs are made high and voice is given a central role.
Another example that sheds light on how non-domination can be furthered by circumscribing the possibilities for exit can be found in how Sweden regulates inheritances. One of the central aims of the Swedish welfare state has been to free individuals from dependence on the good will of their family members, as Berggren and Trägårdh (2006; 2010; see also Trägårdh 2010; see also Sainsbury 1999) have shown persuasively, which makes Swedish public policy in areas that touch on and affect the relative power of family members vis-à-vis each other particularly interesting for scholars interested in investigating the policy implications of the ideal of non-domination.
What is distinctive about Swedish inheritance law is how little discretion it leaves to the person leaving the inheritance to decide how the assets they leave behind are to be distributed. Childrenincluding adult children who are not dependentsare by law entitled to a share equivalent to no less than half of the value of the estate left behind (Ärvdabalk 1958:637). 4 This is the case regardless of whether the child is an adult or a dependent of any kind. It is not contingent on the child demonstrating that they have a financial need for the inheritance or some special moral claim to it. It applies even if there is a will that specifies that all the assets in the estate should be donated to charity, or go to some other beneficiary. Indeed, even the parents' power to distribute the inheritance prior to their deaththrough gifts and the likeis circumscribed. When there is more than one child, should the parents seek to give one of the children most of their assets by, for instance, giving them some of their assets prior to their death (whether in the form of money, or by letting them assume ownership of a house, for example), the other child (or children) is (or are) legally entitled to have the value of those assets included in the estate that is to be divided upon the parents' death. This means that, under certain circumstances, a sibling that has received a large share of an estate in the form of a gift prior to their parents' death would be under a legal obligation to compensate their sibling(s) financially.
If we analyse this policy through the lens of exit, we can see, first, that the exit-costs for the children are reduced through the policy. An adult child who wishes to break with their parents in some way, or do something they know will be met with disapproval and perhaps even ostracismsuch as embracing a religion or lifestyle their parents disapprove ofwill know that, whatever the potential emotional pain or other forms of hardship their parents' reaction might bring, their decision will not result in the future financial penalty of being disinherited. Their future financial well-being will be less dependent on the good will of their parents, and thus the cost of breaking with their parents will be lower than it would be if the policy were not in place.
Second, we can also see that this outcomethe future financial well-being of the children being less dependent on the good will of their parentsis achieved by circumscribing the parents' possibilities for exit. 5 Though they could seek to emotionally punish their adult child in various ways, and though they can decide never to speak with them again, the policy deprives them of the option to disinherit an adult child they wish to separate themselves from (or even to leave them a reduced share). Should they wish to ensure that no money or assets are passed down a child they wish to break with, they would need to either spend down their assets to nothing (if they donate it in a way that makes it looks like they are distributing the assets in their estate to pre-empt it being distributed through an inheritance, the child would be able to demand the assets be included in the estate to be divided after their passing), or they would need to move to a different country. The costs for the parents of a complete exit, in other words, are made higher than they would be otherwise through the laws regulating inheritances.
Though perhaps unusual in the context of policies regulating the relationship between family members, the basic structure of the policyof exit costs being treated as instrument that needs to be adjusted to account for structural disparities in poweris common in other policy areas. It is not uncommon, for instance, for the relationship between employers and employees to be structured in this fashion. Though it varies by country and precise occupation, civil servants and certain other types of employees can in many countries only be fired for cause, for instancesomething that is done, at least in part, to shield the employees from the potentially dominating influence of a powerful employer. This underscores why it is necessary to take into account the relations of power between the parties involved when adjusting the exit-cost policy lever.

The Relationship between Exit and Voice in Relation to Non-domination
As has already been alluded to, the relationship between exit, voice and non-domination is complex. Though exit, voice and non-domination can be mutually reinforcing under some circumstances (Pettit 1997, Shapiro 1999, 2012: Warren 2011, there are also conditions under which the aim of furthering non-domination through a strengthening of the voice mechanismthrough the democratization of relationshipscan come into tension with the aim of furthering non-domination through a low-exit cost approach. Pettit (1997,2014) and Shapiro (1999Shapiro ( , 2012, among other theorists of non-domination, have advocated for reforms that facilitate inclusive participation in the decisions by which people are affected as a means of freeing people from arbitrary power. In Shapiro's words: "I maintain that every domain of human interaction should be subject to democratic conditioning constraints. These vary with time and circumstance, but they always include mechanisms to participate in decision making about the nature of the goods in question and rights of opposition to try to get them changed" (2012, p. 294). As Taylor (2017) has noted, this emphasis on democracy and participationon the voice mechanismcan be seen both in the domain of politics as well as in the economic sphere. Thus, just as Pettit and others have advocated for political reforms that would increase participation and reduce the influence of lobbyists and corporate interests, so in the economic sphere they have advocated for reforms that would give workers more voice and more say in the decisions that affect themby, for instance, making it easier to form unions and force employers to collectively bargain, or by giving union/worker representatives seats on corporate boards. There is, thus, no doubt a lot of merit to the claim that the democratization of different domains of human interaction could serve to free people from arbitrary power, as the granting of voice is one of the mechanisms through which people can be freed from domination.
When it comes to the relationship between exit and voice, exit can serve to strengthen voice and therefore also contribute to the democratization of various relationshipsincluding to democratization of the relationship between spouses and to the relationship between employers and employees (Warren 2011;Hobson 1990). Indeed, where there is no possibility of exit, the likely effectiveness of voice as a mechanism for bringing about change and ensuring that's one's interests are given consideration risks being greatly reduced.
But, under certain circumstances, exit and voice can come into tension. This can be the case, for instance, when the availability of low-cost exit undermines the utility of voice as a mechanism for influencing the views and decisions of another party (see Hirschman 1970). This means that, under certain circumstances, the kind of independence low-exit costs can generate (vis-à-vis a spouse or an employer) may serve to undermine the democratic principle that people should have a say over and a right to contest decisions that affect themthe right to unilaterally exit a relationship, after all, is right not have that decision contingent on the consent or the acquiescence of the other party in the relationship. 6 Working through the relationship between exit and non-domination thus helps us see that a policy approach that focuses primarily on voice as a mechanism for furthering the value of non-domination has some limits. Under certain circumstances, facilitating low-cost exit will undermine voice. In these cases, in effect, policy makers face a choice between two policy alternatives: (1) setting the exit costs low so as to facilitate relationships of mutual independence, or (2) making the exit costs somewhat higher, thereby increasing the level of mutual dependence, and facilitating the use of voice.
To illustrate, consider the example of laws regulating divorce. Countries face the question of how easy or difficult they should make it to get a divorce. If the aim is to ensure that people have a say over decisions that affect themif the aim, in other words, is to democratize the familial relationshipthen it would seem to follow that one would not want divorce to be so easy that it deprives an aggrieved spouse of the opportunity to have their say and to contest the decision to dissolve the marriage. This could be achieved, for instance, by limiting the conditions under which a spouse can file for a no-fault divorce or by mandating a waiting period before a divorce can be finalized. The goal of such policies would not be to make exit impossible. Indeed, where there is no possibility for exit, voice loses some of its impact, as Hirschman (1970) also so insightfully showed (see also Hobson 1990). This is no doubt especially true in the context of a spousal relationship. The goal, rather, would be to keep the exit-costs high enough so that there are opportunities for the mechanisms of voice and contestation to operate. On its face, this type of policy approach would seem to be consistent with the goal of democratizing the marital relationshipperhaps more so than a policy regime that sought to lower the exit-costs as much as possibleand it also seems consistent with an approach to non-domination that emphasizes inclusion and voice under conditions of mutual dependence.
Alternatively, the state could make it extremely easy to get a divorce, enabling either party to file for a no-fault divorce, with no mandated waiting period, no rights of contestation for the other party, and with no attached obligations to pay alimony or child support for either party. Under conditions where neither party is too dependent on the other for their future well-being, such a policy choice would likely further non-domination. But it would arguably come, at least to some extent, at the expense of the kinds of democratic constrains Shapiro (2012) outlinesproviding no formal rights of contestation or appeal. 7 One could perhaps argue that within a spousal relationship, opportunities to exercise voice and to participate in decision-making are always present, at least to some extent. And it is also true, as I have already noted, that where there is no possibility for exit voice risks losing much of its impact as well. But it would nevertheless be a mistake to imagine that nothing is at stake in the policy choice just outlined. If the democratization of familial relationships really is the goal, then there really would be reasons to strengthen voice even if it meant making exit more difficult. This is not, to clarify, I would suggest, a reason to move away from no-fault divorce or to otherwise raise the costs of exiting a marital relationship (in a context where independent sources of financial support and opportunity exist). A (mutually) low-exit cost, mutual independence, policy regime seems preferable to one that prioritises the democratization of the marital relationshipfor reasons I outline in the next section. But, what the above example illustrates is that, in certain contexts, policymakers in effect have to choose between different models for how to try to create conditions conducive to relationships of non-domination, where exit, voice, and democratization are made to play slightly different roles.

Principles for Adjusting Exit-Costs
If the above discussion is correct, the relationship between exit and non-domination is complicated. Depending on the context, exit-costs may need to be lowered or raised to further the value of non-domination, and in certain contexts policymakers face a trade-off between exit-based and voice-based policy mechanisms. Notwithstanding this somewhat complicated picture, there are a number of conclusions that can be drawn regarding the relationship between exit and non-domination that are worth highlighting, which also speak to how the trade-offs that have been identified can be approached and negotiated from a non-domination standpoint.
(1) As a rule of thumb, lowering exit-costs furthers the value of non-domination. It does this by making people less dependent on the good will of the other party (or parties) in the relationship. Low-exit costs can also empower people within a relationship, strengthening their voice and generating a greater degree of structural equality between the parties. (2) Exit-costs need to adjusted, however, to account for asymmetries in the relations of power between employers and employees, the state and the individual, and between parents and children. In these situations, as a general rule, the exit-costs for the more powerful party should be higher than for the party in a more vulnerable position.
(3) In cases where exit and voice conflict, exit-costs should be made low (even at the expense of voice) when the relevant liberty interest depends on being able to change the status quo. Processes that provide for collective decision-making, or that provide rights of contestation, would risk privileging the status quo and therefore jeopardizing the relevant liberty interest. This would imply, for example, generally speaking, that exitcosts should be made low (even if it comes at the expense of voice) when it comes to laws regulating divorce. (4) When the liberty interest depends on being able to maintain the status quo, exit-costs should be set at a higher level and voice should be given primacy. This would apply, for instance, in contexts where independent sources of support and opportunity are lackingsuch as a sufficiently generous social insurance system, a functioning labor market, etc. In these types of situations, the availability of easy exit would privilege the relatively more powerful parties and render those lacking independent resources vulnerable to domination. Exit in these contexts should not be made impossible; rather, exit costs should be adjusted in way sufficient to place the relevant liberty interest on a more secure footing. 8

Conclusion
Whether lowering exit-costs furthers non-domination or not depends both on the background conditionswhether there is a generous social insurance system, whether there are opportunities for employment, etc.as well as on the existing power dynamics within the relevant relationship (be it between an employer and employee, the state and the individual, or between family members). Creating conditions that are conducive to non-domination is thus not easy, and the decisions policymakers face are likely to be complicated and to require careful 8 Whether it and in what ways it makes sense to raise exit costs in these types of contexts will also depend on a range of other factors. Indeed, it would be important to ensure that the policy cure is not worse than the diseaseby rendering the parties dependent on the arbitrary good will of third-party actors (a judge tasked with approving a filing for a divorce, for instance)or, indeed, that it ends up being counterproductive by limiting the opportunities for exit in such a way that it substantially reduces the effectiveness of voice in a context where the democratization of the relationship is essential. In certain contexts, the distributional effects of a policy that seeks to set a relatively high bar for exit should also be considered. If the result of the policy, for instance, is that those with sufficient financial resources are de facto still able to exit easily while those lacking financial resources are effectively barred from doing so, it may make more sense to lower exit costs across the board and use other policy mechanisms to further the relevant liberty interest (see Taylor 2017).
consideration of the relevant facts as well as other moral and value-laden considerations and trade-offs. But as the above discussion shows, there are strong reasons to focus on how different policy alternatives and policy regimes affect the exit-costs people face vis-à-vis employers, family members, and indeed their political community. Lowering exit-costs for people that are in a vulnerable position can be an effective way of reducing their dependence on the good will of employers, spouses, or family members. Yet, there are trade-offs and tensions that it is important to be cognizant of. Because power is a relational good, the most effective way of freeing people from dependence on arbitrary power will in some cases be to raise the exit-costs for the more powerful agent in the relationship. This may have an inhibiting effect on the more powerful agent, and indeed it may make them more dependent on the other party. But as long as they are not put in a position where they are subject to domination, the raised exit-costs will have had the effect of furthering relationships of non-domination. Similarly, while opportunities for exit strengthens the utility of voice in many contexts, low exit-costs can also undermine the utility of voice. In these cases, policymakers in effect must choose between two models, or policy regimes, for how to further non-domination, where exit and voice are made to play different rolesa choice that should be guided by the nature of the relevant liberty interest at stake.
Funding Open access funding provided by Örebro University.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.