1 Introduction

Competition is an indispensable tool for dealing with human ignorance. Individuals are not born with knowledge of the lowest cost means of achieving their goals, the best theories to explain the world around us, or the best way to govern human interaction so as to avoid conflict and promote cooperation. We can err. And in a world of scarcity, errors crowd out alternatives. Societies thrive when individuals are free to undertake many experiments in cooperation, and failed experiments are shut down in a peaceful and orderly manner. F.A. Hayek (1968) refers to this sort of process as competition. Both the freedom to experiment and responsibility for failure are necessary for innovation and progress. Experimentation allows us to discover new ways of doing things, but we must also stand ready to abandon existing approaches.

Competition also enables adaptation. Even if a static state was desirable, it is simply not achievable given that environmental conditions, preferences, and ideas can always change. This is why Hayek (1960) always focused on the struggle between the forces of progress and the forces of decline. Stasis is not a viable option, even if we do not value progress for its own sake. We can only dispense with peaceful forms of competition when we already know the best way of doing things, and the best way of doing things often changes.

While Hayek developed his theory of competition with regard to the production and exchange of goods and services, he later came to apply it to the rules that govern market exchange (Stringham and Zywicki 2011). Following Leoni (1961), Hayek (1973) argues that a long process of legal evolution allowed the West to stumble into a set of legal principles and practices that secured extensive spheres of individual control within which individuals are free to choose. Judges tried out different legal principles as methods for resolving disputes, learning from one another in a decentralized fashion. This accidental emergence of liberal institutions made individuals free to act on their own knowledge and ideas—especially those that were at odds with existing practices—and thus allowed for the West to thrive. Later, Hayek (1988) made competition over rules a centerpiece of his theory of social evolution more generally. Societies that adopted rules of private property and freedom of contract prospered and grew, supplanting societies that relied on rules of common property.

This essay extends Hayek’s analysis of institutions by focusing on a particular aspect of rules: jurisdictional boundaries. To whom should a given set of rules apply? I focus on why this question matters for classical liberals, though many of the arguments below are relevant for a wide range of political theories. Good boundaries—those that enable innovation and adaptation—are in important respects like good methods of production or good scientific theories: they are best discovered through a competitive process. Section 2 introduces the concept of jurisdictional boundary problems, arguing that such problems (a) are ubiquitous, afflicting all sorts of governance arrangements and (b) are subject to the same sorts of knowledge problems that make competition so valuable for directing economic activity.

Edward Stringham’s book Private Governance is especially relevant for raising the issue of how to effectively draw boundaries. Stringham focuses not on the boundaries of government authority but on the formation of private clubs as a source of governance. Clubs too have boundaries, depending on both who is admitted into membership and what being a member entails. Though he is not primarily concerned with boundary problems, Stringham explores a variety of historical cases in which club governance proved to be an effective and peaceful method for adjusting boundaries on the margin. Clubs work well primarily because individuals consent to join and cheaters can be expelled.

But Chapter 13 of Private Governance takes a different turn, in two respects: its subject and its mechanism. Based on Stringham’s work with Zywicki (op. cit.), it offers a Hayekian argument for market-made law. They utilize the Hayekian idea of competition as a discovery procedure to put forward a case for Hayekian anarchism. So rather than relying on the mechanisms of consent and exclusion, Chapter 13 relies on a decentralized process of social learning. This is at least a difference in emphasis on what mechanisms make private law work well. As far as subject goes, the sort of law discussed in Chapter 13 is different from the governance provided by clubs that the rest of the book discusses in at least one important respect: it is meant to adjudicate boundaries themselves and not merely to establish or renegotiate them. As a result, it more directly addresses the problem of boundary conditions.

While both parts of Stringham’s work offer important insights, I argue that the competitive approach taken in Chapter 13 is especially underappreciated. Section 3 distinguishes between two criteria that classical liberals have traditionally used for evaluating boundaries: ex ante consent to jurisdictional boundaries or the ability to exit from them ex post. Both criteria emphasize individual choice, but they come at it from opposite directions. The consent approach evaluates institutional arrangements according to whether individuals have agreed to them ex ante, while the exit approach evaluates institutions according to whether individuals have an exit option. I then make a case that exit-oriented liberalism is underappreciated, especially in its capacity to deal with boundary problems by enabling a competitive discovery procedure.

2 Boundary problems and the irreducible commons

James Buchanan (1975, p. 5) argues that anarchy is ideal but unworkable. Anarchists fail to adequately address a fundamental question: “What are to be the defining limits on individual freedom of behavior?” (ibid., emphasis added). While “ordered anarchy” (p. 9) governs most human interaction, it operates only to the extent that boundary problems about the allocation of decision rights—especially property rights—are settled in an acceptable fashion. “The logical foundation of property lies precisely in this universal need for boundaries between ‘mine and thine’” (p. 13). It should be noted Buchanan has an extensive definition of property rights in mind. Property includes not only concerns bare possession but the uses to which individuals may put the claims that they have. He even extends it to include the definition of legal personhood, from slaves with minimal legal standing to masters with expansive personal spheres that include extensive claims on others. While recognizing that there is wide scope for reciprocal, spontaneous recognition of such boundaries, Buchanan contends that the state is necessary to prevent marginal defections from those accepted limits from unraveling such a tenuous social contract.

Buchanan expresses sympathy for the market anarchist theories of both Murray Rothbard (1973) and Freidman (1973). But he argues that neither successfully addresses these boundary problems (Buchanan 1974; Buchanan 1975, p. 9, footnote 2). Rothbard (1973, p. 37–53) assumes, for instance, that a libertarian legal code predicated on a radical Lockean approach to natural rights is firmly in place before arguing that competing firms could provide effective law and order. What he fails to explain is the process by which such a departure from existing property practices can initially gain a foothold in either his preferred system or the present one. Friedman (Freidman 1973, Ch. 31) admits more readily that market-based law is not necessarily libertarian law, as individual beliefs and preferences matter a great deal in determining what sorts of rights will ultimately get enforced.

Wagner (2007) develops a framework that helps to extend Buchanan’s point. He argues that private property and common property are two different grammars by which human beings relate to one another. The language of private property–“mind and thine”–takes hold to the extent that we take others’ decision rights as given. But the grammar of private property can never fully encapsulate all social interaction, for two reasons. First, the boundaries of private property are always contestable (p. 47). Private property only ever extends to the limits of others’ forbearance. At any given point in time some property claims will be fixed, but individuals often seek to limit or compel the uses to which others may put their property, or to engage in redistribution of decision rights. All of these activities either shift private property into the commons or change who has a given property claim. These activities necessarily invoke the grammar of common property, of “we together.” Second, interaction structured by the grammar of private property always occurs through a medium of common property. In order to approach you with an offer to exchange privately held decision rights, I must be able to initiate contact. Gaining your permission requires that, at some level, the right to seek your permission is held in common.

The boundaries of private property are always worked out in common property, and the rules of common property denote a group of individuals who share governance arrangements. Classical liberals are well versed in the challenges confronting common property arrangements. They often entail a lack of incentives to use resources effectively, and since they are non-severable they do not enable economic calculation. Theoretically and historically, clubs have offered at least a partial solution to both of these problems (Buchanan 1965). Stringham (2015) documents a number of examples of clubs that have effectively provided governance services for their members, including early financial and insurance markets. Clubs function by extending the scope of private property, providing effective and voluntary means for alleviating commons problems when fully individual claims are not functional. But fencing in common property only alleviates commons problems for that which has been fenced in. Clubs push out the boundaries of common property but do not eliminate boundaries altogether. Whether between individual property owners or club providers of governance problems, any non-monolithic society confronts some irreducible legal commons. So while clubs play an important role in dealing with boundary problems, there are still boundary problems between clubs.

The existence of an irreducible legal commons means that ‘privatize everything’ is not a viable approach to securing a liberal order. But this does not mean that we should uncritically accept Buchanan’s preferred approach of a constitutionally constrained state. Even if the best option for dealing with boundary problems is to establish a monopoly referee with incentives to be productive and protective but not predatory, this option is too abstract to be of any real help. Advocating for this strictly formal condition is like saying that we want goods created through low cost methods of production: it tells us nothing about what those methods of production are.

One important consideration that Buchanan’s formal proposal does not tell us is: what are the appropriate boundaries for a governance jurisdiction itself? The reason that a system of club governance does not completely eliminate commons problems is that there are forms of activity that fall outside of its jurisdiction. But the same is true of any liberal form of public governance as well. Only a totalitarian, socialist state that was completely insulated from other jurisdictions would not confront its own boundary problems.

Discovering good jurisdictional boundaries is a problem of overcoming ignorance that confronts both public and private forms of governance. Many factors determine whether a boundary is well drawn. Most importantly, a good boundary is liable to be different for different sorts of governance problems. Governance problems are shaped by technological, organizational, and social factors. Technologically, collective goods have different economies of scale. Spillover effects vary radically in size. Organizational and decision-making costs vary both from one community to the next and for different collective goods problems. Socially, an area with a heterogeneous population is likely better served by smaller jurisdictions than one with a homogenous population. As a consequence of these sources of variation, the efficient jurisdictional boundary for, say, the provision of police services is unlikely to be the same as for governing groundwater usage. Even within the narrower domain of police services, different neighborhoods, industries, and law enforcement functions have varying needs and varying economies of scale. Ostrom (1976), for example, notes that forensic services are liable to have much larger scale economies than regular police patrols. The effective provision of governance is a multidimensional problem of balancing conflicting means and ends no less than the effective provision of private goods and services.

Moreover, boundaries themselves are multidimensional and varied. Some boundaries are best defined geographically, such as those governing a clearly defined common pool resource. Others may best be defined by function, such as the merchant law governing contracts among medieval traders (Benson 1989). Boundaries may be more or less porous to entry and exit. The right level of exclusion or inclusion will vary from one governance problem to the next. And some boundaries may need to be redrawn more frequently, while others can only work to the extent that they are stable. This leads to the most important feature of boundaries from a Hayekian point of view: change.

“Economic problems arises always and only in consequence of change” (Hayek 1945, p. 523). All the features that make for effective boundaries are liable to shift through time with both technological and social changes (Foldvary and Klein 2002). Safer fertilizers, for example, may decrease the area that needs to be monitored to protect a groundwater basin. More effective information technology, by contrast, may increase the efficient scale of police dispatch services. Communities may grow more or less homogenous, changing the costs of collective decision making. Hayek argues that market socialist proposals are flawed because they fail to grapple with the omnipresent reality of change. Similarly, it would be strange if there was a once and for all solution to boundary problems.

Providing effective governance requires more than thinking about abstract issues like externalities, collective action problems, and boundary problems, just as effectively coping with the economic problem requires more than thinking through the implications of a perfect competition model or economies of scale. In both cases, not only scientific knowledge but the more ephemeral knowledge of time and place are important. And in both cases, we can only assess the relevance and validity of such knowledge through some experimental process. In the provision of governance, just as in the provision of goods and services or the generation of scientific knowledge, societies prosper to the extent that there is wide scope for experimentation, and a mechanism for peacefully shutting down failing experiments. Hayek uses a single word to sum up these two conditions: competition.

My aim here is not to outline a specific model of what competition over jurisdictional boundaries might look like. At a minimum, competition would entail a form of competitive federalism. More radical proposals might look like Bruno Frey’s Functional Overlapping Competing Jurisdictions (Eichenberger and Frey 2006), a system of small city states, or full blown private governance as imagined by thinkers like Murray Rothbard and David Friedman. My aim is more modest: to establish that a commitment to such competition is a likely (if not necessary) concomitant of recognizing the existence of boundary problems. To the extent that boundary problems are a binding constraint, providing effective governance may require a commitment to a peaceful method for determining jurisdictional boundaries.

Competition over jurisdictional boundaries may sound utopian, especially among those whose fundamental vision of the world is Hobbesian. Peaceful competition among governance providers could readily, after all, cease to be peaceful. Indeed, this is always a possibility. But there is in the end only one alternative to peaceful competition: some combination of conflict and poorly defined jurisdictions. The prospects for peaceful competition over jurisdictional boundaries may be grim, but this would not be a rebuttal of my argument so much as a counsel of pessimism.

3 Consent and exit: Finding good boundaries

Classical liberals advocate widespread reliance on private property as a method for reducing commons problems, settling conflicts, and incentivizing cooperation. Whether they ground private property rights in natural law or social convention, there is more agreement than disagreement about what such property rights should entail. Different strands of liberalism, however, emphasize different ways of dealing with the establishment and maintenance of boundaries. Some liberals argue that what makes for a good boundary is that individuals have consented to it ex ante. Others argue that what matters is whether individuals can exit ex post.Footnote 1

Drawing a distinction between consent-focused and exit-focused classical liberals only makes sense if a few qualifications are kept in mind. The two orientations overlap to a very large extent, tending to differ in emphasis rather than in substantive judgments. Figure 1 depicts a range of positions from the most consent-focused to the most exit-focused. All of the positions in the middle are likely uncontroversial among most classical liberals. These judgments reflect the ordinary morality of market activity, which in most forms gives wide scope for both ex ante consent and ex post exit. Synchronic agreements, like transactions over hamburgers and haircuts, can be justified in terms of either consent or exit. Most liberals would also allow that some intertemporal exchanges that contain safeguards against reneging are also allowable, e.g., contracts involving an escrow account. Going the other direction, most would also likely rebut the enforcement of specific performance of labor contracts: one can make a laborer pay damages for reneging on a deal, but forcing them to do specific work amounts to servitude. But few classical liberals would hold the positions on either extreme of this continuum: that no contracts are binding or that all contracts (e.g., voluntary slave contracts) should be binding. So rather than two wholly separate approaches we have two tilts—one ex ante and one ex post—for how we understand what it means to be free to choose.

Fig. 1
figure 1

The Consent-Exit Spectrum.

Full size image

But where the middle cases above are largely uncontroversial among those who advocate robust private property rights, differences in tilt become more pronounced in addressing boundary problems. Thinkers with a consent tilt argue that good boundaries are those we have consented to ex ante. Thinkers with an exit tilt argue that good boundaries are those individuals can opt out of ex post. These two positions can be quite different when it comes to analyzing how good rules might come about and to whom they should apply.

Figure 2 divides some prominent classical liberal thinkers into those with a consent tilt and those with an exit tilt. This table is not meant to offer a nuanced reading of each thinkers’ position, as my aim is not to elaborate on the history of liberal thought. Rather, my goal is merely to illustrate consent- and exit-oriented arguments by appealing to salient examples. Every thinker listed above values both ex ante consent and ex post exit to a large extent, but iconic arguments from these thinkers tend to lean in one direction or the other. The table pairs thinkers who otherwise have similar commitments to some aspect or subset of the liberal program in order to illustrate that this distinction cuts across traditional boundaries such as consequentialism vs. deontology, minarchism vs. anarchism, and across different academic disciplines.

Fig. 2
figure 2

Examples of Consent-Tilt and Exit-Tilt Arguments.

Full size image

Consider the differences between Buchanan’s (1975) model of a unanimous social contract and Kukathas’s model of a diverse liberal archipelago (Kukathas 2007). For Buchanan (1962), the only statements of value that count are those expressed in voluntary choices. Economics becomes a tool of moral philosophy by allowing the economist to propose changes in the rules of the political game by which boundaries are set. Proposals are justified to the extent that they approach unanimous approval. For Kukathas, the fundamental freedom is freedom of conscience, which is cashed out in social life by freedom of association and, especially, of disassociation. All associations are partial, so none can claim the absolute allegiance of an individual. Life under existing rules is only sometimes a matter of consent and more often of mere acquiescence: we do not try to leave or are only willing to bear so much cost to change them. But acquiescence is acceptable as long as individuals have robust exit rights. Kukathas goes so far as to claim that a society of competing but restrictive monastic orders is freer than a unified polis governed by those concerned with engendering individual autonomy. But also note that neither thinker is single-minded. Buchanan recognizes the importance of federalism and club governance, implying an appreciation for exit. Kukathas, likewise, values both rights of association and disassociation. But when considering tough cases about the essence of liberty, Buchanan’s argument leans toward ex ante consent while Kukathas’s approach leans towards ex post exit.

Jacob Levy (2015) offers an alternative but in some ways overlapping distinction between strains of the liberal tradition, which he dubs rationalism and pluralism. Rationalist liberals believe in the importance of a uniform legal code administered by a centralized state aimed at protecting individual liberty. Pluralist liberals place a heavier emphasis on the importance of intermediate associations, including both private clubs and local government jurisdictions. Obviously this latter approach has much in common with an exit-focused approach to liberalism, since both entail a commitment to polycentric governance.Footnote 2 But since both centralized states and local groups can be sources of domination or tyranny, Levy argues that we should not try to purge one or the other strand of liberalism and must learn to live with the tension between these two approaches. He also argues that the pluralist strand of liberalism is relatively neglected. Exit-tilt liberalism, I argue, is likewise unduly neglected, especially when it comes to dealing with boundary problems.

Exit may take a variety of forms, and it is not my goal to identify when one or another form is most appropriate. Migration changes who is subject to a law, as does geographic secession or the fragmentation of existing authority along functional lines. But exit need not take such radical forms. Benson (1999) argues that contracting provides an important mechanism for bypassing existing default rules. The forms of exit are also affected by the available forms of entry into the provision of governance. What mix of these forms of exit is most conducive to solving particular governance problems is part of what we want a competitive process to discover. Again, accepting an exit-tilt version of liberalism is not the same as arguing for an absolute right of exit. The optimal porousness of a jurisdictional boundary is one of the very things we wish to discover. So the right of exit may not be uniform from one jurisdiction to the next.Footnote 3

Exit is undervalued because, when it comes to dealing with boundary problems, it facilitates social learning to a greater extent than does ex ante consent. Both innovation and adaptation require dissent from existing ways of doing things. The fact that a jurisdictional boundary is agreed to at time t will not make it appropriate at time t + 1. Institutional entrepreneurs may discover better ways of carving up the social world, or technological conditions may make existing arrangements less functional or just. Exit has several advantages over ex ante consent in dealing with these issues. First, exit lowers the costs of agreement by allowing those with distinct preferences to go their own way. Lower cost dissent means more experimentation with rules, including jurisdictional boundaries.Footnote 4

Second, exit works from within the status quo. The world we inhabit is not governed by institutions founded on consistently liberal principles or consented to by all parties. Consent-focused accounts, in order to genuinely endorse institutional change, have to either use unanimity as a yardstick or imagine that rights are distributed correctly to begin with (Buchanan 1962). Consent-tilt liberals thus have a difficult time dealing with objections to pro-liberty proposals rooted in the reality of historical injustice (Lomasky 2005). Exit has the advantage of being empowering from the present, not only if it was applied in the past. The overwhelming barrier to human liberty and flourishing in the modern world is defunct institutional arrangements that were founded upon and continue to facilitate predation (Acemoglu and Robinson 2012). The dead hand of the past mostly impairs life prospects through the endurance of bad rules. So while exit does not address all distributive concerns of past injustice, the ability to leave is the most valuable for precisely those most afflicted by oppressive institutions (Martin 2015). Placing more emphasis on exit gives liberals an important tool in addressing historical injustice. At the same time, an exit tilt allows classical liberals to maintain a strong stance against existing government transfer programs, because the right to opt out can take normative priority over the expectations that such programs have established for their beneficiaries.

Third, exit-tilt liberalism avoids reliance on hypothetical consent. Since unanimous agreement is not forthcoming regarding most real world institutional changes, a standard philosophical move for consent-tilt liberals is to argue about what individuals would agree to under certain idealized conditions. At the same time, a standard complaint about exit as a mechanism for allowing individuals to choose rules is that it can be very costly, especially when it involves migration. Certain forms of exit are costly, but this can in fact be an advantage for certain sorts of choices. Individuals who can engage in cheap talk can easily hold up beneficial changes in governance arrangements (Kuran 1995; Buchanan and Tullock 1962). Making choice costly elicits genuine preferences. On the other hand, many forms of exit—such as the sort of contracting around default rules that Benson (2011) describes—are not only much less costly than collective choices, but also generate genuine unanimity among the contracting parties.

Finally, an exit-tilt allows liberals to more effectively counter stock charges of atomism that various thinker level against liberalism. Communities can matter a great deal, even unchosen communities like families and religious groups. Many social practices are not exactly coercively imposed but nor are they actively consented to. An exit-tilt liberalism does not need to posit humans that arrive to the social world as premade, autonomous individuals, and to affirm only governance arrangements that such individuals would choose. Rather, it only needs to insist that groups have value only insofar as individuals are willing to stay a part of them. Exit-tilt liberalism can thus allow classical liberals to make an important contribution to debates about multiculturalism that are typically carried between competing camps of progressives (Tebble 2016).

4 Conclusion

In an address to a Cambridge University conference commemorating the 75th anniversary of the publication of Keynes’ General Theory, Paul Krugman distinguishes between “Chapter 12ers” and “Book 1ers” (Krugman 2011). “Chapter 12ers” are those who like Keynes’s philosophical ruminations regarding expectations about the future and beauty contests. “Book 1ers” are those who have more appreciation for the formal models meant to demonstrate the possibility of a shortfall in aggregate demand. We might similarly distinguish two readers of Stringham’s Private Governance. “Part 1’ers” find the work on club governance—which takes a consent-tilt approach focused on contracts and exclusion—most compelling. “Chapter 13ers” find the arguments about Hayekian anarchism—taking an exit-tilt approach focusing on competition and social learning—most compelling. And while there is a great deal of insight in both, I find myself drawn to Chapter 13.

Boundary problems are among the most difficult for any political philosophy to deal with. Buchanan (1975) builds an entire book on them, and it is among his most pessimistic works. Nozick, even after assuming a robust set of individual rights at the outset of Anarchy, State, and Utopia, dedicates a long and technical chapter to the problem of “border crossings.” (Nozick 1974, Ch. 4) And Kukathas, who offers what is arguably the seminal defense of the right of exit, leaves open the question of what I get to take with me when I leave a group (2003, p. 37). I have argued that these may be better thought of as matters of institutional design rather than universal philosophical principles.

The answer to the question—“what makes a good boundary?”—is contingent on the circumstances of time and place. The best boundaries, then, are those that have been discovered through a competitive process and are subject to some degree of challenge to ensure that they continue working. Classical liberals have often emphasized the role of consent in evaluating the institutional processes that generate boundaries. I have argued that an alternative strain of classical liberalism—one that emphasizes ex post exit—offers more promise. Most importantly, exit does a better job facilitating and capitalizing on social learning, allowing for both adaptation and innovation of jurisdictional boundaries.

Despite these advantages, however, it is worth keeping in mind that a tilt towards ex post exit should remain just that: a tilt. There are some disadvantages from an exit-focused approach to evaluating institutions. First, there remains the thorny issue of when individuals should be able to enter into binding agreements, including those involving jurisdictional boundaries. The optimal number of such binding agreements is far from zero. A social order in which criminals can exit from a jurisdiction’s authority after committing a crime is simply not functional. But these qualifications are no different than the general recognition that no moral principle or right is absolute and unconditional, especially when it comes to questions of institutional design and evaluation. Second, exit-tilt liberalism leaves ambiguous the dividing line between voluntary association and local government. The ability to exit a small municipality may be no less than the ability to exit a neighborhood association, even if one resulted from conquest and another through contract.

Exit does not tell us everything about whether a state of affairs counts as voluntary or not. Exit-tilt liberals tend to be more concerned with whether present governance arrangements are polycentric and competitive than with the history of how they came about, which will leave some questions unanswered. Just as we must live with a tension between rationalism and pluralism, then, classical liberals must learn to live with conflicting commitments to ex ante consent and ex post exit, even though I contend that the advantages of exit have typically gone underappreciated. Exit should play a more central role in the body of liberal thought.