Dani Rodrik writes that “The problem with self-enforcing agreements is that they do not scale up.” I want to make a few interrelated points about this, as it often is the “market failure” argument given to justify a state when I confront people with evidence that what Pete Leeson has referred to as anarchy and Dani calls self-enforcing agreements actually can and do create order. Let me say that my own preferred terminology here is “customary law communities” (bottom-up development of rules and institutions as opposed to top-down imposition of rules and institutions under “authoritarian law”). These communities do not just rely on self-enforcement in the sense of unilateral (morality-driven) or bilateral (e.g. tit-for-tat) actions. They often establish third party dispute resolution mechanisms (arbitration, mediation) backed by ostracism threats, and other institutions.
The first point I want to make is that while these communities may be based on geographic proximity, they also may be based on kinship, functional proximity as in a trade association or the “business community,” religion, or any of a number of factors that create repeated dealings and/or reputation effects). Second, a group does not necessarily have to expand geographically to expand opportunities for beneficial interaction. Indeed, if individuals want to interact, but only on some dimensions, or if they want to maintain different sets of rules for different dimensions of interaction, then parallel “localized” mutual support groups may be maintained while a “second order of clustering” is established, facilitating a relatively limited scope for interaction.
A group whose members insist on strictly imposing their own morality and penalties on outsiders would probably be unable to initiate beneficial inter-group interaction. Thus, if people wish to simultaneously facilitate inter-group interaction and impose rules that differ substantially from the norm in other groups, they have strong incentives to inform outsiders of the differences in order to avoid conflict and minimize the difficulty of maintaining non-standard laws. Part of the reciprocal agreements with other groups may be the explicit recognition of differences in laws and procedures for treating conflicts. This in turn implies that as inter-group interactions expand, a hierarchical jurisdictional arrangement may arise. For example, each localized (geographically, functionally) group may have jurisdiction over rules for and disputes between its members. Rules for inter-group interactions can differ from both groups’ internal rules, although they certainly do not have to [members of customary legal systems have incentives to imitate desirable institutions and rules developed elsewhere, so competition and emulation lead to standardization of many rules and institutions across similarly functioning groups, although differences may remain, reflecting preferences of various groups’ members], and disputes between members of a confederation of different groups are settled by some “higher” confederation level adjudication process. Note that these are not “higher courts” where disputes can be appealed from within-group dispute resolution mechanisms (e.g., they are not a supreme authority). Rather, this is a jurisdictional hierarchy defining the role of each adjudication process and allowing for increasingly more distant interactions. This allows for differences between the law applied within groups and between groups; a monopoly in law is not required.
A judgment involving an inter-group dispute will have to be considered to be a fair one by members of both groups, of course. Thus, an equal number of individuals from each group might serve as an arbitration board, or a mutually acceptable third party (i.e., an arbitrator or mediator with a reputation for good judgment) might be chosen. This provides another reason for the tendency toward standardization of rules across parallel groups with similar functions, at least for those functions carried out in the process of inter-group interactions.
Some individual members of each group must recognize the potential benefits of inter-group interaction and be willing to bear the cost of initiating institutional innovations. Furthermore, the resulting innovations must involve more than just dispute resolution, because such interaction faces an assurance problem. Individuals must feel confident that someone from the other group will not be able to renege on a promise and then escape to the protection of that other group. After all, at least initially, repeated game and reputation affects are localized within each group, and there is limited potential for a boycott sanction. Thus, for second order clustering to develop, some sort of inter-group insurance or bonding arrangement becomes desirable, along with an apparatus for inter-group dispute resolution. For instance, as inter-group interaction develops the mutual support group can become a surety group as well. Membership in a group then serves as a signal of reputable behavior, and if a member of a group cannot or will not pay off a debt to someone from the other group, the debtor’s group will. The individual then owes his own group members so the boycott threat comes into play once again.
Limits to the extent of an inter-group network of cooperation are determined by the relative costs and benefits of information about other groups and their legal systems. The costs of establishing inter-group institutional arrangement depend in part on how “distant” the groups are from one another, where distance can be in terms of geographic space, or in terms of the behavioral norms that are relevant to the groups. Thus, extensive interaction between starkly different groups may not arise. However, these limits are stretched as individuals become members of several groups. After all, as Mises (1957: 257) explains, “Man is not the member of one group only and does not appear on the scene of human affairs solely in the role of a member of one definite group…. The conflict of groups is not a conflict between neatly integrated herds of men. It is a conflict between various concerns in the minds of individuals.” Thus, the relatively limited jurisdictions of some customary communities are not as constraining as they might appear to be.
A person may simultaneously belong to many groups that have well established customs (and be subjects to the commands of several rule-making authorities, e.g., as in a formal federalist system of government), so being in one community does not preclude dealing with people in other communities. A person may belong to a trade association, a homeowners association, a religious group, a fraternal organization, and so on, for instance, each with its own rules and governance institutions. The membership of all of these communities can differ, although considerable overlap may also occur, so individuals may deal with other individuals on some dimensions but not on all dimensions. Indeed, in any complex society, there are many distinguishable systems of rules and institutions, and yet people from many of these different systems interact regularly without having to call upon any legal authority. Thus, inter-group cooperation appears to be the norm rather than the exception, and it appears to be quite widespread. And with good reason: as Gluckman suggests, “multiple membership of diverse groups and in diverse relationships is … the basis of internal cohesion in any society.” An all inclusive legal system would eliminate the benefits of competition and emulation and undermine the incentives for innovation, as Berman explains:
It is this plurality of jurisdictions and legal systems that makes the supremacy of law both necessary and possible…. The very complexity of a common legal order containing diverse legal systems contributes to legal sophistication. Which court has jurisdiction? Which law is applicable? How are legal differences to be reconciled? …. The pluralism of … law, … has been, or once was, a source of development, or growth — legal growth as well as political and economic growth. It also has been, or once was, a source of freedom. 
The competitive/cooperative relationship between consensual customary legal systems is driven by the desire to facilitate voluntary mutually-beneficial interactions rather than a desire for legal sovereignty. Thus, many different customary systems can co-exist and interact. An understanding of customary law requires that individuals and their organizations be the points of reference rather than “society” as a whole:
there may then be found utterly and radically different bodies of “law” prevailing among these small units, and generalization concerning what happens in `the’ family or in `this type of association’ made on the society’s level will have its dangers. The total picture of law-stuff in any society includes along with the Great Law-stuff of the Whole, the sublaw-stuff or bylaw-stuff of the lesser working units” (Llewellyn and Hoebel 1961: 28). 
Customary law can be geographically extensive and functionally decentralized (i.e., specialized), in contrast to the law of geographically defined states which tends to be functionally centralized and geographically constrained. Thus, customary law can have different sized jurisdictions for different functions. In some areas of law, economies may be considerably more limited than any state, for instance, so existing political entities are too large geographically [e.g., this applies for many aspects of criminal law] or functionally [e.g., many aspects of domestic commerce may be most effectively governed by diverse trade associations rather than by the state].
In other areas of law, such as international commerce, some of these economies appear to be greater in geographic scope than any existing nation can encompass, although many also are narrow in functional scope, as international trade associations may be the most efficient source of rules and governance for many groups of traders. A customary system of polycentric law would appear to be much more likely to generate efficient sized jurisdictions for the various communities involved — perhaps many smaller than most nations, with others encompassing many of today’s political jurisdictions (e.g., as international commercial law does today). The desire to “scale up” (i.e., existence of economies of standardization in some rules) really provides an argument against state provision of law then, in order to break away from the inefficient artificial political restrictions that exist. That is, unless the contention is that states achieve the efficient scale? But if that is the case, why is it that some of the smallest states (e.g., Iceland, Ireland, Luxemberg, Switzerland) are among the most prosperous, along with some of the largest (e.g., the U.S.), and visa versa. The fact is that I can buy a product from India or Argentina without being part of the same customary community (or ruled by the same state) as the producer (that is, we do not have a common source of recourse) because I buy it from someone I deal with regularly or because I am aware of her reputation, who in turn purchased it from someone she deals with regularly or who has a reputation she is aware of, who in turn bought it from someone he deals with regularly or who has a reputation he is aware of, who in turn … ! It is not one large legal system that matters, but the effective linking of large numbers of parallel customary law communities (or ordered anarchies) with overlapping memberships.
*The following arguments are made in more detail in various publications, including, Benson, B. L. “The Spontaneous Evolution of Cyber Law: Norms, Property Rights, Contracting, Dispute Resolution, and Enforcement without State Involvement,” Journal of Law, Economics and Policy, 1 (Winter 2005): 269-348. “To Arbitrate or to Litigate: That is the Question,” European Journal of Law and Economics, Vol. 8, No. 2, September 1999, pages 91-151; “Polycentric Law Versus Monopolized Law: Implications from International Trade for the Potential Success of Emerging Markets,” Journal of Private Enterprise, 15 (Fall 1999): 36-66; “An Economic Theory of the Evolution of Governance and the Emergence of the State,” Review of Austrian Economics, 12 (November 1999): 131-160; “Knowledge, Trust, and Recourse: Imperfect Substitutes as Sources of Assurance in Emerging Economies,” Economic Affairs 21 (March 2001): 12-17. Therefore, I have omitted references to supporting evidence and theory here, except for a few direct quotes.
 Gluckman, M. (1955) The Judicial Process Among the Barotse of Northern Rhodesia, Manchester: University Press of the Rhodes-Livingston Institute, page 20.
 Berman, H. J. (1983). Law and Revolution: The Formation of Western Legal Tradition, Cambridge, MA: Harvard University Press, page 10. Berman’s “or once was” phrase recognizes that diverse legal systems are increasingly being subjugated by authoritarian legal systems. Indeed, while consensual legal arrangements tend to be characterized by internal stability, they face a significant external threat to stability. The size of consensual groups and second order clusters are constrained by transactions costs, and in many cases such organizations have been unable to resist takeover by groups cooperating in the production of violence. This issue is explored below.
 Llewellyn, K. N. and Hoebel, E. A. (1961) The Cheyenne Way, Norman: University of Oklahoma Press, page 28.