The rule of rules

Rules are central to the constitutional political economy (CPE) approach. On this approach, rules, of a variety of types and forms, are necessary for the emergence of a political and social order, so that all genuine political order is rule-based. The central role of rules within the CPE approach is examined starting from an explicit definitional discussion of the concept of a rule and including discussion of the nature of rule-following behavior, the supply of rules, and rule enforcement.


Introduction
The broad purpose of this paper is to present and discuss the concept of a rule and the associated idea of rule-based political order from the perspective of constitutional political economy (CPE). 1 The basic motivation is simply to pull together an explicit and integrated account of a variety of aspects of CPE's conceptual approach to rules, together with some critical discussion. It might seem that such an account should not be needed; CPE is now a relatively well-established approach and rules are central to that approach, so that we might expect that clear, explicit accounts of the CPE approach to rules would be easy to find. However, in much of the CPE literature fundamental aspects of rules often are taken for granted or dealt with implicitly or by means of analogies. For example, no explicit or detailed definitional discussion of what we might mean by a 'rule' seems to be anywhere in Buchanan's published work. Even in Brennan and Buchanan (1985), perhaps the most explicit and extended discussion of rules within mainstream CPE, constitutional, political and social rules are introduced necessarily narrowly self-interested actors. A third, normative, branch of CPE then seeks to provide a broadly contractarian account of the legitimacy of certain constitutional rules, and the actions and outcomes that arise under those rules. 4 The idea of a nested hierarchy of rules requires further elaboration, not least because it bears on the fundamental CPE distinction between the constitutional level and the "inperiod" or political level. An obvious question to ask is where the boundary between those two levels lies. Two extreme answers to that question can be given. On the one hand, we might identify a very narrowly defined constitution of the sort that might be laid out in the formal constitutional documents of a nation-state, with everything that is not specified in that narrowly drawn document located at the in-period or political level. Such a reading places most rules, including all laws and regulations enacted by the specified constitutional processes as well as less formal rules, at the political level along with individual behavior under the rules. Here, then, the major distinction is between constitutional rules and all other political and social rules and activity.
On the other hand, one might specify a maximally inclusive idea of the constitution, such that the global set of rules is seen as forming the constitution of the relevant society, so that the in-period or political level is identified with individual behavior. Here, then, the major distinction is between the realm of rules and the realm of behavior under the rules.
Both the distinctions between constitutional rules and other rules and between rules and behavior under the rules are important to CPE. The idea of a nested hierarchy of rules operates between those two extreme positions and attempts to preserve both distinctions. In such a nested hierarchy, rather than a simple distinction between the constitutional and the political, or a simple distinction between the realm of rules and the realm of behavior, we maintain a multi-level conception of society.
Consider, for example, an organization like a university: a CPE-inspired analysis of such an organization would recognize that it has a constitution-a set of founding rules that establish the organization and structure its basic form. Within that constitution, the university will promulgate a variety of other rules that regulate activity and the behavior of individuals associated with the university. In that way, one might distinguish between the constitutional level and the in-period level for the university. But at the same time, the university is nested within a broader social setting; operating within and under the laws and regulations of the jurisdictions (national and local) in which it operates, which in turn operate under the national constitution.
Many levels of nested organizations may emerge within a nation-state-with the national constitution (written or unwritten) as the all-encompassing set of rules at the top of the hierarchy. But seen from the perspective of an individual, the rules faced in any particular situation may sit at a variety of levels-some may be constitutional, some political or legal, some deriving from other organizations and some operating at an informal level.
In such a multi-layered nested hierarchy, we can maintain a clear distinction between rules (at whatever level) and individual behavior under the rules, while also recognizing the distinction between constitutional rules and other rules at each level.
The main body of this paper is divided into five further sections. The first four attempt to identify the CPE approach to key aspects of rules and their operation and provide some discussion of that approach. The next section begins with an explicit definition of a rule that attempts to formalize the CPE conception and provides a discussion of that definition and, therefore, of the role of rules within CPE. A linked discussion of the issue of rule-following in CPE is also provided. The next section then discusses the link between the idea of the rule of rules in the CPE tradition and the debate on the rule of law. A third section then addresses the issues associated with rule setting and the supply of rules, rather than the more familiar demand for rules. The fourth section then tackles the relationship between rules and enforcement. A final section offers some comments on the context within which CPE developed and, in particular, its connections with new institutional economics (NIE) and with the Rawlsian enterprise, before offering some final thoughts.

Rules and rule-following
In approaching a definitional discussion of what CPE means by a 'rule' we may take as a starting point Vanberg's (1994, p. 12) statement that, "A social norm or rule is a prescript for how generally to act in certain types of situations." which allows us to distinguish between the type of rules that we take as objects of study within CPE, along with other types of rules, such as those rules that merely are generalizations from empirical observation (as a rule, swallows fly south in the winter), or those that are analytical (the product rule in differential calculus). But Vanberg's definition leaves rather too much unsaid. As an expansion of that idea, I offer what I take to be a simple, explicit statement of Buchanan's implicit definition of a rule: a rule, in the political and social context, is an authoritative, normative influence on individual behavior operating across individuals, circumstance, and time.
Working through the elements of that definition in turn allows us to focus on the major characteristics of a rule from the CPE perspective and some of the issues raised. First, the definition is focused on the political and social context. That is, the primary domain of the CPE approach to rules is inter-personal in nature, 5 thus fitting with the general idea of seeing exchange, rather than choice, as the basic mode of human action. Exchange is an essentially interpersonal phenomenon. Setting the CPE approach to rules in the social domain focuses attention on the role of rules in resolving or dissolving potential inter-personal conflicts and dilemmas. 6 Second, a rule is authoritative both in the sense that it carries significant weight in influencing individual behavior and in the sense that its status derives from acknowledged authority. Here, a rule differs from mere advice or a request, either of which might suggest a course of action without having the authority required by a rule. A rule's authority may derive from any of a variety of sources (or combination of sources), including a moral code or principle, a recognized political or lawmaking process, or the judgement of a recognized magistrate or authority figure, and may operate through intrinsic or extrinsic motivations (or some combination of the two). 7 A key point of debate here relates to the extent 5 Of course, the general form of CPE's analysis of rules likewise may apply to rules that might be argued to be internal to single individuals, such as personal ethics and habits; see, for example, Buchanan (1979Buchanan ( , 1991a. 6 We will not here attempt to analyze the role of rules in relation to each of the wide variety of potential interpersonal conflicts and dilemmas. 7 For example, some rules may provide intrinsic motivation for some individuals, while the regularity of behavior generated by rules may provide extrinsic motivation for others. On the interplay between intrinsic and extrinsic motivations, see Kreps (1997).

3
to which the authority of a rule has to be widely acknowledged for it to count as a rule. If a putative rule has no (or almost no) recognized authority, can it really be a rule? At the other extreme, if we require full (or almost full) recognition of its authority as a defining condition of being a rule, we are limiting the idea of a rule to the class of widely accepted rules. In order to keep open the debate on the degree of acceptance or legitimacy of any rule, while avoiding the first extreme, we might insist that in order for a rule to be a rule at least many of the individuals who are to be regulated by that rule must recognize, at least to some degree, the authority of the rule, even if that recognition falls short of legitimation.
Third, a rule is normative in the simple sense that it is action-guiding. Action guiding rather than action determining; it is not automatically the case that any individual (still less, all individuals) will comply fully with any particular rule (still less, all rules), even if its authority is accepted (note the use of the word 'generally' in Vanberg's definition cited above). To abuse Buchanan's terminology slightly, a rule's normative force is a 'relatively absolute absolute'. 8 A rule has significant influence in determining behavior, and that influence may be deployed in a variety of ways, but however it is deployed, it is not absolute and so is not sufficient to ensure full compliance. It may be that some rules do generate full compliance as a matter of fact, but that is not a general or defining feature of rules.
Fourth, a rule is an influence on individual behavior; its influence may take either of two forms. First, rules may constrain by delimiting acceptable (that is, compliant) behavior. Whether a constraining rule is prescriptive or proscriptive in formulation, it essentially partitions the set of available behavior into rule-compliant and rule-breaching subsets. Second, rules may empower, by granting powers to individuals or particular officeholders. Of course, such powers often are limited or constrained, so that the two types of influence frequently work together. That is particularly true of constitutional rules (whether they constitute a nation-state or some other organization), wherein governance roles are created, empowered and constrained by rules.
Fifth, rules ultimately govern individual behavior. That is, individual behavior is in some way modified by the particular rules in place. However, it is clear that not all rules relate directly to the behavior of all, or even many, of the individuals that comprise the relevant society. Some rules, and particularly the rules of political process that are a main focus of CPE, seem to operate at some distance from the behavior of typical individuals. Often those rules are better seen as secondary, that is, rules that govern the processes that generate primary rules-they do not impact directly on the behavior of most individuals. 9 The structure of primary and secondary rules will be explored in more detail below, but for the moment I simply note that the CPE approach holds that all rules, whether secondary or primary, are to be analyzed and justified by reference to their ultimate effects on individual behavior and, hence, on social outcomes.
Finally, a rule operates across individuals, circumstances and time, so that a rule typically applies to a wide range of particular cases involving a number of people and its application continues over an extended time horizon. A rule brings together a number of specific cases and treats them symmetrically; by spanning time, it creates the basis for expectations of future behavior. While rules can be somewhat flexible, a rule that varies to suit each case, at each moment in time, is scarcely a rule.
We would stress several points relating to the foregoing definition of rules. First, it does not depend in any way on any particular process for the production or modification of rules. That is, it leaves entirely open the question of the supply of rules. As noted already, it is a characteristic component of the CPE approach that rules can be created or varied by the actions and agreements of those who operate under those rules, rules are, to at least some extent, endogenous to the political system over time, but endogeneity does not form part of the definition of a rule.
Second, our definition of rules makes no reference to enforcement. Rules may be enforced in a number of different ways, or not enforced at all, and still be rules.
Third, the definition of a rule is intended to operate in the context of a nested hierarchy of levels, including the constitutional, the political, the organizational, the formal and the informal. As we have noted, a major feature of the CPE approach is its distinction between the constitutional and political (or 'in-period') domains, but the dame idea of a rule operates in both domains.
With our definitional structure in play, the first and most obvious issue raised is the analysis of rule-following and the nature of the connection between rule-following and rationality. While CPE sometimes is caricatured as providing the political dimension of homo economicus, it has long been argued, by a significant group of scholars who would see themselves as associated with the Buchanan tradition, 10 that the strict version of homo economicus building case-by-case optimal decision making on the basis of a utility function that reflects self-interest, is inconsistent with any genuine form of rule-following. 11 For example, Vanberg (1994, p. 14) distinguishes genuine rule-following explicitly from case-by-case compliance: genuine rule-following requires 'a disposition to abide by rules relatively independently of the specifics of the particular situational constraints.' 12 As suggested in our definitional discussion, such a disposition has to involve a respect for rules that is strong, but not absolute. Rule-compliance must itself be the norm, but some allowance must be made for exceptional cases and circumstances. In that way, the approach to rule-following attempts to balance the idea of genuine rule-following behavior, along with an element of discretion that allows individuals to depart from rule-following in at least some circumstances. It may well be the case that, in any given society, most individuals break at least some rules some of the time, and it also may be the case that some rules are broken routinely, but despite those exceptions, the norm of rule-following has to be sufficiently strong and sufficiently widespread if a rule-based order is to be supported.
The element of discretion is important for several reasons, most obviously because it is important to the wider CPE project that individuals should be able to challenge existing rules and consider potential rule changes. While it may be possible to imagine a world in which existing rules are followed fully and automatically and where, simultaneously, criticism of existing rules and the proposal of replacement rules are encouraged; however, seeing existing rules as totally inviolate would seem, in practice, to inhibit debate on reform of those rules.
A second argument for recognizing at least some discretion in rule-following is more normative in nature in that it allows for the individual's own values to be the final guide for behavior in cases that we might think of as involving tragic choices, such as conscientious objection or civil disobedience to an existing rule, particularly if that rule is believed (by the individual concerned) to be illegitimate, fitting well with Buchanan's liberal form of normative individualism. 13 Even if we consider societies believed to be generally wellordered in the sense that most rules are approved of by most of those subject to them, specific circumstances may arise in which conscientious objection is at least allowable. And if we consider societies wherein the rules generally lack the sort of democratic legitimacy that provides the bedrock of normative CPE, at least some allowance for conscientious objection and civil disobedience aimed at constitutional reform seems appropriate. Such issues revolve around the distinction between rules and principles and the nature of the normativity of rules at a fundamental level. Discussion of those complex matters would take us too far from our central theme, but the general position adopted here is essentially that of 'inclusive legal positivism'. 14 Further reasons for restricting the status of rules to the relatively absolute include the general inability to specify rules that appropriately cover all possible circumstances and the difficulty of ensuring that rules never conflict. 15 In a perfectly rule-governed world, rules would offer a complete guide to behavior in all relevant situations. In practice, however, simple rules of the form 'do not X', meaning that no circumstances exist in which X-ing is permissible, are likely to be open to the form of criticism that involves pointing to some specific circumstances in which X-ing is, in fact, appropriate. Exceptions may be accommodated by refining the rule, making it more complex by attempting to specify the set of circumstances in which the rule applies, or the rare circumstances in which it does not. But the practicality of such a process is limited. In practice, rules generally are not complete. It is as if rules have a 'normalcy' clause written into them, so that any particular rule specifies behavior in 'normal' circumstances but allows for some discretion in exceptional cases. That is fine as far as it goes, except that it puts some strain on the idea that genuine rulefollowing should not be subject to case-by-case evaluation. Having to check whether the current situation is 'normal' or 'exceptional' before following a rule seems to reintroduce an element of case-by-case evaluation, but only a very minor element. The idea of genuine rule-following surely can withstand the incorporation of such 'normalcy' clauses.
Similarly, in a perfect world of many rules, all rules would be co-possible in the sense that no two rules ever would conflict. However, in practice, it is difficult, if not impossible, to ensure that no circumstances arise in which rules conflict. Of course, it is possible to form a rule that specifies what should be done in the case of conflict between two other rules, but such rules will themselves be incomplete, and the system of rules as a whole cannot be guaranteed to exclude (rule out) conflict among individual rules.
While the foregoing comments on rule-following might seem clear enough in terms of the primary rules of social and economic interaction-those that structure everyday activities in a tolerably well-ordered society-they may seem somewhat remote from the secondary rules of political procedure that are a main focus of CPE. After all, the rules governing the election of representatives, or constraining their powers, do not seem, at first glance, to depend on any particular disposition of rule-following, or to be subject to any element of discretion. But first glances can be deceptive. The 'constitutional attitude' that includes a commitment to constitutional government, the acceptance of whatever is defined as 'due process' for the transference of power from one government to the next, the acceptance of the idea of limited governmental power, is perhaps the single most significant example of a relevant political disposition stressed within the CPE approach. At the same time, constitutional rules are incomplete in the sense that they remain open to interpretation and re-interpretation and at least some instances arise wherein constitutional conflicts may arise. Of course, some formal constitutions include further rules that set out to govern the process of resolving matters of constitutional interpretation or any constitutional conflicts that may emerge (as, for example, in the case of the US Supreme Court), but such rules can be understood as locating the arena in which discretion may be exercised and by whom, rather than abolishing the need for discretion. It is equally clear that the laws and regulations that materialize from governmental processes also typically involve a degree of discretion-to be exercised by judges and other officeholders as well as by private individuals. 16 So, rules, in the political and social context, and whether they operate at the social, organizational, political or constitutional level, represent authoritative normative influences on individual behavior that operate across individuals, circumstances and time. The rules are accepted by at least many of the individuals who are subject to them and those individuals may be said to be disposed to follow the rules in normal circumstances, while retaining some discretion to depart from rule-following behavior in exceptional circumstances.

Rule of rules or rule of law
We now address the reading of 'the rule of rules' by analogy with the more familiar 'the rule of law'. The analogy suggests both that rules, rather than some other social construct or force, provide the basis of order, just as 'law' provides one basis for order to be contrasted with, for example, military force and, moreover, that support for the rule of rules essentially is normative, as is the support for the rule of law. CPE's response surely must be that the analogy is false, and that 'the rule of rules' points to a much more foundational idea than does 'the rule of law'.
The basic claim is that essentially all social order must depend on the existence and operation of rules: rules literally constitute order. No alternative bases for social order are possible. Of course, many different sets of rules may be possible, and so many different social orders may arise. The mere claim that rules constitute order should not be taken to imply any particular normative judgement about the comparison of alternative rules and their associated orders. The claim is simply that without rules there can be no order, no genuine society, only individuals engaged in something like the Hobbesian war of all against all. 17 Life without rules may or may not be solitary, poor, nasty, brutish and short, but it will certainly be unruly.
The rule of law, or nomocracy, by contrast, is normally taken to refer to a narrowly defined subset of social orders containing at least two elements. 18 First, every person within the relevant group should equally be subject to the law, including those who are involved directly in the process of lawmaking and enforcement, so that no one shall be above the law. Second, law should not be the result of the arbitrary declarations of an individual lawmaker (or small group of lawmakers). The two elements are separate since it is at least possible that arbitrary laws nevertheless might apply to everyone, including the lawmaker. The rule of law therefore is concerned with identifying a particular subset of social orders, governed by a particular type of rules, that satisfies the two normative criteria, whereas the rule of rules simply states the claimed positive or analytic truth that without rules (of whatever content) no meaningful order can exist.
Of course, the CPE tradition generally is supportive of the rule of law. But that is a matter depending on a combination of further positive and normative commitments that fill out CPE's analysis of the nature of the rules that are likely to emerge from the political actions and choices of individuals, together with the commitment to a form of normative individualism that grants legitimacy both to the outcomes of such constitutional choices, and to the political outcomes that emerge from the operation of such constitutions.
The basic idea here is that rules and, therefore, constitutions ultimately are legitimized by the conditions surrounding their acceptance. That conclusion should not be taken to mean that only those rules and constitutional arrangements that can trace their origins to historical acts of relatively free choice by some sort of constitutional convention or other representative body of individuals can be deemed legitimate. Such a test clearly would render almost all political rules and constitutions illegitimate. Rather, the idea is to point to the criterion of agreement as the appropriate test of the normative acceptability of any rule or set of rules in the absence of any transcendent or overarching normative criterion of the sort posited by, for example, utilitarianism or some forms of perfectionist liberalism. Whether a particular rule or set of rules is in fact agreeable always will subject to some doubt and will depend on theoretical or hypothetical analysis as well as more practical and historical considerations. 19 A further issue for the CPE approach to rules and order relates to the balance between rules and discretion in constitutional, political and legal settings. A useful starting point is Scalia's (1989) influential discussion focusing on the appropriate balance between rules and discretion in the hands of judges within the US constitutional system; that is, on the discretionary power of the courts to make (or modify) law. The position he takes follows Aristotle's Politics: Rightly constituted laws should be the final sovereign; and personal rule, whether it be exercised by a single person or a body of persons, should be sovereign only in those matters on which law is unable, owing to the difficulty of framing general rules for all contingencies, to make an exact pronouncement. (Scalia 1989(Scalia , p. 1176 18 For more detailed discussion, see, for example, Tamanaha (2004). 19 Examples of the role of theoretical or hypothetical arguments for the agreeability of particular rules can be found throughout Buchanan's work; see, for example, Buchanan and Tullock (1962), Buchanan and Wagner (1967) and Brennan and Buchanan (1980). 20 'Personal rule' in Aristotle usually is taken to refer to the direct power of a ruler or ruling elite, whereas Scalia's discussion focuses on the powers of judges within a democratic system, but the difference is not material here.
While that position clearly accepts the point made earlier that all systems of rules are incomplete and must leave open some degree of discretion, Scalia goes beyond that observation to argue for the most extensive possible use of rules and correspondingly minimal reliance on judicial discretion.
A contrary view, namely that judicial discretion significantly broader than the minimum warranted, is advanced by Calabresi (1982), who argues that over-reliance on the formal rules of statute law can lead to the perpetuation of bad law in changing circumstances and that a balance between rules and discretion must be struck, providing for more than minimal discretion.
On the face of it, the idea of a set of rules that maximally constrains the power of elected and appointed officials seems to fit well with the idea of limited government power that sits at the heart of the normative element of CPE. Individuals, it might be argued, would not approve of any set of rules that grants officials unnecessary or excessive discretionary power over them. But the crucial words here are 'unnecessary or excessive'. Since constitutions create as well as check power, the limiting idea of 'maximal constraint' could be achieved only by a constitution that empowered no officeholders. It is difficult to see how such a constitution could exist. That some discretionary power in the hands of officials is necessary is a simple fact, following both from the core idea of a constitution as conferring as well as limiting power and from the essential incompleteness of rules. It also is the case, however, that a degree of discretion that may go far beyond that which strictly is necessary can offer benefits to the general population, for example, by making the law and, more generally, the operations of government, more flexible in responding to changing circumstances. The optimal balance between rules and discretion is not obvious. 21 Restrictions on the discretionary powers of officials must be balanced against enhancing the liberties and opportunities available to the general population. When focused on the population at large, normative CPE adopts a classically liberal (some might argue libertarian) approach that emphasizes the liberty of individuals to pursue their own lives under the protection of the law. The broadly contractarian CPE approach to the notion of democratic legitimacy might be summarized as identifying the constitutional arrangements that offer discretionary liberty to private individuals by limiting (but not minimizing) the discretionary powers of elected or appointed officials. The liberty of individuals must be constrained by rules to some extent, just as some discretionary power must remain in the hands of officials and politicians.

The supply of rules
The discussion to this point has focused on the status of rules and their justification but has not addressed the critical issue of the supply of rules, the ways in which rules are generated and modified. Normative CPE is committed to the view that agreement is the only criterion by which to establish the legitimacy of rules. Positive CPE is committed to the view that intentional constitutional design and reform based on wide-ranging agreement is possible in more practical or pragmatic settings. 22 But positive CPE is not committed to the view that rules can emerge only from such an agreement-based process. Rules, and constitutional rules in particular, can emerge and evolve in a wide variety of ways. 23 The distinctions between primary and secondary rules, and between formal and informal rules are of value here. The clearest example of formal, secondary rules is provided by the rules of political procedure that govern the process of making and amending formal primary rules. However, not all primary rules are governed by secondary rules. Some, including many informal rules-norms and conventions-are better understood as operating independently of any process that might be seen as a secondary rule. Of course, the further question arises as to which rules are to be accorded the status of law, and that is the question addressed by Hart's 'rule of recognition', but our focus is on rules in general rather than the specific content of the law, so we will not pursue that question here. 24 The study of the supply of formal, primary rules, laws and regulations governed by secondary rules, may be thought of as the study of the operating characteristics of the relevant secondary rules. The specific laws and regulations that emerge in the United States and the United Kingdom, for example, can be expected to differ, inter alia, because of the differences in the secondary constitutional rules in place in the two jurisdictions. In those and possibly other cases, then, the issue of the proximate supply of that class of formal primary rules seems comparatively unproblematic, although the details may be complex. Essentially, an existing secondary rule (or set of secondary rules) specifies the process by which a broad class of primary rules, laws and regulations are supplied or modified. 25 To go beyond the study of formal primary rules governed by established secondary rules, two rather different cases must be considered. On the one hand, we have the supply of the informal primary rules, norms and conventions that are not themselves governed by secondary rules. On the other hand, we have the supply of secondary rules. I will discuss them in turn.
The following sketch of an account of how we might see a wide class of informal primary rules emerging from the practices and conventions that might arise spontaneously in society is by no means original to CPE. It draws heavily on a tradition that includes David Hume, Max Weber and F A. Hayek, among others. 26 We begin with a convention.
A convention essentially is a social practice that has no directly normative component. On Lewis's (1969) account, a convention is a social practice that resolves a coordination problem, but many scholars extend the idea of a convention to include forms of non-normative social practices, such as cultural traditions, that do not necessarily arise from coordination problems. By contrast with that idea of convention, we argue, with , that a norm involves an ineliminable normative component, so that a norm depends upon shared normative attitudes, whether or not that shared attitude is related to any specific social practice. Of course, many norms are, in fact, related to social practices, and that observation may be important in understanding the origins of some informal norms, as will be discussed below.

3
All that is necessary to establish a convention is that a sufficient number of individuals engage in the relevant social practice and come to accept it as a convention. That is a case in which acceptance by the relevant community is both necessary and sufficient for the existence of the convention. If we all (or almost all) accept that the appropriate mode of address in our society involves, say, the use of given name and family name in a particular order, then that practice simply is the convention, regardless of any normative attitudes that may exist about it. Of course, conventions that arise out of coordination problems may involve slightly richer stories, a convention of that type essentially is a means of selecting one equilibrium from among several possible equilibria. Some story may be told about how one equilibrium rather than another emerged as the convention, but whatever the details of that story may be, the essential point is that the convention exists once it is accepted by a sufficiently large group. The link between widespread acceptance and the establishment of a conventional rule is clear: general acceptance of a common practice constitutes a convention. Of course, acceptance and agreement differ, so that we are not at the point of concluding that widespread agreement is a necessary part of a convention's formation, but a clear connection exists between the idea of widespread acceptance and the scope for agreement.
Informal norms are rather more complicated. One argument is that at least many informal social norms derive from conventions that, once established, engage appropriately with underlying normative attitudes. So, for example, we might view driving on the left (or the right) as a simple convention arising from a coordination problem, but then recognize that coordinating our driving habits successfully engages our existing normative attitudes relating to injuring others (as well as ourselves), so that driving on the left (or right) takes on the additional character of a norm. Indeed, we might continue that line of argument to suggest that it is the robust normative attitudes associated with the successful coordination of driving that then supports the further transition from an informal norm to a formal norm or law requiring us to drive on the left (or right). In that way we can see a possible transition from a social practice to a convention, then to a norm, and on to common law. That reasoning represents the possibility of the supply of rules emerging from the bottom up.
But note two points. First, and most simply, it is not necessarily the case that all conventions become norms by the same route-the distinction between a convention and a norm persists. Second, such reasoning does not explain the normative attitudes underlying the norm. It simply argues that an appropriate and widespread normative attitude can serve to impart to a convention the additional character of a norm and, in at least some cases, carry it on into a formal rule or law. Two distinct elements are required to make the story work: first, the widespread acceptance of a social practice necessary to establish the original convention and, second, the relevant normative attitudes adding the normative impetus in transforming the convention into a norm or law.
Turning now to the set of secondary rules, the rules that are constitutional at some level in the nested hierarchy, we might first point to the fact that some constitutions (such as that of the United States) include specific and detailed provisions for their own amendment and reform, while other constitutions (such as that of the United Kingdom) make no procedural distinction between constitutional rulemaking and everyday lawmaking, so that the same rules of procedure apply. Those facts suggest that we might see the supply of even constitutional rules as itself being rule-governed in at least some cases, and so look to the properties of the relevant procedural rule to understand the supply of at least some constitutional rules. And, indeed, that reasoning seems to be an appropriate basis for the relatively modest study of the process of constitutional reform in such societies, i.e., societies that already have relatively well-established and operative constitutional structures that are at least widely accepted.
One interpretation of CPE is that indeed it is this sort of modest, but still important, exercise. An exercise first in understanding both the impact of any given constitution on social and political outcomes and the possibilities for constitutional reform and development in societies that may be described broadly as constitutional democracies. Also, an exercise in evaluating alternative possible constitutional reforms from a broadly contractarian standpoint. This interpretation of CPE seems entirely consistent with Buchanan and Tullock (1962), which distanced itself explicitly from any attempt to explain the origins of constitutional government itself and sought to analyze marginal constitutional reforms in a broadly democratic setting.
To move beyond such a relatively modest, but valuable role, CPE needs a more general account of the ultimate supply of rules, and particularly secondary constitutional rulesan account that in some way grounds the emergence of constitutions themselves and provides an account of their normativity. Two major possibilities arise here that may operate separately or together. The first possibility involves the continuation of the story told above concerning the transition from social practice to convention to norm to common law. The idea here is that even constitutional rules may be supplied from the bottom up. Such a story might point to mutual advantage as the guiding principle underlying the emergence of the various forms of rule and extend the argument from common law to statute law and, ultimately, to constitutional law. One might refer to Buchanan (1975) as providing a starting point for this more ambitious understanding of CPE (see Brennan and Eusepi 2013), seeking to outline the range of possible constitutional structures lying 'between anarchy and Leviathan'.
But such an account would be incomplete unless it can argue that all rules should be understood as developing from the bottom up, from some underlying social practice. It likewise would have to recognize that the story originates in accounts of social interactions among individuals that display a variety of structures (prisoner's dilemmas, coordination games, and so on). Since many of those game forms generate multiple equilibria, some of which may be fragile in the face of underlying parameter shifts, the emergent constitution in any given society might be both somewhat arbitrary (that is, merely one of a number of constitutions that might have emerged) and somewhat fragile. In that more general setting, there can be no presumption that 'marginal' changes to an existing constitution can be adopted. Nor can we presume that any marginal changes that might emerge will tend to lead either to any particular form of constitution over time, or to greater constitutional stability.
If the story of the emergence of constitutions from underlying traditions and conventions is to be developed more fully, it also would have to tackle more detailed discussion of the relationship between the normative attitudes that support norms and the idea of widespread acceptance and agreement that drives the story forward. Broadly speaking, those two components provide the normative and behavioral foundations both for a positive account of the development of rules and the normative justification of rules arising from the idea of democratic legitimacy.
The second possibility is that the supply of at least some rules operates from the top down rather than the bottom up, so that rather than emerging from the various detailed social practices in a variety of contexts, at least some rules are created by a process that operates directly at the higher, constitutional level. Any such process would seem to depend on at least some pre-existing society, and therefore on some pre-existing set of rules, so that it would seem that any top-down supply of rules is likely to occur only when and where at least some significant rules already have emerged from the bottom up. The idea of a radical constitutional convention creating new secondary rules in an environment where 1 3 a dense social network of rules already exists, is one possibility here, but my use of the adjective 'radical' is important. To represent a top-down supply of rules, an example must not be one in which a constitutional convention is part of the existing constitutional apparatus. If so, the example simply would represent a rule change governed by existing secondary rules. A radical constitutional convention is one that is neither mandated or recognized under existing (constitutional) rules. As such, it cannot claim legitimacy in any straightforwardly procedural sense; any legitimacy that emerges must be earned in some other way. The founding constitutional process of the United States may seem to provide a real-world example of the top-down supply of rules, but the more general discussion of the how such a top-down process operates and the source of the normativity attaching to the outcome of such a process is by far the least developed and the least satisfactory aspect of CPE.
That CPE does not offer anything like a full answer to the questions of the origins of constitutional government or of political obligation should not be surprising. Nor does it limit the value of the more modest reading of CPE as an approach that provides a systematic study of complex structures of primary and secondary rules within broadly democratic societies, both in terms of their implications for social outcomes and in terms of their potential for reform.

Rules, enforcement and a rule-based order
The idea that genuine rule-following behavior requires a widespread disposition favoring compliance, rather than individuals relying on case-by-case evaluations of behavioral options, might seem to suggest that rule enforcement largely is irrelevant. By contrast, it might be suggested that unenforced rules are not really rules at all, but mere normative suggestions. A balancing thought might be that enforcement provides a means of securing compliance from individuals who are not genuine rule-followers (in relation to a particular rule in particular circumstances), so that compliance is of two sorts: genuine or dispositional compliance and compliance that depends on enforcement and the threat of sanction.
Before exploring that thought a little further, we must recognize two general points about enforcement. First, enforcement comes in a wide variety of forms. Many informal rules and norms are in fact enforced despite the fact that they may not be enforced by any central authority. The simple idea that some norms and informal rules are enforced by mechanisms such as social pressure involving patterns of approval and disapproval or loss of esteem is a powerful reminder that enforcement goes well beyond the police or the courts. The idea of informal enforcement is particularly important in the context of bottomup arguments about the supply of rules because some pattern of informal enforcement is likely to be involved along the path that leads from a simple social practice towards law. That is, enforcement is likely to be a characteristic of relatively informal rules even in the absence of any specific enforcement agency.
Second, enforcement of a rule itself relies upon the operation of and compliance with other rules. That is most obvious in the case of formal, state-operated systems of enforcement, where formal rules are required to generate the institutions of enforcement (e.g., police, courts) and regulate their behavior. Compliance with those rules may then also be enforced, setting up the standard enforcement regress captured in the classic question, 'Quis custodiet ipsos custodes?' The same issue arises in less formal settings wherein, for example, individuals have to be disposed to disapprove of others who fail to comply with some particular social norm and approve of those who do comply, showing their approval or disapproval even at some cost to themselves. In that way, the rules surrounding enforcement may be seen as special cases of the more general class of rules.
The last two points raise two connected puzzles: why are rules so widely enforced (formally or informally) if the disposition of rule-following is common, and how can enforcement work in situations where the disposition of rule-following is uncommon? Addressing those two puzzles together points to the balance between genuine rule-following and rule-following in the shadow of enforcement that is likely to exist in any reasonably stable social order. 27 Clearly, if the disposition to follow rules was universal and strong, enforcement would be both easy to organize and unnecessary: easy to organize because all individuals could be relied upon to play their part in enforcement mechanisms by following the relevant rules; unnecessary for the obvious reason that no individuals would deviate from the rules even in the absence of enforcement. By contrast, if the disposition to follow rules is rare and weak, enforcement would seem to be required, but would be difficult to organize since we could not rely on rule-following behavior by those charged with enforcement. So, even allowing for enforcement, a reliable and stable rule-based order can exist only if the disposition to follow rules is sufficiently wide-spread and sufficiently strong to (a) allow genuine compliance in a significant proportion of cases and (b) allow the successful operation of enforcement rules to strengthen compliance with other rules still further. In short, a rulebased order with enforcement still depends on the widespread, though not universal, disposition of rule-following. Enforcement, on this account, feeds off of genuine compliance and can enhance it, but cannot operate successfully in its absence. 28 At the same time, the power to enforce rules places very substantial power in the hands of officials and politicians, and so is open to very obvious abuse. One might expect governments to rely too heavily on their coercive powers and so potentially undermine genuine rule-following and democratic legitimacy. At the constitutional level in any broadly democratic system, therefore, we might expect to see particular attention to the institutionalization of enforcement powers so as to guard against such abuse to the greatest degree possible. 29

Context and concluding comments
Constitutional political economy's (CPE's) focus on rules stands in sharp contrast to the mainstream economic approach of the later twentieth century but was much more in tune with other parts of the intellectual landscape of the time. The focus on rules was a key part of both the New Institutional Economics (NIE), associated with Oliver Williamson, Ronald Coase and Douglass North, and the Rawlsian account of the basic structure of society. At the same time, the idea of a rule-based order (as well as that phrase) was in widespread use in the international context in the period following the Second World War, when the three intellectual movements were developing.

3
The common ground on rules between CPE and NIE is very clear and noted widely. 30 For example, North (1991, p. 477) writes that, "Institutions are the rules of the game in a society; more formally, they are the humanly devised constraints that shape human interaction. Thus, they structure incentives in exchange, whether political social or economic", a statement that might equally be attributed to the CPE and NIE approaches. The differences between CPE and NIE largely are matters of focus, with NIE's focus on the institutional framework of the market and CPE's on the wider political constitution. While NIE shares CPE's commitment to the idea that rules and institutions are manmade, NIE generally does not take up the broadly contractarian normative stance associated with Buchanan. More recently, both CPE and NIE influences can be found in the burgeoning comparative political economy literature (Persson and Tabellini 2005;Robinson 2006, 2012).
Buchanan's general approach to rules and the idea of a rule-based order likewise relate closely to the Rawlsian enterprise. 31 It is now standard to note the connection between Buchanan and Rawls in terms of their shared commitment to a form of contractarianism, but I would point to their potentially deeper shared commitments to the idea of a rule-based order, recognition of the direct link from rules to institutions, and focus on the properties of rules and constitutions or basic institutions as the primary objects of analysis. One of the key features of Rawls's work that chimed with Buchanan was its focus on the basic institutional structure and the rule-based order that it supports. A fundamentally shared vision of society as constituted by its basic rules and institutions is quite distinct from a commitment to any variant of the contractarian method; the two commitments differ in both nature and scope. The former is a conceptualization of the basic subject matter of politics, the latter a specification of a mode of normative analysis.
Although I never have found any direct reference to Rawls (1955) in Buchanan's published work, it displays a deep parallel with Buchanan's use of the distinction between the choice of rules and the choice of actions under the rules, even if Rawls's (1955) discussion is more utilitarian than contractarian. Rawls (1955) presents two concepts of rules that really are two approaches to the idea of justification in a rule-based order. The key distinction, as with Buchanan, is between (a) justifying a practice as a system of rules to be applied and, possibly, enforced, and (b) justifying a particular action under those rules. The broadly utilitarian argument advanced by Rawls (a) is forward looking-a rule is justified by its anticipated impact or consequences on society in the future, even though (b) it may be backward looking as, for example, a particular punishment under an enforced rule is justified by reference to the past action of breaking the rule. The link to Buchanan's fundamental distinction between constitutional choice and in-period political choice is clear enough.
While the definition of a rule, in the political and social context, as an authoritative normative influence on individual behavior acting across individuals, circumstances, and time seems simple enough, the discussion here reveals that rules come in wide varieties of type, primary and secondary, formal and informal, operating at various levels in a nested hierarchy. While individual scholars differ on details, the CPE approach to rules accords them a very special status: rules are seen as the essential foundation of all social order and, therefore, as the primary focus of study.
Even if all order is rule-based, different rules may be expected to create very different orders and very different social outcomes. Social outcomes are seen as emerging from individual behavior within the prevailing rules (whatever those rules might be), so that any attempt to influence social outcomes should be directed at reform of the underlying rules. At the same time, the CPE approach takes rules themselves to emerge from, and be subject to reform through, the political actions and choices of individuals, opening up the practical possibility of the reform of the underlying rules.
The constitutional, rule-based, perspective characteristic of CPE is at its strongest when operating in the relatively modest arena of political analyses of constitutional democracies. In that arena, across which many significant differences materialize, the basic commitments of both positive and normative CPE connect well with the political realities, and the marginal approach to institutional design seems both warranted and productive. The operating characteristics of particular political, social and constitutional rules often are complex and may depend crucially on other rules that are in place. At the same time, the analysis of the social and political pressures for reforming existing institutional and constitutional rules, along with the normative evaluation of alternate potential reforms from the perspective of democratic legitimacy, is challenging but feasible.
The CPE perspective, seen as a mode of positive analysis, also may apply in settings that are not themselves constitutional democracies. It is still the case that the focus on the operating properties of the rules in place and the study of possible reforms to those rules, often will provide valuable insights. Non-democratic societies may be regarded as rulebased social and political orders, and reforms of the rules may still be considered to be either by mechanisms that already exist within the constitution or by some other bottom-up or top-down process. However, it is less clear that the normative aspect of CPE connects well with non-democratic societies. Of course, alternative rule-regimes nevertheless can be evaluated by reference to the criteria of democratic legitimacy, evaluated, that is, as if from the outside. But such an external evaluation may not connect with established patterns of normativity within the society under study, and so may not play any role in the analysis of the potential internal dynamics of the constitution under study, or the internal justification of rules.
The CPE perspective is less well developed, and weaker, when attempting to provide the more ambitious and deeper account of the origins of constitutional democracy or the justification for such regimes or for particular democratic structures. Whether the strategy of the argument is bottom up or top down, or some mix of the two, the broadly contractarian normative structure of CPE carries a heavier load and cannot altogether escape the criticisms of contractualist political philosophy more generally.
The idea of a rule-based politics in a more general rule-based social order is the starting point for CPE analysis, recognizing as it does that no society is possible without rules and no politics is possible without society. At least some collective decision making, and the resolution of the various interpersonal conflicts and dilemmas are necessary in any society worthy of the name, and any means of satisfying those requirements and so providing the basis for a society will involve rules. The attempt to study the implications of any particular set of rules, and the possibility for reforming them, is fundamental. The justification of rules is equally fundamental but much more challenging.