Introduction

“Law” is just a label. What we count as law depends not on some inherent truth, but rather on what we are trying to explain. Anthropologists are primarily interested in documenting the empirical content of the rules of a particular culture, and less interested in distinguishing between categories of rules. They therefore count as law whatever rules produce socially sanctioned punishment in response to violations (e.g., Malinowski [Reference Malinowski1926] 2013; Hoebel [Reference Hoebel1954] 2006). Legal positivists are primarily interested in developing a conceptual account of law that keeps a clean separation between law and morality in order to provide a stable framework for the objective normative evaluation of a consistent social phenomenon and to explore the nature of the authority of law. This approach counts as law the rules produced and enforced according to established rules of validity by whatever government is recognized by a population, even if the citizens of that state have no voice or live in fear of state-perpetrated violence: rules are created and enforced in accordance with pre-established rules of validity (e.g., Hart [Reference Hart1961] 2012; Raz Reference Raz2009). Natural law theorists are engaged in a more affirmative normative activity: determining what is necessary for a social practice aimed at achieving common goods to achieve its goals. For them, the label law is only properly attached to the actions of a government that respects minimal obligations of fairness and decency to its citizens (Finnis Reference Finnis2011; Waldron Reference Waldron2011).

Both Fred Schauer (Reference Schauer2015), in The Force of Law, and Richard McAdams (Reference McAdams2015), in The Expressive Powers of Law: Theories and Limits, are primarily interested in an empirical account of how law works in practice. Schauer's focus is on the role of coercion—imposing sanctions that cause an individual to change his or her behavior from what he or she would otherwise choose—which he claims is central to the concept of law on empirical grounds. “Although we know that a legal system could in theory exist without sanctions and without coercion,” he argues, “we know as well that, with somewhere between few and no exceptions, no such legal systems actually exist” (63). McAdams's focus is on coordination. His claim is that law sometimes works without any coercion, inducing changes in behavior through expressive mechanisms, either by proposing a focal point in a coordination game or by providing information about the practical (not legal) consequences of behavior and/or attitudes of the community toward behavior.

To the extent that Schauer and McAdams disagree, then, it is about the empirical relevance of those cases in which law coordinates without coercing. However, the points of tension are few: McAdams is not seeking to explain all of law and is, I think, happy to agree that the purely expressive account of law is limited in scope. And I think Schauer is not taking aim at those cases where law coordinates without coercion, but rather the more expansive claim, propounded by many legal theorists (and typified by H. L. A. Hart), that the concept of law is centrally concerned with law's capacity to give moral agents reasons to comply voluntarily, and not the actual coercive force of law.

Both books contribute to the important project of moving our theoretical accounts of law onto the terrain of behavioral theory and evidence and I applaud them both for this. Both contribute to the richness of our understanding of how law works in practice and urge that our theoretical accounts converge with this empirical reality.

The question I want to focus on in this comment, however, is the underlying premise in both about what counts as law in the first place. Neither McAdams nor Schauer really treats this question as problematic. Both take as a starting point, implicitly, the empirical context of current legal regimes and particularly those stable, well-ordered regimes that characterize the advanced West. This is fine on its own terms: both McAdams and Schauer have told us more about these legal settings. But I want to suggest that this is an overly narrow framing, and it replicates a shortcoming of the dominant approach to legal theory, with its focus on the internal authority of law, that both McAdams and Schauer are challenging.

To my mind, the reason for building a social scientific theory of law is to gain insight into the practical demands of building legal systems that are effective in achieving the goals of law: providing a platform for a productive and efficient economic system, implementing fair and effective systems of governance, and ensuring the protection of the rights of individuals and groups to decent treatment, security, and opportunity. To the extent that we build up our empirical accounts of law based on observations about how law works in stable, advanced settings, then, we run the risk of failing to account for how law is made to work like this. In empirical jargon, we could say we commit the error of sampling on the dependent variable: identifying what works in cases in which it does, but not looking to see whether there are examples when “what works” does not. To the extent that our empirical base is our current context, we limit our ability to predict how law might be made to work in very different contexts.

The problem of how legal order is achieved is a critical one for theorizing about law and legal institutions. When social order is accomplished through legal rules and institutions, the population living under that order can enjoy the principal benefit of law: the capacity to introduce new law or change existing law to improve human well-being. That is what law does: it gives us a policy instrument and a means of shifting behavior to, for example, limit harm from economic production, share a country's wealth better, or prevent discrimination against a vulnerable minority. Social order grounded in organic social norms, in contrast, only changes slowly if at all, and not in ways that can be the subject of deliberate choice (Hadfield and Weingast Reference Hadfield and Weingast2013).

Yet half the population of the world lives in environments that lack reliable legal order (United Nations 2008). And despite the widespread recognition that stable rule of law is critical to promote economic growth, democratic governance, and the protection of human rights, and despite billions spent on rule-of-law building projects across the globe, efforts to secure legal order in poor and developing countries have largely failed (Carothers Reference Carothers2006). A lack of a robust social scientific account of “what law is” and how it is generated and stabilized is a key reason for this failure (Hadfield and Weingast Reference Hadfield and Weingast2014b).

Nor is the need for careful theory about law limited to the problem of building wealth, security, and justice in poor countries. In advanced settings, we face the increasingly urgent problem of adapting legal systems that are hard-pressed to manage the growing speed, complexity, and global nature of the digital economy (Hadfield Reference Hadfield2017). Furthermore, with increasing global interconnectedness, we face the growing puzzle of how to secure international legal order in the absence of global government. Without an account of law that is freed from our current institutional approaches to producing legal order, we have little with which to design new methods for managing the future and confronting the challenge of international legal order.

The problem of developing an account of law and legal order that is not rooted in the institutions of advanced and stable legal regimes is one that my coauthor Barry Weingast and I have explored in a series of recent papers (Hadfield and Weingast Reference Hadfield and Weingast2012,Reference Hadfield and Weingast2013,Reference Hadfield, Weingast, Galiani and Sened2014a,Reference Hadfield and Weingast2014b, forthcoming see also Carugati, Hadfield, and Weingast Reference Carugati, Hadfield and Weingast2015). Our work dovetails in important ways with that of both McAdams and Schauer: our account emphasizes both coercion and coordination as critical elements at work in legal order. But, for reasons I explore below, I think our account goes further toward meeting the challenge of figuring out how to build law where it has failed and how to adapt law where it is under pressure from complexity and globalization. This is not necessarily a criticism of either Schauer or McAdams—it may be their purpose only to explain how law works in advanced and stable legal regimes—but it is a limitation that I think bears noting.

Coercion

Schauer's (Reference Schauer2015) principal claim is that, even if there are instances in which people comply with the law because it is in their interest to do so (e.g., to coordinate with others) or because they feel an obligation to respect the authority of law, the representative case of law is the case in which someone complies because if he or she does not, a government sanction is likely (enough) to result. This is the taken-for-granted understanding of law that we find throughout most economics and positive political science (Hadfield and Weingast 2014b): people choose to obey the law when the expected benefits of violation are outweighed by the expected penalties. And like that literature, Schauer assumes that when we identify a system of penalties as a legal system, the entity that imposes the penalties is government.

I agree with Schauer that any account of law has to be built up for the ordinary case in which law functions to change people's behavioral choices from the ones they might otherwise have adopted to ones that, collectively, we would prefer them to choose. The expressive function of law that McAdams (Reference McAdams2015) explores in masterful detail is important but, as I discuss below, in the form McAdams presents it—as coordination largely of primary behavior such as whether to continue a contest over resources or not—it is not as central as the coercive function. Nor is the case in which people comply with law out of an internalized respect for the law a useful starting point as an empirical matter; this traditional legal philosophical stance is, of course, Schauer's principal target. Law is an important human invention precisely because it provides a community the opportunity to identify behaviors that it wants to deter and others that it wants to encourage and a means for achieving those behaviors: the threat of penalty for noncompliance.

Where I part ways with Schauer is with the presumption that the imposition of penalties for failure to comply with legal rules is exclusively a function of government. Schauer (Reference Schauer2015) does not delve deeply into this question. He says, for example, “there is something important in the common observation that law depends on the state's ultimate monopoly of the legitimate use of force” (129), but we do not really learn what is important, or essential (since he is after the essence of law), about coercion coming from a state with a monopoly of legitimate force.

This is, however, a critical inquiry if our project is to understand how law works as a general theoretical matter, and therefore to understand how to make it work in places where it currently does not. The presumption that legal order rests on the exercise of coercion by a state with a monopoly on legitimate force is, as Weingast and I have argued, a major reason for the failed policy prescriptions followed in rule-of-law-building efforts in poor and developing countries (Hadfield and Weingast 2014b). If our theoretical framework starts with the well-established state with a monopoly on legitimate force, then our prescription for law building can only be to establish a stable state with a monopoly on legitimate force first and then get that state to enact laws according to established criteria and enforce them. Now, maybe this theory is correct and the fact that this is a major Catch-22 for rule-of-law building—you can have stable legal order once you establish a stable, legally defined government—may simply be the way the world is. But Weingast and I think this is wrong, and empirically wrong.

First, we can identify many instances of relatively stable legal regimes that exist without relying (exclusively or at all) on the state to supply the penalties for rule violation. We discuss in Hadfield and Weingast (Reference Hadfield and Weingast2013) and Carugati, Hadfield, and Weingast (Reference Carugati, Hadfield and Weingast2015), for example, some canonical examples: medieval Iceland, the world of European medieval merchants, California during the gold rush, and ancient Athens. The international legal order is another critical example—and indeed one that Schauer (Reference Schauer2015) discusses, albeit relatively briefly (136, 161–62).

Second, even in the presence of a stable and powerful government, much of the coercive force that incentivizes people to comply with legal rules comes not from centralized, state-imposed penalties, but from decentralized penalties supplied by private individuals and entities. Substantial amounts of contract compliance, for example, are motivated by fear of private penalties such as loss of business and reputational harm, as Stewart Macaulay (Reference Macaulay1963) originally documented, and as my work with Iva Bozovic has shown in more recent settings (Hadfield and Bozovic forthcoming). And, indeed, we should expect, for good theoretical reasons, that this is so: the cost of imposing public penalties with sufficient likelihood in all instances in which we intend to modify behavior against self-interest is exponentially high. To keep the expected penalties high enough, such a regime either must devote huge resources to detecting and prosecuting violations, or it must impose outrageous penalties on those few who are caught (Hadfield and Weingast forthcoming).

A theory of law based on coercion must, therefore, account for the role of privately supplied coercion. Governments, of course, have come to play a central role in implementing penalties, but there is no shortage of examples where the state's role has been limited to providing a clear statement of what the rules are, leaving enforcement to private means. Indeed, the fully centralized enforcement apparatus of the state is a relatively recent phenomenon in human history, more recent than the emergence of distinctively legal order (Hadfield Reference Hadfield2017). Schauer (Reference Schauer2015) does address the issue of social norms and notes, correctly, that “insofar as the effect [of law on behavior] is produced by way of triggering the coercive force of social norms, then coercion still occupies center stage in explaining how law affects behavior” (148). But he maintains the idea that government is the starting point when he goes on to observe that “in this context we might think of law as simply outsourcing its coercive power to private actors.”

Historically, however, the pattern seems to have been reversed: most enforcement in emerging legal orders was supplied by private efforts (again, medieval Iceland is a sharply drawn example, as is ancient Athens) and it is only over time that enforcement is gradually shifted to a centralized organized enforcement apparatus. At a minimum, therefore, for historical accuracy, we need a theory of law that does not start with state enforcement. More importantly, a theory of how law emerges without a stable centralized enforcement system (as it originally did in the now stable West) will prove far more useful for developing law-building advice for the poor and developing countries of the world where states are weak and failing.

Once we recognize the need for a theory of law to account for the role of private enforcement efforts, we also bring into focus a critical theoretical challenge: distinguishing between forms of social order. Every society has some form of social order, some means of ensuring that behavior follows reasonably predictable patterns, and most of these rely to some extent on coercive penalties, particularly of the milder form of social criticism and exclusion of rule breakers. The question is: What is distinctive about legal order, if anything? The answer is easy if we are focused on government-created and enforced rules; this is the standard means of distinguishing social norms from legal rules in the social science literature (see, e.g., Ellickson Reference Ellickson1991; Dixit Reference Dixit2004). But if the means of enforcement of legal rules include private and social sanctions—the same types of sanctions that enforce social norms—the question of what is distinctive about law is more problematic. And yet there is a strong view that law matters: there is widespread consensus that the best explanation for the failure of growth in poor and developing countries is the absence of good governance, by which most commentators generally mean the rule of law (see Easterly Reference Easterly2006; Collier Reference Collier2007; Acemoglu and Robinson Reference Acemoglu and Robinson2012).

Schauer's treatment of this question seems to me to be too glancing. He recognizes that “one of the things that an account, or even a theory, of law must explain is the widely accepted fact that the systems of social norms on the one hand and of law on the other are understood to be different” (2015, 142). His response reaches to something important: “Social norms … are not systematic in the way law is. Collections of norms tend to be just that—collections—and it is thus not surprising that few of such collections contain secondary as well as primary rules, and, more specifically, that few of them contain rules of recognition, rules of change, and rules of adjudication” (143). I agree that the formality with which legal rules are identified and changed is key to their distinctive status, but what I find missing in Schauer's treatment of this issue is an account that ties these distinctive attributes to the theoretical centrality he places on coercion.

The rules of validity he is appealing to are, of course, those of the legal positivists (Hart [Reference Hart1961] 2012), who say that the presence of rules of validity is sufficient to identify a regime as legal and who see no essential role for coercion. So it is not clear why Schauer points to rules of validity as the mark that distinguishes social norms from legal rules. More troubling, Schauer does not address the question of how the rules of validity are enforced. If rules are primarily observed because of the threat of coercion, coercion is generally supplied by a government with a monopoly of legitimate force, and if the rules enforced by government are legal and not merely social because they are subject to rules of validity, then why does the government follow the rules? The legal positivists resolve the question by saying that the rules of validity ultimately are not legal rules but social rules, implemented because officials voluntarily choose to observe them. But it is not clear to me that Schauer can accept this answer if he wants to say that coercion is central to the concept of law.

The approach in Hadfield and Weingast (Reference Hadfield and Weingast2012) avoids this dilemma because we do not treat government-supplied coercion as a defining attribute of law. We do not attempt to distinguish legal from other normative social orders on the basis of the enforcement mechanism. Instead, we focus on the classification function—the means of determining what is and is not punishable in a society. Law, we argue, is distinctive because the classification function is performed by a centralized and formalized institution, rather than supplied organically by means of an equilibrium generated by repeat interaction.Footnote ¹ Enforcement, however, may be supplied entirely by decentralized means. Indeed, we argue that it is only when there is a substantial role for private enforcement that we can systematically expect that a centralized classification institution will follow its own rules (Hadfield and Weingast forthcoming). Our account of the distinctiveness of law is then closely tied to our analysis of coercion and, in particular, the challenge of coordinating and incentivizing private participation in the enforcement of publicly defined rules. For this reason, while I agree with Schauer's critique of the dominant literature in legal philosophy, which discounts the role of coercion in legal order, I think his account falls short of a full theory of the role of coercion.

Coordination

In The Expressive Powers of Law, Richard McAdams (Reference McAdams2015) grabs hold of the social order–legal order continuum at the other end. McAdams is focused on law when it operates in ways that are deeply rooted not in the institutions of government enforcement, but in the ordinary enforcement of social interactions. The coordination problems he analyzes can be resolved by social institutions as well as by legal institutions, by social norms as well as by legal rules. And, indeed, McAdams builds his account of how law can serve as a focal point to coordinate behavior by appeal to the nonlegal mechanisms that perform this function: the person who jumps into the intersection to direct traffic when the lights go out, the shared knowledge and local saliency that makes the central train station a natural place to meet when no other plans have been made.

McAdams is clearly right about the capacity for legal rules and institutions to coordinate behavior. The Expressive Powers of Law provides a terrific overview of the multiple ways in which law does this, ranging from the well-worn case of determining the side of the road on which we drive to the more powerful and generic case of resolving the hawk-dove dilemma, which can arise in routine disputes and that explains the power of third-party arbiters, who lack enforcement power, to resolve disputes effectively. McAdams has written the definitive account of this role for law, with careful treatment of the multiple forms in which it can arise and careful treatment of the limits of the account.

My principal concern with the approach McAdams is taking is the limited theoretical scope of his account. I think his account of the role of coordination in understanding legal order is not ambitious enough. The expressive function of law he identifies only operates in those particular settings in which individuals are playing a pure coordination game, meaning that they each achieve higher payoffs when they coordinate than when they fail to coordinate. They may have conflicting preferences about which coordination equilibrium they would like to pick, but they still both prefer to coordinate on the other's preferred solution than not at all. This is a restrictive set of cases. The hawk-dove dilemma, for example, while applying to a wide array of possible disputes (over property, child custody, contract breach, international borders, etc.), is still a special case in which it happens to be the case that both disputants prefer to accept the short end of the stick than continue fighting. However, lots of disputes are more intransigent than this: even if A is told clearly that the law resolves the dispute in B's favor, A may still prefer to continue fighting than to relent and go along with the solution in B's favor. For many routine disputes, settlement on the expected legal outcome is a consequence of the credible threat that the law will impose that solution coercively.

McAdams is not trying to supply a theory that explains all of law, and so the limited scope of his treatment of coordination is not a flaw in the work. However, I do think that this work has missed a more fundamental relationship between law and coordination. That is the role of law in coordinating not primary behavior—how much is owed under a contract, who gets the kids, what side of the road to drive on—but rather in coordinating decentralized, third-party efforts to enforce legal rules—shunning the business partner who breaches contracts, criticizing the parent who ignores a custody order, condemning the driver who drives the wrong way on a one-way street. McAdams (Reference McAdams2015) does give brief treatment to the coordination of third-party enforcement, referencing my work with Weingast (117–19), but he does so by treating the problem of how to coordinate third-party enforcement as merely another instance of the pure coordination game. I think this is insufficient. Third-party enforcement is a central feature of any legal system, precisely because—as I noted above in my critique of Schauer—any legal system that achieves deterrence at reasonable cost (and maybe any legal system in which centralized institutions respect the rule of law [Hadfield and Weingast forthcoming]) must draw to a significant extent on decentralized enforcement efforts. It is because these enforcement efforts are largely collective—not unilateral—that coordination is a central, not merely incidental, element of legal order.

Indeed, there is perhaps no more critical question to resolve in a theory that seeks to generate insight about how to build legal order where it is absent or fragile than the question of how coordination of third-party enforcement around a central classification institution is achieved. The poor and developing countries that desperately need legal order are precisely those that possess governments that are poor—lacking resources for effective public enforcement—and weak, often corrupt—lacking authority and legitimacy with the citizenry. The limitation I see in McAdams's approach is that it tells us that if payoffs are structured so that individuals prefer to coordinate on an announcement of a rule by a government than fight, and/or if ordinary individuals face incentives to coordinate with their fellow citizens to enforce the rules announced by a government, then government coordinates behavior around its rules. The hard issues are two: (1) How are payoffs structured in this way? and (2) Why does the population coordinate on a government rule rather than some other focal point (such as tribal loyalties or an aspiring despot)? A coordination account that begins with the neat payoff matrix of a battle of the sexes game or a prisoner's dilemma does not help us when payoffs fail to line up. Nor does it help us figure out how to secure a centralized rule maker in the role of coordinator. But those are the circumstances in which societies find themselves in the struggle to build the rule of law.

Weingast and I take this as the central challenge of a positive theory of legal order—to explain how incentives are structured to induce people to participate in third-party enforcement of a central classification institution and to identify what attributes that institution has to possess in order to fulfill this function. Our model, for this reason, begins not with the assumption that payoffs are structured to generate a coordination game, but with the most economical assumption we have in positive theory: people are self-interested maximizing agents with no a priori interest in coordinating with others. We emphasize that securing coordination of third-party enforcement, therefore, poses not only a coordination challenge, but also, more fundamentally, an incentive-compatibility constraint: people will only participate in third-party punishment if they judge that by doing so they, personally, will be better off in the long run.

Our emphasis on the incentive compatibility constraint yields a remarkably rich set of insights. For example, our approach suggests that the likelihood that an institution proposing legal rules will gain the enforcement support of a population is a function not of the content of individual rules (the solution to a particular game, such as a dispute over a border or use of an asset), but rather on the totality of the collection of rules that an institution manages to weave together under the label law. In our model, the question for the residents of ancient Babylon, for example, was not “am I willing to follow and/or help enforce the rule about who pays for damage to a neighbor's crops when a poorly maintained dam breaks,”Footnote ² but, instead, “am I willing to follow and/or support Hammurabi's Code?” This predicts that a set of rules has to contain benefits for all those who are necessary for stable enforcement—a characteristic we call qualified universality.

Qualified universality is just one of several legal attributes that our model predicts that a central classification institution will have to possess if it is to coordinate third-party enforcement successfully. Some of these attributes come from the coordination function per se—clarity and common knowledge, for example. But a number are implications of the need to incentivize participation—that is, to create the payoffs under which someone prefers to coordinate more than not. Incentive compatibility predicts, for example, not only qualified universality in the content of the set of rules, but also long-run stability, neutrality, and open processes. These attributes help generate the belief among individuals that the costs they bear enforcing rules in the near term, which may be irrelevant or even contrary to their personal interests, will pay off in the long run when they are the beneficiary of the deterrence that stable community enforcement of the rules provides.

Our approach also helps illuminate another issue that is elided in McAdams's coordination game framework: the choice to coordinate on the legal rules announced by a government rather than some alternative focal point. McAdams, drawing on his work with Janice Nadler, does take care to state that for law to coordinate it must be that there are no “stronger, competing focal points” (2015, 62). But he focuses on particular instances in which strong competing focal points are absent or disrupted (such as where norms about gay marriage or smoking have been disrupted by social movements), not how law manages to displace strong competing focal points.

The displacement of alternative focal points, however, is the principal challenge of creating, or maintaining, legal order. Moreover, displacement is not a case-by-case problem, but a wholesale challenge in explaining both the history of the West and the failure of legal order in many parts of the developing world. Weingast and I are far from a solution to this ourselves, but our framework does emphasize that the challenge of securing legal order is fundamentally one of shifting allegiance in third-party enforcement efforts from an alternative or preexisting focal point. Ancient Athens, for example, was possibly the first society to transition from social order based on organic social norms, elite rule makers and magistrates, and tyrants to legal order based on neutral, somewhat democratic, institutions.Footnote ³ As we argue in Carugati, Hadfield, and Weingast (Reference Carugati, Hadfield and Weingast2015), this feat required convincing ordinary Athenian citizens to shift from choosing whom to assist in a dispute on the basis of friendship, family, fear, or enmity to choosing on the basis of what the Athenian jury had decided.

We argue that we can understand otherwise puzzling features of the Athenian legal system—such as its willingness to allow litigants, who had to identify a particular written law in order to bring a lawsuit, to argue whatever seemingly irrelevant reasons they wanted to try and convince the hundreds of jurors to decide in their favor. In our account, allowing litigants to appeal to considerations irrelevant to the charged violation (e.g., an alleged contract-breacher's history as a prostitute or a lazy son) assured potential participants that the process was seen as open and not a challenge to preexisting social norms. Without that assurance, they might well have been unwilling to give up their longstanding ways of thinking about how to resolve a dispute properly and who to support.

McAdams focuses explicitly on the idea that law in stable settings coordinates because it is law—but this does not help us understand how the distinctive quality of law is achieved. If we focus on how law works in stable legal orders, the question is not in the foreground. McAdams does discuss the problem of legitimacy at some length and offers, with characteristic care not to overreach, a few possibilities: “we might be more inclined to use law to solve our higher-order coordination problem [on which focal point should we coordinate] to the extent it is legitimate and to the extent it is more legitimate than the alternative expressions” (2015, 127), and citizens and government enforcers might coordinate in an assurance game if citizens “gain utility (pride or a sense of belonging) by contributing when others contribute or perhaps lose utility (guilt, alienation) by free-riding” and legitimate governments are perceived to contribute and not free ride when they enact laws that are legitimate in the sense of pursuing general and not special interests and respecting procedural fairness (133–34).

The real puzzle, however, is not why legitimate governments coordinate. It is how legitimate governments capable of coordinating can be created. Weingast and I think the answer to that question lies in a deeper understanding of the phenomenon of coordination, particularly of third-party enforcement, itself. For this reason, I think that it is important that our theorizing about how law works not begin with the stable, powerful, and legitimate governments we enjoy in the affluent West, but rather, more fundamentally, with the raw materials of social order from which such governments can be built.