The Invisible Hand in Legal Theory
What do the separation of powers, free speech, the adversary system of litigation, criminal procedure, the common law, and property rights have in common? In all of these cases and others, legal theorists have offered invisible-hand justifications for legal institutions, comparing them to explicit economic markets. Invariably, other theorists sharply criticize the invocation of the invisible hand. Yet the debates are largely localized, with few comparisons across contexts and no general and accepted account of how invisible-hand justifications might work in legal theory. In general, although there is a literature that defines the structure of invisible-hand explanations, the normative use of invisible-hand arguments as a justification for institutions, in law or elsewhere, is much less explored.
This essay has two aims. The first is to identify general conditions under which an invisible-hand justification of legal institutions will succeed, in the modest sense that it is internally coherent and plausible (whether or not true in fact). The second aim is to identify several theoretical puzzles about invisible-hand justifications that cut across the local contexts in which such justifications are offered. I believe these puzzles are, in fact, genuine and irreducible dilemmas, which arise from the very structure and nature of invisible-hand reasoning. Moreover, they can arise in the context of explicit economic markets as well. Although in some cases invisible-hand justifications fail because the analogy to markets fails, these dilemmas arise when the analogy works only too well.
Section 1 defines invisible-hand justifications, illustrates their use in various contexts in legal theory, and specifies general conditions for such justifications to be cogent, whether or not true or useful.
Section 2 discusses the dilemma of norms. Norms of truth-seeking, public-regarding action or altruism can both promote and undermine the workings of the invisible hand in the relevant legal institutions. Although it is abstractly possible to identify the conditions under which one effect or the other will predominate, the relevant norms are lumpy and cannot be fine-tuned sufficiently well to capture all, but only, the conditions under which norms are beneficial.
Section 3 discusses the dilemma of the second best. In a range of interesting cases, partial compliance with the conditions for an invisible-hand justification produces the worst of all possible worlds. The result is that such justifications have a knife-edged quality: either a move to full compliance or a move to full rejection will prove superior to the intermediate case of partial compliance. Where this is so, institutional designers face an all-or-nothing choice.
Section 4 discusses the dilemma of verification. In many cases, theorists claim that an invisible-hand process functions as a Hayekian discovery procedure, producing information that a centralized decisionmaker could not obtain. The question whether that claim holds is empirical, but it begs the question to judge the success of the invisible-hand process by comparing it to the body of information held by any single mind (such as the analyst’s) or by some centralized institution. It follows that invisible-hand justifications resting on informational benefits will have a speculative quality, but by the same token it is difficult to assemble decisive evidence against them.
1. Invisible-Hand Justifications
The most famous invisible-hand justification was pioneered by Adam Smith1 and later elaborated in neo-classical economics: under conditions of perfect competition, markets will produce allocative and productive efficiency, and (hence) Pareto efficiency. In Friedrich Hayek’s variant of the argument for markets, the market coordinates the decentralized information and tacit knowledge distributed throughout society, coordinating individual plans better than could any central planner.2 Legal theorists have adapted these arguments, and their relatives, to offer invisible-hand justifications for a range of legal institutions. Here are some of the most prominent, although the list is by no means comprehensive:
● Influenced by Adam Smith and other figures of the Scottish Enlightenment, James Madison argued that the separation of powers would protect liberty, as the unintended byproduct of a system in which “ambition [is] made to counteract ambition.”3
● Following Madison, modern theorists have argued that the separation of powers system produces a “spontaneous order” through Coasean bargaining among lawmaking institutions.4
● Madison and other framers of the American constitution were suspicious of political parties, on the traditional civic-republican ground that parties are harmful “factions.” However, following Hume, Madison recognized that the cure for the evils of party might be competition among multiple parties.5 Under pluralism in a large federal republic, each party serves as a check upon the others;6 the unintended byproduct of their interaction is freedom from majoritarian oppression and promotion of the “general good.”7
● Justice Oliver Wendell Holmes, Jr., adapting an idea developed by John Stuart Mill,8 argued for free speech on the ground that “the best test of truth is the power of the thought to get itself accepted in the competition of the market.”9 Truth arises as an unintended byproduct of the market for ideas.
● A close relative of Holmes’s approach is Alexander Bickel’s analysis of the struggle between government and journalists over the disclosure of classified government information. Bickel argued for a regime in which government may prosecute officials who leak government information, but would be barred by free speech principles from prosecuting reporters for publishing such information (absent a narrow category of very serious harms, for example to ongoing military operations). That regime, he suggested, would result in an “unruly contest” between a government biased toward secrecy and journalists biased toward disclosure; the contest would produce an “optimal” level of social disclosure overall.10
● The American system of criminal procedure has been compared to a market, in which legislatures set background entitlements and the “price” of crime is set by decentralized bargaining over sentences among prosecutors, defense lawyers and judges.11
● James Fitzjames Stephen argued that that although “the inquisitorial theory of criminal procedure is beyond all question the true one,” and a trial “ought to be a public inquiry into the truth,” nonetheless “it may be, and probably is, the case, that in our own time and country, the best manner of conducting such an inquiry is to consider the trial mainly as a litigation, and to allow each party to say all that can be said in support of their own view; just as the best means of arriving at the truth in respect of any controverted matter of opinion might be, to allow those who maintained opposite views to discuss the matter freely and in public.”12
● In a generalization of Stephen’s claim, one main argument for the adversary system of litigation is that decentralized production of information by competing parties will produce a closer approximation to truth than an inquisitorial system. As a corollary, some legal ethicists have argued that the adversary system excuses advocates from moral obligations that would otherwise attach,13 although others hotly disagree.14
● Law and economics theorists have suggested that under certain conditions the common law will evolve towards efficiency, in the sense that the rules maximize aggregate wealth. In the classic models, efficiency arises because inefficient precedents impose deadweight losses and are thus more likely to be challenged.15 A more recent variant considers conditions under which the common law evolves towards a type of informational efficiency, in the sense that all material distinctions are incorporated into the legal rules.16 In both classes of models, efficiency arises through an invisible-hand process, as the unintended byproduct of action by litigants and judges.
● The economist Harold Demsetz suggested17 that “property rights emerge when the social benefits of establishing such rights exceed their social costs.”18 Although Demsetz’s original argument took no clear position on whether this development required a “conscious endeavor,”19 later reconstructions posit an invisible-hand process in which efficient property rights evolve as the unintended consequence of decentralized action.20
Put side by side, these arguments are highly heterogeneous, yet they also display important common features. The common ground is that in every case some good arises as an unintended byproduct of decentralized action. There are also notable differences among them, the major one being that the various justifications point to different goods, such as liberty, welfare, and truth. To sort all this out I will begin by attempting to elicit the conceptual structure of invisible-hand justifications – as briefly as possible, for it is true of both invisible-hand explanations and invisible-hand justifications that “[t]he definitional details of what counts as ‘invisible hand’ are less interesting than the particular theories.”21
I suggest that an invisible-hand justification combines (1) an explanation that identifies an invisible-hand process with (2) a value theory that identifies some social benefit arising from the invisible-hand process and (3) a mechanism that explains how the invisible-hand process produces that benefit. As to condition (1), I will follow Ullmann-Margalit in supposing that an interesting invisible-hand explanation is one that takes as an input the preferences (or desires or goals) and beliefs of agents,22 and that yields as an output a structured pattern of behavior.23 The invisible-hand explanation substitutes for an explanation from design, based upon the intentional action of some designer(s).24 A hallmark of invisible-hand explanations is that none of the participants within the system need intend to bring about to the structured pattern that arises from their actions.25
However, this does not entail that the invisible-hand system must itself emerge through an invisible-hand process, although it may do so. The emergence and the operation of invisible-hand systems are different questions.26 Some explicit markets, for example, grow up through the decentralized action of many participants, whereas others arise as the result of intentional action, as when regulators design and impose a system of emissions trading. Thus top-level managers at Hewlett-Packard set up an internal prediction market in which mid-level managers would place bets on the firm’s sales in future periods – bets that, when aggregated by the prediction market, in effect collected and transmitted to their superiors valuable information that was previously dispersed throughout the firm.27
The list of invisible-hand justifications in legal theory is heterogeneous on this dimension. Some arguments, such as the claims that the common law and property rights evolve towards efficiency, are diachronic and focus on the genesis of legal institutions or rules. Others are synchronic and focus on the operation of legal institutions or rules at a given time, regardless of how those institutions or rules originated. The Madisonian argument for the separation of powers attempts to justify an invisible-hand arrangement that results from an intentional act of constitutional design, rather than arising through an invisible-hand process. Yet the same argument could be used to justify a separation of powers regime that evolved organically, as in legal systems that have no written constitution.
An explanation that satisfies condition (1) posits an equilibrium or spontaneous order; it thus excludes random or chaotic behavior. However, merely satisfying condition (1) need not at all imply that the spontaneous order is in any way desirable or beneficial.28 The Tragedy of the Commons, for example, is a spontaneously-arising structured equilibrium of waste and inefficiency. Accordingly, condition (2) requires the theorist to posit some normative good that the spontaneous order produces, turning the explanation into a justification.
On this dimension as well, invisible-hand justifications are highly heterogeneous. In the neo-classical argument for markets, the relevant good is welfare, which markets are said to maximize. In a more recent version, the argument is that the market mechanism conduces to freedom, understood as the realization of individual opportunities and autonomy.29 In the Madisonian arguments for the separation of powers and for a plurality of parties, the main good is political liberty, although Madison also assumed that freedom from factional oppression would conduce to “justice and the general good.”30 In the Holmesian argument for the marketplace of ideas, the relevant good is truth or, less grandly, information; the same goods are relevant under Stephen’s argument for the adversary system of litigation.
Moreover, these goods may be combined in complex ways. The Hayekian variant of the argument for explicit markets has a hybrid character: welfare is still the ultimate good that markets produce, but instead of being produced directly by the consummation of all desired trades as in the neoclassical argument, welfare is produced indirectly by the market’s ability to exploit dispersed information and even generate information that would not otherwise exist. This contrast, between a welfarist and an informational argument for markets, is analogous to the contrasting arguments for the efficiency of the common law: the standard argument focuses exclusively on wealth-maximization through the weeding out of inefficient rules, while a more recent hybrid argument suggests that the common law evolves towards informational efficiency.
Given all this heterogeneity, it is reasonable to question whether there is even a coherent category of invisible-hand justifications in the first place. I believe that the common structure of these arguments, in which the relevant goods are produced as the unintended byproduct of decentralized action, gives rise to important common problems, but the proof of this must emerge from the subsequent discussion. Most of the points I will make in Sections 2 through 4 are addressed to the common structure of the arguments and thus do not depend upon whether the argument has a diachronic or synchronic character, or upon the nature of the good to which the argument appeals. That said, however, the dilemma discussed in Section 4 – the justification of competition as a discovery procedure creates empirical questions which by definition cannot be tested, at least in interesting cases – applies by its terms only where truth or information is the relevant good.
Condition (3) requires the analyst to supply a mechanism that aligns the structured pattern or spontaneous order of condition (1) with the social benefit identified to satisfy condition (2). For concreteness, and to set a benchmark case, consider the economic argument for free markets in ordinary goods. In the standard welfare-economic interpretation of Smith’s cursory references to the “invisible hand,”31 the operation of the price system in perfectly competitive markets will produce long-run allocative and productive efficiency in which net social benefits are maximized. By definition, the conditions of allocative and productive efficiency also entail Pareto efficiency. In this argument, the price system is the relevant mechanism (condition 3) that ensures that the behavior of actors under perfect competition produces an equilibrium (condition 1) that maximizes social welfare (condition 2).
In a number of invisible-hand justifications, this third condition either is not clearly satisfied or else is clearly not satisfied. As an example of the former problem, the argument that decentralized bargaining in the American system of criminal procedure creates a “market price” for crime32 does not clearly posit any mechanism that aligns individual behavior with the relevant good, here social welfare. The sentences themselves, whether set in terms of jail time or in monetary fines, are numerical, continuously graded and plausibly commensurable, so that the analogy to the price system is at least somewhat plausible. However, it is hardly clear that this suffices to align private and social costs. Genuinely irrational actors, insensitive to price changes, might be selected out of markets through bankruptcy but selected into the criminal system by their inability to respond to the law’s deterrent signals. Although changes in the “price” of crime through increased sentences will affect rational criminals and thus change behavior at the margin,33 if the fraction of rational criminals is very small, those changes will also be marginal in the colloquial sense. Furthermore, punishment through incarceration, as opposed to fines, builds in a systematic misalignment between private and social costs, because the punishment itself requires social expenditures for prisons and guards.34 That extra social cost cannot be entirely folded into the private cost of crime by increasing the expected sentence, because the social cost would then increase as well – unless the costs of running the prison system are all fixed rather than marginal costs, which is implausible.
More generally, the system of criminal procedure might better be described not as a genuine market but as a system of sequential decisionmaking by autonomous monopolies, in which the prosecutor has near-unchecked discretion over whether to charge a crime, the jury has near-unchecked discretion over whether to acquit, and the judicial system has near-unchecked discretion over final review of convictions. Although the prosecutor bargains with the defendant’s lawyer over the conditions of the plea, the relevant “prices” are set by the parties’ anticipation of what the judge and jury might eventually do, and there is no bargaining between the prosecutor and the jury, the prosecutor and the judges, or the jury and the judges. In competitive markets, by contrast, the price at which A will buy from B is defined by the terms A and B could agree to with another buyer and seller, not by the terms some third party would impose upon them. Even if there were simultaneous bargaining among prosecutor, jury, judge and defendant, the theory of bargaining under multilateral monopoly is notoriously indeterminate, so it would be hard to say whether a system of this sort conduces to efficiency however defined. The market analogy here is thus noticeably attenuated.
An example of clear failure to satisfy the third condition is Demsetz’s original argument for the evolution of efficient property rights, which posited conditions under which such rights would be socially efficient, but offered no mechanism to explain why the relevant groups would move toward the efficient regime. Later analysts have attempted to supply such a mechanism by positing various invisible-hand processes conducing to efficiency, yet there are equally plausible interest-group accounts suggesting that property rights need not be efficient at all.35 The latter accounts parallel a standard argument that the common law need not evolve to efficiency. Repeat players, such as organized interest groups, have superior access to courts or differential stakes in judicial precedent and will bias the selection of cases for litigation, thereby skewing the precedents that result.36
More controversially, I believe that Madison’s invisible-hand argument for a system of separated powers clearly fails to satisfy the third condition. Although stemming rather directly from Adam Smith,37 Madison’s argument is dissimilar in a critical respect: it does not specify any mechanism that aligns the “private” costs and benefits to institutions with social costs and benefits.38 The decentralized bargaining among institutions that characterizes the separation of powers contains nothing remotely resembling a well-defined system of explicit or implicit prices, not even in the broad sense in which criminal sentences are prices. There is no systematic reason to think that this sort of bargaining will produce efficient outcomes, somehow defined, or other benefits such as the protection of liberty.
The Coase theorem holds that where transaction costs are zero, actors will bargain to efficient outcomes whatever the initial allocation of entitlements. In the limiting case, the Coase theorem implies that any institutional structure is irrelevant, because actors will simply bargain around it. By contrast, the separation of powers presupposes that standing institutional forms matter, and cannot be costlessly dissolved and reformed on an ad hoc basis.39 The consequence, however, is that a system of separated powers creates externalities that cannot always be internalized through bargaining; separated institutions may inflict externalities upon one another that represent real social losses, not merely transfers.40
In any event, of course, the transaction costs of bargaining within a system of separated powers are much greater than zero; they include all manner of posturing, pandering, bluffing, brinkmanship, and holdouts. What are the properties of such a system? The legislature, president and judiciary do bargain repeatedly over similar issues, and this produces something that vaguely resembles a marketplace for policies. Yet there are two major stumbling blocks to believing this marketplace to be efficient, in any sense. First, the institutions are themselves aggregates of individual officials with competing agendas, a great deal of agency slack between their own interests and that of the institution as such, and, especially in Congress, severe costs of coordination and collective action.41 It thus requires an elaborate analysis of particular cases to decide whether bargains among the institutions efficiently serve even the interests of their members, let alone the interests of the ultimate principals – the citizens or voters.
Second, even if institutions were internally unified, severe commitment problems cripple the marketplace for policies. There is no institution external to government that can enforce agreements reached through institutional bargaining;42 this absence of enforceable contracting remits the parties to self-help and tit-for-tat mechanisms in which cooperation is merely one equilibrium. These two problems reinforce one another, because the turnover of personnel within institutions undermines the reputational mechanisms that can sometimes produce long-term cooperation even where agreements are unenforceable. Under these conditions there is no reason to think that institutional bargaining over policies produces efficient results, and there can be no political Coase theorem.43
All this suggests that there is no general welfarist argument for the separation of powers, but as we have seen the relevant good may be political liberty, not welfare. Perhaps Madison’s argument should be understood to claim that the separation of powers will protect liberty in some rough-and-ready way, by raising the costs of enacting new legislation. On this argument, the transaction costs of the system are a virtue, not a vice.