PRAGMATISM AND ITS IMPLICATIONS FOR INSTITUTIONAL ANALYSIS

We advance a pragmatist argument. It thus will help if we sketch what we mean by “pragmatist.” In doing so we try to adhere to “the pragmatic rule” which holds that “in order to discover the meaning of [an] idea ask for its consequences” (Dewey 1948, 163). Our conception of pragmatism especially highlights its consequences for thinking about the political–economic institutions that govern contemporary societies.

Three features are central to the pragmatism we endorse. First, pragmatists are committed to anti-skepticism. As Charles Peirce insisted, “complete doubt” of the Cartesian sort is impossible, a mere philosophical pretension. He did, however, countenance the sort of “real and living doubt” that we experience in the face of problematic situations and that pressures us to settle our beliefs. This is “real doubt” precisely insofar as we have reason to question our beliefs (Peirce 1958, 40, 98–101). For pragmatists, then, it is not just belief but also doubt that stands in need of justification (Putnam 1995, 20–21). Second, pragmatists embrace fallibilism. They acknowledge that even those beliefs or commitments that we hold most confidently may turn out to be false or confused. They insist that whatever certainty we may attain requires actively organizing experience in ways that subject our beliefs to challenge (Putnam, 20–21, 71).

Arguably, these first two characteristics, in combination, are what makes pragmatism a distinctive intellectual position (Putnam 1995). Anti-skepticism and fallibilism combine to sustain an instrumentalist or problem solving approach to ideas and institutions. On a pragmatist view, our ideas, principles, practices and institutions simply are tools for navigating a social and political world that is shot through with indeterminacy. Roughly, one can say that pragmatists treat those tools that prove most reliable over time as “true.” But all tools—even our most reliable ones—can fail or prove inadequate or inappropriate. Real doubt arises in just such problematic situations, in circumstances, that is, where some of our extant beliefs or practices or policies or institutions encounter significant, often unforeseen difficulties. Such circumstances, though, are unlikely to leave us bereft of all resources—it is highly unlikely that all of our beliefs and practices and institutions will fail simultaneously. That said, a pragmatist stance encourages us to place pressure on our cherished beliefs and institutions. It encourages a broadly experimental approach both to inquiry and, more importantly for present purposes, to politics. This approach, in turn, sustains a strong commitment to democratic institutional arrangements.

Some contemporary pragmatists will find this last point controversial because they insist that their philosophical views on matters of knowledge and meaning have few, if any political implications (e.g., Posner 2003, 44, 55; Rorty 1998, 228–43). We clearly do not deny that pragmatism is a philosophical view. We do, however, find the inference regarding politics unpersuasive. Here we take our cue from Dewey who observed that if “the pragmatic idea of truth has itself any pragmatic worth, it is because it stands for carrying the experimental notion of truth that reigns among the sciences…over into political and moral practices…” (Dewey 1998, 2, 110). Such experimentalism implies that pragmatists are not committed in advance to any specific moral stance, political action, or social policy.²

Yet, for reasons that will become clear momentarily, pragmatism verges on incoherence if it entails literally nothing politically.

Here we arrive at the third commitment that pragmatists make, namely, to consequentialism. This comes through clearly in the “pragmatic rule” mentioned earlier. We take this recommendation to be general in the sense that it applies, for example, to theoretical and ethical concepts, to public policies, to social and political and economic institutions and, most especially, to broad intellectual commitments such as the fallibilism and anti-skepticism that characterize pragmatism itself. In short, Dewey's assessment of the political worth of the properly philosophical commitments that pragmatists make simply applies this consequentialist impulse reflexively. Contemporary pragmatists can deny the political import of their views only by downplaying or abandoning their own consequentialist commitments.

Our mention of consequentialism might be taken as an invitation to identify pragmatism with some version of utilitarianism. This would be a mistake. In fact, pragmatists and utilitarians adopt quite different approaches to problems of judgment and justification. Pragmatists are nonfoundationalists and so deny that some nonpolitical vantage point exists from which we might render and assess political judgments. Utilitarians hold out hope that “utility” can provide just such a foundation. This is not the place to explore this difference in intellectual strategy. It is enough to indicate that on their own terms, utilitarians encounter grave difficulties once they acknowledge that there are broad expanses of politics in which interpersonal comparisons of utility are unreliable or impossible. For in such circumstances endorsing utility maximization simply amounts to endorsing democratic processes and attending to the conditions necessary to render the outcomes that such processes produce justifiable (Riley 1990).

This last point is important. Some contemporary pragmatists take the problems with utilitarianism (or other views that presume a neutral metric such as “wealth maximization”) as warrant for endorsing a “truncated form of consequentialism.” They claim that pragmatists are consequentialist only up to a certain point but then subscribe to some standard of “reasonableness” (Posner 2003, 65–67, 74–75). Such pragmatists might, however, learn from utilitarians themselves who insist that, once we despair about the reliability or possibility of interpersonal comparisons of utility, and thereby of the existence of some neutral metric by which to assess consequences, we must both identify the conditions necessary for democratic processes to operate properly and insist that those conditions be met. As we shall see, this is a considerable but not impossible task. And it is one that does not differ in kind from the effort to identify and implement the conditions presupposed for the operation of other common institutional forms, say efficient markets.³

Pragmatist consequentialism, then, is not “truncated” by a commitment to “reasonableness.” It is constrained or tempered, though, in particular applications by an awareness that consequences can be assessed only in light of some set of, in principle identifiable, initial conditions. In any given instance a major component of a pragmatist's assessment of theories or policies or institutions or practices will consist of an inquiry into what those conditions are and whether they obtain.

PRIVILEGING MARKETS I: THE TEXTBOOK ACCOUNT

Like other social institutions, markets are mechanisms for coordinating ongoing social interactions (Kreps 1990, 5–6, 187–98; Lindblom 2001). Yet markets occupy an unusual, indeed an unusually and unjustifiably privileged place in extant theories of how institutions emerge, operate, and change. This becomes clear when we treat markets as institutions.

It is important to distinguish at the outset among three concepts that economic and political theorists too often conflate. These are exchange, markets, and competition. Exchange is pervasive in social life. It occurs whenever two or more individuals transfer goods and/or services among themselves. Exchange, however, does not automatically imply market interactions. The latter emerge from the need to predictably coordinate exchange in multilateral, increasingly impersonal settings (Lindblom 2001, 37). Only once markets begin to operate under certain specifiable conditions do we observe the attractive normative features typically attributed to them. These conditions operate in various ways to establish and sustain varying degrees of competition among parties to exchange relations.⁴

So, although exchange is a mechanism of coordination common to market systems, we can reasonably expect the latter to operate in normatively attractive ways only under conditions that approximate true competition. One virtue of standard microeconomic analysis is that it clearly identifies these conditions.

In basic textbook versions, microeconomic analysis can in general show that, under specifiable conditions, (1) an equilibrium generated by market competition always yields a Pareto efficient allocation, and (2) that any such allocation constitutes an equilibrium. These results, roughly speaking, are the fundamental theorems of welfare economics (Kreps 1990, 199–202, 286–89). They establish the existence and attractiveness of market equilibrium. This is a remarkable convergence: “You should now be hearing choirs of angels and choruses of trumpets. The ‘invisible hand’ of the price mechanism produces equilibria that cannot be improved upon” (Kreps, 200). This, in turn, prompts us to ask what it means to say economists “can” establish these things? Here two subsidiary questions arise. What, more precisely, are economists doing? And how robust are the conditions under which their analytical results hold? These questions highlight why each of the three tasks facing political theorists is important.

Consider the first question. What economists offer is an analysis of the existence and qualities of equilibrium outcomes that does not explain how markets actually generate such nice equilibria. In other words this demonstration “doesn't provide…any sense of how markets operate. There is no model here of who sets prices, or what gets exchanged for what, when, and where” (Kreps 1990, 195–8). Instead, economists offer “a reduced form solution” that “describes what we imagine will be the outcome of some underlying and unmodeled process” (Kreps, 195, 187). Standard microeconomic analysis, in other words, offers little understanding of precisely how “market/exchange mechanisms” actually operate (Kreps, 195, 190).⁵

Thus the claim that economic agents will find their way to equilibrium in a decentralized process is a “rather heroic assertion” and, by implication, it “seems natural to think that we could increase (or decrease) our faith in the concept of Walrasian equilibrium if we had some sense of how markets really do operate.” Progress on this task can be made “only if we are more specific about the institutional mechanisms involved” in market interactions (Kreps, 187, 190).

Economists have generated insightful work on that score. Yet the “exploration of more realistic models of markets is in relative infancy” (Kreps 1990, 190, 195, 197). Moreover, that work underscores difficulties in simultaneously explaining how the price mechanism generates equilibrium outcomes and in persuasively demonstrating how equilibria attained in particular settings retain the attractive normative properties of competitive equilibria that economists can establish in the general case.⁶

What if we want to try to explain how the price mechanism generates an equilibrium in particular cases? First, moving to “partial” equilibrium analysis is not terribly helpful insofar as it too neglects underlying institutional mechanisms (Kreps, 263–64, 286). In essence it simply amounts to shifting the markets surrounding those under scrutiny into the same domain as what Kreps (263–64) identifies as the “larger social and political environment which general equilibrium takes as given.” Second, we cannot assume that the particular case will have the attractive normative properties that the general equilibrium argument establishes. That, however, is what partial equilibrium analysis does. Because it focuses on some particular realm of exchange it takes all others as given. It thus assumes that the interactions under consideration in fact meet the restrictive conditions presupposed by “nice” equilibrium outcomes without considering the other interactions that a general equilibrium argument would encompass (Kreps, 279–84). But from the vantage point of general equilibrium analysis this process of extraction clearly is artificial and thus is a theoretical move for which the analyst must offer a plausible defense. Otherwise there is no reason to expect that whatever equilibrium emerges in particular cases is normatively attractive in even efficiency terms. To appreciate the difficulties here consider our second question.

Economists can establish the existence of normatively attractive equilibria only under reasonably restrictive conditions. Some of these conditions are purely technical and we set them aside (Kreps 1990, 199–200, 287). Others revolve around questions about the adequacy of efficiency as a normative criterion.⁷

Although efficiency is an anemic normative criterion we grant for purposes of argument that markets can be justified in terms of their efficiency enhancing effects.⁸ We thus set these conditions aside as well.

What we do care about at the moment are several further conditions necessary to the argument that market interactions generate normatively attractive equilibria. These conditions essentially ensure that market exchange is competitive in the sense that participant's choice exerts no significant influence on the choices of other agents. In other words, they insure standard microeconomic models are populated by agents who are free and equal. These conditions are significantly less easy to relax than those we have mentioned thus far (Myles 1994, 48–49). They fall into two categories.

The first category concerns the role of information and time in market exchange (Kreps 1990, 193–95). These conditions can be relaxed with some strain if we incorporate futures markets and increasingly clairvoyant agents (Kreps 1990, 216f; Myles 1995, 48–49).⁹

Awareness of the constraints that information and time impose on economic exchange prompt economists to focus on the “transaction costs” involved in processes of search and monitoring. This in turn provided the impetus for a large literature on economic institutions (Kreps 1990). This literature largely is concerned with the ways that information and time constraints produce inefficiencies (Eggertsson 1990; North 1990; Williamson 1985). What is less often noted is that although transaction costs affect all parties to economic exchange, their burdens typically are distributed unequally. For example, information imperfections typically confer advantage on one party to an exchange in ways that clearly subvert the equal standing of other parties and thereby threaten the normative attractiveness of the resulting outcomes. This is especially troubling where persistent or systemic asymmetries give rise to similarly persistent, systemic maldistributions in outcomes. As we establish in subsequent sections, the way that the literature on institutions typically approaches its explanatory tasks implies that institutions function to mitigate such asymmetries. To the extent that it does so, this literature tacitly but systematically intermingles its normative and explanatory tasks.

The second category of conditions that concern us are meant to establish the sort of competitive environment needed to ensure that individuals choose independently by precluding such things as externalities, monopolies, and public goods. These conditions are meant to ensure that agents who populate these models are parametric in the sense that they all are more or less equally situated price takers (Kreps 1990, 202, 264–65; Myles 1995, 20). Absent such conditions not only will market competition potentially generate an inequitable distribution along the efficiency frontier (the normative problem we set aside above), but also it generally cannot be relied on to generate efficient outcomes at all (Kreps, 202–4, 288–92).

Both categories of conditions are necessary for markets to operate in “textbook” fashion. Thus we do not question whether these conditions must hold in order for economists to establish their analytical results. We simply wish to highlight the highly conditional nature of those results and to suggest how awareness of those conditions prompts us to ask how they are embodied in actual economic institutions and whether and, if so how, those institutions operate to shape economic exchange into competitive markets. Likewise when we question whether these conditions obtain, we are not challenging the economist's preferred normative criterion efficiency in the name of some other criterion (e.g., equity) but rather questioning whether the economists' normative claims are warranted on their own terms. We then, of course, also need to ask whether there exist alternative nonmarket means of coordinating ongoing interactions that are preferable in normative terms—whether those be terms of efficiency or otherwise.

Here we return to the claim that microeconomic analysis presumes further inquiry into the “institutional framework” of market interaction (Kreps 1990, 190). This is an ambiguous claim. Standard accounts readily acknowledge that the proper operation of competitive markets presupposes an exogenous institutional framework including well-defined property rights and enforceable contracts (Kreps, 263; Myles 1995, 20, 48–49). Such accounts less commonly acknowledge that insofar as they advance reduced form analyses they do not properly address the task of explaining how informal institutional frameworks of roles, norms and prerogatives define the scope and operation of actual markets. It is a truism that markets cannot themselves produce institutional arrangements of the first sort. As we argue in what follows, it is highly unlikely that they can generate institutions of the second sort. But we cannot understand how markets in fact operate without specifying how informal institutions and decentralized exchange interpenetrate to establish the sort of “trading environment” that allows us to treat markets as institutions (Lindblom 2001; Swedberg 1994).

At this point a skeptical reader might well object. She might suggest that what we really have established is a virtue of standard microeconomic analysis. For such analysis prompts us to seriously and systematically consider the role institutions of various sorts play in structuring economic exchange into more or less competitive markets. In one sense such a skeptical reader would be correct. Standard microeconomic analysis privileges markets in a relatively benign sense insofar as they treat other institutional forms primarily as responses to or remedies for market failures or imperfections. On this view markets serve as the default mechanism for coordinating social interactions, whereas various alternative institutional mechanisms represent “means of achieving the benefits of collective action in situations where the price mechanism fails” (Arrow 1974, 33). The problem arises when theorists unselfconsciously transport this relatively benign privilege from its native analytical environment into their properly explanatory and normative enterprises. This is clearest when theorists characterize a range of social and political institutions—families, the rule of law generally or enforceable contracts and property rights in particular, various constitutional provisions such as federal arrangements or separation of powers, and even democracy—as “market preserving” (Weingast 1995) or “market supporting” (McMillan 2002, 11). For this characterization sustains efforts to explain and, tacitly, to justify institutions primarily in terms of how they contribute functionally to the operation of markets. It is this way of privileging markets that we find objectionable.

PRIVILEGING MARKETS II: ACCOUNTING FOR INSTITUTIONAL EMERGENCE

Consider an abstract game theoretic treatment of institutions. Calvert views an institution as a set of rules that is sustained as an equilibrium in an underlying repeated game (Calvert 1995a, 228–33; 1995b).¹⁰

He understands the basic underlying game as an indefinitely repeated, N-person prisoner's dilemma in which players are randomly matched for pairwise interactions. He then models a set of increasingly complex versions of this game in each of which he identifies threshold values for the discount parameters of players, the size of N and, eventually, the costs of communication that sustain a variety of cooperative equilibrium outcomes, both non-institutional and institutional. These outcomes all involve some more or less elaborate variation on conditional cooperation. They range from what we might call specific reciprocity in which mutual cooperation remains noninstitutional, through the sorts of generalized reciprocity characteristic of informal social norms, and on through increasingly formalized institutions that incorporate first multilateral communication, then centralized communication, and finally centralized communication coupled with protections against higher order malfeasance.

This is in many ways a remarkable demonstration. As Calvert himself acknowledges, however, he does not actually explain the emergence of any social or political institution. His analytical exercise generates an existence result that establishes the possibility of equilibrium institutions by identifying the conditions under which they can emerge. The difficulty, of course, is that as in nearly any repeated game where players discount the future the games he examines generate multiple equilibria. This unavoidable coordination problem sets the basic explanatory task for institutional theories. Which, if any, of the available equilibrium institutions actually will emerge from social and political interaction? How might we best explain the way institutions are generated? Under what conditions will such institutions most likely persist and operate over time? These are crucial questions if we aspire to fulfill the explanatory ambitions of institutional theories (Calvert 1995a, 228). They are questions that Calvert explicitly does not address (Calvert 1995b, 70, 80–82).¹¹

These questions are important, however, because they underscore how various rational choice accounts of institutional emergence privilege markets.

There exist at least three distinct rational choice explanations for the decentralized emergence of social institutions (Knight 1995). The first is a theory of arbitrary coordination on social conventions (i.e., Schotter 1981). The second is a market-based theory of exchange and selection through competition (i.e., North 1990). The third is a bargaining theory, which explains the emergence of institutions in terms of the asymmetries of power in a society (i.e., Knight 1992). On each account social institutions resolve the strategic indeterminacy generated by multiple equilibria: any of the available equilibrium institutions might coordinate social interaction, but actors need to identify and pursue a common set. Each account starts from the microlevel, arguing that social institutions emerge as equilibrium outcomes in repeated interactions among small numbers of rational actors. Processes of institutional development culminate when one set of rules emerges as the common set of institutions for that population. Over time the maintenance and stability of that institutional arrangement is a function of the persistence of the conditions under which it was originally created. While the three approaches share the view that social institutions can emerge from decentralized interactions, they identify very different mechanisms to explain institutional emergence, maintenance and change.

The social conventions approach is animated by the idea that institutions coordinate beliefs on mutually beneficial outcomes. This explanation of institutional genesis rests on an account of how conventions are recognized and coordination is achieved. It builds on Schelling's (1960) conception of salience. The causal mechanism central to convention accounts is the existence of focal points that coordinate social expectations. Some feature of the social context in which the relevant interaction occurs establishes the salience of a particular cooperative outcome. Convergence on the focal point leads to the emergence of a convention that coordinates ongoing interactions. The key explanatory claim is that institutional rules evolve arbitrarily. The coordination process is unconsciously driven by whatever salient factor focuses the attention of the actors. The interests of any particular actor or group of actors do not determine the nature of the final institutional arrangement and so the explanation does not invoke the distributional differences the different alternatives will generate.

The contract and selection account, like the convention approach, is grounded in the logic of mutual benefit. When any two social actors perceive that they can achieve benefit from mutual exchange, they must agree on the terms of that exchange. Among these terms are the rules governing the actions of the parties during the course of the exchange relationship. In the context of the theory, these contracts constitute the social institutions produced by social actors to facilitate the achievement of socially beneficial outcomes. But these contracts can take many forms and serve merely as the set of possible institutional forms in a society. Market-like competition serves as the dynamic selection mechanism that determines the survival of various institutional forms on grounds of reproductive fitness. This is the logic behind the Alchian (1950) model of evolutionary competition employed by most economic analyses of institutional genesis. On this account, competition selects the institutional form that maximizes the aggregate benefits of the exchange. In doing so, it neutralizes arbitrary asymmetries of power among the actors, asymmetries that some might use to steer institutional selection processes in the direction of alternative, more distributionally biased outcomes.

Finally, the bargaining approach explains institutional genesis primarily in terms of distributional consequences. On this account, social institutions are a by-product of strategic conflict over substantive social outcomes. The mechanism of selection among the possible institutional alternatives is bargaining among the actors over the various alternatives. Institutional development is a contest among actors to establish rules that structure outcomes to those equilibria most favorable for them. As in any bargaining situation, some factors distinguish the actors, thus influencing the bargaining outcome in favor of one of the parties. Knight (1992) develops a theory that invokes resource asymmetries in a society as the main factor explaining how bargaining over social institutions is resolved. Resource asymmetries influence the willingness of rational actors to accept as credible the bargaining demands made by other actors. In explaining how particular social institutions are established, bargaining consists in attempts by players to commit to particular rules of behavior where the credibility of any such commitment is conditional on asymmetries in the resources available to the bargainers. Social actors who have fewer resources simply will be more inclined to acquiesce to commitments made by those who are better endowed. In this way, the existence of resource asymmetries in a society can significantly influence the selection of equilibrium institutions.

The plausibility of each explanatory account depends, in large part, on whether we expect the underlying conditions assumed by the respective models to actually obtain. The coordination approach will perform best when all actors derive equal benefit from the institutional rule. The contract and selection approach will offer the best explanation when highly competitive conditions obtain. The bargaining approach will produce the best account when actors have asymmetrical bargaining power and derive unequal benefits from the institution. The more fully these respective conditions are realized empirically, the greater the explanatory power each proposed account will have. Although we do not argue the case here, the bargaining account is more robust than the convention and contract/selection accounts precisely because the initial conditions it presupposes are considerably less restrictive that those presupposed by its rivals (Knight 1995).¹²

For present purposes it will suffice to point out that the contract/selection approach explicitly and the convention approach implicitly presuppose restrictive initial conditions that approximate those presupposed by theories of efficient operation of markets.

The relative robustness of the bargaining account has important implications for both explanatory and normative claims about the priority of markets as an institution of social coordination. These implications become clear when we differentiate two ways that market institutions enter into competing explanations of institutional genesis. Markets, or market-like mechanisms, enter into the convention and contract/selection accounts as part of the explanation to the extent that each account invokes competitive pressures that select among the institutional alternatives. Markets also enter into those accounts as part of the explanandum, as one of the range of possible institutional alternatives for coordinating social interactions. Those who argue for the normative priority of markets relative to other institutional forms presuppose that the following positive claims about these two features of the explanation actually hold. First, the conditions for the existence of market-like competitive pressure are generally prevalent empirically and readily reproducible in decentralized social environments. Second, impartial market institutions that neutralize the effects of power asymmetries are the types most likely to be produced by a decentralized process of institutional emergence and change.

The bargaining account calls both of these claims into question. As we saw in the last section, the conditions necessary for the existence of market-like competitive pressure are highly restrictive. In particular, these conditions obtain most fully where strategic interdependence among actors is minimal. It thus is extremely unlikely that those conditions will govern the decentralized emergence of social institutions. It follows that the types of institutions most likely to emerge from such a decentralized process will systematically favor the distributional interests of dominant social groups rather than neutralizing the effects of resource asymmetries in ways that would facilitate effective market competition.

Because markets generate normatively attractive outcomes only under highly restrictive conditions, our confidence in them as a means of coordinating social interaction rests on our estimation of the extent to which in particular circumstances those conditions actually obtain or might be brought about. Because those conditions of free and equal action must have institutional embodiment, microeconomic analysis of markets quite naturally leads us to take up the task of explaining how institutions emerge. In this section we have argued that the most robust account of institutional emergence actually undermines just the sort of confidence required to sustain the propensity of rational choice theorists to privilege markets. In fact we expect that in most cases the institutions, including market institutions, that emerge from decentralized social interaction will embody sorts of asymmetries that are quite at variance with conditions of free and equal participation that the efficient operation of markets presupposes.

THE BURDEN OF JUSTIFICATION AND THE PRIORITY OF DEMOCRACY

There is, despite common practice, no compelling reason to privilege markets in the explanation or normative assessment of institutional arrangements. Yet our argument against the priority economic and political theorists commonly accord to markets raises clear and especially vexing difficulties for the task of justifying social institutions generally and democratic institutions in particular. If actually existing social and political institutions typically are by-products of strategic interactions characterized by asymmetries of power, they are morally arbitrary. How, then, can we hope to justify any given or proposed institutional arrangement?

This is a heavy burden of justification. We see two general strategies that one might adopt in hopes of meeting it. One might argue that some alternative institutional form should receive default priority. Alternatively, one might reject the idea that any institutional form properly enjoys default priority and embrace more fully and explicitly the plurality of institutional mechanisms. We opt for the latter strategy. In doing so we acknowledge that the burden of justification, extending to the full array of available institutional forms, is an unavoidable implication of our “realist” explanation of institutional development. Moreover, we acknowledge that that account imposes rather tight constraints on the possible ways we might approach our normative task. In particular it undermines confidence in strategies that seek to derive justification for any institutional arrangement from the purportedly attractive processes by which it emerges. Conversely, it prompts us to focus our justificatory strategy in broadly pragmatist fashion on the consequences we anticipate different institutional forms to produce and the conditions under which we can expect them to do so. In this sense we endorse what might be called a constrained or tempered consequentialism.

There exists a plurality or range of feasible institutional forms. In addition to markets and democratic decision-making procedures this includes, but is not limited to, bureaucracy, adjudication through courts, private associations, economic hierarchies, and social norms. Once we recognize this plurality, the task of persuasively ascribing priority to any institutional form in any particular setting appears quite daunting. This in turn raises the difficult task of discerning how, in such circumstances, any heterogenous constituency might determine which array of institutional forms to use to coordinate ongoing interactions across various domains. Such decisions will raise difficult political questions to which we return next. Here it is important to reiterate two claims that underscore just how arduous this task actually is. First, as standard textbook accounts make clear in the case of markets, any available institutional form will generate normatively attractive outcomes only under particular (and in principle specifiable) conditions. Second, as the realist implications of our bargaining account suggest, in any given circumstance these conditions are unlikely to either govern the processes by which institutions actually emerge or characterize the arrangements that those processes generate. The central perplexity of institutional reform is whether any among the available institutional forms provides the sorts of resources necessary to address this acute, systemic discrepancy. Processes of institutional reform, then, highlight the normative task of identifying a decision-making mechanism capable of allowing relevant constituencies both to assess the extent to which, in particular settings, actual conditions approximate those that various institutional forms presuppose if we expect them to generate normatively attractive outcomes and, crucially, to reflexively monitor whether the conditions of its own effective operation actually obtain.¹³

Democratic institutional arrangements differ from other available political and economic institutional forms precisely to the extent that they are capable of performing these tasks. And it is on just this basis that we rest our case for the priority of democracy. This is a large claim and we do not exhaustively defend it here. We do, however, wish to address briefly several matters that it raises.

The distinctiveness of democratic arrangements stems from the two basic mechanisms—voting and argument or debate—through which they operate. Voting generates normatively attractive outcomes just insofar as it is constrained by commonly recognized conditions of freedom and equality.¹⁴

Unfortunately, the results of social choice theory make us painfully aware of the difficulties—in the form of unstable or cyclical aggregate outcomes—that emerge when we impose a set of minimal and apparently unobjectionable conditions of equality and freedom on aggregation procedures. Here is where argument or debate plays a crucial role in democratic institutions. Political argument too generates normatively attractive outcomes only insofar as it is underwritten by conditions of freedom and equality. When it operates in circumstances approximating these conditions, argument or debate constitutes what now commonly is called democratic deliberation. This is not the place for a full-fledged discussion of democratic deliberation.¹⁵ The important point is that, quite felicitously, political debate can operate to mitigate qualms about democracy expressed by theorists who focus too exclusively on mechanisms of voting. This can occur in at least two ways: democratic arrangements facilitate outcomes that are “better” than those that might emerge from alternative institutional forms and they do so in a reflexive manner. We thus justify democratic arrangements in pragmatist fashion in terms of the consequences they produce under specifiable constraints.

What do we mean when we say that one institutional form will resolve such a political problem “better” than another? From our pragmatist perspective democratic arrangements accomplish this in two senses. As a way of eliciting information regarding the provision of public goods democratic institutions afford comparative advantage over either markets or technocratic, market-mimicking institutional forms of the sort depicted in the literature on mechanism design.¹⁶

Viewed on their own terms, democratic arrangements also generate outcomes that are better in a broadly cognitive or epistemological sense.¹⁷ In a polity whose members unavoidably are characterized by an irreducible diversity of interests, commitments, and attachments, processes of political argument— especially under initial conditions that approximate those required to characterize such debate as democratic deliberation—provide the sort of competitive environment in which, minimally, participants can convey information about proposals and counter-proposals that might otherwise remain unavailable.

What do we mean when we say that democratic institutions operate in a reflexive manner? The reflexivity of democratic arrangements derives from the fact that political argument—again, increasingly so to the extent that it transpires under conditions of freedom and equality—requires relevant parties to assert, defend, and revise their own views and to entertain, challenge, or accept those of others. It derives, in other words, from ongoing disagreement and conflict. In this respect democratic institutions clearly differ from market interactions where, in the case of disagreement, parties simply “exit” to seek more favorable terms of exchange elsewhere, whereas in cases of agreement they simply trade without concern for the efficiency or distributive features of aggregate outcomes. In this sense, too, it resembles judicial decision making where (in the ideal) arguments are exchanged under established, fair conditions. But in legal settings too the parties—judges, juries, litigants—typically differ in the degree to which they are concerned about “the law” as an entity or structure writ large (as opposed to their interests in settling the case at hand) and are institutionally constrained in their ability to affect broad rules that will impact future consequences.

The key here is that market interaction and legal decision making are decentralized in the sense that those party to any particular interaction are charged neither with reviewing nor with monitoring in an ongoing way the conditions under which institutions of exchange and legal disputation themselves operate nor with assessing the consequences of how broader institutional arrangements operate. This hopefully is clear from our discussion of markets. We argue in the next section that it is true as well of judicial decision making. It is just here that democratic institutions enable a level of reflexivity unavailable in other institutional forms. For political argument not only allows members of a democratic polity to collectively revisit past substantive decisions but also allows them to collectively reconsider and revise the terms of their ongoing interactions. Moreover, because democratic arrangements structure political argument so as to allow the emergence of new ideas, perspectives, and interests and, thereby, new constituencies and oppositions, this reflexivity has an endogenous, dynamic dimension.

This propensity to inclusiveness is an attractive normative feature of democratic arrangements. But it raises a potential problem when we recall the practical effects that political theorists commonly ascribe to political argument, namely, that it can mitigate the sort of instability that social choice theorists show threatens outcomes generated by voting. More precisely, under democratic conditions political argument might induce just the sort of common or shared understanding of what is at issue in particular circumstances that makes coherent (i.e., noncyclical) aggregate outcomes possible.¹⁸

Given this, an apparent tension emerges within democratic institutions between their normatively attractive inclusiveness and their hoped for practical effect of inducing stable outcomes (Knight and Johnson 1994). This tension, however, is much less acute than it might at first appear. In practical terms it is less useful to characterize political debate as inducing agreement than to see it as structuring disagreement.

Political argument operates to mitigate the instability that plagues voting by allowing participants to try to persuade one another, where persuasion contributes to the formation of beliefs and preferences in a quite specific sense. Persuasion operates, on the view we advance, less to directly influence the substantive content of individual beliefs and preferences than to impose conceptual restrictions on the structure of individual preferences. It need not always succeed in so doing but, when successful, this amounts to establishing agreement among the relevant population regarding the dimensions of political conflict. It is a mistake, though, to focus on the agreement here. We must instead keep our eye on the conflict. That is because even when political argument succeeds in establishing a shared assessment of the relevant dimensions of conflict, it leaves considerable space for at least two sorts of persistent disagreement. First, the fact that members of some population agree on the dimensionality of conflict in no way entails that they agree substantively on any particular issue. So, for instance, the fact that the parties to some controversy might agree that what is at stake is a matter of national defense rather than, say, facilitating interstate commerce (e.g., as in the argument over the U.S. interstate highway system) this in no way means they agree regarding how much to spend on that issue. Indeed, it is entirely possible that clarifying the dimensions of conflict can sharpen disagreement along that dimension. Second, the persistence of substantive disagreement does not exhaust the room for conflict in democratic arrangements. That is because we have for some time known that aggregate stability presupposes far less than consensus over even the dimensions of conflict. As many as twenty-five to thirty percent of a population can dissent from a commonly held view regarding what is at stake in any political conflict without subverting the possibility that voting on the matter will generate a non-cyclical collective outcome (Niemi 1969).

The grounds for our claim that democracy is reflexive, then, emerge when we see how, when it is successful, it structures disagreement and, thereby, potentially fosters still further disagreement. The irreducible diversity of interests, commitments and attachments that gives rise to politics in the first place means that members of a population have an interest in monitoring the institutions that coordinate their ongoing interactions. Because democratic institutions afford space for the two kinds of disagreement we identify, they also provide the agents who populate them with ample opportunity to challenge the terms of their ongoing interactions.

Several important things follow on this view. First, in many domains this reflexivity will make democracy inappropriate—in the sense of being cumbersome and ineffective—as a first-order mechanism for coordinating ongoing social interaction. Conversely, it is a virtue of competing institutional forms such as markets and common law adjudication that they allow us to more effectively coordinate social interaction precisely because they are not reflexive in the sense we indicate. So, democracy has no more presumptive privilege as a first-order mechanism for coordinating social interactions than other institutional alternatives. However, it nevertheless enjoys a second-order priority precisely because it allows constituencies to monitor and maintain the conditions presupposed by effective operation of prevailing institutional arrangements—including, crucially, the conditions that govern its own effective operation.

It also follows that the process by which constituencies must decide which institutional forms they will rely on and for what purposes is unavoidably political. Here there is room, at least, for disagreement over both outcomes and initial conditions. With markets, for example, efficiency competes with equity, freedom and other considerations, as defensible criteria for assessing aggregate outcomes. Likewise, because actual conditions on the ground will only ever approximate the ideal conditions presupposed by the effective operation of markets, legitimate disagreement will persist over whether particular goods or services—say, sex or reproductive capacity—ought to be available on markets in the first place (Satz 1992, 1995). Markets hardly are unique in these respects. The same issues will arise with alternative institutional forms, including democratic ones, with the consequence that social actors cannot reasonably expect a neutral, apolitical metric that might be applied to such decisions. Even under ideal conditions there is no reason to expect consensus on such matters as how to weight competing normative criteria or how to judge whether empirical conditions sufficiently approach those necessary to insure normatively attractive outcomes. The fact of pluralism we recognized in the introduction suggests that there will always be space for legitimate disagreement about such matters.

Skeptical readers may find this emphasis on politics and disagreement an unsatisfactory conclusion for a pragmatic analysis of social institutions. They may question why pragmatists would focus on possibility results for ideal typical institutions rather than on the actual effects of institutions operating within a context that departs from optimal conditions. They therefore might recommend that we instead examine actual conditions in hopes of opening avenues for further theoretical analysis, something akin to comparative-static analysis of markets for other institutional alternatives. They might reason that by doing so, we might find, for example, that markets are more robust than democratic decision making to deviations from optimal operating conditions.¹⁹

And this could give greater support to claims about the appropriateness of using markets to coordinate various social interactions.

Such a skeptical response misunderstands two important aspects of our argument. First, even if we could conduct such robustness tests at the theoretical level, the results would only bear on the choices for first-order tasks, that is, on questions such as whether we should rely on markets or on bureaucracy to coordinate a specific domain of activity. It would not affect the argument about the second-order priority of democratic decision making, an argument that rests crucially on an assessment of what tasks institutions can and cannot accomplish when they are operating effectively. Second, and more importantly, such skeptics err in thinking that questions about robustness can be resolved at the theoretical level. Theoretical analysis can make important contributions to our understanding of issues of institutional performance but alone it cannot provide definitive answers to questions of actual effect (Posner 2003.) Such answers can only come from the cumulative experience of using the various institutions at our disposal. And this further highlights the importance of institutional experimentalism in generating knowledge about the relationship between institutional performance and changing social conditions.

This in turn supports our specific claim about the priority of democracy. Reflexivity is a crucial feature of experimental procedures. We trust or value experimental outcomes insofar as they emerge under proper conditions and we expend much effort in monitoring those conditions. The priority of democratic institutional arrangements emerges because they embody the sorts of reflexivity needed to monitor the conditions under which alternative institutions and practices, including democratic procedures themselves, generate normatively attractive outcomes. But, in keeping with our commitment to a plurality of institutional forms, it is important to highlight one, hopefully obvious implication of our discussion to this point: the reflexivity of democratic decision making will often allow constituencies to decide to coordinate their ongoing social interactions across expansive domains through various nondemocratic institutional forms.

DOES JUDICIAL DECISION MAKING REPRESENT A BETTER ALTERNATIVE?

Our argument for the second-order priority of democracy rests on the comparative claim that democratic decision-making has properties other institutional forms either lack or possess to a lesser degree.²⁰

One potential challenge to our argument is that some other institutional form can reflexively monitor and assess the conditions of effective institutional performance in ways comparable or superior to those we ascribe to democratic mechanisms. Any such comparison would primarily focus on the different forms of communication facilitated by different types of institutions. We have already addressed and rejected one potential alternative, the market. The crux of our argument was that markets do not facilitate the types of communication necessary for the tasks of institutional assessment and monitoring. In perfect markets prices convey all of the information necessary for effective choice. Communication between actors is about particular exchanges and not about the general functioning of the market itself. As communication extends to other types of issues, markets tend toward imperfection and inefficiency.

Here we briefly consider the analogous argument for judicial institutions: to what extent can courts adequately undertake the tasks of monitoring conditions and assessing consequences in the ongoing process of effective institutional performance? To fairly answer this question we must assess the courts, as we do with the market and democracy, as characterized in their most favorable, yet plausible, justification.²¹

Hart (1994) offers such an account, a positive conception of the rule of law that highlights the distinctive institutional features of judicial decision making.

On Hart's account, both the role of the courts as well as the standards of their effective operation are established by what he calls the rules of recognition, the set of secondary rules that dictate how the primary rules, the laws, are to be implemented, administered, and changed (Hart 1994, 94–96). The rules of recognition distribute authority among various legal and political institutions over the different tasks involved in maintaining the legal system. Within this distribution of labor, courts are primarily responsible for adjudicating controversies over how law is to be applied in particular cases. Rules of recognition are effective just so long as members of the relevant community voluntarily accept them (Hart 1994, 114–15). He thus links effective institutional performance to shared views about the legitimacy of the various institutions.

For Hart courts, typically in the form of unelected judges, derive legitimacy from the particular nature of their most common form of decision making, the application of primary rules to new fact situations. He acknowledges that there is always room for discretion in the application of rules, but insists that rules remain crucial even in such cases. He argues that discretionary judgments “must not disguise the fact that both the framework within which they take place and their chief end product is one of general rules. These are rules the application of which individuals can see for themselves in case after case, without further recourse to official direction or discretion” (Hart 1994, 136). Hart thus clearly emphasizes the “framework” in which judges act, highlighting the importance of institutional procedures as a precondition for effective performance. Although courts rely on numerous unique procedures vis-à-vis other institutional forms, most important here are those that define (1) the scope of the judicial process and (2) the nature of acceptable judicial argument. Both are intended to maintain the rule-application focus of judicial decision making.

The scope of judicial decision making is limited in at least two important ways. First, the agenda-setting power in judicial institutions is largely exogenous. Although judges often are authorized to refuse to hear various kinds of cases, they typically cannot initiate causes of action. They must instead rely on interested parties to bring controversies before the court. Second, participation in the process is limited in most jurisdictions to those actors who satisfy restrictive definitions of who constitutes an interested party. Rules of standing limit participation to those who can demonstrate that they are directly affected by the controversy at hand. Although standing may be defined differently across jurisdictions, the accessibility of participation under even the most expansive definitions restricts the scope of relevant interests.

In addition to constraining the scope of decision making, judicial procedures also restrict the types of claims that legitimately can influence court decisions. Hart's justification of the court as an institution rests on the idea that judges will render impartial decisions (Hart 1994, 207). This criterion restricts both the types of arguments that parties can employ when pressing claims and the ways judges can justify their decisions. According to Hart, an impartial judge ideally will justify her decision in terms of predetermined rules. In response to the criticism that citing precedent merely rationalizes decisions that are made on other grounds, he notes that “some judicial decisions may be like this, but it is surely evident that for the most part decisions, like the chess-player's moves, are reached either by genuine effort to conform to rules consciously taken as guiding standards of decision or, if intuitively reached, are justified by rules which the judge was antecedently disposed to observe and whose relevance to the case in hand would generally be acknowledged” (Hart 1994, 141).²²

This characterization of judicial institutions clearly is incomplete. It is sufficient, however, for the purpose of assessing whether courts might effectively monitor institutional performance. It is useful to distinguish three features of this task. The first two involve judicial oversight of other institutions, specifically monitoring conditions of effective performance and assessing the ongoing consequences of institutional choice. The third involves self-assessment, monitoring conditions and assessing outcomes related to their own performance.

Courts arguably might afford the appropriate venue for monitoring conditions presupposed by other institutions. This, for example, is what constitutional courts normally do when they address questions of constitutional rights. Rights can be conceptualized as institutional protections for individual activities and prerogatives in various institutional contexts. They enable individuals or groups to claim that their participation is being infringed in some important way. In such instances courts must ascertain whether or not effective participation, as constitutionally defined, is being adversely affected in the particular case.

Where the rules defining effective participation are clear, courts can normally do an adequate job of monitoring such conditions. As Hart suggests, courts are best at applying predetermined rules to particular sets of facts. However, courts may not be the best or only mechanism for monitoring conditions. Waldron (1999), for instance, argues that the protection of basic rights is too important to be left to unelected judges. He insists that when there are issues of rights at stake in a democratic society, the right to participate in the resolution of the controversy is of prime importance. There is room for disagreement here. Waldron's claim presumes a prior definition of how we define who is affected when some basic rights are infringed. If we define the impacted group narrowly to include only those who are directly affected by the infringement under question, judicial decision making may, given our characterization of such processes, allow for adequate resolution of particular cases. If, on the other hand, we define rights controversies broadly, arguing that such claims are always challenges to the basic definition of effective participation, then the scope of the challenge is much more expansive and the courts may prove inadequate to the task.²³

When we move on to consider how courts might handle the task of assessing institutional outcomes, we encounter the real weakness with claims that courts can satisfy the requirements of institutional effectiveness. While courts might be able to adequately monitor conditions, they are not well suited to assess the consequences of institutional choice. The relevant questions in assessing effective performance are: “to what extent is the institution satisfying its intended goals?” and “would another institutional form do a more effective job?” These are general questions the answers to which involve basic considerations of institutional goals as well as criteria of effectiveness. The institutional restrictions on judicial decision-making render courts inadequate for the effective consideration of these questions.

Institutional assessment consists in ongoing processes of experimentation. These processes must take account of the full array of interests, attachments and commitments in a society by affording them equal opportunity of access to political influence (Knight 2001; Knight and Johnson 1996, 1997). If the range of interests is restricted arbitrarily the outcome of assessment processes correspondingly will be suspect. The constraints operative in judicial decision making—constraints that restrict arguments to invoking and applying predetermined rules—preclude consideration of many issues relevant to the assessment of institutional consequences.²⁴

Similarly, restrictions on accessibility to judicial procedures circumscribe the participation of potentially relevant parties. Unlike controversies over rights violations that may be conceptualized in terms of the specifics of a particular case, assessment questions generally affect broad segments, if not all members, of the community. Therefore, constraints on participation will tend to overly restrict the range of relevant factors taken into account in the assessment of institutional consequences.

Most importantly, courts are restricted in the types of remedies that they typically seek to implement. These normally consist of sanctions aimed at forcing individuals to comply with the dictates of predetermined rules. Because judicial decision making consists primarily in applying rules, courts are ill-equipped to pose broadly normative questions about a particular policy or institution? Simply put, when effectively undertaking their legitimate role, courts are not reflexive in the sense that questions regarding the general validity of either institutional forms or procedural constraints remain beyond their purview. Their basic prerogatives are set by either constitutional or legislative pronouncements.

This reflexivity deficit is even more pronounced when we consider whether courts might satisfy the requirements of self-monitoring and self-assessment. The factors that prevent courts from effectively assessing the consequences produced by other institutions are even more relevant to self-assessment considerations. Restrictions on their agenda and on standing to participate diminish the likelihood that courts will be challenged to examine their own effectiveness. The same barriers restrict the range of interests that might influence such an assessment. Finally, the emphasis on rule-application constrains the methods by which courts might address problems within the judicial system. Together these institutional restrictions undermine the ability of the courts to be a self-regulating institution. Put simply, when the court is effectively performing its first-order functions, it must rely on other institutions for an adequate assessment of its own performance.

In summary, courts are appropriate for resolving particular controversies (including particular violations of institutional conditions), but are inadequate to the task of ultimately assessing the requirements of general institutional performance (the reconsideration of conditions as well as the assessment of institutional outcomes).²⁵

To the extent that courts might be remodeled to better undertake these more general tasks, they would have to be refashioned to look more like institutions of democratic decision-making.