Toward a theory of precedent in arbitration.

• Author: Weidemaier, W. Mark C.

Do arbitrators create precedent? The claim that they do not recurs throughout much of the arbitration literature. Instead, arbitration often is viewed as an ad hoc forum in which arbitrators do justice (at best) within the confines of particular cases. As an empirical matter, however, it is increasingly clear that, in some arbitration systems, arbitrators often cite to other arbitrators, claim to rely on past awards, and promote adjudicatory consistency as an important system norm. Much like courts, then, arbitrators can (but do not always) create precedent that guides future behavior and provides a language in which disputants, lawyers, and adjudicators can express and resolve grievances.

This Article provides a theoretical foundation for understanding the conditions under which precedent will (or will not) arise in arbitration. It identifies three considerations that may account for the development of precedent across a range of arbitration systems: (1) whether the system is structurally conducive to the creation of precedent; (2) whether arbitral precedent benefits the parties by filling gaps in (or displacing) state-supplied law; and (3) whether arbitrators are likely to be viewed as legitimate producers of law within the relevant context. After explaining the relevance of these considerations, the Article explores how they might apply in different arbitration contexts and sets forth a research agenda capable of shedding light on arbitration not only as a mechanism for resolving disputes, but also as a mechanism for generating robust systems of privately made law.

TABLE OF CONTENTS INTRODUCTION I. ARBITRATION AS CAPABLE OF GENERATING PRECEDENT A. A Traditional View (with Caveats): Arbitration as Particularized, Ad Hoc Decision Making B. The Creation of Arbitral Precedent: Three Case Studies 1. ICSID as an Evolved System of International Investment Law 2. International Commercial Arbitration's Weaker System of Precedent 3. An Example from the United States: Labor Arbitration 4. Precedent's Role Across Different Systems of Arbitration II. CONSIDERATIONS RELEVANT TO THE EVOLUTION OF ARBITRAL PRECEDENT A. Structural Characteristics as Necessary (Even Sufficient?) Conditions of Precedent 1. The Surprisingly Common Use of Reasoned Awards 2. Accessibility, Not Publication 3. Might Reasoned, Accessible Awards Be Enough? B. Filling Gaps in, or Displacing, State-Supplied Law 1. Arbitral Precedent as Gap-Filler 2. Arbitral Precedent as a Tool for Displacing State-Supplied Law C. Attitudes Concerning Arbitrators' Legitimacy as Producers of Law 1. Precedent's Uncertain Place in a System of Arbitration 2. Evolving Attitudes in ICSID and Labor Arbitration 3. Shaping Conceptions of the Arbitrator's Role a. Arbitral Precedent as Legitimacy-Seeking Strategy b. A Counterexample: Employment Arbitration in the United States c. The Arbitrator's Role in Fashioning Norms Concerning Arbitral Precedent III. MORE SPECIFIC HYPOTHESES ABOUT THE OPERATION OF ARBITRAL PRECEDENT CONCLUSION INTRODUCTION

Do arbitrators create precedent? The claim that they do not recurs throughout the arbitration literature. Yet this claim conflicts with a small but growing body of evidence that, in some arbitration systems, arbitrators frequently cite to other arbitrators, claim to rely on past awards, and promote adjudicatory consistency as an important system goal. Thus, although not every system of arbitration generates precedent, some clearly do.

Both theoretically and empirically, however, arbitral precedent remains a poorly understood phenomenon. As a result, assessments of arbitration's lawmaking potential vary significantly. At one end of the spectrum, some associate arbitration with confidentiality and secrecy and assert a conflict between these characteristics and the production of law. Thus, "[w]hatever else arbitration may be, it is not 'law'--the kind of findable, studiable, arguable, appealable, Restateable kind of law" that courts produce. (1) By contrast, other conceptions of arbitration's lawmaking capacity are expansive, even raising the possibility that some systems of arbitration inevitably yield "substantive results that have a systemic character." (2) If so, the question becomes whether "modern-day arbitrators fashion a commercial, antitrust, employment, maritime, securities, and contract law?" (3)

There are at least two reasons why such disparate conceptions of arbitration persist. First, despite long-standing interest in the topic, little effort has been made to identify the conditions under which arbitral precedent might arise. Too often, arbitration is portrayed as a unitary phenomenon--one that either is or is not capable of generating precedent. By failing to accommodate the diverse array of arbitration practices, the literature fails to yield testable hypotheses concerning the creation and use of precedent in arbitration. Second, even if there were well-articulated theoretical reasons to believe that some arbitration systems generate precedent, the limited empirical evidence makes it difficult to compare arbitrator behavior across systems.

This Article begins the process of filling these gaps. It provides a theoretical foundation for understanding the conditions under which precedent is (or is not) likely to evolve in arbitration. I use the term "evolve" because few if any systems of arbitration are designed with the intent to create a body of precedent. To the contrary, arbitral precedent typically arises, if at all, in systems intended "merely" to resolve disputes.

By referring to arbitral precedent I do not mean that past awards determine the outcome of future disputes. They do not. Nor do I argue that awards necessarily constrain the discretion of future arbitrators. As I will explain, there are cases in which, for very pragmatic reasons, an arbitrator may have little choice but to follow past awards. (4) There undoubtedly are other cases in which past awards play a less substantial but still material constraining role. But arbitral awards need not serve this constraining function to constitute precedent. The extent to which judicial precedent constrains judges is itself a matter of debate. (5) Yet judicial precedent remains an important legal and social phenomenon, shaping the arguments lawyers make, the explanations adjudicators provide, and serving as a focal point around which parties can order their affairs. (6) Where it exists, the same can be said about arbitral precedent. (7)

Note that this definition means that I am primarily interested in precedent as an observable phenomenon, even though precedent may sometimes operate in ways that cannot readily be observed. As an example, consider a system in which arbitrators decide cases but do not provide any explanation for their decisions and do not make their awards available to anyone but the parties. (8) Within the system, of course, arbitrators are familiar with their own past decisions and may strive to maintain consistency across cases. Moreover, when arbitrators sit in panels of three, they may share information about previous decisions. In each scenario, knowledge of past decisions may shape a decision made today. We might therefore describe the arbitration system as "precedential" even if it produces awards that obscure the operation of precedent and even if the disputants themselves are unaware that precedent exists.

Such unobservable forms of precedent, however, are not the focus of this Article. As I use the term, arbitration generates precedent if awards have some observable relevance to the future conduct of system participants. (9) For example, parties might order their affairs, and lawyers might structure their arguments, around rules announced in past awards. Likewise, arbitrators may justify their decisions, at least in part, by invoking past awards or principles deducible from past awards. Indeed, I focus in particular on how arbitrators justify their decisions, for the practice of citing to and engaging with past awards suggests that system participants invest those awards with "normative authority" and that arbitrators view themselves as engaged in a lawmaking enterprise. (10)

I adopt this definition for two reasons. First, although the existing literature often does not define the term, the definition approximates the one implicitly used by most authors. Second, the definition emphasizes precedent's important functional qualities, including the possibility that precedent facilitates private ordering by articulating rules that parties expect future arbitrators to follow. As another example of these functional qualities, consider the possibility that precedent may legitimize the result of the arbitration for the losing party, perhaps by suggesting that the result is justified by some normative criterion--say, the belief that similarly situated litigants should receive equal treatment--that the losing party is likely to accept. If it is to serve these and other important functions, arbitral precedent must be observable to the relevant constituency. To use an obvious example, third parties cannot easily structure their behavior around a rule of arbitral precedent if they do not know that the rule exists.

Part I of this Article briefly recounts the existing debate over arbitration's capacity to generate precedent. To frame the discussion that follows, Part I also offers several examples of systems in which arbitrators' awards appear to have precedential force. Although fundamentally different in design and purpose, these diverse systems of arbitration illustrate the wide range of contexts in which precedent may evolve and provide clues into how an arbitration system's characteristics may shape this process.

Part II then suggests that, notwithstanding the diverse range of arbitration contexts, a core set of considerations may help explain whether arbitral precedent is likely to evolve. Part II identifies three such considerations: (1)...