首页 | 农村土地改革 | 法经济学茶座 | 法经济学论坛
     网站首页≯ 程序法≯
仲裁
文/Bruce L. Benson

ARBITRATION
Bruce L. Benson
DeVoe Moore Professor and Distinguished Research Professor
Department of Economics, Florida State University
© Copyright 1999 Bruce L. Benson

Abstract

Within the economics literature on arbitration, some studies, generally
considering commercial disputes, emphasize arbitration-litigation relationships.
For instance, commercial arbitrators often rely on customary commercial law
rather than national law, suggesting that arbitration can be a jurisdictional
alternative to litigation rather than simply a procedural one. In support of this,
the literature demonstrates that arbitration is a potential source of precedent,
and is viable without judicial backing. Others studies, mostly dealing with labor
disputes, focus on negotiation incentives given alternative forms of compulsory
interest arbitration. Arbitration selection processes’ implications for arbitrator
behavior and the relative ‘quality’ of arbitration are also considered.
JEL classification: K41, J52
Keywords: Arbitration, Customary Law, Private Sanctions, Precedent, Interest
Arbitration, Exchangeability

1. Introduction

The law and economics literature on contracts and contract enforcement has
focused almost exclusively on judicial adjudication (Rubin, 1995, p. 113),
despite the fact that the vast majority of contracts are never adjudicated.
Outside the realm of law and economics there is a large literature on alternative
dispute resolution (ADR) including arbitration, for instance, indicating that
even when disputes cannot be resolved through negotiation or mediation, they
are directed away from national courts and into arbitration, at least for some
large categories of contracts. Lew’s (1978, p. 589) detailed examination of the
evidence on international commercial contracts concludes that around 80
percent of these contracts had arbitration clauses at the time of his study, for
example, and that over time, ‘more and more [international traders] ... turn to
arbitration’. More recent studies confirm this trend: Casella (1992, p. 1),
Berger (1994, p. 12) and others report that about 90 percent of all international
trade contracts contain arbitration clauses. Similarly, within the United States,
arbitration under the auspices of various commercial organizations, or by
independent arbitrators, perhaps from the American Arbitration Association
(AAA), resolve at least three times as many commercial disputes as the
common law courts do (Auerbach, 1983, p. 113).

Arbitration of disputes between employers (both government and private)
and unionized employees has also been routine (and even compulsory for
government employees, as well as for some private sector employees, when
negotiation proves inadequate) for several decades in the United States.
Furthermore, while non-union employees’ disputes were almost never
arbitrated before 1970, growing numbers are now resolved by arbitrators (Ware,
1996, p. 1). Arbitration is also used for disputes between businesses and
customers. For instance, the New York Stock Exchange formally provided for
arbitration in its 1817 constitution, and it ‘has been working successfully ever
since’, primarily to rectify disputes between Exchange members and their
customers (Lazarus, et al., 1965, p. 27). The Council of Better Business
Bureaus (BBB) operates arbitration programs for consumers in many parts of
the United States, several automobile manufacturers have contracts with the
BBB to arbitrate car owners’ complaints, AAA arbitrators annually resolve
thousands of insurance claims, the National Association of Home Builders
offers AAA arbitration of buyers’ complaints against association members,
medical malpractice arbitration, begun in 1929, is on the rise, and so on
(Denenberg and Denenberg, 1981). Non-contract civil disputes are also shifting
to arbitration in the United States, in part to avoid litigation costs such as
delays due to congested government courts (Bloom and Cavanagh, 1987, p. 35).
Indeed, a new private-for-profit court industry, developing since 1979, offers
a wide variety of ADR procedures to resolve all kinds of disputes (there were
more than 50 such firms in the United States 1992, most with offices in several
states (Ray, 1992, p. 191)). These firms are attracting growing numbers of
customers (as well as profits and investors (Phalon, 1992, p. 126)), including
many who do not contractually stipulate ADR prior to the dispute arising.

While arbitration has not attracted much attention in the core of the law and
economics literature, there are some exceptions (for example, Landes and
Posner, 1979; Bernstein, 1992; Shavell, 1995). Furthermore, expanding ‘law
and economics’ to include some contributions from ‘new institutional
economics’ and labor economics, reveals considerable analysis of various
aspects of arbitration, thus providing the foundations upon which a law and
economics approach to understanding arbitration’s functions can be built.
There is a much larger literature on arbitration outside of economics, of course,
including several journals exclusively concerned with arbitration or ADR, but
this review focuses on research by economists or by legal scholars who have
adopted a law and economics approach (and at times, contributions from the
larger literature referred to in this research), thus leaving out some issues that
may be attracting attention in the larger literature and/or of potential but yet
unexplored interest from a law-and-economics perspective.

The economics literature on arbitration divides, roughly, between labor
arbitration and commercial arbitration. However, within this literature, a
different (but correlated) division is also apparent. In particular, some studies,
mostly concerned with commercial arbitration, emphasize relationships
between arbitration and litigation, while other economics research, primarily
in the labor literature, focuses on arbitration’s influence on negotiation
incentives (explorations of the arbitration process are found in both literatures).
The following presentation is organized in reflection of this division. It appears
that the reason for the differences in focus arises, at least in part, because
commercial arbitration is generally an ex ante voluntary decision by both
parties to contractually specify arbitration over litigation in the event of a
dispute, while the most widely studied examples of labor arbitration by
economists (interest arbitration dealing with public sector employees) are
compulsory under statute law. This is probably due to the fact that data on some
compulsory arbitration systems can be obtained relatively easily, while data on
general grievance arbitration is not nearly as accessible. As a consequence,
however, within the economics literature, voluntary commercial arbitration is
often depicted as a cooperative endeavor to minimize the costs of dispute
resolution, while labor arbitration tends to be characterized as a much more
adversarial process. In reality, like commercial arbitration, most private-sector
labor arbitration arises through collective bargaining contracts rather than
through compulsory statutes, the primary exception being parts of the
transportation industry governed by the Railway Labor Act so, as indicated
below, the impression taken from the labor economics literature on arbitration
may be quite misleading.

First, some of the reasons for why law and economics scholars should be
interested in arbitration are discussed in Part 2. An important one, at least for
commercial law, is that arbitrators often resolve disputes under customary
commercial law and/or trade association rules rather than under the statute
and/or precedent law of a particular nation. That the choice of arbitration is a
jurisdictional issue rather than simply a procedural one might not be accepted
if, as is frequently claimed, arbitration is not viable without judicial
enforcement of arbitration agreements and rulings, and/or arbitration is simply
a compromising process and not a source of legal interpretation and precedent.
Therefore, the arguments and evidence regarding these two ‘arbitration versus
litigation’ issues are explored in Sections 3 through 6. The
arbitration/negotiation relationship as it is depicted in the labor economics
literature, primarily in consideration of the effect of alternative forms of
compulsory arbitration on incentives to negotiate, is examined in Sections 7
through 10. Arbitration selection processes and their implications for arbitrator
behavior and the relative ‘quality’ of arbitration are examined in Section 11,
and concluding comments appear in Section 12.

2. Why Should Arbitration be Studied?

There are many reasons for more careful consideration of arbitration in the law
and economics literature. For instance, the similarities between arbitration and
civil litigation are considerable, so perhaps we can learn something about more
complex dispute resolution processes by studying arbitration (Ashenfelter,
1987, p. 342; Bloom and Cavanagh, 1987, p. 353). The differences between
arbitration and litigation may also be instructive. For example, unlike judges
and juries, arbitrators tend to specialize in particular types of disputes. A
knowledgeable specialist can render a decision more quickly and with less
information transfer from, and therefore less costs incurred by, the disputants,
and is less likely to make an error, but Ashenfelter (1987, pp. 342-343) raises
another point: since disputants typically have veto power in arbitrator selection,
arbitrators competing for business have incentives to develop expertise and
render unbiased decisions echoing those of past arbitrators. Thus, as explained
below, considerable empirical evidence suggests that arbitrator decisions are
statistically ‘exchangeable’. In contrast, US juries are selected for their lack of
knowledge in an effort to assure ‘unbiased justice’ (fairness), but the
exchangeability of expert arbitrators’ decisions suggests an alternative way to
achieve bias-free justice while reducing the likelihood of error or bias:
competition between ‘experienced jurors’.

Perhaps a more important reason for considering arbitration is to gain a
fuller understanding of contracting and contract law. In this regard, the focus
on judicial adjudication in the law and economics literature on contracting
would be appropriate if court precedents and statutes as interpreted by a
nation-state’s courts (as well as court-interpreted trade agreements between
nations) was the exclusive source of law shaping contracts. This is not the case,
however, at least in the area of commercial contracting (Rubin, 1995; Benson,
1989, 1998a, 1998b; Bernstein, 1992, 1996). For instance, Lew’s (1978, pp.
582-583) examination of international commercial arbitration records
demonstrates that arbitrators often ‘denationalize’ disputes, looking instead to
a ‘non-national and generally accepted rule or practice appropriate to the
question at issue ... developed through the concerted efforts of those concerned
with and participating in international commerce’ (Lew, 1978, pp. 582-583).
That is, as Berman and Dasser (1990, p. 33) explain, international arbitrators
‘refer to international commercial custom, including contract practices in
international trade, as a basis for their award’, and no nation’s law is applied
unless the contract specifies it or the parties expect it (also see Trakman, 1983,
and Benson, 1992b).

Standards of business practice and usage within trade associations and other
commercial groups (or a trade association’s explicit internal ‘legislation’
(Bernstein, 1996)) are also the sources of many of the basic rules under which
contracts are drawn up and disputes are arbitrated within many countries.
Bernstein’s (1992, p. 126) examination of the systematic rejection of
state-created law by the diamond industry in favor of its own internal rules
(including arbitration institutions and privately produced sanctions) provides
a very revealing example: the New York diamond merchants’ ‘Board of
Arbitrators does not apply the New York law of contracts and damages, rather
it resolves disputes on the basis of trade custom and usage’. Furthermore,
despite Bernstein’s (1992, p. 116) contention that ‘the diamond industry is
unique in its ability to create and, more important, to enforce its own system of
private law’, the same abilities actually exist in many other commercial groups.
The fact is that commercial arbitration makes ‘the courts secondary recourse
in many areas and completely superfluous in others’ (Wooldridge, 1970, p.
101). Thus, while many contend that law is what judges say it is, this is only
true in contracts to the extent that a judge is expected to have the last word on
the contract; when adjudication is not sought, perhaps due to the availability of
arbitration and the benefits it produces (and/or private sanctions to discourage
adjudication, an issue discussed at length below), other behavioral rules can
and often will control the contracting parties’ conduct (Rubin, 1995, pp. 118,
124; Bernstein, 1992, 1996; Benson, 1998b, 1998c, 1998d).

Furthermore, if government courts are occasionally called upon, thus
influencing the expectations of parties to particular kinds of contracts,
‘influence is different from control... [and] One cannot assume, moreover, that
the influences flow only in one direction’ (Rubin, 1995, p. 124). Some of the
new rules for business behavior that develop through evolving practices,
contractual negotiations, and arbitration are likely to be recognized and adopted
by judges (Chen, 1992; Benson, 1989, 1990a, pp. 227-229, 1998b), for
instance. But even when the courts do not recognize custom as a relevant
source of law (perhaps because they conflict with precedents or statutes), a
judicial decision does not have to be treated as a permanent determinant of
behavioral rules. A contract can specify different types of behavior along with
a general arbitration clause, thereby nullifying the future application of the
judicial ruling (De Alessi and Staaf, 1991). Thus, an understanding of the
evolution of contract law, at least in many areas of commercial activity,
requires recognition of arbitration, not only as an ADR, but at least in
conjunction with contracts, as a mechanism for nullifying statutes and
precedents - that is, as an alternative to legislation and judge-made precedent
(Wooldridge, 1970, p. 101; Benson, 1989, 1990a, 1990b, 1992a, 1992b, 1995a,
1998b, 1998c, 1998d; Bernstein, 1992, 1996; Rubin, 1995). This is in sharp
contrast to the view that arbitration is essentially a procedural issue (for
example, Brunet, 1987), of course, but it is reinforced below when the potential
for precedent setting through arbitration is discussed. Before this potential for
law creation through arbitration can be fully appreciated, however, an
understanding of what motivates the use of arbitration in commercial contract
disputes is required.

3. Why is Commercial Arbitration Chosen Rather than Litigation?

Any dispute is ‘adversarial’ of course, but in the commercial area, the voluntary
use of arbitration is motivated by positive economic incentives. Private
arbitration appears to be an attractive substitute for litigation, in part because
arbitrators can be selected on the basis of their expertise in matters pertinent to
specific disputes, as noted above. Government judges need have no such
expertise, so arbitration reduces the uncertainty associated with judicial error
and/or bias. This specialization also means that arbitration can be accomplished
faster, less formally, and often with less expense than litigation because the
parties do not have to provide as much information to the arbitrator as they
would to a judge/jury. Another benefit arises when government court time is
allocated by waiting, since delay can be devastating to a business. Other
potentially important benefits include: (a) arbitration is generally less
‘adversarial’ than litigation so it is more likely to allow continuation of
mutually-beneficial repeated-dealing relationships; (b) if desired, privacy can
be maintained; and (c) businessmen may wish to avoid the application of
state-made law by agreeing to something in a contract that would be overturned
in a government court, since an arbitrator looks to the contract first and the
contract can specify the relevant law for deciding a case (for example, business
practice and custom rather than the statutes or judicial precedents of a
particular nation). For elaboration on some or all of these points, see
Mentschikoff (1961), Lazarus et al. (1965), Williamson (1979), Trakman
(1983), Ashenfelter (1987), Berman and Dasser (1990), Benson (1989, 1990b,
1992b, 1995a, 1998a, 1998b, 1998c), Bernstein (1992, 1996), Currie (1994),
Shavell (1995), and Rubin (1995). The potential of mutual gains (lower costs
of dispute resolution and/or a larger ‘pie’ to share in the long run) means that
there clearly is a ‘cooperative’ element to arbitration as compared to litigation.

Despite widespread recognition of at least some of the benefits listed above,
a frequent contention is that the defendant can always refuse to engage in
arbitration or to accept an arbitration ruling, so the plaintiff must be willing
and able to seek judicial (government) enforcement in order to have a credible
threat to induce acceptance (see, for example, Willoughby, 1929, p. 56; Lazarus
et al., 1965, pp. 31, 125; Landes and Posner, 1979, p. 247; Domke, 1984, p. 27;
and Shavell, 1995, p. 3). If this is correct then the contention that arbitration
often involves a choice between legal systems may be at least partially
undermined. After all, if the primary threat backing arbitration is litigation,
then for it to be credible arbitration procedures and awards will have to be
acceptable to judges.

In sharp contrast, Charny (1990, pp. 409-412) maintains that when a
‘community of transactors recognizes an authoritative nonlegal decisionmaker’
such as an arbitrator, ‘nonlegal sanctions’ will induce the members of the
community to accept arbitration and comply with the arbitrator’s judgement;
thus, arbitration and nonlegal sanctions are ‘a perfect substitute for legal
enforcement’. These nonlegal sanctions are essentially the ‘private’ (as apposed
to government-imposed) sanctions discussed in the large economics literature
suggesting that bond-posting or hostage-taking (for example, Klein and Leffler,
1981; Williamson, 1983; Kronman, 1985), including the potential loss of
reputation (for example, Kreps, 1990; Ellickson, 1991; Milgrom, North and
Weingast, 1990; Klein, 1997), provides powerful sources of credibility, and
Charny (1990) explains that this analysis also applies for commitments to
arbitrate (also see Wooldridge, 1970; Trakman, 1983; Auerbach, 1983;
Bernstein, 1992; Benson, 1989, 1992b, 1995a, 1998a, 1998b, 1998c). Perhaps
more important than these negative threats, however, are various positive
incentives associated with relation-specific reciprocities that arises in repeated
dealings (Fuller, 1964, p. 24; Axelrod, 1984; Ellickson, 1991; Trakman, 1983,
p. 10; Benson, 1989, 1998c). Indeed, one potential long-term benefit from
accepting an unfavorable arbitration award is the reciprocal commitment by a
trading partner to accept low cost arbitration and abide by an unfavorable
judgement in any future dispute. Of course, private sanctions reinforce such
incentives, so the combination can provide strong incentives to arbitrate even
without an added ‘legal’ (that is, government-imposed) threat (indeed,
reciprocities are valuable assets which can be threatened, and therefore, part of
the private sanctions arsenal).

4. Does Commercial Arbitration Rely on Reciprocities and Private
Sanctions, or on Legal Sanctions?

Those who see legal sanctions as necessary to back arbitration cite United
States arbitration history as ‘evidence’ in support of this view. In light of the
competing hypotheses that reciprocities and private sanctions create strong
incentives to arbitrate, Benson (1995a) re-examines this alleged ‘historical
evidence’: that arbitration supposedly was not in widespread use in the United
States prior to passage by New York (1920), New Jersey (1923), the federal
government of Pennsylvania (1927) and California (1925), Oregon (1925),
Massachusetts (1925, 1927) of statutes commanding the common law courts to
enforce arbitration agreements and rulings (examples of this contention include
Willoughby, 1929, p. 56, and Lucas, 1987, p. 55). Prior to this, it is often
alleged (incorrectly, as explained below) that agreements to arbitrate were
generally not considered binding under common law until these statutes were
passed, and that hostile judges felt free to overturn arbitration decisions if one
of the parties chose to litigate (examples include Lazarus et al., 1965, p. 18;
Horwitz, 1977; Murray, Rau and Sherman, 1989, p. 435; Allison, 1990, p. 11).

A relative lack of litigation in the United States regarding arbitration issues
appears to be the evidence that many have looked to in contending that
commercial arbitration was not widely used before 1920. It is true that litigation
over issues arising in arbitration increased dramatically following passage of
the 1920s statutes (see Sturges, 1930), but court records do not provide a clear
picture of the historic level of arbitration because the vast majority of
arbitration decisions are never appealed, and the statutes themselves may have
increased the propensity to appeal rather than the propensity to arbitrate, as
explained below. Moreover, an examination of newspapers, merchant letters,
and the records of organizations providing arbitration services, such as the New
York Chamber of Commerce, clearly demonstrate that commercial arbitration
actually was in widespread use in each of the British Colonies almost three
centuries before modern arbitration statutes were passed (Auerbach, 1983;
Jones 1956; Smith, 1961, pp. 180-188; Odiorne, 1953, 1954). After the
revolution, arbitration remained in wide use in all of the states (Auerbach,
1983; Jones, 1956, p. 219; Smith, 1961, pp. 180-188; Odiorne, 1953, 1954).
Why? Because ‘Not only did courts, according to one New York merchant,
dispense “expensive endless law”; they were slow to develop legal doctrine that
facilitated commercial development’ (Auerbach, 1983, p. 33).

Common law judges in America were hostile toward arbitration during the
eighteenth and early part of the nineteenth century, as they exhibited an
increasing willingness to overturn arbitrators’ decisions for issues relating to
either law or fact is evident (see Benson, 1995a for cases and references). The
use of commercial arbitration developed during the colonial and
post-revolutionary periods in spite this hostility, however, suggesting that court
backing is not a prerequisite for such arbitration. Later in the nineteenth
century, a trend toward less hostility can be detected in several state courts
(MacNeil, 1992; Benson 1995a), but arbitration continued to evolve both in
states where the courts were relatively receptive and states where judicial
acceptance was evolving more slowly (Benson, 1995a). For instance, as New
York merchants organized into various associations and exchanges, provisions
were always made for the arbitration of disputes among members (Jones, 1956,
p. 214), despite New York’s maintenance of the common law doctrine of
revocability of arbitration clauses in statutes longer than many other states.

The volume of evidence regarding the widespread and growing use of
arbitration is particularly heavy for the last four decades of the nineteenth
century (Jones, 1956, pp. 214-215; Wooldridge, 1970. p. 101; Auerbach, 1983;
Macneil 1992; Benson, 1995a). Indeed, these developments suggest that
arbitration was being substituted for litigation, since this period witnessed
rising litigation costs due to court congestion and trial delay, as well as
increasing uncertainty costs regarding the credibility of state courts’ implicit
commitments to support business contracts as ‘the growth of the regulatory
state unsettled advocates of commercial autonomy who turned to arbitration as
a shield against government intrusion’ (Auerbach, 1983, p. 101; also see
Benson, 1995a). Rising litigation costs increased incentives to establish
contractual arrangements and organizations (for example, trade associations,
commercial exchanges) to govern the process of dispute resolution and insure
against litigation (for example, arbitration clauses, arbitration institutions,
institutionalized private sanctions). Thus, by the end of World War I,
arbitration had clearly made the courts irrelevant for contract disputes by a
large segment of the business community in the United States (Wooldridge,
1970, p. 101).

Perhaps private sanctions and reciprocities are sufficient to encourage
disputants to live up to most ex ante agreements to arbitrate, but to date, most
arbitration apparently arises under preexisting agreements (or mandates in the
case of compulsory interest arbitration discussed below). Does this mean that
arbitration cannot be agreed to after a dispute has already developed, unless a
strong (that is, state-backed) sanction can be brought to bear? An alternative
explanation for the apparent lack of ex post arbitration may simply be Shavell’s
(1995, p. 9) point that the benefits of arbitration are greater when agreed to ex
ante than when established ex post. However, he also notes that an ex post
agreement to arbitrate can still reduce dispute resolution costs (Shavell, 1995,
p. 9). Other potential benefits from ex post arbitration agreements include the
possibility that the dispute will be resolved more amicably (arbitration is a less
adversarial procedure), that risks of judicial error are reduced, and/or that the
parties prefer the application of alternative law (for example, custom, the rules
designed in a trade association). Thus, arbitration should also be attractive ex
post, and perhaps legal sanctions to explicitly encourage such arbitration would
be beneficial. However, there appears to be an evolving trend toward more ex
post arbitration even without direct efforts to encourage it through legal
sanctioning. For instance, nonmembers of the diamond merchants’
organization who have a dispute with members often request ex post that the
diamond industry’s arbitration board hear their cases, and this is done if both
parties sign an ex post agreement to arbitrate (Bernstein, 1992, p. 126).
Furthermore, the recent and rapid development of private for-profit courts that
resolve business disputes, personal injury disputes, divorces, construction
warranty disputes, disputes over loan defaults, and so on, as they arise (Pruitt,
1982; Benson, 1990b; Ray, 1992; Phalon, 1992), and the growing scope of
arbitration of non-contract civil disputes, both suggest that arbitration can be
quite attractive even for disputants who do not precommit, and that private
institutions are evolving to provide such services.

Additional evidence that reciprocities and/or private sanctions are sufficient
to support a great deal of commercial arbitration at least, comes from
international commerce. International contracts often expressly exclude
litigation and almost always refer disputes to arbitration, and international
trade disputes that cannot be resolved through negotiation are almost always
arbitrated (for example, Lew, 1978; Trakman, 1983; Berman and Dasser, 1990;
Casella, 1992; Berger, 1994; Benson, 1990b, 1992b, 1998c). Certainly, statutes
in several nations, as well as various international treaties and agreements
mandate court enforcement of international arbitration agreements, but the fact,
as David (1985, p. 357) stresses, is that:

A sense of fair dealing on one hand, the respect for public opinion on the other, also
the fear of being criticized by one’s own community or that non-performance may
be interpreted as evidence that one’s business is in a critical financial situation are
reasons why arbitral awards are most generally promptly and willingly executed by
business people. Such conduct may also be encouraged if the losing party is led to
believe that the community will not remain passive in case of non-performance, but
that sanctions may be imposed if the award is not executed.

Arbitration is also attractive because international traders generally assume
that national courts will not enforce obligations derived solely from customary
commercial law (Chen, 1992, p. 100). Indeed, during the medieval period this
was particularly true, and as a result, the medieval ‘Law Merchant’ was an
almost pure system of privately-arbitrated customary commercial law backed
by private sanctions (for example, Trakman, 1983; Berman, 1983; Benson,
1989, 1990b, 1998b, 1998c, forthcoming; Milgrom, North and Weingast,
1990).

5. What are the Consequences of Legal Sanctions for Arbitration?

Rejection of the hypothesis that government sanctions are necessary for all
successful commercial arbitration does not definitely follow from the refutation
of the evidence typically supporting it. Indeed, a middle ground is obviously
also possible that may well be consistent with all of the evidence. Rubin (1994,
p. 4), for example, suggests that ‘private parties can use many available
mechanisms to make agreements self-enforcing ... [but] the law can facilitate
the use of these mechanisms ... [by enforcing] arbitration clauses in contracts
if parties insert such clauses’. Perhaps positive incentives from reciprocities
and/or private sanctions are not strong enough to support arbitration for some
(many?) commercial transactions and, therefore, statutes mandating court
backing are, on net, beneficial, as is often claimed (for example, Rubin, 1994;
Shavell, 1995a; Ware, 1996)? Frequently, however, those who call for
arbitration agreements and awards to be enforced by the judicial system look
only to the benefits of state backing without considering potential costs. The
fact is that when reciprocities and private sanctions provide sufficient sources
of credibility for transactors, state-imposed sanctions may actually raise
transactions costs (Ashe, 1983; Auerbach, 1983; Trakman, 1983; Benson,
1989, 1990b, 1995a, 1998b; Charny, 1990, pp. 426-429; Bernstein, 1992, pp.
154-157) (note that this appears to be consistent with a similar argument from
the labor arbitration literature discussed below, that the availability of
compulsory arbitration can raise the transactions costs of negotiations). For
instance, an enormous number of court cases were filed as businessmen tried
to determine what characteristics of arbitration would be considered “legal” by
the courts (Sturges, 1930), and this has continued: the early 1980s were still
witnessing a ‘growing number of court challenges to arbitration awards’ (Ashe,
1983, p. 42).

Ashe (1983, p. 42) suggests that the increase in appeals reflects the
increasing use of lawyers in arbitration because losing attorneys have a stronger
tendency to circumvent the arbitrator’s decision than does the losing party who
tends to have greater ‘allegiance to the system of arbitration itself’ when
lawyers’ ‘advice’ is not involved (that is, there is a principal/agent problem
between attorneys and their clients). It is clear that lawyers were rarely involved
with commercial arbitration in the United States before the 1920s (indeed,
Auerbach, 1983 and Benson, 1995a, contend that the arbitration statutes were
passed due to the lobbying efforts of Bar Associations, because lawyers saw
arbitration as a growing threat to their business and sought to ‘legalize’ it in
order to gain a role in the arbitration process), but as Charny (1990, pp. 388,
403-405) explains, when state-imposed sanctions are available, the demands
placed on formal contract writing are increased. If reciprocities and private
sanctions alone apply, particularly within narrowly focused commercial
organizations, less formality in contracting is required because the parties are
intimately familiar with business practice and custom in their particular area
of transactions; they understand what a general statement in a contract means,
and they can choose an arbitrator with similar intimate understanding.
However, a judge is much less likely to have such an understanding, so a
contract that may face judicial scrutiny will have to be much more specific and
formal in order to avoid a high probability of judicial error (Charny, 1990, pp.
385, 404). Thus, after the modern statutes were passed, some businesses, forced
to pay attention to the prospect of judicial review, had to make their arbitration
processes compatible with statute and precedent law including court procedure.
In order to do so, they had to consult lawyers in drafting contracts and involve
lawyers in arbitration (for instance, lawyer representation before AAA
arbitration tribunals rose from 36 percent in 1927 to 70 percent in 1938, 84
percent in 1942, and 91 percent in 1947 (Auerbach, 1983, p. 111; Benson,
1995a); also see Bernstein, 1992, p. 156 for related discussion about the
growing use of lawyers in diamond industry arbitration).

The increased likelihood of appeal reflects more than the increasing reliance
on lawyers, however. Cohen (1921, p. 150) observed, after New York
Arbitration Act’s passage in 1920, that this statute

establishes legal machinery for protecting, safeguarding and supervising commercial
arbitration. Instead of narrowing the jurisdiction of the Supreme Court it broadens
it ... Instead of being ousted of jurisdiction over arbitration, the courts are given
jurisdiction over them, and ... the party aggrieved has his ready recourse to the
courts.

When the state asserts that it is the source of authority and sanctions for
arbitration agreements it implies that rulings from arbitrated commercial
disputes are less decisive than they otherwise would be, weakening incentives
to abide by arbitrated settlements explicitly subject to potential appeal.
Furthermore, there is the possibility that upon appeal a judge will find private
sanctions themselves to be illegal (for example, as a restraint of trade). For
instance, in Paramount Lasky Corporation vs. United States (282 U.S. 30
(1930)) an explicit agreement to boycott designed to back an arbitration system
was struck down. A group of motion picture producers had agreed to place an
arbitration clause in all contracts with motion picture exhibitors, and to boycott
any exhibitor who refused arbitration or refused to accept an arbitration ruling.
Thus, government sanctions can undermine the ability or incentives to use
private sanctions (Ashe, 1983, p. 42; Bernstein, 1992, p. 156; Benson, 1989,
1995a, 1998d), and even stifle the development of trust relationships and
organizations from which reciprocities and private sanctioning mechanisms
often spring (Benson, 1989, 1995a, 1998d; Charny, 1990, p. 428). So the
possibility that government sanctions to back arbitration may benefit some
parties for whom reciprocities are uncertain and private sanctions provide weak
threats does not imply that such sanctions are, on net, beneficial, because they
may raise costs to others for whom reciprocities and private sanctions are or
could be strong enough.

6. Are Precedents Produced Through Arbitration?

Even if the net effect of court backing of arbitration is negative, however, there
may be another reason for judicial review. For instance, Brunet (1987, p. 19)
contends that ‘the output of conventional litigation should be viewed as a public
good - society gains more from litigation than would be produced if litigation
were left to the private market’. This presumably is the case because the results
of arbitration and other forms of ADR are ‘internal’ to the parties involved
(Brunet, 1987, pp. 14-15) so they will either not produce precedent, or will
underproduce precedent. Similarly, Landes and Posner (1979, pp. 238, 239,
245) argue that ‘because of the difficulty of establishing property rights in a
precedent, private ... judges might deliberately avoid explaining their results
because the demand for their services would be reduced by rules that, by
clarifying the meaning of the law, reduce the incidence of disputes’; that ‘a
problem is that a system of voluntary adjudication is strongly biased against the
creation of precise rules of any sort’; and further, that commercial arbitration
is ‘not a source of rules or precedents’. Thus, the increasing use of arbitration
apparently is eroding ‘the guidance function of the law’ (Brunet, 1987, p. 23).

Secrecy is often a characteristic of arbitration. For instance, Bernstein
(1992, p. 124) explains that within the diamond industry ‘As long as judgments
are complied with, the fact of the arbitration as well as its outcome are officially
kept secret’. But this ‘official’ secrecy does not mean that precedents are never
created, or that too few rules exist. The parties to the dispute will certainly
consider the arbitration result in future dealings under similar circumstances,
for instance, and probably have to explain them to trading partners. But more
importantly, a diamond bourse (trading club) is ‘an information exchange as
much as it is a commodities exchange’. As one author put it, ‘the bourse
grapevine is the best in the world. It has been going for years and moves with
the efficiency of a satellite communications network. ... Bourses are the
fountainhead of this information and from them it is passed out along the
tentacles that stretch around the world’ as each local bourse is part of an
umbrella organization that, among other things, arbitrates disputes between
members of different bourses, enforces arbitration judgments from other
bourses, and facilitates the establishment of uniform trading rules throughout
the industry (Bernstein, 1992, p. 121). Under such a circumstance, ‘official’
secrecy is probably not much of a constraint on the spread of important
information about an arbitration ruling that might provide new precedent. It is
clear that in the diamond industry arbitration results do ‘become known
through gossip’ (Bernstein, 1992, p. 126) at any rate.

Bernstein (1992, pp. 126-127) contends that when diamond industry
arbitrators hear complex cases ‘it is difficult to determine what substantive
rules of decisions are applied’, and such observations may appear in support of
the view that arbitration does not produce effective precedent. However, as
Fuller (1981, p. 90) explains, ‘[e]ven if there is no statement by the tribunal of
the reasons for its decision, some reason will be perceived or guessed at, and
the parties will tend to govern their conduct accordingly’. Thus, precedent of
a sort may well be produced even in such an environment. Furthermore, and
importantly, it must be recognized that within a customary legal system there
are a number of ways for new rules to evolve (for example, through unilateral
adoption of behavior that is observed and emulated, through bilateral
negotiation and contracting with resulting contract clauses spreading and
becoming standardized), besides through precedent (Benson, 1998b), so when
a particular arbitration process does not appear to be designed to produce
precedent it simply may mean that precedent is a relatively unimportant source
of new rules for the relevant group, or that circumstances do not change often
enough to require new rules. And importantly, if situations change, making
precedent more important, the group is also free to change its arbitration
procedures. Thus, as Bernstein (1992, p. 150) explains, diamond dealers have
begun to recognize that ‘The lack of written decisions and a tradition of stare
decisis makes it difficult to determine in advance the type of sanctioned
behavior. In order to increase predictability, many bourses in the world
federation have relaxed the norm of complete secrecy. Arbitrators publish
written announcements of the principles used to decide novel cases while
keeping the parties and other identifying facts secret’. Private dispute resolution
mechanisms are very flexible and diverse. They can even accommodate the
demands for precedent setting while still meeting demands for privacy.
However, privacy is not always as important as some critics of arbitration seem
to believe.

Landes and Posner (1979), Brunet (1987) and others who see in arbitration
a failure to create external benefits appear to take the characteristics of some
arbitration and extrapolate them to all arbitration, assuming that arbitration is
a much more homogeneous commodity than it really is. But if external benefits
are significant there are strong incentives to internalize them, so when
precedents become important institutional adjustments are likely to be made.
Indeed this provides traders with one of the incentives to form groups such as
trade associations and diamond bourses which clearly can internalize the
benefits of precedents. Within such an organization, it is easy to imagine a
contractual arrangement that creates incentives to minimize disputes by setting
clear precedents (for example, consider arbitrators who compete to receive a
lump-sum fee under a contract to handle all of the disputes between members
of a particular organization over a particular period of time, or arbitrators
chosen from the membership of an association who have high opportunity costs
if they encourage excessive disputes in the form of time spend away from their
business pursuits, and who also personally benefit from clear precedents). Thus,
those arbitrators who, as characterized by Landes and Posner (1979, p. 240),
‘tend to promulgate vague standards which give each party to a dispute a
fighting chance’ actually do less business. Fuller (1981, pp. 110-111) contends
that incentives are exactly the opposite, for instance: ‘Being unbacked by state
power ... the arbitrator must concern himself directly with the acceptability of
his award. He may be at greater pains than a judge to get his facts straight, to
state accurately the arguments of the parties, and generally to display in his
award a full understanding of the case’.

An obvious hypothesis is that when the environment is a dynamic changing
one in which the need for existing rules may be frequently inadequate guides
for a new dispute, arbitration rulings are likely to be recorded and/or made
known to the relevant group (which certainly is not likely to be the entire
‘public’, of course, but in general, that is not necessary or even desirable - the
idea that a single universal system of law is somehow superior to polycentric
law with parallel, as well as overlapping and competitive jurisdictions is not
consistent with either theory or reality (Benson, 1990a, 1990b, 1992a; Berman,
1983)). This in turn creates incentives for arbitrators to make careful rulings
based on past business practices, customs and precedents established within the
relevant group. For support of this hypothesis, consider international
commercial arbitration. When no clear rule applies from the ‘private
international law systems from which the parties come’ arbitrators must
determine the appropriate new rule (Lew, 1978, p. 584). To do so, ‘in their
desire to ‘denationalize’ the award and make it acceptable and fair, arbitrators
often try to show the inherent correctness of their decision in the award itself
... arbitrators may refer to several rules to show how they all lead to the same
result’ (Lew, 1978, pp. 584-585). Thus, international arbitration has
characteristics often attributed to the common law, as arbiters look to past
rulings, practices, traditions, and usage in extending the law to new issues -
that is, in producing new precedent based on older law. In fact, the general
view of international arbitrators is that, ‘owing no allegiance to any sovereign
State, international commercial arbitration has a special responsibility to
develop and apply the law of international trade’ (Lew, 1978, p. 589).

It appears that the misallocation of resources under arbitration is not be as
great as Landes and Posner (1979) and Brunet (1987) predict. Indeed, the lawmaking
consequences of private arbitration led Wooldridge (1970, p. 104) to
suggest that its substantial growth in this century has involved a ‘silent
displacement of not only the judiciary but even the legislature’. Furthermore,
Bernstein (1992, p. 117) points out that ‘nonlegal norms trump legal rules in
a given [voluntary] market only where market participants find that keeping to
the industry norms advances their own self-interest. The private regime must
be Pareto superior to the established legal regime in order to survive’. (In
addition to their contentions that private judging underproduces precedent,
however, Landes and Posner, 1979 and Brunet, 1987, p. 21 argue that the
common law produces efficient rules as contended by Rubin, 1977 and Priest,
1977, but the validity of this argument has also been subject to considerable
attack; this literature is not discussed here, but see for instance, Rizzo, 1980a,
1980b; Aranson, 1986, and Benson, 1990a, 1990b, 1992a, 1996).

7. Are there Important Differences in Labor Arbitration Institutions?

Non-compulsory labor arbitration in the private sector is also growing, as
collective bargaining agreements include arbitration clauses much like many
contracts between businessmen do. The development of repeated dealings
between unions and firms with institutionalized collective bargaining along
with strike and lockout options appears be a sufficient stimuli for arbitration of
many grievances and perhaps even contract disputes when negotiations break
down. The incentives of the decision-makers must also be recognized.
Arbitration may be relatively attractive to union leaders, for instance, because
of their legal status. They are in a position of representing everyone in the
union, including those who opposed their election and perhaps even
campaigned against them. Indeed, they have a legal duty to represent all
members, and therefore, in order to avoid a breach of fair representation
lawsuit, union officials may pursue grievance arbitration in cases that they
would prefer not pursue or would negotiate in the absence of the antagonistic
position held by the member with the grievance. Arbitration of such cases is
clearly more attractive than the alternatives (a strike or a lawsuit) that may be
much more costly for the union as a whole, including the union leader’s
supporters. Thus, grievance arbitration may offer union leaders an effective way
to avoid controversy and maintain the appearance (and indeed, the reality) of
fairness (this suggests that the availability of grievance arbitration may reduce
strikes, litigation, and negotiation - the latter is a subject that is widely
considered in the labor arbitration literature, and discussed below).

There also appears to be incentives to structure grievance arbitration so that
benefits of precedents can be internalized. After all, both labor unions and
corporate management are likely to recognize that repeated disputes over the
same type of issues are costly, and therefore that arbitration which establishes
clear precedent is desirable. In this light, consider the incident reported by
Bloom and Cavanagh (1986, p. 412): ‘a system known as expedited arbitration
was adopted by labor and management in the basic steel industry in 1971.
Under this system, unresolved employee grievances that do not require
precedent-setting rulings are arbitrated by a rotating panel of young,
inexperienced arbitrators (mostly lawyers) who decide the case for a relatively
small fee within two weeks of the decision to arbitrate’. When a resolution is
expected to be particularly valuable because it will set a precedent, more
experienced, more expensive arbitrators are chosen (factors determining the
choice of arbitrators are discussed below). This system was established because
of rising costs and increasing delays due to a relative shortage of experienced
arbitrators. While it has not yet attracted ‘an in-depth study’, it is spreading
(Bloom and Cavanagh, 1986, p. 412). United Steel Workers contracts outside
the basic steel industry now include it, and it has also been included in some
United Mineworker contracts, and is employed in the US Postal Service. Thus,
it is instructive for at least three reasons. First, it points to the fact that within
an ‘industry’ (the group of employers and the union, not simply the two parties
in a particular dispute) the benefits of precedent can be recognized and
internalized, even in some types of labor arbitration. Second, it reinforces the
fact that arbitration is flexible enough to produce a mix of dispute resolution
procedures and outcomes. Third, it suggests that even labor arbitration may
have a cooperative element, in contrast to the perception one draws from the
economics literature on the subject (Non-union employees’ disputes are also
being resolved by arbitrators in growing numbers (Ware, 1996, p. 1)). This
perception reflects the fact that the focus of the economics literature on labor
arbitration has been on compulsory interest arbitration.

Compulsory interest arbitration is almost exclusively a ‘public sector’
phenomena (baseball arbitration and some transportation sector arbitration are
among the few exceptions). Yet many of the publications in this
interest-arbitration literature begin with a brief discussion of the widespread use
of arbitration in commercial areas, perhaps also recognizing that arbitration
between individuals and their employers is frequently used to resolve
private-sector labor grievances, implicitly suggesting that inferences from their
empirical and/or theoretical examination of compulsory public-sector
arbitration apply under other institutional arrangements as well. This clearly
is not always the case, however. For instance, the compulsory-arbitration focus
of the literature might create the impression that labor arbitration in general
must rely on state backing or mandates. Much of the literature does not appear
to recognize that the institutional environment might create different
incentives, although there are a few exceptions. Bloom and Cavanagh (1987,
p. 355) and Currie (1989, pp. 365-366) note that private sanctions and
expectations of reciprocities may be quite weak for some government employers
whose incentives are tied to particular constituencies that may not view an
arbitrated (or negotiated) outcome as desirable, for example. Currie (1989, p.
366) also suggests that significant principal/agent problems may arise when
imperfectly-monitored self-interested union leaders are bargaining with
imperfectly-monitored self-interested government officials (private-sector
managers are not perfectly monitored either, but the incentives to monitor
public officials are much weaker (Benson, 1995b)). When publicly employed
laborers are protected by civil service ‘tenure’ the potential for reciprocities and
private sanctions influencing their behavior may also be sharply limited. Thus,
the institutional background may be very important, and expanding the scope
of the statutes (for example, to mandate arbitration in place of private sector
strikes and lockouts) might alter incentives to develop and use labor arbitration
institutions voluntarily, as evidenced by some of the things that have happened
in commercial arbitration since arbitration statutes were passed in the 1920s.

Compulsory arbitration can take different forms, but it often means that
strike and lockout options are precluded for public employees by the state, so
if one party requests arbitration the other can be forced to accept it. Thus,
Stevens (1966), in an early and influential contribution to this literature,
describes compulsory arbitration as a bargaining tool in a purely distributional
dispute (for example, a zero-sum dispute over the distribution of a fixed ‘pie’
to be divided between employers and a union) that has much in common with
a strike or a lockout. In particular, if arbitration is costly then when it is
threatened, it can induce the other party to negotiate. Furthermore, he explains
that arbitration mechanisms can be designed to encourage negotiated
settlements (discourage arbitration), and as demonstrated by the empirical
research discussed below, this clearly appears to be the case. He then suggests
that arbitration should be designed to impose costs on bargainers in order to
encourage voluntary settlements. In fact, following Stevens, it is often
contended that the most successful form of compulsory arbitration ‘is one that
is not used’ (Long and Feuille, 1974, p. 202). A recommendation that
arbitration be made more costly in order to encourage negotiation obviously
requires a strong normative belief that negotiated settlements are more
desirable than arbitrated settlements (Bloom and Cavanagh, 1987, p. 354;
Currie, 1989, p. 364). This norm seems to permeate the economics literature
on labor arbitration (with important exceptions, like Bloom and Cavanagh,
1986, 1987; Ashenfelter, 1987, and Currie, 1989). Indeed, as Currie (1989, p.
364) explains, ‘most industrial relations experts believe that settlements
reached by the parties themselves are superior to those imposed by an
arbitrator’, but without much theoretical basis for the belief. There is some
theoretical basis for the conclusion, of course. In particular, as Stevens (1966)
alleges, standard interest arbitration as a compromising process tends to ‘chill’
bargaining as it creates incentives for negotiators to misrepresent their true
underlying profit- and/or utility-maximizing positions, expecting arbitrators to
split the difference between their extreme demands. However, there are
alternative theoretical explanations for the relationship between arbitration and
negotiation that can produce the same empirical fact and quite different
normative implications.

Law and economics scholars might anticipate that arguments for
encouraging negotiations are based on Coase’s (1960) famous theorem, that if
the transactions costs of bargaining are not prohibitive an efficient allocation
of resources is achieved through voluntary negotiation. However, the decision
to arbitrate obviously implies positive transactions costs for bargaining (perhaps
in the form of uncertainty for the union representative regarding the ‘political’
or legal consequences of a negotiated settlement that might be perceived as
biased or unfair to some of the membership), and in that context the typical
inference drawn from Coase is to lower the cost of negotiating (for example, by
clarifying property rights, or in labor arbitration, perhaps by allowing the
aggrieved party rather than the elected union leaders to choose a negotiator),
not raise the cost of the alternative. Thus, for instance, in the diamond industry,
Bernstein (1992, p. 124) explains that parties who submit a dispute to
arbitration must first participate in a conciliation proceeding before a
conciliation panel that attempts to help the parties negotiate a settlement. Only
if such third-party aided negotiation fails can the dispute go to arbitration. This
is clearly a mechanism intended to encourage negotiation without raising the
cost of arbitration, in sharp contrast to Stevens’ (1966) proposal, and an
estimated 85 percent of the diamond-industry disputes submitted to arbitration
are settled during this conciliation procedure.

Moreover, as Williamson (1979) explains, characteristics of the transaction
that determine transactions costs (for example, degree of asset specificity,
frequency of the transaction, uncertainty) determine the efficient organizational
choice, given ‘market’, ‘bilateral’ (direct negotiation and contracting), and
‘trilateral’ (use of a third-party), or ‘ownership-integration’ governance
options. Under these circumstances, if arbitration is used it may be because the
transactions costs of negotiation are relatively high, so fair arbitration is a more
efficient means of achieving a settlement. Thus, encouraging negotiation by
raising the costs of arbitration can reduce efficiency. As Posner (1986, p. 366)
notes in his discussion of summary jury trials and other ADR options, the
objective should not be to maximize negotiated settlements but to minimize the
total costs of the system of achieving settlements, whether voluntarily or
through third-party procedures.

Perhaps the points made here do not apply to Stevens’ (1966) arguments
about compulsory labor arbitration, however. The disputants themselves
presumably have not voluntarily designed this governance arrangement if it is
mandated by statute (in all likelihood, public sector unions and employers
lobbied to influence the statutes, of course, but the statutes may not allow the
transaction-specific adjustments in governance organizations that would lead
to efficient outcomes (Benson, 1995b)). Compulsory arbitration still may be a
cooperative outcome if it is the low-cost dispute resolution mechanism relative
to a particular high-transactions-cost negotiation, however. Indeed, Bloom and
Cavanagh (1986, p. 409) point out that ‘labor practitioners place greater
emphasis [than economic theorists] on arbitration as a mechanism for helping
disputants identify and reach efficient outcomes. In their view, arbitrators are
professional gatherers and processors of information who play a highly
constructive role in a bargaining process which is better treated as a cooperative
attempt at problem solving than as direct economic conflict’. Furthermore,
Bloom and Cavanagh (1986, p. 421) conclude, from their study of the selection
of arbitrators (discussed below), that ‘the similarity of union and employer
preferences for different arbitrators suggests that collective bargaining
functions more cooperatively than most existing models indicate’. Doubt is cast
on the theoretical underpinnings of this hypothesis by considering the
assumption that arbitrators split the difference between offers. Farber (1981)
raises important theoretical questions regarding the assumption, noting that
even empirical evidence of difference splitting would not be sufficient to prove
that this is the objective of the arbitrator since parties are likely to position their
offers around the expected arbitration award. That is, ‘the expected arbitration
outcome shapes the parties’ bargaining positions rather than the reverse’
(Farber, 1981, p. 70). Gibbons (1988) goes even further, modeling arbitration
in the context of a three-party game in which arbitrators also learn about the
state of the employment relationship from the parties’ offers.

8. Does Final Offer Arbitration Encourage Negotiation (or Discourage
Arbitration)?

Stevens (1966) and several others propose what has come to known as final
offer arbitration (FOA) to overcome the hypothesized tendency for
compromising arbitration to ‘chill’ negotiation. FOA requires the arbitrator to
choose one or the other of the two sides’ final offers rather than choosing some
compromise between them. The idea is that the negotiators will have incentives
to move toward relatively reasonable positions before arbitration, but in doing
so they are presumably more likely to reach a negotiated settlement on their
own. FOA has been implemented in labor arbitration through statute law in a
number of places, and as a consequence, it has attracted a good deal of
theoretical analysis.

Crawford (1979) provides one of the earliest formal theoretical
examinations of FOA. He notes that if the arbitrator’s exogenously determined
notion of what a settlement should be is known to the two parties then FOA and
conventional arbitration will lead to the same outcome in a zero-sum
distributional game. Under conventional arbitration neither party would settle
for a distribution different from the arbitration outcome so all awards have to
correspond to that outcome whether they are achieved through negotiation or
arbitration. If the arbitrator selects the final offer closest to his preferred
outcome, the same is true, as one party can always improve on any final offer
by offering the arbitrator’s preferred settlement. Of course, given positive
arbitration costs (that is, if use of arbitration makes the process a negative-sum
game), arbitration should actually not be used at all. This might appear to
desirable, given the arguments by Stevens (1966), Long and Feuille (1974) and
others, but in this case the parties are simply agreeing on what the arbitrator
would award anyway, so arbitration is actually the determining factor, not
negotiation (Ashenfelter and Bloom, 1984, p. 112).

Crawford (1979) also draws on Schelling’s (1963) model of commitment,
suggesting that parties find it advantageous to commit to a position that is
costly to disavow. Given uncertainty about the strength of the parties’
commitments, the potential for commitment by both parties can lead to a
disagreement and therefore to arbitration. Factors that increase the payoff to
commitment increase the likelihood of disagreement. Still, the model predicts
that FOA should not increase the likelihood of negotiation. Indeed, Crawford
(1979, p. 152) suggests that ‘FOA may create a Prisoner’s Dilemma situation
and thereby prevent agents who would otherwise cooperate from agreeing on
an efficient settlement’. Of course, in a repeated game setting the prisoner’s
dilemma outcome of non-cooperation is less likely (Axelrod, 1984), so if the
interaction generating the dispute is not viewed by the negotiators as a one-shot
game this result may not hold. Nonetheless, Crawford’s framework does not
appear support the contention that FOA should increase the propensity to
bargain.

Actually, Crawford (1979) does raise one point that implies that FOA can
be relatively ‘costly’. If negotiations break down, neither final offer may
actually be efficient (or equitable, depending on the objectives of the arbitrator
and/or the nature of the dispute - the outcome of a zero sum dispute may reflect
the arbitrator’s view of equity, for instance). If there are multiple issues
involved in the dispute, for example, both final offers may involve inefficient
mixes of solutions. Constraining the arbitrator to choose one or the other
would, therefore, guarantee an inefficient solution. Crawford (1979) concludes
that ‘multiple FOA’ is superior to single FOA arbitration, but under
conventional arbitration the arbitrator has a full range of outcomes to consider,
so an efficient solution is also possible. If conventional arbitrators are expected
to search for an efficient solution (or an equitable distribution in a zero-sum
bargain) rather than to split the difference, and if bargaining costs are high,
FOA could easily ‘encourage’ more bargaining and more costs than
conventional arbitration does, but in this case, the result actually involves
higher transactions costs. Thus, whether FOA is desirable or not depends on
what bargainers expect arbitrators to do, which in turn should depend on what
they actually do Bloom (1981, p. 243).

Farber and Katz (1979), and Farber (1980, 1981) offer a different view of
compulsory arbitration than Crawford (1979). They assume that the parties are
uncertain about arbitrators’ preferences and explain that the likelihood of a
bargain depends on the parties’ expectations about the arbitration award and
their risk preferences. For instance, similar expectations about arbitration
outcomes and significant arbitration costs can produce a contract zone, and
Farber and Katz (1979) suggest that a larger the contract zone increases the
probability of a bargaining settlement. Divergent expectations (for example,
overly optimistic expectations by at least one party), on the other hand, can
prevent successful bargaining. However, Farber and Katz (1979) conclude that
uncertainty about arbitration’s outcome coupled with risk aversion on the part
of at least one party, can also produce a contract zone, while greater certainty
about the arbitration award, and hence less risk, may mean that no contract
zone exists and arbitration results. Thus, in keeping with Stevens’ contention
that arbitration should be structured to encourage negotiation, Farber and Katz
(1979) express concern that uncertainty will decline as parties gain experience
with arbitration, and that arbitration will become more prevalent over time (this
is one variant of the ‘narcotic effect’ hypothesis discussed below). In addition,
the relative impact of FOA and conventional arbitration depend on the parties’
attitudes toward risk and their expectations about arbitrator preferences, among
other variables.

In line with Stevens (1966), Kochan (1980) contends that FOA is more
effective at imposing costs on parties if they fail to reach a negotiated
settlement. After all, the expected award from FOA is a probability weighted
average of two points while the expected conventional award is a probability
weighted average of a distribution of potential awards, and by removing the
ability of an arbitrator to compromise, the relatively low risk middle part of this
distribution is removed. FOA appears to be riskier, and as Farber and Katz
(1979) emphasize, uncertainty regarding the behavior of the arbitrator is a
source of arbitration costs. But this reasoning is flawed (Farber, 1980;
Crawford, 1982; Farber and Bazerman, 1987, 1989). For instance, Farber and
Bazerman (1989) point out that the risk of FOA can be manipulated by
manipulating offers, and FOA eliminates the high-risk tails of the distribution
as well as the low-risk middle. Indeed, they use existing estimates of arbitration
behavior, assuming constant absolute risk aversion utility functions and
identical normal prior distributions on the arbitrator’s concept of an appropriate
award, and conclude that contract zones are unambiguously larger under
conventional arbitration than under FOA. Given that identical-expectation
contract zones are relatively large, it should take a greater degree of divergent
expectations to offset the potential for bargaining under conventional
arbitration. Thus, if larger contract zones imply that a voluntary settlement is
more likely, negotiation should be more prevalent under conventional
arbitration.

In contrast to the standard view, however, Bloom (1981) explains that wider
contract zones do not necessarily imply a greater likelihood of successful
negotiations (a point also made by Crawford, 1981). After all, there may be
substantial direct costs of negotiations, as well as uncertainty about settlement
points within the contract zone. Indeed, it may be easier to agree if there is one
point in the contract zone that both parties prefer to arbitration than if there are
a large number of alternatives. Thus, raising the costs and/or uncertainty
associated with arbitration may not lead to more negotiated settlements. In this
light, Farber and Bazerman’s (1989) finding of larger contract zones under
conventional arbitration does not necessarily imply more negotiations, a fact
that they recognize. However, they point to another of their results, contending
that the best explanation for relatively more negotiated settlements under FOA,
as hypothesized by Stevens (1966), comes from their finding that the
expected-utility-maximizing-Nash-equilibrium last offers given to an arbitrator
(where a negotiator is indifferent between the last offer and the expected
arbitration award) are closer under FOA than conventional arbitration. Thus,
they conclude that strategically, when equilibrium offers are farther apart the
parties are more reluctant to make concessions. If this is the case, then FOA
does not appear to raise the cost of arbitration, although it does influence
bargaining behavior.

An alternative hypothesis, apparently not found in the literature but
tentatively proposed above, is that when bargaining costs are high arbitration
becomes relatively attractive because arbitrators are expected to search for the
efficient (or equitable) solution to a dispute rather than simply splitting the
difference, and that FOA constrains such a search so FOA is less attractive
(more costly) than conventional arbitration. Note that this hypothesis is
consistent with recent empirical findings by Burgess and Marburger (1993) that
‘FOA awards tend to be of “low quality” in the sense that they lie outside the
range of negotiated settlements’ in their study comparing baseball player
salaries determined through negotiation versus those determined through FOA
arbitration. Indeed, A number of empirical testable hypotheses follow from this
theoretical literature, and a growing body of empirical research addresses many
of them.

9. Do Arbitrators Split the Difference?

The ‘split-the-difference’ assumption appears to be key to the conclusion that
conventional arbitration chills negotiations (Feigenbaum, 1975; Feuille, 1975;
Anderson and Kochan, 1977; Starke and Notz, 1981). In this regard, Bloom
(1986) finds that on average, arbitration decisions in his sample appear to be
mechanical compromises between the parties’ final offers, but he also
emphasizes that there is in fact a substantial amount of unexplained variance
in arbitration awards, so that the overall conclusion need not apply to any
particular case. His examination of written arbitration decisions also finds a
correlation between final offers and the facts in each case (a finding supported
by several studies discussed below). That is, negotiators tend to adjust final
offers to be consistent with the facts that arbitrators are expected to look at, as
Farber (1981) suggests, and therefore arbitrators can use the parties’ final offers
as a source of information about facts. Furthermore, Bazerman and Farber’s
(1985) study of wage awards by 64 actual arbitrators in 25 simulated
conventional arbitration cases clearly suggests that arbitrators look at both final
offers and the facts of the case, weighing the facts more heavily than the offers.
The facts also are increasingly important as the offers diverge, contradicting
‘the naive split-the-difference view of arbitrator behavior’ (Bazerman and
Farber, 1985, p. 76).

Farber and Bazerman (1986) report the results of a similar study comparing
arbitrator behavior under FOA and conventional arbitration. They assume that
the same criteria, determining awards based on facts rather than offers,
characterizes arbitrator behavior under different arbitration schemes. Their
findings support the assumption: again, conventional arbitration awards are
weighted averages of the facts and offers with weights depending on the
reasonableness of the offers, and final offer awards are also determined by the
facts and the offers. Thus, it appears that arbitrators are generally consistent in
their behavior under different arbitration regimes (Farber and Bazerman, 1986,
p. 842). Ashenfelter and Bloom (1984) reach similar conclusions about the
virtually identical decision processes for arbitrators under conventional
arbitration and FOA, using a data sample drawn from New Jersey where public
employee disputes can be decided under conventional arbitration if the parties
agree to it, but FOA if they do not. Arbitrators, it seems, do much more than
split the difference.

10. Does Conventional Interest Arbitration ‘Chill’ Negotiations?

Despite the significant evidence that arbitrators do not simply split the
difference, there does appear to be empirical evidence of a chilling effect to
conventional arbitration. One source of evidence (although there is in fact no
strong theoretical linkage (Currie, 1989, p. 364)), is the hypothesized ‘narcotic
effect’ (Kochan and Baderschneider, 1978, p. 431) of third-party dispute
resolution. It is contended that when compulsory arbitration is available parties
become reliant upon it rather than negotiating. Early studies conclude that such
an effect exists because, for example, wage settlements appear to be less
responsive to market conditions when third-party procedures are employed (for
example, Kochan and Baderschneider, 1978; Butler and Ehrenberg, 1981, and
Auld, et al., 1981). Bloom and Cavanagh (1987, p. 355) point out that the
simple descriptive statistics which appear to support the hypothesis are far from
conclusive (also see Anderson, 1981), however, as a complete test of the
hypothesis would require establishing serial correlation in the use of arbitration
after controlling for heterogeneity across bargaining units.

Currie (1989) provides what may be the most sophisticated empirical
analysis of the ‘narcotic effect’ hypothesis to date. She finds, controlling for a
number of other factors, that bargaining units who used arbitration in the
previous round of negotiations are at least 10 percent more likely than other
bargaining units to use it in the current round, thus supporting the
narcotic-effect hypothesis. Yet, the total number of arbitrations in previous
years (her sample included 35 years of compulsory arbitration by teachers in
British Columbia) has a negative effect on the probability of arbitration, and
variables intended to capture attitudes toward risk, changes in the degree of
uncertainty associated with arbitration awards, and differing beliefs about
awards do not have significant effects on the probability of using arbitration.
Thus, Currie (1989, p. 378) concludes that the typical explanations for a
narcotic effect do not appear to hold, so ‘A theoretical explanation of this
“stylized fact [positive state dependence]”, perhaps in the context of a dynamic
model of arbitration, is badly needed’. Indeed, the result is consistent with the
idea suggested above that an experience with arbitration simply reveals that the
net benefits of arbitration are higher than previously believed, so when the
transactions costs of negotiation prove to be high, those with experience tend
to opt for arbitration relatively more quickly. A very similar point is made by
Bloom (1981, p. 243), when he observes that if the split-the-difference
hypothesis does not hold and arbitration appears to chill negotiation it may be
because ‘the parties view this mechanism [compulsory arbitration] as the least
costly alternative for establishing a contract’. Of course, if this is the case, the
‘chilling’ of negotiations may well be appropriate from an efficiency
perspective, and designing arbitration mechanisms to raise the cost of
arbitration is inefficient (another possible explanation for increasing relative
use of arbitration might be that union leaders perceive an increasing probability
of breach of fair representation lawsuits in light of the perceived increases in
the propensity to litigate in the United States, leading union officials to pursue
more grievance arbitration in cases that they would prefer to negotiate in the
absence of this perceived threat).

There is yet another set of studies that probably provides the strongest
evidence of a potential chilling effect of conventional arbitration: the growing
literature comparing negotiation under FOA and conventional arbitration. It
appears that FOA increases the probability of a negotiated settlement (Feuille,
1975; Neale and Bazerman, 1983; Farber and Bazerman, 1989). Thus, Stevens
(1966), Farber and Katz (1979), and others who suggest that the nature of the
arbitration process influences the likelihood of a bargained outcome appear to
be correct. But does this support the contention that arbitration should be
structured to encourage negotiation or that FOA is desirable? The answer still
depends on the relative costs and benefits of negotiated and arbitrated
settlements, as Bloom (1981) suggests. In this regard, the evidence of arbitrator
behavior, including evidence from the FOA literature, is instructive. Some
characteristics of arbitration behavior have already been discussed, such as the
fact that they apparently do not split the difference. However, in order to fully
appreciate the reasons for this and other characteristics of arbitrator behavior,
we must first look at the arbitrator selection process.

11. How are Arbitrators Selected?

Many critics of arbitration and other forms of ADR contend that these private
alternatives to litigation will be biased. The proposed reasons for such bias
vary, however. Some contend that arbitrators are easily corruptible, so the bias
is in favor of the disputant with the most financial resources: ‘If the rendering
of verdicts is to be independent of the relative wealth of the litigants, then the
provision of judicial services naturally requires separation of the
decision-makers gain from that of each litigant. This fact either requires heavy
regulation or it requires public provision of the judge directly’ (Mabry, et al.,
1977, p. 83). A different bias is seen by Landes and Posner (1979, p. 254) who
contend that ‘it might seem that competition would lead to an optimal set of
substantive rules and procedural safeguards. But this is incorrect. The
competition would be for plaintiffs, since it is the plaintiff who determines the
choice among courts having concurrent jurisdiction of his claim. The
competing courts would offer not a set of rules designed to optimize dispute
resolution but a set designed to favor plaintiffs regardless of efficiency’. Brunet
(1987, pp. 31-40) simply concludes that arbitration’s procedures bias results
away from ‘quality’, ‘accurate’ results because they do not incorporate as much
information as litigation does (Currie, 1994, makes a similar point except that
she focuses on the relative use of information in arbitration and negotiation,
and she is much more tentative in her normative conclusions). Essentially, he
sees ADR procedures as too ‘informal, ambiguous, and not administered in a
managerial fashion’ and lacking effective discovery processes (Brunet, 1987,
pp. 31-33).

All such arguments involve a failure to recognize the potential for creating
an arbitrator selection process to avoid such biases. Indeed, as Bloom and
Cavanagh (1986, p. 409) explain, ‘one of the most important characteristics of
arbitration systems is that they may be designed in different ways’. Not
surprisingly, arbitration selection mechanisms actually vary widely. For
example, within some organizations a single arbitrator or panel is appointed for
a set period to arbitrate all disputes between members. Of course, prescreening
occurs because these arbitrators are chosen from a competitive pool by the
association through its membership approved selection process. For instance,
in the diamond industry, arbitrators are elected from the organization’s
membership for two year terms (Bernstein, 1992, pp. 124-125). A different
alternative is that contracting parties or organizations preapprove a list of
arbitrators, and if a dispute arises, an arbitrator is chosen from the list by some
preset mechanism (for example, random selection, rotating selection, selection
by a third party such as a governing board of the association). Yet another
common selection system gives the parties to a dispute the resumes of odd
numbered list of arbitrators from a larger preselected group (for example,
preselected by a trade association, or provided by the AAA), with each party
having the power to successively veto names until one remains. Thus, a second
level of prescreening is added at the disputant level, contributing ‘to the
legitimacy of the arbitrator and his award in the eyes of the parties’ (Bloom and
Cavanagh, 1986, p. 409). The parties are also given the arbitrator’s résumés,
so they know about experience, training, the nature of awards given in the past,
and so on. A similar practice, the subject of Bloom and Cavanagh’s (1986)
empirical study discussed below, is also common: the parties are given a list
and résumés of seven potential arbitrators with the power to veto three and rank
the other four, and the arbitrator who is not vetoed by either party and has the
highest combined rank is chosen. Another common practice is for both sides
of the dispute to provide a list of, say, six arbitrators, and each can then veto
any or all of the names on the other party’s list; if all names are vetoed each
provides another list and the process is repeated (clearly, this procedure
requires that both parties want to arbitrate, so they do not continue to provide
unacceptable names). All such systems guarantee the appointment of an
arbitrator without requiring explicit agreement by the two parties while still
allowing for prescreening of the potential arbitrators.

In light of the fact that selection mechanisms allow veto, each party might
be expected to support at least some candidates that are expected to be
unbiased. In fact, it is widely contended that arbitrators are chosen for their
expertise, reflected by their experience and training, as well as for their
consistency in deciding cases solely on their merits, and for their impartiality
(for example, Elkouri and Elkouri, 1985), but before Bloom and Cavanagh’s
(1986) study, little statistical evidence existed to verify this. They examine the
selection of arbitrators for 75 public employee disputes in New Jersey. A
number of findings are of interest. First, in many cases the union and the
employer have similar rankings, suggesting that their preferences for arbitrator
characteristics are at least moderately similar. Second arbitrators with perceived
biases tend to be eliminated. Thus, arbitrators that exhibit a tendency over time
to favor labor are given poorer rankings (generally vetoed) by management and
vice versa. Similarly, economists get low rankings by labor, relative to lawyers
and labor relations practitioners, perhaps because ‘economists are likely to be
heavily influenced by efficiency conditions’ (Bloom and Cavanagh, 1986, p.
418) such as the advantages of competition, mobile labor, and other factors that
labor unions tend to undermine. Third, both employees and employers show
significant preferences for experience, controlling for impartiality and training.
Clearly, while it is not possible to prove that experience reflects arbitratorspecific
characteristics such as expertise, it probably does because experience
both adds to expertise and is a reflection of past popularity. Fourth, while a
number of variables representing impartiality, experience and training appear
to be significant, their magnitudes are small, suggesting ‘that the parties are
relatively indifferent to many of the arbitrators in the New Jersey system’
(Bloom and Cavanagh, 1986, p. 419). It would appear that the preselected
group of 70 arbitrators that each panel of seven in this sample is drawn from
are actually perceived to be quite similar. As Ashenfelter (1987) explains, this
is precisely what should be expected.

12. Conclusions: ‘Exchangeability’ and the Benefits of Arbitration

Given that disputants have the same information about arbitrators and the
power to veto any arbitrators that are expected to be biased (Ashenfelter, 1987)
explains that arbitrator decisions should be statistically ‘exchangeable’.
Furthermore, a number of independent empirical studies tend to verify this
exchangeability hypothesis (for example, Ashenfelter and Bloom, 1984;
Bazerman and Farber, 1985; Farber and Bazerman, 1986; Ashenfelter, 1987;
Bloom, 1986; Faurot and McAllister, 1992; Burgess and Marburger, 1993).
Consider for example, Faurot and McAllister’s (1992) study of salary
arbitration in major league baseball. They hypothesize that an arbitrator will
attempt to base decisions on the same criteria with the same weights as other
arbitrators in order to avoid being vetoed, and find that the expected value of
the arbitrator’s notion of a fair settlement is significantly influenced by a
number of fact-based variables specified in the contract (the Basic Agreement):
player performance in the last year, career performance and consistency, recent
club performance, and previous compensation (Faurot and McAllister, 1992,
p. 710; also see Burgess and Marburger, 1993).

Furthermore, while some evidence might appear to suggest that arbitration
is biased, a careful examination generally reveals the opposite. For instance,
Ashenfelter (1987, pp. 343-345) discusses a sample of New Jersey’s FOA
system for police disputes in which the employers win only about a third of the
decisions. However, both conventional arbitration and FOA are used in this
system, so they can be compared. Recall the evidence discussed above that
arbitrators choose conventional awards using a consistent criteria based on the
facts, and that FOA arbitrators use the same criteria, choosing the final offer
closest to their preferred award. In this light, Ashenfelter (1987) compares
conventional and FOA awards and finds that on average, union offers were
more reasonable than the employee offers (perhaps because union bargainers
were more risk averse, or perhaps because employers were making
unreasonable offers for political reasons), so the disproportionate number of
union wins reflects bargaining positions, not arbitrator bias, and is actually
consistent with the exchangeability hypothesis. He also examines a set of data
from the Iowa FOA system for public employees where, in contrast to New
Jersey, the employer offers were more frequently chosen, but in this case it
appears that the employers’ offers were, on average, the ‘more reasonable’.

It is exchangeability, Ashenfelter (1987, p. 341) suggests, that underlies the
continued acceptability of labor arbitration systems. Indeed, given the political
influence of unions and bureaucratic organizations (Benson, 1995b), it is hard
to imagine that compulsory arbitration statutes could survive if they were not
perceived by both parties to be beneficial relative to alternatives (strikes,
lockouts, litigation, negotiation when transactions costs are high). The view in
the commercial arbitration literature that arbitration is a positive-sum process
appears apply to compulsory labor arbitration as well.

The private sector is responding to the demand for resolution of a wide
variety of disputes by offering a wide variety of specialized dispute resolution
options. Specialization has desirable consequences in the production of most
goods and services. It is reasonable to expect that this is also true for the
production of justice. Indeed, many of the efficiency-based arguments against
arbitration (for example, that it will produce ‘low quality’ biased judgments,
that it cannot efficiently produce precedent) are undermined by the flexibility
of arbitration and other ADR options. Furthermore, while there is evidence that
appears to support some arguments against at least some arbitration, such as
the hypothesis that its availability raises the transactions costs of negotiation,
it appears that the evidence is also consistent with other hypotheses. Thus, the
evidence to date does not provide strong reasons to be discouraged by the fact
that arbitration is being substituted for litigation and legislation.

 


中国法律经济学网登载此文出于学术研究之目的,绝不意味着中国法律经济学网赞同其观点或证实其描述。以上内容仅供研究者学习与交流,无意侵犯版权。如有侵犯您的利益,请告知。我们将尽快删除。

加入日期:2007/1/13浏览次数:2261
发表评论
名号:
内容:
验证: 1304
法律经济学网
联系站长: 柯华庆 lawgame@263.net 京ICP备09028584号
北京市昌平区中国政法大学法学院(102249)
本网站由卡卡鱼网提供技术支持 网站总访问量:1996356