The Flight from Arbitration: An Empirical Study of Ex Ante Arbitration Clauses in the Contracts of Publicly Held Companies

Volume 56 Issue 2 Winter 2007: Symposium - Is the Rule of Law Waning in America?

DePaul Law Review DePaul Law Review

Article 7

The Flight from Arbitration: An Empirical Study of Ex Ante The Flight from Arbitration: An Empirical Study of Ex Ante

Arbitration Clauses in the Contracts of Publicly Held Companies Arbitration Clauses in the Contracts of Publicly Held Companies

Theodore Eisenberg

Geoffrey P. Miller

THE FLIGHT FROM ARBITRATION: AN EMPIRICAL STUDY OF EX ANTE ARBITRATION CLAUSES IN THE CONTRACTS OF PUBLICLY HELD COMPANIES

Theodore Eisenberg* and Geoffrey P. Miller**

Informed parties bargaining for their mutual advantage will tend to agree to provisions that maximize the social surplus. Such bargaining includes provisions regarding the resolution of disputes that might arise under the contract. Thus, if a form of alternative dispute resolution, such as binding arbitration, provides greater social benefits than litigation, the dynamics of the process should tend to induce the parties to include a clause submitting future disputes to arbitration.' This Article studies the actual contracting practices of large, sophisticated actors with respect to arbitration clauses. We examined over 2800 contracts, filed with the Securities Exchange Commission (SEC) in 2002 by public firms, for the presence of contract terms requiring arbitration. Little evidence was found to support the proposition that these parties routinely regard arbitration clauses as efficient or otherwise desirable contract terms. The vast majority of contracts did not require arbitration; only about 11% of the contracts did.

Although the results reported here test hypotheses about the frequency of arbitration clause use, the findings do not provide definitive answers as much as they generate more hypotheses. The surprisingly low frequency of arbitration clauses, and their varying frequency across contract types, generate questions about the characteristics of

the parties, their contracts, and their attorneys that should be the object of future modeling and research. For now, we interpret our findto ings as evidence arbitration, encounter obstacles to negotiating mutually satisfactory contract terms that include arbitration clauses, or some combination of these factors.

that sophisticated actors prefer

litigation

Part II of this Article sets forth the hypotheses about arbitration clauses that were tested. Part III describes the data, and Part IV reports the results. Part V discusses these results, and Part VI offers some tentative conclusions.

A. Arbitration Clauses Generally

One view of arbitration suggests that sophisticated parties bargaining to maximize surplus from trade would adopt binding arbitration clauses in their contracts nearly as a matter of course.2 Arbitration is often said to be cheaper than litigation.3 It can be completed more

quickly,4 it can lower risks,5 it can involve limited discovery,6 it avoids the risk of overly generous awards handed out by judges and juries,7 and it can be used in large class actions.8 Procedural disputes can be resolved by informal, flexible procedures.9 Limited judicial review of arbitration awards reduces the costs of the procedure,10 although it also increases the risk of an erroneous award.11

Costs and risks can be further reduced because parties have choices in arbitration that extend beyond the simple decision of whether or not to arbitrate. The parties also enjoy broad autonomy to select the arbitrator. The parties can assure themselves ex ante that the forum for resolving their disputes will be relatively unbiased. Such assurances may not be available in litigation, where judges and juries may come to the case with prejudices and preconceptions that the parties cannot control. Some argue that because arbitrators compete to be selected by the parties, they have an incentive to develop reputations for fidelity to the parties' positions.12 Furthermore, parties may specify the rules governing the arbitration proceedings,13 control the schedule of hearings,'4 specify the applicable law,15 specify the rules of evidence, and determine the law governing the interpretation of the arbitration clause itself.'6 The parties may even be able to determine contractually the finality of the arbitrator's decision17-although this is subject to dispute.'8

More subtly, it has been noted that arbitration can confer an adjudicative benefit relative to litigation because arbitrators may be better than judges at detecting substandard performance by a contracting party.19 In such cases, an arbitration clause's ex ante existence en-

courages parties to render proper performance. The expertise of arbitrators may supply another benefit that warrants sacrificing the benefits of litigation. Where contracts are highly standardized-as in the case of pooling and servicing or trust agreements-arbitration may add value compared to litigation because expert arbitrators selected by the parties might be able to give a better informed interpretation of the highly specialized terms contained in these agreements. Uncertainty about arbitration is further reduced because, as a legal matter, little doubt exists as to the enforceability of the arbitration clauses in our sample. The Federal Arbitration Act (FAA) provides that arbitration agreements are enforceable "save upon such grounds °20 This as exist at law or in equity for the revocation of any contract. statute preempts state laws and operates to the full extent of Congress's power under the Commerce Clause.21 A plaintiff wishing to avoid arbitration in the face of an arbitration agreement must show that a ground "for the revocation of any contract" applies to the arbitration agreement.22 The preference for arbitration expressed in the FAA is enthusiastically endorsed by many courts as part of a general trend towards encouraging the "outsourcing" of dispute resolution proceedings.23 Some courts have even gone so far (perhaps questionably) as to enforce arbitration agreements involving parties who were not privy to the contracts containing the clause.24

But skeptical views of arbitration exist, even among proponents of the procedure. Aside from adjudication costs, some regard arbitration as potentially inferior to litigation.25 The parties must pay the arbitrators; if the chosen form of arbitration involves two party-selected arbitrators and a neutral one, this means paying three highly compensated

professionals. Judges and juries, in contrast, are paid by the government.26 A trial usually has a definite beginning and end, and in the case of jury trials, the process is compressed because of the need to maintain the empanelled jury. A party's legal fees in arbitration are not necessarily lower than they are in litigation.27 Nor are arbitrations necessarily quicker than trials.28 Arbitration proceedings are subject to the schedule of the arbitrator. Proceedings can be interrupted for months at a time, only to be reconvened at significant cost to the parties. Unlike judges or juries, the arbitrator, whether paid hourly or daily, has short-term economic incentives to prolong the proceedings.29 Further, if an arbitrator issues an award that is not supported by the evidence, it may be difficult or impossible to correct the error. The potential lack of effective appellate review makes arbitration risky relative to litigation.30 Arbitrators may lack the ability to grant emergency or interim relief of the sort that is available to judges under their equitable powers. Cases that might be too expensive to litigate may be affordable to arbitrate; a defendant might therefore find itself paying on a claim in arbitration that would not typically generate a lawsuit.31 The parties cannot assume that arbitration proceedings will be confidential or exempt from discovery in later litigation.32 Thus, some commentators encourage attorneys drafting arbitration clauses not to include boilerplate provisions, but rather to think carefully about what provisions should be in the agreement.33

Notwithstanding these skeptical views, the bulk of authority seems to agree that arbitration is a more efficient dispute resolution procedure than litigation. This generates the hypothesis that, in a data set of contracts freely agreed to by sophisticated parties, we will almost

always observe arbitration clauses being used.34 Thus, if an arbitration clause is absent from an agreement, it is reasonable to infer that the parties either preferred litigation over arbitration, were unable to reach agreement despite the perceived benefits of arbitration, or were simply indifferent.35

The incidence of arbitration clauses, if it is significantly less than 100%, provides a rough measure of the degree to which arbitration is perceived to provide greater efficiency in dispute resolution than litigation. Thus, we hypothesize that sophisticated parties will nearly always opt for binding arbitration.

B. Arbitration Clauses in International Contractual Settings

To the degree that arbitration clauses are not uniformly adopted by sophisticated parties, it is useful to examine the reasons why parties select arbitration or litigation. The difference between domestic and international contracts is one possible reason. Arbitration is viewed as particularly valuable in the case of international contracts. Arbitration is said to have become "the preferred mechanism for resolving international disputes.''36 One reason for the widespread use of arbitration in commercial contracts is that people distrust the legal systems in foreign countries.37 International arbitration awards also offer greater certainty of enforcement than domestic legal judgments. While other countries might not, and often do not, afford full faith and

credit to the judgments of U.S. courts,38 arbitral decrees are widely recognized and enforced under the New York Convention.39 The parties' choice of law may be entitled to greater respect in international arbitrations than in domestic ones, where the court or the arbitrator may refuse to follow the designated law because it does not bear a reasonable relationship to the parties or the transaction.40 Thus, a widely held, though not universal,41 view is that arbitration clauses are particularly beneficial in the case of.contracts between parties located in different countries. We hypothesize that binding arbitration will be observed more frequently in international contracts than in domestic contracts.

C. Standardized Contracts and Industry Practices

Some commentators have suggested that arbitration clauses will be preferred, relative to litigation, when the rules of decision are clear and unambiguous.42 Further, by making standardized provisions subject to binding arbitration, a repeat player can avoid the risk of being subject to offensive nonmutual collateral estoppel in disputes with

new parties under identical contract terms.43 This suggests that we will observe binding arbitration terms in standardized contracts. Arbitration clause usage rates are likely associated with industry contracting practices, which may be only partly captured by a measure of contract-type standardization. It is likely that some industries have developed a practice of including arbitration clauses to a greater extent than other industries. These considerations suggest the hypotheses that arbitration will be more commonly observed when contracts are standardized than when they are negotiated individually, and that arbitration clause usage rates will vary by industry.

D. Choice of Law

The decision to require arbitration may relate to the law chosen to govern the arbitration decisions. It is difficult to separate ex ante whether arbitration was chosen in a particular contract because other considerations dictated a particular choice of law, or a particular law was chosen because of a desire to provide for arbitration. Regardless of the direction of causation, contracting parties may regard some state laws as providing desirable or undesirable arbitration-related features. Therefore, one might expect to observe an association between the presence of an arbitration clause and choice of law or other factors that will influence the choice of law.

Two classes of reasons relating to the choice of law may influence the decision to include an arbitration clause. First, parties may agree to arbitrate based on the perceived efficiency of a state's laws and courts. In states in which litigated cases proceed expeditiously, contracting parties may feel less pressure to include arbitration clauses. States with slow or inefficient case processing may increase incentives to include arbitration clauses.

Second, state laws regarding arbitration can influence the incentives to include arbitration clauses. For example, California allows an arbitration proceeding to be stayed pending the outcome of related litigation.44 It is also less receptive than most other states-including New York-to arbitration clauses banning class actions.45 These features

might cause parties designating California law to include or not to include an arbitration clause if they regard these provisions as desirable or undesirable. New York's highest court has held that awarding punitive damages in an arbitration proceeding violated public policy,46 a decision that has not been widely emulated elsewhere,47 but is still cited by New York courts.48 If state law varies with respect to the enforceability or other treatment of particular arbitration provisions, we would expect that the incidence of arbitration clauses would vary depending on the ties between the contract in question and a state.

Our models below treat the decision to include an arbitration clause as a function of the ties between the contract in question and a state. The direction of the association, like the direction of causation, may also be uncertain. Parties who strive for certainty in arbitration may prefer the laws of states that limit judicial authority to influence the substance or operation of arbitration proceedings. Parties who strive for flexibility or fairness in arbitration may be less reluctant to embrace the laws of states more willing to allow judges to influence or review arbitration. It is thus difficult to predict whether a particular state's laws will be associated with the increasing or decreasing presence of arbitration clauses.

What does seem clear is that the perceived attitudes of state courts toward arbitration, and the particular contents of a state's laws, can influence the. decision to include an arbitration clause. For present purposes, we hypothesize that, as described above, California's histor-

ically flexible attitude towards judicial involvement in arbitration49 will attract arbitration clauses on the ground that flexibility may promote fairness in a particular case at the cost of some certainty. We do not mean to suggest that particular arbitration clauses, such as those relating to class actions, are necessarily relevant to each of the contracts studied here. We do suggest that perceptions of a state court's general willingness to strictly enforce arbitration clauses, as reflected in state law, may influence decisions to include arbitration clauses. We hypothesize that the incidence of arbitration will vary geographically, and there will be more arbitration clauses for contracts with California connections than for contracts with New York connections.

E. Relational Contracts

time. Without detailed knowledge of

If arbitration is less disruptive and adversarial than litigation, then we would expect. to see arbitration selected more frequently in relational contracts in which the parties expect to work together for a substantial period of the relationship between the contracting parties, one cannot precisely identify which contracts are relational and which are not. But we believe some contract categories are more likely to be relational than others. The contract categories are described in Part III. For now, we identify the following categories that tend to be relational: credit commitments, employment, licensing, pooling and servicing agreements, security agreements, and trust agreements. In contrast, asset sales, bond indentures, mergers, securities purchase agreements, and underwritings are less likely to be relational. Settlement agreements are too ambiguous to lump with one group or the other. We hypothesize that arbitration clauses will be observed more frequently in relational contracts and less frequently in one-off contracts.

F Prior Research

Little information exists about arbitration clause incidence in sophisticated contracts; we know of no study examining as broad a range of contracts as this study.50 The empirical evidence that does exist

suggests that arbitration has been expanding as a dispute resolution mechanism.5' The overall case load of the American Arbitration Association grew "from about 1000 cases in 1960 to more than 17,000 in 2002," with a decline to 15,800 in 2003.52 International arbitration is also reported to be increasing.53 A 1998 Cornell University study of Fortune 1000 companies yielded responses from more than 600 companies. The study found that alternative dispute resolution (ADR) processes are well established in corporate America, widespread in all industries, and used for all types of disputes. ADR practice was not haphazard or incidental, but rather seemed integral to a systematic, resolve disputes.54 long-term change Whatever the growth rate, the Cornell study also found that arbitration is not as ubiquitous as some might believe. About 80% of the companies stated that they had used arbitration "occasionally" or less often; only 7.5% reported using arbitration frequently.55

the way corporations

Several studies look to the incidence of binding arbitration of employment disputes in the workplace. Contrary to the predictions of theory, these studies find that employers have not overwhelmingly adopted binding arbitration procedures over litigation, despite their perceived benefits. A 1995 General Accounting Office Study of 1448 establishments subject to the reporting requirements of the Equal Employment Opportunity Commission (EEOC) found that only 9.9% had adopted binding arbitration.56 Alexander Colvin's studies of the telecommunications industry found that only 14% to 16% of firms

adopted binding employment arbitration.57 Stewart Schwab and Randall Thomas's study of CEO contracts in large public firms found that 41.6% had arbitration clauses.58

Other studies have examined the incidence of arbitration clauses in franchise agreements. An informal review of franchise contracts by Christopher Drahozal found that about half of Minnesota franchise agreements provided for arbitration.59 He found that the agreements that did provide for arbitration typically involved major national franchisers such as McDonalds. He seemed surprised that the incidence of arbitration agreements was so low.6° Drahozal, in a study co-authored by Keith Hylton, found a statistically significant, negative correlation between the percentage of company-owned units in a franchise network and the probability of an arbitration clause appearing in the franchise contract.61 Both of these results suggest that arbitration clauses will appear in most of the contracts studied here.

In areas other than employment and franchising, studies find low rates of arbitration clause presence. Florencia Marotta-Wurgler's recent study of 597 end-user software license agreements found low rates (about 6%) of arbitration clauses.62 And Elizabeth Rolph, Erik Moller, and John Rolph found that only 9% of a sample of California physicians and hospitals used arbitration agreements.63

With respect to international contracts, there are estimates that as many as 90% have binding arbitration provisions.64 Drahozal and Richard Naimark report that, from 1993 to 1996, 15 of 17 (88.2%) international joint venture agreements in the SEC EDGAR database included an arbitration clause.65 Alessandra Casella reports that officials of the Netherlands Arbitration Institute indicate that more than 80% of private international contracts have arbitration clauses.66 But

it is recognized that systematic knowledge of international arbitration is thin.67

The data consist of thirteen categories of contracts contained as exhibits to Form 8-K, "current report" filings with the SEC. A form 8-K must be filed by SEC-reporting firms to disclose certain material corporate events that have not previously been reported by the company. We searched the Form 8-K filings and coded information about any contract that fit into Table l's categories. The resulting sample consisted of 2858 contracts.

The types of contracts studied are listed in Table 1. Each of the contract types in Table 1 was reported to be, or to be related to, a material corporate event. For twelve contract categories, six months of contracts were studied, covering the period from January 1, 2002 to June 30, 2002. For merger contracts, the study covered a seven-month period from January 1, 2002 to July 31, 2002. The slightly expanded period for merger contracts exploits our earlier detailed work on choice of law and choice of forum in merger contracts.68 Most of the contract types are self-explanatory. Pooling and servicing contracts are used in mortgage pass-through and other asset-backed securities arrangements; they represent agreements under which an owner transfers receivables to a trustee who holds title to, and collects the income from, the assets and passes the funds through to investors.69 Trust agreements establish these trusts and define various powers and responsibilities.70

Securities purchase agreements were the most frequent contract type and accounted for 16.1% of the total contracts. Credit-related contracts-bond indentures, credit commitments, pooling and servicing agreements, and security agreements-accounted for another 20.3% of the contracts. Merger contracts were about 14% of the sam-

Asset sale/purchase (314) Bond indentures (155) Credit commitments (216) Employment (111) Licensing (48) Mergers (411)

Pooling and servicing (173) Securities purchase (460) Security agreements (37) Settlements (72) Trust agreements (48) Underwriting (351) Other (362)

pie, although they did have one extra month of coverage in the data. Together, the contract types offer a rich variety of relations. Several types, including the credit-related contracts, obviously involve substantial financial institutions. Asset sale/purchase and merger contracts involve corporate restructurings. Settlements involved the resolution of disputes. Employment contracts offer insights into dispute resolution contract terms in agreements between key employees and large corporate employers.

As noted in our previous study of choice of law and forum in merger contracts, the contracts in our sample are attractive objects of study.71 Because the contracts are material events in the lives of publicly traded corporations, the contracts are a reasonable sample of what sophisticated parties specify ex ante regarding arbitration. The importance of the contracts to corporate operations suggests that the contracts receive care and attention when they are negotiated and drafted. Moreover, because the contracts are written before disputes arise, we can be reasonably confident that in most cases the contracting parties did not anticipate the precise nature of the disputes that might arise, and therefore would not know whether a particular term would help or hurt them in the event of a conflict. These characteristics support our hypothesis that the contract terms we observe represent what the parties regard as reasonably efficient, ex ante allocations of rights and duties.

We analyzed each contract to determine whether it contained a binding arbitration clause. We report results.for the basic frequency of arbitration clauses by contract type as well as results subdivided by

whether there was a non-U.S. party to the contract. Although the type of contract involved is the strongest indicator of whether or not the contract contains an arbitration clause, we also explore the following secondary influences: the degree of standardization for the type of contract and the contractual category's relational status, choice of law, place of incorporation, place of business, and attorney locale.

A. Arbitration Clauses

Table 2 reports the number and percentage of contracts that contain arbitration clauses. The overwhelming majority, about 89%, do not mandate arbitration.72 Whatever arbitration's supposed efficiencies, sophisticated actors are not flocking to it in a broad range of important contracts. Some contract types-pooling and servicing agreefor arbitration. ments and Others-bond indentures, underwriting agreements, and security agreements-almost never did. But material variation exists across contract types. For example, 37% of employment contracts and 33% of licensing agreements provided for arbitration. Clearly, the nature of the contract is associated with whether the parties agree ex ante to arbitrate.

trust agreements-never provided

The results support the view, expressed by corporate lawyers, that the decision to include an arbitration provision in a commercial contract is complex and cannot be determined across the board. The infrequency of arbitration clauses across the entire range of categories also suggests that sophisticated contracting parties may not, in fact, believe in the purported benefits of arbitration over litigation.

B. Arbitration Clauses in International Contracts

Table 3 reports the frequency of arbitration clauses in contracts involving a non-U.S. party. Consistent with our hypothesis, arbitration clauses appear more frequently in international contracts than in domestic ones. Clauses appear at about twice the domestic rate when the contract includes a non-U.S. party, and the result is highly statistically significant (p < 0.001). More surprisingly, however, the international contracts, like the domestic contracts, contain a low absolute rate of arbitration clauses: only about 20% of international contracts

contain them. This finding contrasts with predictions that arbitration is the dispute resolution mechanism in international contractual settings, and that the vast majority of international contracts provide for binding arbitration.'

Table 4 breaks down the results in Table 3 by contract type. The first row reported for each contract type shows the proportion of contracts containing an arbitration clause, subdivided by domestic (first numerical column) and international (second numerical column) contracts. For example, the first row in Table 4 shows that 19.0% of the domestic merger contracts and 18.6% of the international merger contracts had arbitration clauses. The second row for each contract type shows the number of contracts in the sample. For example, the second row in Table 4 shows that there were 368 domestic merger agreements and 43 international merger agreements.

Although the rate of arbitration clauses in merger contracts does not vary by domestic/international status, it does significantly or nearly significantly vary for other contract types. Securities purchase contracts (p = 0.078), licensing contracts (p = 0.027), assets sale or purchase contracts (p = 0.067), and underwriting contracts (p = 0.040) show statistically significantly (p < 0.05) or nearly statistically significantly (p < 0.10) increased rates of arbitration clauses in international contracts compared to domestic contracts. Perhaps the most striking figure in Table 4 is that 63.6% of the international licensing contracts mandate arbitration-the only figure that conforms to the received wisdom about the frequency of arbitration clauses in international contracts. However, the relatively small number of contracts on which the figure is based (7 of 11) suggests that caution is warranted in reaching conclusions based solely on these data.

C. Secondary Sources of Variation in Rates of Arbitration Clauses

The varying frequency of arbitration clauses suggests the possible utility of more detailed modeling of whether a contract contains an arbitration clause. This Section explores five possible influences on the presence of arbitration clauses: the degree of standardization in the type of contract being considered, the contract type's relational status, the choice of law in the contract, the place of business of a party to the contract, and the locale of the attorneys who appear in the Form 8-K report that contains the contract. We first explore each influence separately and then combine them in regression models.

No non-U.S. party

Non-U.S. party

0.096 2583

0.202 272

0.106 2855

1. Standardization and Industry Practice

As shown in Table 2, the rate of arbitration clauses varies substantially depending on the type of contract. One possible reason for this variation is the degree of standardization. Highly standardized contracts may be perceived as having a low risk of litigation. Standard

terms create a common understanding that can reduce the disputes over contractual meaning that tend to spark litigation. Arbitration clauses may not be regarded as being necessary or worth sacrificing the perceived benefits of litigation.73 On the other hand, where contracts are highly standardized-as in the case of pooling and servicing or trust agreements-it might be the case that arbitration could add value because expert arbitrators selected by the parties might be able to give a better informed meaning to the highly specialized terms contained in these agreements, or because arbitration eliminates the risk of a judicial interpretation that could function as precedent or even be the basis for collateral estoppel in later cases.

The degree of contract standardization is not a commonly defined metric. We constructed an objective measure of standardization by examining the distribution of choice of law patterns for each contract type. A class of contract that regularly designates a particular choice of law has likely achieved a degree of standardization beyond that of a class of contract that designates many different choices of law. It is, of course, possible for contracts to be completely standardized in all other terms and vary only in their choice of law. But this is not likely to be the general tendency. Furthermore, as shown below, the results of categorizing according to our standardization measure appear plausible. Highly regularized financial transactions such as pooling and service agreements and trust agreements score high in standardization. Less regularized contracts, such as settlements and licensing agreements, score low by our standardization measure.

In measuring the concentration of choice of law provisions, it is helpful to note that approximately 70% of the contracts designate New York (47%), Delaware (14%), or California (7%) law as the governing law. The choice of law for other contracts is widely dispersed across many locales and these contracts can be excluded without sacrificing a reasonable measure of concentration. No other locale accounts for even 4% of choice of law provisions. We use the degree of concentration among the three leading states to classify contracts as having high, medium, or low rates of standardization.

Figure 1 shows, for each of twelve contract categories (the contract category Other is excluded from this analysis), the distribution of choice of law among New York, Delaware, and California. It shows

that bond indentures, credit commitments, underwriting contracts, pooling and service agreements, security agreements, and trust agreements all have high choice of law concentrations. For example, nearly all bond indentures and underwriting contracts designate New York law as the governing law. Trust agreements most frequently select Delaware law as the choice of law. The others most frequently select New York law. We treat mergers, securities purchase agreements, and asset sale purchase agreements as having medium concentrations. Each has a concentration (or density74) of over 0.5 in one state according to the histograms, but these three contract types are not nearly as concentrated as the highly concentrated group. Settlements, employment contracts, and licensing agreements have low choice of law concentrations. None has a density of 0.5 in the histograms.

Choice of law concentration is strongly associated with the presence of an arbitration clause. Almost no contracts with high choice of law concentrations have arbitration clauses (0.9%). In contrast, 16.3% of contracts with medium choice of law concentrations have arbitration clauses, and 29.9% with low choice of law concentrations have arbitration clauses. Our standardization index clearly conveys some information about the relationship between the type of contract and the inclusion of an arbitration clause.

The standardization measure has an advantage over individualized contract types as an explanatory variable in modeling the presence of an arbitration clause. The standardized measure does not require excluding contract types that are so standardized that they exhibit no choice of law variation. Table 2 shows that pooling and service contracts and trust agreements never contain arbitration clauses. This lack of variation precludes modeling the decision to have an arbitration clause by using a dummy variable for these contract types.

We hypothesized that arbitration clause rates would vary by industry.75 A reporting company's SEC filing includes its Standard Industry Classification (SIC).76 The SIC system yields many industry categories that contain too few firms to allow for reasonable statistical

analysis. We therefore regroup the SIC categories into seventeen reasonably sized classifications.77

Table 5 reports the rate of arbitration clause use by these classifications. The incidence of arbitration clauses varies substantially by industry; the table as a whole yields a statistically significant difference (p < 0.001). The "Finance, insurance, real estate" grouping is the largest and has an extremely low rate of arbitration clauses (about 4.7%). The highest rate of arbitration clause use is in the "Industrial machinery and equipment" grouping (25%).

2. Relational Contracts

We found little evidence that relational contracts, as best we could code for that characteristic, are more associated with the inclusion of arbitration clauses. The six contract types that we estimated to be more relational than others-credit commitments, employment, licensing, pooling and servicing agreements, security agreements, and

trust agreements-had arbitration clauses in 64 of 633 (10.1%) contracts. The five contract types that we estimated to have been less relational than other contract types-asset sales, bond indentures, mergers, securities purchase agreements, and underwritings-had arbitration clauses in 195 of 1691 (11.5%) contracts. The direction of the effect is the opposite of what we expected but the result is not close to being statistically significant (p = 0.332). Results do not change materially if we characterize trust agreements as ambiguous and include them in neither the relational nor nonrelational categories. The relational category had a 10.9% arbitration clause rate and the nonrelational category an 11.5% rate, an insignificant association (p = 0.698). We did not include a variable for relational contracts in our subsequent analyses.

3. Choice of Law

Independent of its association with standardization, choice of law information may help explain the pattern of arbitration clauses. State law treatment of arbitration provisions could influence parties' decisions to include arbitration clauses. The state law chosen, as distinct from the bare concentration of choice of law, may reflect underlying views about the respective desirability of arbitration and litigation. If the parties believe a particular state's law is highly efficient, that might be viewed as reducing the costs of litigation and providing a reason not to include an arbitration clause.78 Because choice of law is concentrated in three states, we report the rates of arbitration clauses for four choices of law: Delaware, New York, California, and Other.79

Table 6 shows, for each contract type and choice of law locale, the proportion of contracts containing arbitration clauses. For mergers, settlements, employment contracts, licensing agreements, and credit commitments, California law is associated with a higher proportion of arbitration clauses than New York or Delaware. California has a higher rate of arbitration clauses in five of the seven contract types in which California is designated more than three times as the choice of law. Overall, about 24% of contracts that chose California law contain arbitration clauses, compared to 12% of contracts that chose Del-

aware law, 4% of contracts that chose New York law, and about 16% of contracts that chose Other law.

Pooling service

0.120 133 0.000 1 0.167 12 0.167 78 0.333 6 0.200 5 0.313 32 0.000 6

0.000 17

0.000 39 0.068 88

0.120 417

0.174 69 0.007 138 0.231 13 0.053 188 0.333 12 0.300 10 0.134 82 0.000 105 0.003 327 0.000 149 0.000 22 0.000 7 0.038 185

0.040 1307

0.373 51 0.000 4 0.273 11 0.152 46 0.357 14 0.500 8 0.233 30 0.111 9 0.000 1

0.000 3

0.143 42

0.237 219

0.212 66 0.000 12 0.111 36 0.162 148 0.380 79 0.320 25 0.194 170 0.042 96 0.000 23 0.000 7 0.167 12 0.000 2 0.095 147

0.162 823

Table 6's breakdown by contract type shows that the relative prominence of arbitration clauses in California is not solely a consequence of different contract types being associated with different state laws. Regression analysis confirms this in a model that simultaneously con-

3 0 DEPAUL LAW REVIEW

trols for contract type and choice of law.80 California contracts also contain arbitration clauses at significantly higher rates than New York contracts.

4. Place of Business

In addition to choice of law and contract type, one might expect a

business's location to influence, the decision to include an arbitration clause. Business location is associated with where events affecting a business might occur. Thus, the location of events influences choice of law and forum. Place of business is often associated with choice of law,81 but they are not always the same.

Designating a single place of business for each contract requires considering the nature of the contract. For some contracts, such as merger contracts, two places of business are plausible: that of the acquiring company and that of the acquired company. For such contracts, we used what one would normally expect to be the dominant place of business. For example, for merger contracts, we used the acquiring company's place of business. Table 7 shows the business locale chosen for twelve types of contracts. Because of the varied nature of the Other category, we excluded such contracts from this analysis.

Table 8 reports, for each contract type and place of business, the proportion of contracts with arbitration clauses. To keep the output manageable, we report results only for states that were the place of business in at least fifty contracts. States with fewer contracts are included in the residual category.

Table 8 confirms the Table 6 findings that arbitration clauses appear at a greater rate in contracts with California ties. Table 8 shows for place of business what Table 6 shows for choice of law. For every type of contract with more than a handful of arbitration clauses, California's arbitration clause rate (first numerical column) exceeded the arbitration clause rate for all places of business combined (last numerical column). For example, 23.7% of California merger contracts had arbitration clauses, compared to 19% for all states combined. Table 8's Total row shows that only Pennsylvania had a higher

Place of business

Mergers Bond indentures Settlements Securities purchase Employment contracts Licensing Asset sale purchase Credit commitments Underwriting Pooling service Security agreements Trust agreements

Acquiring company's place of business Issuer's place of business Reporting company's place of business Issuer's place of business Employer's place of business Licensor's place of business Buyer's place of notice location Principal lender's designated office Issuer's place of business Depositor's place of business Registrant's place of business Registrant's place of business

rate of contracts with arbitration clauses, 21.1%, than California's 15.7%. But Pennsylvania had only about 20% as many business locations as California.

We recognize, of course, that choice of law locales and business locales are not independent of one another. The central result is that using either locale shows that California has relatively high arbitration clause rates.

5. Attorney's Place of Business

An attorney's location could influence whether an arbitration clause is used if local practice or custom dictates whether to include an arbitration clause. Attorney locale is likely associated with choice of law and place of business.82 Nevertheless, there may be an attorney effect with arbitration clauses in addition to choice of law and place of business effects. For example, in merger contracts, New York attorneys are more likely than other attorneys to specify a litigation forum.83 than information about choice of law or place of business. Attorney locale is available only for law firms mentioned in the Form 8-K, so we are limited to using those law firms appearing in the form with useable addresses. And we could not always link an attorney mentioned with a principal party to the contract.

Information about attorney

sketchier

locale

is

~o

~

w Z

o6

~

20071

Table 9 reports the rate of arbitration clauses for each of twelve contract types (excluding_ Other), broken down by attorney locale. Attorney locale is more widely dispersed than choice of law. As shown above, Delaware, New York, and California account for almost 70% of the choices of law, but they account for less than 50% of the attorney locales. Even taking into account the greater likelihood of missing data for attorney locale (note the high number of locales coded as Other in Table 9), New York and California are the leading attorney locales, but they represent only a small portion of the contracts. Table 9's last row indicates that New York can be identified as an attorney locale in 269 of 2396 contracts; California can be identified as an attorney locale in 182 of 2396 contracts. Interestingly, Delaware was the attorney locale in less than 50 contracts. The large Other category makes us reluctant to conclude whether attorney locale is more or less concentrated than business locale.

Despite the sketchiness of the attorney data, they tend to confirm a key Table 8 finding about the relationship between locale and the use of arbitration clauses. For every type of contract with more than a few arbitration clauses, California's arbitration clause rate (first numerical column) exceeded the arbitration clause rate for all attorney locales combined (last numerical column). For example, 24.3% of California merger contracts had arbitration clauses compared to 19% for all states combined. Table 9 shows that only Massachusetts had a higher rate of contracts with arbitration clauses, 18.2%, than California's 18.1%. But Massachusetts had only about 30% as many attorney locales as California. As in Table 8, Pennsylvania had a relatively high rate of arbitration clauses.

6. Regression Results

The previous subsections separately assess the relationship between arbitration clause presence in a contract and (1) presence of a nonU.S. party to the contract, (2) degree of contract standardization and industry, (3) choice of law, (4) place of business, and (5) attorney locale. This subsection models arbitration clause presence as a simultaneous function of these characteristics.

Each model in Table 10 is a logistic regression model in which the dependent variable equals one if an arbitration clause is in the contract and otherwise equals zero. Model (1) contains a dummy variable for a non-U.S. entity being a party to the contract and dummy variables for the degree of standardization as discussed above with respect to Figure 1. A low degree of standardization is the reference category. Model (2) adds dummy variables for choice of law, with Delaware as

DEPAUL" LAW REVIEW

the reference category. Model (3) adds, but does not report, dummy variables for SIC codes as reported in Table 5. Model (4) adds, but does not report, dummy variables for place of business as described in Table 8. Model (5) adds, but does not report, dummy variables for attorney locale as described in Table 9. Model (6) uses dummy variables for each contract type in lieu of the standardization dummy variables and includes dummy variables for SIC codes.

All models confirm the bivariate analysis results. Contracts without a U.S. party have a higher incidence of arbitration clauses. Contracts with medium standardization are less likely to have arbitration clauses than contracts with low standardization. Contracts with high standardization tend not to have arbitration clauses compared to contracts with medium or low standardization. Contracts with medium stand-

ardization tend'not to have arbitration clauses compared to contracts with low standardization.84

Contracts with California ties'tend to include arbitration clauses more than contracts without California ties. Contracts with' New York ties tend to exclude arbitration clauses compared to contracts lacking New York ties. These results are statistically significant or nearly. so for all variables in all models. The marginal significance of the California law dummy variable in models (4) and (5) may be due to the multicollinearity of that variable with the California business and attorney locale dummy variables included in the models.

Note that model (6) has fewer observations than the other models. This is because pooling and service agreements and trust agreements drop out of the model due to a lack of variation; they never contain arbitration clauses. The model nevertheless confirms the non-U.S. party and state choice of law results that emerge in the other models. The coefficients on the contract-type dummy variables in model (6) are also consistent with the results suggested by using the standardization dummy variables in models (1) to (5). For example, the highly standardized bond indenture, credit commitment, and underwriting contracts are all negatively, and significantly, associated with the presence of an arbitration clause compared to the reference category, merger contracts. Employment contracts and licensing agreements, with relatively low degrees of standardization, are positively associated with the presence of arbitration clauses, compared to merger contracts. Model (6) thus establishes that the coding of contract types into our three standardization categories does not distort the relationship between contract type and the presence of an arbitration clause. Table 10's models also suggest that, of all the variables we explore, most of the information about the presence of an arbitration clause is embodied in the non-U.S. party variable, the choice of law variables, and the standardization variables. In models (4) and (5), a test of the hypothesis that the place of business dummy variables are jointly equal to zero yields p-values that preclude rejecting the hypothesis. Similarly, in model (5), one cannot reject the hypothesis that the attorney locale dummy variables are jointly equal to zero.

The marginal effects of the explanatory variables for model (3) are shown in Table 10's far right column. They report the estimated change in the probability of the presence of an arbitration clause as an explanatory variable changes from zero to one, while assigning the

other explanatory variables their average values. The marginal effects should be interpreted against the background of the small overall probability, about 0.11, of an arbitration clause appearing in a contract. So, for example, the 0.031 increase in the probability of an arbitration clause when a non-U.S. party is present is about a 30% increase in the probability of an arbitration clause, given the low rate of arbitration clauses. From this perspective, all of the statistically significant marginal effects are substantial, with the effect of high standardization being extraordinarily large at -0.157.

This is the first study of its kind; it is thus important to recognize its limitations. First, the contracts we study exist in a small slice of time, so we would ideally like information for periods before and after the first half of 2002. Second, the variation across the contract types in the rates of arbitration clauses suggests that more sophisticated modeling of the decision to include such clauses could be fruitful. The details of the relationships between the contracting parties and the motivations of those drafting the clauses could and should be studied. Third, while corporate actors may be expected to seek efficient solutions, they do act through agents. If litigation raises costs, some of those costs go to attorneys who play a prominent role in drafting contracts. The agency issues raised by our findings should be the object of interesting research. Subject to these limitations, and recognizing the need for future research, we explore some of the possible implications of our results.

A. The Paucity of Arbitration Clauses

The most striking result is the paucity of arbitration clauses, even in international contracts. Our results contradict some received wisdom but are consistent with the 1998 Cornell University survey finding that relatively few large corporations use arbitration frequently,85 and a later Cornell survey finding substantial variation in corporate approaches to conflict resolution.86 The finding of greater use of arbitration clauses in employment contracts, as reported in Table 2 and

confirmed in Table 10's regression models, is consistent with the Cornell survey.87 The infrequency of arbitration in financial contracts confirms the observation of William Park that banks prefer to litigate, rather than arbitrate, loan defaults.88

Our international contract arbitration clause findings show substantial variation by contract type. The highest arbitration clause rate is shown in Table 4 to be 63.6% for international licensing contracts. This high rate suggests that our findings, although generally showing low rates, are not necessarily inconsistent with Drahozal and Naimark's reported 88.2% arbitration clause rate for international joint venture agreements.89 Both high rates are consistent with the influence of standardization on the decision to include an arbitration clause. Licensing and joint venture contracts are likely to be less standardized than most other contracts.

Why are arbitration clauses so infrequently used? The paucity of such clauses may partially reflect the view of corporate counsel that the decision whether to include binding arbitration in an agreement is not one that can be made across the board, but rather depends on the needs and circumstances of the parties.90 That arbitration is not called for in every instance, however, would not explain why it is observed so rarely. Perhaps the reason has something to do with education and inertia; transactional attorneys may not be fully aware of the dispute resolution options.91 But we doubt that the sophisticated attorneys who draft these major commercial contracts would be unaware of the availability of binding arbitration as an alternative to litigation. Parties might be electing to omit arbitration clauses on the theory that if a dispute arises, they can agree to submit the matter to arbitration at that time. However, ex post arbitration is not an effective substitute for arbitration agreed to ex ante; once a dispute has arisen, one party

will typically have a strategic reason to demand litigation.92 Parties may also encounter difficulty in proposing or negotiating a mutually acceptable arbitration clause. A party may fear that proposing such a clause will generate concerns about its intended contract performance or about its perception of the other party's likelihood of performance. Another plausible explanation is that the benefits of arbitration have simply been oversold. Arbitration may not be as efficient a method for resolving disputes as its most ardent supporters claim.93

B. Arbitration Clause Rates and Perceptions of Court Fairness

Similarly, the disadvantages of litigation may have been exaggerated. One reason for shunning litigation is the supposed unpredictability and unfairness of adjudication. If these reasons are in fact at work, one might expect a strong association between the perceived quality of state civil justice systems and the rate of arbitration clauses in contracts connected to a state by incorporation or place of business. Incorporating in a state and having a substantial place of business in a state should be associated with an increased risk of adjudication in that state's courts. Our data allow us to rank states based on the arbitration clause rates of contracts connected to the states. The Chamber of Commerce of the United States annually ranks state civil justice systems.94 One measure of whether fear of state court systems explains arbitration clause rates is whether states that are perceived as having the least fair court systems have the highest rates of arbitration clauses.

Figures 2 and 3 show the relationship between a state's rate of arbitration clauses and the overall ranking of state liability systems by large corporations, as surveyed by the Chamber of Commerce.95 The Chamber's overall ranking is based on the scores for several topics,96

including treat!,ent of tort and contract litigation, judges' impartiality, judges'. competence, juries' predicta.biity, and juries' fairness.97 Figure 2 reports the relationship between arbitration clause rates in our data based on state of incorporation and the Chamber's overall state ranking. Figure 3 reports the relationship between arbitration clause rates in our data based on place of business and the Chamber's overall state ranking. Place of business for a contract is assigned as described in connection with Table 7. Place of incorporation for a contract is assigned using the same criteria. For example, in merger contracts, the acquiring company's place of incorporation is used.

~NV

If there were a positive association between civil justice perceptions and arbitration clauses, the data points in the figure should flow from lower left to upper right. The lower left of the figures, near the origin,

~NJ

corresponds with high rankings in civil justice systems and low rankings in arbitration clause rates.98 The upper right corresponds with low rankings in civil justice systems and high rankings in arbitration clause rates. The expected outcome is not observed. Both figures indicate that no significant association exists between rates of arbitration clauses and the Chamber's rankings. The patterns appear to be essentially random, though some caution is necessary because several states have few observations. If the Chamber's ranking methodology has validity, then the figures suggest that arbitration clauses are being used for reasons other than perceived interstate variation in the performance of the judicial system.99

DEPAUL LAW.REVIEW

C. Arbitration Clause, Rates and State Laws and Courts

The associations between state contacts and the use of arbitration clauses, shown in Tables 6, 8, 9, and 10, suggest the need for further research into the relationship between state law and arbitration clause use. One factor we suggested as possibly explaining use of arbitration clauses, the speed with which state courts process cases, does not appear to be helpful in explaining the observed pattern of arbitration clauses.

Available information about state court case processing indicates that California cases resolved by trial are resolved more quickly than New York cases. The Bureau of Justice Statistics studied all 2001 state court trial terminations in forty-six of the nation's seventy-five largest counties.100 The study included New York County (Manhattan) and nine California counties.10' For the nine large California counties, the study included 1264 trials. The mean time from trial to verdict was 601 days, with a median of 510 days. The large New York County had 309 trials, with a mean time from trial to verdict of 1245 days and a median time of 1043 days. These dramatically faster disposition times in large California counties persist when one controls for the different mix of case types in the two states.

Thus, if contracting parties are attracted to arbitration in states in which adjudication times are relatively slow, we would expect parties entering contracts with New York ties to be attracted to arbitration compared to parties entering contracts with California ties. Yet we find just the opposite. Parties are drawn to arbitration more in California than in New York. Indeed, California's processing time for tried cases was noticeably faster than the forty-six county mean and median. For all forty-six counties, the mean time from filing to verdict was 753 days and the median was 629. Yet our data show that California contracts tend to include more arbitration clauses compared to other states. We interpret this as evidence that substantive legal rules tend to have a greater influence over decisions to include arbitration clauses than case processing efficiency does.

D. Observed Arbitration Clause Rates Compared to Consumer Contracts

Our core arbitration clause finding contrasts with the reportedly widespread use of mandatory arbitration clauses in certain consumer contracts.102 Business interests defend consumer arbitration as a "less-expensive, more efficient alternative to a court dispute resolution."'03 Some suggest that arbitration clauses in consumer contracts may be used for some other purpose, such as a mechanism to avoid dispute resolution by foreclosing class actions,104 or to gain an advantage in dispute resolution over parties who cannot realistically negotiate.105 Our study suggests the value of further inquiry into the advantages and disadvantages of using binding arbitration in standardized consumer contracts.'06

We present evidence that large corporate actors do not systematically embrace arbitration. International contracts include arbitration clauses more than domestic contracts, but also at a surprisingly low rate. Our results have implications for the justifications for the widespread use of arbitration clauses in consumer contracts. If the reasons that some have advanced to support the use of arbitration in the consumer context-that it is simpler and cheaper than litigation-are cor-

is surprising that public companies do not seek

rect, it these advantages in disputes among themselves. In the simple economic view, our results suggest that corporate representatives believe that litigation can add value over arbitration.

--- FOOTNOTES ---

INTRODUCTION

* Henry Allen Mark Professor of Law, Cornell Law School.

** Stuyvesant P. Comfort Professor of Law, NYU Law School. We thank Edward Brunet, Christopher Drahozal, John Duffy, Samuel Estreicher, Mark Geistfeld, Florencia Marotta-Wurgler, Henry S. Noyes, Stephen Ware, and participants at the Twelfth Annual Clifford Symposium at DePaul University College of Law for comments, and Natalie Erbe, Jeremy Masys, Sergio Muro, Hilel Pohulanik, Whitney Schwab, and Cathy Weist for research assistance.

1. See, e.g., Keith N. Hylton, Agreements to Waive or to Arbitrate Legal Claims: An Economic Analysis, 8 Sup. CT. ECON. REV. 209, 222 (2000) (arguing that if the option to litigate reduces the joint wealth of contracting parties, market forces will push them toward alternative forms of dispute resolution); Steven Shavell, Alternative Dispute Resolution: An Economic Analysis, 24 J. LEGAL STUD. 1, 5 (1995) (contending that "parties would tend to adopt ADR if it would lead to mutual advantages").

II. HYPOTHESES AND EXISTING RESEARCH

2. See, e.g., NAT'L ARBITRATION FORUM, THE CASE FOR PRE-DISPUTE ARBITRATION AGREE-

MENTS: EFFECTIVE AND AFFORDABLE ACCESS TO JUSTICE FOR CONSUMERS EMPIRICAL STUDIES AND SURVEY RESULTS 1 (2004), available at http://www.adrforum.com/rcontrol/documents/ ResearchStudiesAndStatistics/2004EmpiricalStudies.pdf [hereinafter CASE FOR PRE-DISPUTE ARBITRATION AGREEMENTS] (asserting that "[m]any independent studies ... have confirmed the comparative benefits of arbitration (versus lawsuits) for both businesses and consumers").

3. See, e.g., Allied-Bruce Terminix Cos. v. Dobson, 513 U.S. 265, 280 (1995) (citing H.R. REP. No. 97-542, at 13 (1982), as reprinted in 1982 U.S.C.C.A.N. 765, 777); Gilmer v. Interstate/Johnson Lane Corp., 500 U.S. 20, 31 (1991); Thomas J. Stipanowich, ADR and the "Vanishing Trial": The Growth and Impact of "Alternative Dispute Resolution," 1 J. EMPIRICAL LEGAL STUD. 843, 876 (2004). Arbitration proponents frequently tout its benefits relative to litigation. See, e.g., Eric J. Mogilnicki & Kirk D. Jensen, Arbitration and Unconscionability, 19 GA. ST. U. L. REV. 761 (2003). One author listed the "inherent advantages" of arbitration:

(1) the preparation, filing, and service of claims and responses are much easier and faster, (2) relevant and reliable information can be or is more effectively exchanged between parties before a hearing, (3) there is no need for costly and lengthy motion proceedings, (4) parties can, when they are able to, represent themselves or choose to have an attorney represent them, (5) the rules are much fewer in number and much easier to understand, (6) the rules of evidence readily allow relevant and reliable information and do not restrict or complicate the admission of evidence, (7) the hearing procedures are more informal and less complex, (8) the arbitrator or administrative judge is a procedural and substantive law expert, (9) the arbitrator and administrative judge have more flexibility in allowing the parties to present the information they need to present, (10) the type of hearing can vary and allow parties and witnesses to participate by written documents, telephone, e-mail, video conference, and by personal appearance, (11) the day and time of the hearing can be specifically scheduled assuring parties that their case will be heard as scheduled, and (12) the entire proceeding, from beginning to end, may take a few months instead of years.

Roger S. Haydock, Civil Justice and Dispute Resolution in the Twenty-First Century: Mediation and Arbitration Now and for the Future, 27 WM. MITCHELL L. REV. 745, 748-49 (2000). Studies of arbitration in the employment context tend to confirm some of these beliefs. See Theodore

Eisenberg & Elizabeth Hill, Arbitration and Litigation of Employment Claims: An Empirical Comparison, 58 Disp. RESOL. J. 44 (2003-2004) (finding faster case processing in arbitration): U.S. GEN. ACCOUNTING OFFICE, ALTERNATIVE DISPUTE RESOLUTION: EMPLOYERS' EXPERIENCES wITH ADR IN THE WORKPLACE 19 (1997) (finding that employers' legal fees were generally lower in arbitration); Lewis L. Maltby, The Projected Economic Impact of the Model Employment Termination Act, 536 ANNALS AM. ACAD. POL. & SOC. ScI. 103, 117 (1994) (finding that legal fees in employment arbitrations were surprisingly low).

4. See, e.g., Eisenberg & Hill, supra note 3, at 51; Mogilnicki & Jensen, supra note 3, at 765-66; see also Gary G. Mathiason & Pavneet S. Uppal, Evaluating and Using Employer-Initiated Arbitration Rules and Agreements, in ALI-ABA COURSE OF STUDY MATERIALS: EMPLOYMENT DISCRIMINATION AND CIVIL RIGHTS ACTIONS IN FEDERAL AND STATE COURTS 875, 894 (1994) (average arbitration case resolved in less than half the time required for civil litigation).

5. Shavell, supra note 1, at 5. 6. But discovery may not be dispensed with altogether, even in the commercial context. See

Cole v. Burns Int'l Sec. Servs., 105 F.3d 1465, 1482 (D.C. Cir. 1997).

7. Employers fear such jury awards. See, e.g., Alexander J.S. Colvin, From Supreme Court to Shopfloor: Mandatory Arbitration and the Reconfiguration of Workplace Dispute Resolution, 13 CORNELL J.L. & PUB. POL'Y 581, 585 (2004). Eisenberg and Hill found that in comparing arbitration and litigation, median awards in the employment cases studied were roughly similar, but mean awards were much higher in litigation, which suggests that in a few cases courts awarded much higher damages than in similar arbitrated cases. Eisenberg & Hill, supra note 3, at 49-51. 8. Stephen J. Ware, The Case for Enforcing Adhesive Arbitration Agreements-With Particu-

lar Consideration of Class Actions and Arbitration Fees, 5 J. AM. ARB. 251. 261 (2006).

9. See Allied-Bruce Terminix Cos. v. Dobson, 513 U.S. 265, 280 (1995); Gilmer v. Interstate/ Johnson Lane Corp., 500 U.S. 20, 31 (1991); Donna K. McElroy, Compulsory Arbitration Agreements ...Issues Concerning the Enforcement of Compulsory Arbitration Agreement Between Employers and Employees, 31 ST. MARY'S L.J. 1015, 1021 (2000).

10. The Federal Arbitration Act specifies when a commercial arbitration award may be

vacated:

(1) Where the award was procured by corruption, fraud, or undue means. (2) Where there was evident partiality or corruption in the arbitrators .... (3) Where the arbitrators were guilty of misconduct in refusing to postpone the hearing, upon sufficient cause shown, or in refusing to hear evidence pertinent and material to the controversy ....

(4) Where the arbitrators exceeded their powers, or so imperfectly executed them that a mutual, final, and definite award upon the subject matter submitted was not made.

9 U.S.C. § 10(a) (2000). Some courts have recognized nonstatutory grounds for vacating awards. See, e.g., United Paperworkers Int'l Union v. Misco, Inc., 484 U.S. 29, 43 (1987) (public policy); Montes v. Shearson Lehman Bros., Inc., 128 F.3d 1456, 1458 (11th Cir. 1997) (arbitrary and capricious); Tanoma Mining Co. v. Local Union No. 1269, United Mine Workers of Am., 896 F.2d 745, 749 (3d Cir. 1990) (manifest disregard for the law).

11. Parties can try to work around this risk by providing contractually for judicial review of arbitral decisions, but the efficacy of such contracts is not well established. See, e.g., Christopher R. Drahozal, Contracting Around RUAA: Default Rules, Mandatory Rules, and Judicial Review of Arbitral Awards, 3 PEPP. DisP. RESOL. L.J. 419, 426-33 & nn.19-22 (2003); Stephen J. Ware,

"Opt-in" for Judicial Review of Errors of Law Under the Revised Uniform Arbitration Act, 8 AM. REV. INT'L ARB. 263, 265 & n.21 (1997).

12. See Bruce H. Kobayashi & Larry E. Ribstein, Contract and Jurisdictional Freedom, in THE

FALL AND RISE OF FREEDOM OF CONTRACT 325, 329 (F.H. Buckley ed., 1999).

13. See, e.g., Green Tree Fin. Corp.-Ala. v. Randolph, 531 U.S. 79, 94-95 (2000) (Ginsburg, J., concurring in part and dissenting in part) (noting that the contract could have specified that arbitration would be governed by the rules of the American Arbitration Association).

14. See, e.g., AM. ARBITRATION ASS'N: COMMERCIAL ARBITRATION RULES AND MEDIATION PROCEDURES E-8 (amended and effective Sept. 15, 2005) (noting that arbitrator may schedule additional hearings within seven business days after the initial hearing).

15. See, e.g., Hackett v. Milbank, Tweed, Hadley & McCloy, 654 N.E.2d 95, 100 (N.Y. 1995); Bruce L. Benson, To Arbitrate or to Litigate: That Is the Question, 8 EUR. J.L. & ECON. 91, 92 (1999); Lee H. Rosenthal, One Judge's Perspective on Procedure as Contract, 80 NOTRE DAME L. REV. 669, 673 (2005) (stating that "[t]he parties may select state law and may specifically exclude the application of federal law"). The ability to select an applicable law is potentially limited by the principle that the law chosen must bear some reasonable relationship to the parties or the transaction. RESTATEMENT (SECOND) OF CONFLICT OF LAWS § 187(2); see also Curtis 1000, Inc. v. Suess, 24 F.3d 941, 948 (7th Cir. 1994).

16. See, e.g., Pedcor Mgmt. Co. Welfare Benefit Plan v. Nations Pers. of Tex., Inc., 343 F.3d 355, 361 (5th Cir. 2005) (citing Ford v. NYLCare Health Plans of the Gulf Coast, Inc., 141 F.3d 243, 248 (5th Cir. 1998 .

17. See, e.g., LaPine Tech. Corp. v. Kyocera Corp., 130 F.3d 884 (9th Cir. 1997) (enforcing the standard for judicial review set forth in the parties' agreement to arbitrate), overruled by Kyocera Corp. v. Prudential-Bache Trade Servs., Inc., 341 F.3d 987 (9th Cir. 2003) (en banc).

18. Compare Bowen v. Amoco Pipeline Co., 254 F.3d 925, 933-37 (10th Cir. 2001), and Kyocera Corp., 341 F.3d at 1000, with Gateway Techs., Inc. v. MCI Telecomms. Corp., 64 F.3d 993, 996-97 (5th Cir. 1995).

19. Shavell, supra note 1, at 6.

20. 9 U.S.C. § 2 (2000). 21. Allied-Bruce Terminix Cos. v. Dobson, 513 U.S. 265, 272, 277 (1995). 22. 9 U.S.C. § 2. 23. See Judith Resnik, Procedure as Contract, 80 NOTRE DAME L. REV. 593 (2005). 24. Acosta v. Master Maint. & Constr., Inc., 52 F. Supp. 2d 699, 706-07 (M.D. La. 1999); eyoub ex rel State of La. v. Am. Tobacco Co., No. 97-1174, 1997 U.S. Dist. LEXIS 24219 (W.D. La. Sept. 11, 1997); cf. Marathon Oil Co. v. Ruhrgas, A.G., 115 F.3d 315, 320-21 (5th Cir. 1997), vacated by 145 F.3d 211 (5th Cir. 1998). Contra Merrill Lynch Inv. Managers v. Optibase, Ltd., 337 F.3d 125, 127-28, 131-32 (2d Cir. 2003) (holding that the investor was not entitled to compel arbitration with its investment adviser, where the investor had an arbitration agreement with its broker-dealer that referenced affiliates, but had no arbitration agreement with the investment adviser itself); Zimmerman v. Int'l Cos. & Consulting, Inc., 107 F.3d 344, 346 (5th Cir. 1997) (refusing to subject nonsignatories to arbitration clauses); In re Talbott Big Foot, Inc., 887 F.2d 611, 614 (5th Cir. 1989).

25. See, e.g., Edward Brunet, Seeking Optimal Dispute Resolution Clauses in High Stakes Employment Contracts, 23 BERKELEY J. EMP. & LAB. L. 107 (2002) (challenging the popular conception that arbitration and mediation will be optimal for contracts between employers and highly skilled employees).

26. See Henry S. Noyes, If You (Re)Build It They Will Come: Contracts to Remake the Rules of Litigation in Arbitration's Image, 30 HARV. J.L. & PUB. POL'Y (forthcoming 2007) (questioning whether arbitration proceedings are cheaper than trials, given the need to pay the arbitrators).

27. See, e.g., Tom Arnold, Delay and Cost Booby Traps in Arbitration Practice and How to Avoid Them, in ALI-ABA COURSE OF STUDY MATERIALS, ALTERNATIVE DISPUTE RESOLUTION: HOW TO USE IT TO YOUR ADVANTAGE 99, 103 (1996) (describing an arbitration that lasted over seven years and generated more than $60 million in attorney's fees).

28. See Noyes, supra note 26.

29. See id.

30. See id. 31. Ware, supra note 8, at 261 (noting that small meritorious claims, some of which might not

be cost effective in litigation, may see higher awards in arbitration).

32. Richard C. Reuben, Confidentiality in Arbitration: Beyond the Myth, 55 U. KAN. L. REV.

(forthcoming 2007).

33. See, e.g., Tony Starr, Choosing Between Litigation and Arbitration, BOSTON B.J.,

Mar.-Apr. 2005, at 31.

34. We do not attribute this view to any particular scholar of arbitration, but rather advance it

as a hypothesis against which to test the evidence.

35. The parties can, of course, agree to arbitration after a dispute has arisen. Unless the parties agree to arbitration before a dispute arises, however, it is doubtful that they will be able to achieve the benefits of arbitration by subsequent agreement. Once a dispute has arisen, it is likely that strategic factors will cause one of the parties to refuse to arbitrate, even if the other makes a request. See CASE FOR PRE-DISPUTE ARBITRATION AGREEMENTS, supra note 2 (arguing that once a dispute has occurred, at least one of the parties is likely to perceive litigation as preferable to arbitration, and accordingly, parties in real-life disputes rarely submit their disputes to arbitration); David Sherwyn, Because It Takes Two: Why Post-dispute Voluntary Arbitration Programs Will Fail to Fix the Problems Associated with Employment Discrimination Law Adjudication, 24 BERKELEY J. EMP. & LAB. L. 1 (2003). Accordingly, if the parties wish to guarantee that their disputes will be decided by an arbitrator, they must, as a practical matter, include an arbitration clause in the underlying contract.

36. Susan D. Franck, The Role of International Arbitrators, 12 ILSA J. INT'L & COMp. L. 499,

499 (2006).

37. See, e.g., Christopher R. Drahozal, Privatizing Civil Justice: Commercial Arbitration and the Civil Justice System, 9 KAN. J.L. & PUB. POL'Y 578 (2000). But empirical evidence about the mistreatment of foreigners by U.S. courts is lacking. See Kevin M. Clermont & Theodore Eisenberg, Commentary, Xenophilia in American Courts, 109 HARV. L. REV. 1120 (1996); Kevin M. Clermont & Theodore Eisenberg, Xenophilia or Xenophobia in American Courts? Before and After 9/11, 4 J. EMPIRICAL LEGAL STUD. (forthcoming 2007), available at http://ssrn.com/abstract=923595.

38. The Hague Convention on Choice of Court Agreements was concluded on June 30, 2005. 44 INT'L LEGAL MATERIALS 1294 (2005). If, it is ratified by enough countries, it may provide a viable alternative to arbitration for resolving international commercial disputes.

39. United Nations Convention on the Recognition and Enforcement of Foreign Arbitral Awards, June 10, 1958, 21 U.S.T. 2517, 330 U.N.T.S. 38. There are 137 signatories to the New York Convention. William W. Park, The International Currency of Arbitral Awards, 739 PRACTISING L. INST. LITIG. & ADMIN. PRAC. COURSE HANDBOOK SERIES 585, 591 (2006). The Convention has been implemented in the United States. To Implement the Convention on the Recognition and Enforcement of Foreign Arbitral Awards, Pub. L. No. 91-369, 84 Stat. 692 (codified as amended at 9 U.S.C. §§ 201-208).

40. See Cindy G. Buys, The Arbitrators' Duty to Respect the Parties' Choice of Law in Com-

mercial Arbitration, 79 ST. JOHN'S L. REV. 59 (2005).

41. International arbitration carries with it potentially significant costs, including the fees of the private arbitrators and the costs of the dispute resolution forum (if any). See Linda Silberman, International Arbitration: Comments from a Critic, 13 AM. REV. INT'L ARB. 9 (2002). Some claim that the virtues of international arbitration have eroded as arbitration becomes more like litigation. Franck, supra note 36, at 500-01. It is even claimed that international arbitration may not be less expensive than litigation, although the relative advantage of arbitration over litigation likely varies across court systems. CHRISTIAN BOHRING-UHLE, ARBITRATION AND MEDIATION IN INTERNATIONAL BUSINESS 140-48 (1996) (cited in Franck, supra note 36, at 500 n.4) (noting that arbitration is not less expensive but that it may be quicker). A recent article argues that some classes of international disputes have become increasingly judicialized. Andrea Schneider, Not Quite a World Without Trials: Why International Dispute Resolution Is Increasingly Judicialized, J. DisP. RESOL. (forthcoming 2007), available at http://ssrn.com/abstract= 920510 (arguing that judicialization of international disputes has occurred in the areas of human rights, economic and trade disputes, and border and maritime disputes).

42. William M. Landes & Richard A. Posner, Adjudication as a Private Good, 8 J. LEGAL

STUD. 235, 249 (1979).

43. See Kenneth S. Abraham & J.W. Montgomery, III, The Lawlessness of Arbitration, 9

CONN. INS. L.J. 355, 364 (2003).

44. Cronus Invs., Inc. v. Concierge Servs., 107 P.3d 217, 228-29 (Cal. 2005) (holding that a statute permitting stay of arbitration pending outcome of related litigation does not contravene the FAA).

45. Discover Bank v. Superior Court of L.A., 113 P.3d 1100, 1116 (Cal. 2005) (limiting authority to preclude class-based activity); see also Kristian v. Comcast Corp., 446 F.3d 25, 64-65 (1st Cir. 2006) (invalidating prohibitions on treble damages, attorney's fees and costs, and class actions, and permitting waiver of Massachusetts state law antitrust claim); Jenkins v. First Am.

Cash Advance of Ga., L.L.C., 400 F.3d 868 (11th Cir. 2005) (discussing the question of arbitrability presented by ban on class actions); Livingston v. Assocs. Fin., Inc., 339 F.3d 553 (7th Cir. 2003) (same); Whitney v. Alltel Commc'ns, Inc., 173 S.W.3d 300, 309-11 (Mo. Ct. App. 2005) (finding the class action arbitration ban unconscionable). Contra Tsadilas v. Providian Nat'l Bank, 786 N.Y.S.2d 478, 480 (N.Y. App. Div. 2004) (finding the arbitration provision waiving class actions enforceable).

46. Garrity v. Lyle Stuart, Inc., 40 N.Y.2d 354 (1976) (holding that the arbitration award of punitive damages violates public policy). California courts did not follow Garrity. See Baker v. Sadick, 208 Cal. Rptr. 676, 683-84 (Cal. Ct. App. 1984).

47. See Mastrobuono v. Shearson Lehman Hutton, Inc., 514 U.S. 52 (1995) (finding that the FAA preempts state law); Barnes v. Logan, 122 F.3d 820, 822-24 (9th Cir. 1997) (upholding punitive damage award under Minnesota law); Todd Shipyards Corp. v. Cunard Line, Ltd., 943 F.2d 1056, 1062 (9th Cir. 1991) (upholding punitive award in arbitration conducted under arbitration rules that allow punitive damages); Raytheon Co. v. Automated Bus. Sys., Inc., 882 F.2d 6, 12 (1st Cir. 1989) (same); Dandong Shuguang Axel Corp. v. Brilliance Mach. Co., No. C 004480 SC, 2001 WL 637446, at *6 (N.D. Cal. June 1, 2001) (recognizing that the Supreme Court has severely limited Garrity); PaineWebber, Inc. v. Landay, 903 F. Supp. 193, 204 (D. Mass. 1995) (upholding punitive damages award rendered under National Association of Securities Dealers (NASD) rules); Drywall Sys., Inc. v. ZVI Constr. Co., 747 N.E.2d 168 (Mass. App. Ct. 2001) (rejecting Garrity).

48. In re Mohawk Valley Cmty. Coll., 814 N.Y.S.2d 428, 429 (N.Y. App. Div. 2006) (quoting

Garrity in noting that an award imposing punitive damages violates public policy).

49. See also Elizabeth Rolph, Erik Moller & John E. Rolph, Arbitration Agreements in Health Care: Myths and Reality, 60 LAW & CONTEMP. PROBS., Winter 1997, at 153, 159 (noting that California's "legal policies have been very supportive of arbitration agreements").

50. For a comprehensive study of arbitration clauses in end-user software license agreements, see Florencia Marotta-Wurgler, "Unfair" Dispute Resolution Clauses: Much Ado About Nothing?, in "BOILERPLATE": FOUNDATIONS OF MARKET CONTRACTS (Omri Ben-Shahar ed., forthcoming 2007). For a call for more empirical research in the area of contract law, see David V. Snyder, Go out and Look: The Challenge and Promise of Empirical Scholarship in Contract Law, 80 TUL. L. REV. 1009 (2006).

51. See ELIZABETH ROLPH, ERIK MOLLER & LAURA PETERSEN, ESCAPING THE COURTHOUSE: PRIVATE ALTERNATIVE DISPUTE RESOLUTION IN Los ANGELES 18-20 (1994) (reporting that the private ADR case load in the Los Angeles area grew by an average of 15% per year from 1988 to 1993, while the court system case load was stable); Bruce L. Benson, An Exploration of the Impact of Modern Arbitration Statutes on the Development of Arbitration in the United States, 11 J.L. ECON. & ORG. 479, 490 (1995).

52. Stipanowich, supra note 3, at 873. 53. See Data on Filings with International Arbitration Institutions, in TOWARDS A SCIENCE OF INTERNATIONAL ARBITRATION: COLLECTED EMPIRICAL RESEARCH app. 1 at 341 (Christopher R. Drahozal & Richard W. Naimark eds., 2005) (showing that filings in a collection of international arbitration institutions increased from 1392 in 1993 to 2577 in 2003).

54. David B. Lipsky & Ronald L. Seeber, The Appropriate Resolution of Corporate Disputes: A Report on the Growing Use of ADR by U.S. Corporations 8 (1998) (Cornell Inst. on Conflict Resolution), available at http://digitalcommons.irl.cornell.edu/icrpubs/4.

55. Stipanowich, supra note 3, tbl.19 at 880 (summarizing Lipsky & Seeber, supra note 54, at

8).

56. GEN. ACCOUNTING OFFICE, EMPLOYMENT DISCRIMINATION: MOST PRIVATE-SECTOR EMPLOYERS USE ALTERNATIVE DISPUTE RESOLUTION 21 (1995). For a study using EEOC and other data to suggest new paths for empirical research in employment arbitration, see David Sherwyn, Samuel Estreicher & Michael Heise, Assessing the Case for Employment Arbitration: A New Path for Empirical Research, 57 STAN. L. REV. 1557 (2005).

57. See Colvin, supra note 7, at 586-87 & tbl.l. 58. Stewart J. Schwab & Randall S. Thomas, An Empirical Analysis of CEO Employment

Contracts: What Do Top Executives Bargain for?, 63 WASH. & LEE L. REV. 231, 234 (2006).

59. Drahozal, supra note 37. 60. Id. 61. Christopher R. Drahozal & Keith N. Hylton, The Economics of Litigation and Arbitration:

An Application to Franchise Contracts, 32 J. LEGAL STUD. 549, 575 (2003).

62. Marotta-Wurgler, supra note 50. 63. Rolph, Moller & Rolph, supra note 49, at 171. 64. KLAUS PETER BERGER, 9 TRANSNATIONAL ECONOMIC LAW 8 n.62 (1993) (citing A.J. VAN

DEN BERG ET AL., ARBITRAGERECHT 134 (1988 .

65. Commentary to Stephen R. Bond, How to Draft an Arbitration Clause (Revisited), in To-

WARDS A SCIENCE OF INTERNATIONAL ARBITRATION, supra note 53, at 57, 59.

66. Alessandra Casella, On Market Integration and the Development of Institutions: The Case of International Commercial Arbitration, 40 EUR. ECON. REV. 155, 156-57 (1996). But cf Bar-

bara Koremenos, If Only Half of International Agreements Have Dispute Resolution Provisions,

III. THE DATA SET

Which Half Needs Explaining?, 36 J. LEGAL STUD. 189 (2007) (reporting a study in which about half of public international agreements have dispute resolution clauses).

67. Introduction: Empirical Perspectives on International Commercial Arbitration, to ToWARDS A SCIENCE OF INTERNATIONAL ARBITRATION, supra note 53, at 3, 3 ("[M]uch of what we do know about international arbitration is based on anecdote ... study.").

rather than careful empirical

68. Theodore Eisenberg & Geoffrey Miller, Ex Ante Choices of Law and Forum: An Empiri-

cal Analysis of Corporate Merger Agreements, 59 VAND. L. REV. 1975 (2006).

69. See, e.g., Circuit City Credit Card Master Trust, Current Report (Form 8-K), exhibit 4.2 (Jan. 31, 2002). See generally Thomas E. Plank, The Security of Securitization and the Future of Security. 25 CARDOZO L. REV. 1655 (2004).

70. See, e.g., First Consumers Nat'l Bank, Current Report (Form 8-K), exhibit 4.3 (Jan. 31,

2002).

TABLE 1. TYPES OF CONTRACTS STUDIED (NUMBER OF CONTRACTS IN PARENTHESES)

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts.

IV. EMPIRICAL RESULTS

71. Eisenberg & Miller, supra note 68, at 7.

72. Some observers have noted that arbitration is not the preferred form of dispute resolution among corporate lawyers, despite its purported benefits. See Rebecca Callahan, Arbitration v. Litigation: The Right to Appeal and Other Misperceptions Fueling the Preference for a Judicial Forum (ExpressO Preprint Series, Working Paper No. 1248, 2006), available at http://Iaw. in southern bepress.com/expresso/eps/1248 California).

(discussing a survey of 400 business

lawyers

TABLE 2. SUMMARY OF ARBITRATION CLAUSE PRESENCE BY CONTRACT TYPE

Mergers (N)

Bond indentures (N)

Percent (%) Settlements (N) Percent (%)

Securities purchase (N)

Employment contracts (N)

Percent (%) Licensing (N) Percent (%)

Asset sale purchase (N)

Credit commitments (N)

Percent (%) Underwriting (N) Percent (%)

Pooling & servicing (N)

Security agreements (N)

Trust agreements (N)

Other (N)

333 81.02 154 99.35 60 83.33 406 88.26 70 63.06 32 66.67 253 80.57 211 97.69 350 99.72 173 100 35 94.59 48 100 429 92.86

2554 89.36

78 18.98 1 0.65 12 16.67 54 11.74 41 36.94 16 33.33 61 19.43 5 2.31 1 0.28 0 0 2 5.41 0 0 33 7.14

304 10.64

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts.

TABLE 3. ARBITRATION CLAUSE PRESENCE BY PARTY STATUS

U.S. parties only (N)

Non-U.S. party (N)

2334 90.36 217 79.78

2551 89.35

249 9.64 55 20.22

304 10.65

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. Contracts in which either party is a non-U.S. entry are coded as Non-U.S. party contracts.

TABLE 4. ARBITRATION CLAUSE PRESENCE BY PARTY

STATUS AND CONTRACT TYPE

Pooling & servicing

0.190 368 0.007 151 0.15 60 0.105 382 0.377 106 0.243 37 0.175 268 0.020 196 0 337 0 170 0.056 36 0 48 0.064 424

0.186 43 0 4 0.25 12 0.182 77 0.2 5 0.636 11 0.304 46 0.050 20 0.071 14 0 3 0 1

0 0.167 36

0.190 411 0.006 155 0.167 72 0.118* 459 0.369 111 0.333** 48 0.194* 314 0.023 216 0.003** 351 0 173 0.054 37 0 48 0.072** 460

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. Contracts in which either party is a non-U.S. entry are coded as Non-U.S. party contracts. The first row for each contract type shows the proportion of contracts containing arbitration clauses. The second row shows the number of contracts used to compute the proportion. The * indicates domestic/international arbitration clause rate difference is significant at p < 0.10; the ** indicates p < 0.05. P-values are based on Fisher's exact test.

73. At least one commentary has suggested, in the context of insurance policies, that arbitration clauses are more likely to be seen in nonstandardized contracts. Abraham & Montgomery, supra note 43, at 358 ("Few standard-form primary commercial liability and property insurance policies issued by American insurers contain arbitration clauses. Certain non-standard insurance policies, in contrast, do contain mandatory arbitration clauses.").

74. The height of a bar in a relative frequency histogram represents the density of data points in the histogram cell that the bar is drawn over. If a cell centered at x has width w and contains k data points, the height of the bar is h(x) = (k/n) x (11w), which is directly proportional to the density of points in the interval, where density = kw.

75. This hypothesis was suggested by Florencia Marotta-Wurgler in comments to an earlier

draft.

76. See U.S. Securities and Exchange Commission, CF SIC Code List, http://www.sec.gov/

info/edgar/siccodes.htm (last visited Feb. 2, 2007).

FIGURE 1. DISTRIBUTION OF CHOICE OF LAW BY CONTRACT TYPE

bond indentues

setlements

surities purchase

.5

Law chosen

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. Each histogram shows the distribution of contracts by choice of law. Contracts designating Delaware, New York, or California law are included. All other contracts are coded as designating an Other choice of law and not used in the standardization computation.

77. We started with the twenty-five industry groups used by the U.S. General Accounting Office, see U.S. GEN. ACCOUNTING OFFICE, PUBLIC ACCOUNTING FIRMS: MANDATED STUDY ON CONSOLIDATION AND COMPETITION app. iv, tbl.12 at 111 (2003), available at http:// www.gao.gov/new.items/d03864.pdf. These were then reduced to seventeen industry groups, see Theodore Eisenberg & Jonathan R. Macey, Was Arthur Andersen Different? An Empirical Examination of Major Accounting Firm Audits of Large Clients, 1 J. EMPIRICAL LEGAL STUD. 263 (2004).

TABLE 5. ARBITRATION CLAUSE STATUS BY INDUSTRY

SIC-based groups

Mineral industries (N)

Construction industries (N)

Manufacturing (N)

Transportation and utilities (N)

Communications (N)

Wholesale trade (N)

Percent (%) Retail trade (N) Percent (%)

Finance, insurance, real estate (N)

Services (N)

Instruments and related products (N)

Food and kindred products (N)

Paper and allied products (N)

Chemicals and allied products (N)

Industrial machinery and equipment (N)

Electrical and electronic equipment (N)

Transportation equipment (N)

No SIC listed or SIC missing (N)

89 88.12 22 84.62 141 89.24 104 96.30 98 91.59 77 90.59 87 89.69 755 95.33 469 84.50 57 83.82 33 97.06 19 90.48 130 83.33 75 75.00 112 86.82 25 86.21 261 89.38

2554 89.36

12 11.88 4 15.38 17 10.76 4 3.70 9 8.41 8 9.41 10 10.31 37 4.67 86 15.50 11 16.18 1 2.94 2 9.52 26 16.67 25 25.00 17 13.18 4 13.79 31 10.62

304 10.64

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts.

78. Similar considerations might support using choice of forum information instead of, or in addition to, choice of law. But many fewer contracts specify choice of forum than choice of law. At least 1700 contracts did not specify a choice of forum whereas nearly all specify a choice of law. See Eisenberg & Miller, supra note 68.

79. For contracts that designate more than one state's law (for example, New York law gov-

erns except where Delaware law applies), we used the first-mentioned state.

TABLE 6. ARBITRATION CLAUSE PRESENCE BY

CHOICE OF LAW AND CONTRACT TYPE

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. The first row for each contract type shows the proportion of contracts of that type containing an arbitration clause. The columns break down the results by choice of law locale. The second row for each contract type shows the total number of contracts of that type with or without arbitration clauses. All contracts not designating Delaware, New York, or California law are coded as "Other."

80. In a logistic regression model with arbitration clause as the dependent variable and dummy variables for contract type, law chosen, and non-U.S. party as the explanatory variables, the coefficient on California law being chosen is statistically significant (p = 0.035). Delaware law served as the reference category. But the lack of variation in the presence of arbitration clauses for pooling and service agreements and trust agreements precludes including them in the model. California's difference from Other was significant (p = 0.018). California's difference from New York was also significant (p < 0.0001).

81. See generally Eisenberg & Miller, supra note 68.

TABLE 7. PLACE OF BUSINESS BY CONTRACT TYPE

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. Excludes contracts categorized as Other.

82. See generally id. 83. See generally id.

F- --

e6

• -

0 0NN

, 0

" 6

6T

6D

F6C

000::

<

I )

m6

C" C5

,.-)

u"

cc

p.4

Mo C,

d cc

w5 CD

-t

0 0

0C0,

0 0

) 0)

0 0D

CD cc 0 , C5 CD CD 6

8D , D C

C CD 0 ,= CD

,85 0 66

I=C

0 C)

I=

0)

C>~

OQ

d > 66

-5

6D

o 6

C N

z ED8 do

-5

c5

'

o 0

C>0CD ,::5

o O.N

N D 0D c5

c5 8

8r~

8 c5

N

j

C, 8

,:58"

'0

D 0D

C>

)

0)

CD 0

CD 0

z 0 F.

" 05

0CD ,: D

C,

C) C0 o 6

00 66

C) CD

C>

0=

" C)

8 CD-

6C>

22 o

-D

2A . 0

0 CD

" C

0 8

e

J )i.

<

C D

-

4l CD

'M

0 CD

ND CD

S0z

,

=0

CA CIO m

)

'44Z).

' m 0

=

t!)

4.0

TABLE 9. ARBITRATION CLAUSE PRESENCE BY ATTORNEY LOCALE AND CONTRACT TYPE

Pooling service

0.243 74 0.000 4 0.400 5 0.104 48

1.000 1 0.292 24 0.000 5 0.000 14

0.056 18 0.000 2 1.000' 1 0.200 10 0.000 1

0.000 9 0.000 4 0.000 5

0.118 17 0.000 1

0.200 15 0.667 3

0.429 7 0.000 4 0.000 7

0.090 67 0.000 13 0.000 6 0.117 60 0.143 7 0.667 3 0.133 45 0.000 14 0.000 51

0.188 16 0.000 1 0.000 3 0.143 7 0.429 7

0.400 5 0.000 3 0.000 8

0.133 15 0.000 7

0.118 17 1.000 2 0.000 1 0.182 11 0.000 10 0.000 26

0.000 7

0.000 1

0.000 1

0.000 3

0.225 204 0.008 127 0.158 57 0.112 303 0.363 91 0.302 43 0.192 213 0.028 176 0.004 240 0.000 173 0.080 25 0.000 48

0.181 182

0.078 51

0.182 55

0.082 269

0.180 50

0.090 89

0.109 1700

0.190 411 0.006 155 0.167 72 0.117 460 0.369 111 0.333 48 0.194 314 0.023 216 0.003 351 0.000 173 0.054 37 0.000 48

0.113 2396

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. The first row for each contract type shows the proportion of contracts of that type containing an arbitration clause. The columns break down the results by attorney locale. The second row for each contract type shows the total number of contracts of that type with or without arbitration clauses. All locales that are attorney locales in fewer than fifty contracts, and unknown locales, are designated Other.

TABLE 10. LOGISTIC REGRESSION MODELS OF

PRESENCE OF AN ARBITRATION CLAUSE

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Dependent variable = 1 if contract has an arbitration Model (3) clause

marginal effects

(2.75)

0.497** 0.535** 0.562** 0.575** 0.583** 0.655** (2.69) -0.809** -0.695** 0728** -0741** -0.760** (4.91) -3.790** -3.315** -3.217** -3.170** -3205**

(2,84)

(4.10)

(3.98)

(4.11)

(2.88)

(2.85)

(3.31)

(4.11)

(10.36)

(8.31) -0.583* (2.49) 0.578* (2.36) 0.187 (0.93)

(7.96) -0.539* (2.26) 0.608* (2.44) 0.191 (0.93)

(7.87)

(7.82) -0.635** -0.608* (2.63) 0.541+ (1.90) 0.241 (1.14)

(2.50) 0.529+ (1.84) 0.235 (1.10)

Non-U.S. party

Medium standardization

High standardization

New York law

California law

Other law

Constant

Observations SIC code dummies Place of business dummies Attorney locale dummies

0.031* (2.24) -0.032** (3.63) -0.157 (8.82) -0.024* (2.14) 0.034+ (1.88) 0.009 (0.89)

-0.473+ (1.85) 0.586* (2.30) 0.116 (0.51) -3.001** (2.88) -0.233 (0.63) -0.521* (2.30) 0.902** (3.23) 0.632+ (1.73) -0.048 (0.21) -2.216** (4.58) -3.807** (3.67) -1.194 (1.51) -0.981* (2.49)

2082 Yes

-0.919** -1.055"* -0.496 (6.27) (4.73) 2395 No

2303 No

(1.15)

-0.278 (0.56) 2303 Yes

-0.241 (0.46)

Source: SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 1, 2002 to June 30, 2002 for all contract types other than mergers and Jan. 1, 2002 to July 31, 2002 for merger contracts. Absolute values of robust t, z statistics in parentheses. + significant at 10%; * significant at 5%; ** significant at 1%. Reference categories are low standardization, Delaware as the choice of law, and mergers as the contract type in model (6). Model (6) omits pooling service and trust agreements because they do not vary-all lack arbitration clauses. Models (3) through (6) include, but we do not report, dummy variables for SIC codes as reported in Table 5. Models (4) and (5) include, but we do not report, dummy variables for place of business as reported in Table 8. Model (5) includes, but we do not report, dummy variables for attorney locale as reported in Table 9. Column (7) reports the change in probability of observing an arbitration clause based on the presence of a variable, while constraining the other variables to their mean values.

84. Hylton's observation that loan contract provisions are clear and well understood by courts is consistent with our standardizationi variables results, which suggest that more standardized contracts are less likely to include arbitration clauses. Hylton, supra note 1.

V. DISCUSSION OF RESULTS

85. Stipanowich, supra note 3, tbl.19 at 880. 86. See generally DAVID B. LIPSKY ET AL., EMERGING SYSTEMS FOR MANAGING WORKPLACE

CONFLICT: LESSONS FROM AMERICAN CORPORATIONS FOR MANAGERS AND DISPUTE RESOLUTION PROFESSIONALS (2003); David B. Lipsky et al., An Uncertain Destination: On the Development of Conflict Management Systems in U.S. Corporations, in ALTERNATIVE DISPUTE

RESOLUTION IN THE EMPLOYMENT ARENA: PROCEEDINGS OF NEW YORK UNIVERSITY 53RD ANNUAL CONFERENCE ON LABOR 109 (Samuel Estreicher & David Sherwyn eds., 2004).

87. Stipanowich, supra note 3, tbl.21 at 881 (citing Lipsky & Seeber, supra note 54, tbl.6 at 11).

88. William W. Park, Arbitration in Banking and Finance, 17 ANN. REV. BANKING L. 213,

215-16 (1998).

89. Commentary to Stephen R. Bond, supra note 65, at 59.

90. Brad S. Karp, The Litigation Angle in Drafting Commercial Agreements, 1459 PRACTISING

L. INST. CORP. L. & PRAC. COURSE HANDBOOK SERIES 343 (2004).

91. See Stipanowich, supra note 3, at 893 (observing that "[a]lthough 'trial lawyers' have often had experience with mediation and other approaches, corporate transactional counsel tend to have had little or no experience with these choices-and are reluctant or unable to incorporate discussions about dispute resolution in commercial contract negotiations"); see also Celeste M. Hammond, The (Pre)(As)sumed "Consent" of Commercial Binding Arbitration Contracts: An

Empirical Study of Attitudes and Expectations of Transactional Lawyers, 36 J. MARSHALL L. REV. 589, 613 (2003) (stating that many transactional lawyers are unsure about the operation of arbitration).

92. See Shavell, supra note 1, at 9-10; supra note 35. 93. Stipanowich, supra note 3, at 894-95. 94. See, e.g., U.S. CHAMBER OF COM., STATE LIABILITY SYSTEMS RANKING STUDY: FINAL

REPORT STUDY No. 14966 (Jan. 2002), available at http:/www.instituteforlegalreform.com/

resources1012202.pdf [hereinafter U.S. CHAMBER OF COM., STATE RANKING]. The Chamber and other business groups use the Chamber's ranking studies to try and influence courts to restrict causes of action and constrain legal actions against the business community. See, e.g., Brief of the Chamber of Commerce of the United States et al. as Amici Curiae in Support of Defendant-Appellant, Henry v. Dow Chem. Co., 701 N.W.2d 684 (Mich. 2005) (No. 125205); Amicus Curiae Brief of Wisconsin Manufacturers and Commerce, Wischer v. Mitsubishi Heavy Inds. Am., Inc., 673 N.W.2d 303 (Wisc. Ct. App. 2002) (No. 99-CV-6553). For prior use of the Chamber's ranking in an empirical analysis, see Marcel Kahan, The Demand for Corporate Law: Statutory Flexibility, Judicial Quality, or Takeover Protection?, 22 J.L. ECON. & ORG. 340, 348 (2006).

95. CHAMBER OF COM., STATE RANKING, supra note 94, tbl.3 at 14.

96. Id. at 8 n.1.

FIGURE 2. ARBITRATION CLAUSE RANK AND STATE LIABILITY

SYSTEM RANK (ARBITRATION CLAUSE RANK BASED ON

PLACE OF INCORPORATION)

30 -CTWA

r 20 lFL

OMI , 15 0 -,

44 E 10

MO

UT

NH

GA

OR

NJ

LA

CO

VA

I

OH

NC

MD

AZ

Chamber of Commerce rank

14966

Source: U.S. CHAMBER OF COM., STATE LIABILITY SYSTEMS RANKING STUDY: FINAL REPORT STUDY No. (Jan. 2002), available at http://wwwinstituteforlegalreform.com/resources1012202.pdf; SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 2002 to June 30, 2002 for all contract types other than mergers and Jan. 2002 to July 31, 2002 for merger contracts. Low numerical rankings in the Chamber's system correspond to more favorably ranked state liability systems. We include states specified as a place of business in at least ten contracts. The Chamber ranking for the remaining states is not included and the 26 included states were re-ranked from 1 to 26. Ties in rank are not broken.

97. Id. tbl.5 at 17-18. Other factors include treatment of class actions, punitive damages, timeliness of summary judgment or 'dismissal, discovery, and scientific and technical evidence. Id.

FIGURE 3. ARBITRATION CLAUSE RANK AND STATE LIABILITY SYSTEM RANK (ARBITRATION CLAUSE RANK BASED 'ON

PLACE OF BUSINESS)

25CA "

20 -

-15 o

AZ

UT

WA 25 -

CO

OR

NH "NH

NC

GA

OH

FL

MO

LA

VA

CT

MI

MD

NV

Chamber of Commerce rank

Source: U.S. CHAMBER OF COM., STATE LIABILITY SYSTEMS RANKING STUDY: FINAL REPORT (Jan. 2002), available at http://wwwinstituteforlegalreform.com/reSTUDY No. 14966 sources1012202.pdf; SEC EDGAR database, LEXIS EDGAR PLUS database, Jan. 2002 to June 30, 2002 for all contract types other than mergers and Jan. 2002 to July 31, 2002 for merger contracts. Low numerical rankings in the Chamber's system correspond to more favorably ranked state liability systems. We include states specified as a place of business in at least ten contracts. The Chamber ranking for the remaining states is not included and the 26 included states were re-ranked from 1 to 26. Ties in rank are not broken.

98. Low numerical rankings in the Chamber's system correspond to more favorably ranked systems. Lower numerical rankings for arbitration clauses correspond to lower arbitration clause rates.

99. In contrast to the absence of an association between use of arbitration clauses and the Chamber's liability system ranking, we do find an associatioA between the use of jury trial waiver clauses and the Chamber's ranking of the fairness of states' juries. Theodore Eisenberg & Geof-

frey P. Miller, Do Juries Add Value?: Evidence from an Empirical Study of Jury Trial Waiver Clauses in Large Corporate Contracts (Cornell Law Sch. Legal Studies Research Paper Series, Working Paper No. 06-044, 2006), available at http://ssrn.com/abstract=946465 (showing an association between jury trial waiver rates and ranking of the fairness of juries).

100. For a summary of the data and methodology, see THOMAS H. COHEN, U.S. DEP'T OF JUSTICE, TORT TRIALS AND VERDICrS IN LARGE COUNTIES, 2001 (2004), available at http://www. ojp.usdoj.govfbjs/pub/pdf/ttvlcOl.pdf.

101. The California counties included were Alameda, Contra Costa, Fresno, Los Angeles, Or-

ange, San Bernardino, San Francisco, Santa Clara, and Ventura. Id. app. C at 12.

VI. CONCLUSION

102. Consumer Reports observes that the largest credit card issuers "insist on arbitration." ConsumerReports.org, Avoid Arbitration (Jan. 2004), http://www.consumerreports.orglcrol personal-finance/avoid-arbitration-credit-card-creditcard-companies-madatory[sic]-arbitration/ index.htm. Consumer Reports also states that most community banks and credit unions that issue credit cards do not require arbitration. Consumer Action similarly reports that 21 of 47 banks (45%) surveyed in 2005 required consumers to settle disputes using arbitration. Consumer-Action, 2005 Credit Card Study Shines Light on Industry Practices (July 28, 2005), available at http:/www.consumer-action.org/press/artices/new-credit-card-study-shines-light-onindustry-practices/. On the other hand, in the perhaps unusual setting of software end-user licensing agreements, Marotta-Wurgler found an even lower rate of mandatory arbitration clauses (6%) than is evidenced in our sample of sophisticated business contracts. Marotta-Wurgler, supra note 50.

103. Marc Furletti, Mandatory Arbitration Clauses in the Credit Card Industry 1 (Jan. 2003) (reporting on workshop led by Alan S. Kaplinsky, chair of the Consumer Financial Services Group, Ballard Spahr Andrews & Ingersoll, LLP).

104. See generally Richard A. Nagareda, Aggregation and Its Discontents: Class Settlement

Pressure, Class-Wide Arbitration, and CAFA, 106 COLUM. L. REV. 1872 (2006).

105. See generally Richard M. Alderman, Consumer Arbitration: The Destruction of the Com-

mon Law, 2 J. AM. ARB. 1 (2003).

106. Common sense and life experience with mobile phones suggest that disputes with mobile phone suppliers exist. If arbitrations are very rare, even when they constitute the only available dispute resolution mechanism, their scarcity may be evidence that arbitration clauses serve not to efficiently resolve disputes but to shelter corporate conduct that might be challenged if a costeffective, aggregate dispute resolution mechanism was available.