Introduction
Suppose a plaintiff files a class action against a defendant, but the parties had signed an arbitration clause that does not say whether it allows or bars class actions. This common fact pattern raises two questions. The first is known as “clause construction”: Did the parties agree to class arbitration or only bilateral arbitration? The second is whether a court or an arbitrator should perform clause construction.
During the 2010s, both questions were unsettled. The U.S. Supreme Court decided several controversial opinions that interpreted the Federal Arbitration Act (FAA) to require judges to enforce class arbitration waivers even when plaintiffs pursued low value claims that would either be aggregated or abandoned.
It was unclear, however, (1) whether an arbitration clause that did not mention class proceedings allowed them, and (2) which decisionmaker should resolve the issue.
Then, in 2019, the Supreme Court tried to put the clause construction dilemma to rest. In Lamps Plus, Inc. v. Varela, the Court declared that there must be an “affirmative ‘contractual basis’” for class arbitration and that neither “silence” nor “ambiguity” is sufficient.
According to the conventional wisdom, Lamps Plus slammed the door on anyone interpreting an arbitration clause that did not mention class procedures as authorizing them.
This Piece revisits the subject seven years after the Court’s game-changing ruling. It uses Lamps Plus to assess one of the most fiercely debated topics in alternative dispute resolution: whether arbitrators follow the law.
Concern that arbitration is “lawless” goes back a century.
But several factors impede any attempt to test this theory. For one, arbitration awards are often issued without published reasoning and kept confidential.
And even if one could peek behind this curtain, systemic differences in the types of disputes that are litigated and arbitrated would problematize efforts to draw meaningful inferences from deviations in outcomes.
Finally, if simply collecting arbitral awards is difficult, gathering empirical evidence about the lawlessness thesis is nearly impossible because it requires making subjective decisions about whether an arbitrator resolved a case “correctly.” Thus, it is not surprising that the literature on point relies on rough proxies and is “ultimately inconclusive.”
Yet clause construction cases are different. For starters, their reasoned awards are accessible: To protect the due process rights of absent class members, the American Arbitration Association (AAA) requires arbitrators to explain their determinations in published decisions.
Moreover, these matters are practically identical whether they appear in courts or in arbitration. Clause construction raises the same neatly packaged issue of contract interpretation and thus facilitates apples-to-apples comparisons between judges and arbitrators.
Lastly, the sweeping nature of Lamps Plus’s holding largely—although, admittedly, not entirely—sidesteps chin-stroking questions about what the law is and whether arbitrators applied it faithfully.
To shine fresh light on the claim that arbitration is lawless, the Piece analyzes forty-two judicial opinions and twenty-six arbitral awards that were issued between April 24, 2019, when the Court decided Lamps Plus, and December 31, 2024. It finds that judges saw the clause construction issue as “straightforward.”
In their eyes, Lamps Plus prohibited them from discovering implicit consent to arbitrate on a class basis. Thus, their task consisted of reading the relevant contract to confirm that it “never mentions class arbitration.”
In stark contrast, 27% of arbitrators held that a generic arbitration provision embodied a tacit agreement to allow class proceedings.
This suggests that although arbitral lawlessness may not be widespread, it is very real.
Three clarifications may be helpful. First, this Piece is a sequel to an Article published in 2019 called Clause Construction: A Glimpse Into Judicial and Arbitral Decision-Making, which analyzed 150 clause construction orders generated by courts and arbitrators between June 2010 and February 2019.
During this period, an adjudicator could defensibly find that an arbitration clause implicitly allowed plaintiffs to aggregate claims.
But Lamps Plus made reaching such a decision next to impossible.
Thus, although this Piece covers the same terrain as Clause Construction, it does so to investigate an issue that the earlier article could not: whether arbitrators obey “a change in controlling law.”
Second, clause construction inverts the normal dynamic in forced arbitration. Typically, defendants move to compel arbitration and plaintiffs fight tooth and nail to stay in the courts.
These tactics stem in part from the belief that arbitrators, who charge by the hour and are chosen by the parties, have economic incentives to cater to the repeat-playing companies that can hire them in the future.
Yet when the issue is clause construction, the current runs in the opposite direction. Arbitrators have financial reasons to favor plaintiffs: Ordering class arbitration starts the meter running on a long, complex, and profitable multiparty dispute.
Third, this Piece will not use the word “silent” to describe an arbitration clause that says nothing about class actions. As this Piece will explain, what it means for an arbitration clause to be “silent” about the permissibility of class proceedings is contested.
Instead, this Piece will refer to arbitration agreements that do not expressly address the topic as “generic.”
The Piece proceeds in three Parts. Two of them provide background: Part I surveys the lawlessness hypothesis and Part II describes the Supreme Court’s clause construction jurisprudence. Part III then uses recent clause construction rulings to contrast how judges and arbitrators conceptualize the same narrow issue.
I. The Lawlessness Hypothesis
This Part sketches the debate over whether arbitrators follow the law. It shows that there is widespread disagreement about the quality of arbitral decisionmaking. It also explains why the lawlessness theory has been so hard to verify or disprove.
Congress passed the FAA a century ago.
The statute built the infrastructure to allow merchants to adjudicate fact-bound disputes without judicial involvement.
One of its key provisions is section 10, which streamlines conflict resolution by permitting judges to vacate awards only for extraordinary defects.
For example, arbitrators cannot “exceed[] their powers” by deciding an issue that the parties did not even agree to arbitrate.
But outside of such gateway limitations, section 10 makes court oversight of arbitrators “so limited as to be little better than a rubber stamp.”
For decades after the FAA took effect, it was widely assumed that arbitrators did not obey the law. Not only did section 10 insulate awards from judicial review, but arbitrators—who were not always lawyers—were infamous for using “equitable rather than legal principles.”
As a result, even the FAA’s draftsperson, Julius Henry Cohen, cautioned that arbitration was “not the proper method for deciding points of law of major importance.”
In 1953, the Supreme Court endorsed this sentiment by creating the nonarbitrability doctrine, which exempted federal statutory claims from the FAA, in part, because arbitrators lacked “judicial instruction on the law.”
Accordingly, there was little doubt that arbitration was “essential[ly] ‘lawless[].’”
But early studies of awards painted a more complicated picture. For example, a 1948 Note in the Harvard Law Review examined three hundred awards from the AAA’s commercial arbitration docket.
The student author found that arbitrators obeyed simple legal rules. As they put it, “[n]o one need explain to an arbitrator the social utility of enforcing contractual obligations, in order to convince the arbitrator that a breach of contract should be compensated.”
Yet the Note also concluded that because arbitrators displayed less allegiance to complex black-letter principles, the “[r]esults in arbitration are sometimes different from those the courts might have reached on the same facts.”
Similarly, in 1960, Professor Soia Mentschikoff conducted a survey of AAA arbitrators and reported that 80% of them “thought that they ought to reach their decisions within the context of the principles of substantive rules of law, but almost 90[%] believed that they were free to ignore these rules whenever they thought that more just decisions would be reached by so doing.”
Both pieces of scholarship observed that the AAA discouraged arbitrators from memorializing their reasoning out of fear that “courts might upset an award which purported to decide an issue according to law but applied wrong principles.”
This issue became more important in the second half of the twentieth century, when the FAA was expanded beyond its original scope. Although the statute was supposed to be merely a procedural rule for federal courts,
the Supreme Court held that it applies in state court and preempts state laws that are hostile to arbitration, elevating it to substantive federal law applicable in both federal and state courts.
Likewise, the Court disavowed the nonarbitrability doctrine, reasoning that arbitration was just as hospitable to plaintiffs as litigation:
By agreeing to arbitrate a statutory claim, a party does not forgo the substantive rights afforded by the statute; it only submits to their resolution in an arbitral, rather than a judicial, forum. It trades the procedures and opportunity for review of the courtroom for the simplicity, informality, and expedition of arbitration.
Businesses inserted arbitration clauses into millions upon millions of consumer and employment contracts, making the word “alternative” in the phrase “alternative dispute resolution” seem like a misnomer.
This trend rekindled interest in whether arbitration is a kind of Wild West. Some critics reasoned this in the abstract, arguing that the process must exist in a legal vacuum because it is not subject to meaningful judicial review.
Others cited deeply misguided awards that courts had vacated as a sign that “arbitrators frequently apply statutes in ways that appellate courts would conclude constitute reversible error.”
Several prominent scholars pushed back against the Wild West paradigm. For instance, Professor Christopher Drahozal noted that the little existing concrete data about arbitration does not necessarily show that it dispenses lower quality adjudication than the court system.
Drahozal acknowledged that surveys like Mentschikoff’s suggest that arbitrators sometimes play fast and loose with controlling authority.
Yet Drahozal observed that research into judges and juries has uncovered the same basic phenomenon: “a significant willingness to disregard the law.”
Indeed, in one study, nearly half of all jurors surveyed stated that they would ignore a judge’s instructions to serve the interests of justice; in another, nearly 75% of federal appellate judges said that they would follow their own views over the closest precedent in borderline cases.
For these reasons, Drahozal argued that “the available empirical evidence to date provides at best weak support for the view that arbitration is ‘lawless.’”
Similarly, in 2012, Professor W. Mark C. Weidemaier compiled perhaps the most detailed account of arbitral decisionmaking and found strong parallels to the way courts resolve disputes.
Taking advantage of the fact that some arbitration institutions and legal research platforms make certain kinds of awards available, Weidemaier scraped together 848 arbitral rulings from securities, employment, labor, and class cases (including ninety-seven pre-Lamps Plus clause construction decisions).
Weidemaier described the dispute resolution process in these cases as “judging-lite”:
[A]rbitrators who write reasoned awards behave much like judges, especially when hearing statutory (as opposed to contract) disputes. They write detailed awards that make extensive use of precedent, although perhaps to a slightly lesser degree than judges. Citations to judicial opinions also dominate the arbitration awards. . . . On the whole, [these findings] . . . undercut the view that arbitration involves a qualitatively different kind of decision-making than judging.
Although Weidemaier conceded that he had no way to gauge how often arbitrators resolved matters “correctly,” he argued that because they “wrote reasonably lengthy decisions that were substantially devoted to legal analysis and that made ample use of precedent,” it did not seem as though they “render[ed] ad hoc decisions.”
Thus, Weidemaier’s study suggested that there may not be meaningful differences between courts and arbitrators.
In sum, concern that arbitration is “lawless” is nearly as old as the FAA and has long been a flashpoint between arbitration’s critics and proponents. But no one has found a way to test how often arbitrators reach demonstrably wrong decisions. The next Part explains why the law that governs clause construction has evolved in a way that can fill this void.
II. Clause Construction
This Part offers a primer on clause construction. It shows that the Court’s topsy-turvy case law on the subject culminated in Lamps Plus, which tried to ensure that generic arbitration clauses do not allow class actions.
The Supreme Court first encountered the nexus of class actions and the FAA in 2003’s Green Tree Financial Corp. v. Bazzle.
Two groups of borrowers filed class actions against Green Tree.
They had signed loans that included generic arbitration clauses.
A South Carolina trial court ordered the parties to arbitrate on a class basis, and the arbitrator issued multimillion dollar awards in favor of both classes.
On appeal, the South Carolina Supreme Court held that the contracts allowed class arbitration because they did not bar it.
The Court granted certiorari to decide whether the FAA “prohibits class-action procedures from being superimposed onto an arbitration agreement that does not provide for class-action arbitration.”
Writing for just three other members of the Court, Justice Stephen Breyer reasoned that whether the contracts allowed class arbitration was an issue for the arbitrator.
Justice Breyer reasoned that arbitrators are “well situated” to answer a question that revolves around the meaning of the contract and “what kind of arbitration proceeding the parties agreed to.”
In turn, because only the trial court had engaged in clause construction, Justice Breyer remanded the matter so that the parties could “obtain[] the arbitration decision that their contracts foresee.”
Justice John Paul Stevens provided the decisive vote, concurring in the judgment but also dissenting in part.
He would have ruled that “nothing in the [FAA] . . . precludes” finding that a generic arbitration clause permits class proceedings.
Although he agreed that “[a]rguably the interpretation of the parties’ agreement should have been made in the first instance by the arbitrator,”
he remarked that the decision to allow class arbitration was correct on the merits, and therefore he would have preferred to affirm the award instead of sending it back to the arbitrator.
Bazzle helped class arbitration briefly go mainstream. Although it was a plurality opinion, five Justices had agreed both that the FAA did not bar class arbitration and that the default rule is that arbitrators perform clause construction.
As a result, the AAA announced that it would handle class claims.
It created procedural rules for these disputes, empowered arbitrators to perform clause construction, and mandated that these awards be reasoned and published.
Between 2004 and 2009, AAA arbitrators released 135 such rulings, 70% of which allowed class arbitration.
But the pendulum swung violently in the other direction in 2010, when the Court decided Stolt-Nielsen S.A. v. AnimalFeeds International Corp.
AnimalFeeds and Stolt-Nielsen signed a maritime shipping contract with a generic but broad arbitration provision that applied to “[a]ny dispute arising from [its] making, performance or termination.”
Later, AnimalFeeds filed an antitrust class action against Stolt-Nielsen, and the parties agreed that a panel of three arbitrators should undertake clause construction.
AnimalFeeds argued that the contract allowed class claims for broadly two reasons. First, AnimalFeeds asserted that public policy supported permitting plaintiffs to aggregate complaints when damages were small and thus no individual would have an incentive to prosecute the matter.
Second, AnimalFeeds cited Bazzle for the proposition that the default rule when a “clause is silent on the issue of class treatment” is that class actions are authorized.
AnimalFeeds’s lawyers expanded on this point at the oral argument in arbitration, declaring that “the parties agree that when a contract is silent on an issue there’s been no agreement that has been reached on that issue, [and] therefore there has been no agreement to bar class arbitrations.”
The arbitrators agreed with AnimalFeeds.
They observed that arbitrators in similar cases had read generic arbitration clauses to permit class actions,
and that these awards “rel[ied] on the same type of broad wording” that this provision boasted.
The case eventually reached the Supreme Court, which reversed the arbitrators’ ruling.
Writing for the majority, Justice Samuel Alito explained that the arbitrators had impermissibly invoked their own “view of sound policy regarding class arbitration.”
Recall that AnimalFeeds’s counsel had contended that there was “no agreement” about the propriety of class actions and thus the contract did not prohibit them.
Focusing exclusively on the first part of this argument, Justice Alito opined that AnimalFeeds had admitted that “there was ‘no agreement’” either to permit or to preclude class arbitration.
In turn, he continued, because there was no shared understanding on the topic, the panel could not have grounded its ruling in the language of the arbitration provision; rather, it must have “rested its decision on AnimalFeeds’ public policy argument.”
According to Justice Alito, this violated the tenet that “the task of an arbitrator is to interpret and enforce a contract, not to make public policy.”
Finally, Justice Alito opined that, despite Bazzle, the award could not be defended as applying a default rule that generic arbitration clauses allow class proceedings.
He explained that the FAA embodies a vigorous preference for individualized dispute resolution:
Consider just some of the fundamental changes brought about by the shift from bilateral arbitration to class-action arbitration. An arbitrator chosen according to an agreed-upon procedure . . . no longer resolves a single dispute between the parties to a single agreement, but instead resolves many disputes between hundreds or perhaps even thousands of parties. . . . The arbitrator’s award no longer purports to bind just the parties to a single arbitration agreement, but adjudicates the rights of absent parties as well.
For these reasons, Justice Alito held that “[a]n implicit agreement to authorize class-action arbitration . . . is not a term that the arbitrator may infer solely from the fact of the parties’ agreement to arbitrate.”
It would be hard to exaggerate the confusion that Stolt-Nielsen spawned. What did Justice Alito mean when he opined that “the parties’ mere silence” could not “constitute[] consent to resolve their disputes in class proceedings”?
Some courts concluded that Justice Alito was referring to the language of the contract—the fact that it did not expressly license class arbitration.
Accordingly, these judges held that Stolt-Nielsen “squarely foreclose[d] the possibility that the class claims are arbitrable”
in such cases, and that generic arbitration provisions “operate[] as an implied waiver of the plaintiff’s class claims.”
This Piece will call this the textual silence reading of Stolt-Nielsen. But other judges focused on the fact that Stolt-Nielsen involved a deeply idiosyncratic version of “silence”: AnimalFeeds’s concession that the clause was silent in the sense that “there was ‘no agreement’” about class arbitration.
In turn, this confined Stolt-Nielsen to its odd facts: Indeed, all a plaintiff had to do to distinguish the opinion was not to repeat AnimalFeeds’s ill-advised admission.
This theory gathered momentum, as several appellate courts affirmed arbitral awards allowing class claims even if the agreement did not “incant[] ‘class arbitration’ or otherwise expressly provide[] for aggregate procedures.”
This Piece will refer to this as the stipulated silence view of Stolt-Nielsen.
The Supreme Court circled back to the topic in 2013’s Oxford Health Plans LLC v. Sutter.
John Sutter, a doctor, sought to represent a class of other physicians whom Oxford had failed to reimburse properly for costs.
The parties’ contract stated that “[n]o civil action concerning any dispute arising under this [a]greement shall be instituted before any court, and all such disputes shall be submitted to final and binding arbitration.”
Sutter and Oxford entrusted clause construction to the arbitrator, who reasoned that the disputes that the clause sends to arbitration are the same universal class of disputes the clause prohibits as civil actions, which therefore must include class claims.
Oxford eventually sought review by the Court, and argued that the award violated Stolt-Nielsen by finding that it and Sutter had “authorize[d] the use of class procedures simply by agreeing to arbitrate their disputes.”
But Justice Elena Kagan upheld the arbitrator’s handiwork.
Justice Kagan emphasized that when, as here, the parties had assigned clause construction to the arbitrator, the question was “whether the arbitrator (even arguably) interpreted the parties’ contract, not whether he got its meaning right or wrong.”
Justice Kagan then explained that the decision passed this lenient test because it “focused on the arbitration clause’s text, analyzing (whether correctly or not makes no difference) the scope of both what it barred from court and what it sent to arbitration.”
Finally, she dismissed Stolt-Nielsen’s significance by adopting the stipulated silence theory of that opinion:
We overturned the arbitral decision there because it lacked any contractual basis for ordering class procedures, not because it lacked . . . a “sufficient” one. The parties in Stolt–Nielsen had entered into an unusual stipulation that they had never reached an agreement on class arbitration. . . . In that circumstance, we noted, the panel’s decision was not—indeed, could not have been—‘based on a determination regarding the parties’ intent.’ . . .
Here, the arbitrator did construe the contract (focusing, per usual, on its language), and did find an agreement to permit class arbitration.
This Piece’s predecessor, Clause Construction, was published six years after Oxford Health.
Its dataset was created by searching Westlaw, Lexis, PACER, and the AAA website for opinions and awards interpreting generic arbitration clauses that were decided between June 1, 2010, and February 15, 2019.
That research found that only two of forty-four judges (4.5%) held that the provision authorized class proceedings, but fifty-eight of 106 arbitrators (54.7%) came out the other way.
This gulf stemmed from the fact that courts usually adopted the textual silence interpretation of Stolt-Nielsen, but arbitrators believed that “[a]n express provision authorizing class arbitration is not required.”
In addition, arbitrators often used two tools to interpret prosaic language as authorizing class arbitration. First, they relied on the breadth of the arbitration clause. The paradigmatic “broad” arbitration clause covers “any” or “all” future disputes, and arbitrators held that this expansive commitment encompasses class claims.
Second, arbitrators invoked the maxim of contra proferentem, which construes contractual ambiguities against the drafter.
Although contra proferentem’s normative basis has never been entirely clear, the consensus appears to be that it is a penalty default rule designed to encourage clarity and precision in drafting.
Then, in April 2019, the same month Clause Construction appeared in print, the Supreme Court doubled down on the textual silence view of Stolt-Nielsen in Lamps Plus.
Frank Varela filed a class action against his employer, Lamps Plus, after the company allowed a hacker to steal his data and obtain a fraudulent tax return in his name.
The parties had signed a generic but comprehensive arbitration provision that applied to “all claims or controversies . . . past, present or future.”
They agreed that a judge, not an arbitrator, would undertake clause construction.
A federal district court and the Ninth Circuit adopted the stipulated silence view of Stolt-Nielsen, held that the provision was ambiguous on the issue of class arbitration, and invoked contra proferentem to allow class arbitration.
Speaking through Chief Justice John Roberts, the Supreme Court reversed.
The Court assumed for the sake of argument that the clause was ambiguous.
Yet the Court reasoned that mere ambiguity is not enough to permit class arbitration, citing Stolt-Nielsen for the proposition that “[c]lass arbitration is not only markedly different from the ‘traditional individualized arbitration’ contemplated by the FAA, it also undermines the most important benefits of that familiar form of arbitration.”
Lastly, the Court held that the FAA preempts the use of contra proferentem to christen class proceedings in arbitration.
As the Justices explained, the against-the-drafter principle seeks to achieve the policy objective of encouraging careful authorship.
In turn, because it “seeks ends other than the intent of the parties,” it violates the maxim that the parties must consent to class arbitration.
Lamps Plus makes it dramatically more difficult—in fact, arguably impossible—for a decisionmaker to interpret a generic arbitration clause to allow class claims. For one, the Court repudiated the stipulated silence view of Stolt-Nielsen.
Chief Justice Roberts reasoned that “Stolt-Nielsen controls” and that his holding “follows directly from our decision in Stolt-Nielsen.”
These remarks would make no sense if Stolt-Nielsen only taught that class arbitration is impermissible in the bizarre situation when the parties agree that they did not reach consensus about that issue. Accordingly, when Lamps Plus instructs decisionmakers that “[n]either silence nor ambiguity” can be a hook for class arbitration, it must mean “neither [textual] silence nor ambiguity.”
In turn, because generic arbitration clauses are textually silent (or at best ambiguous) about whether they allow class actions, one can plausibly read Lamps Plus as making the outcome of clause construction foreordained.
But even if Lamps Plus did not create a one-size-fits-all rule, it closed (or at least narrowed) the two most common paths that decisionmakers had taken to allow class arbitration. For one, recall that after Stolt-Nielsen, arbitrators had relied on the breadth of the arbitration clause to imply consent to class proceedings.
In Lamps Plus, however, a far-ranging agreement to arbitrate “all claims” did not overcome the pull of individualized arbitration.
This should give an adjudicator pause before predicating class arbitration on the sweep of the arbitration provision. Likewise, although arbitrators once cited contra proferentem to punish drafters for failing to clarify whether class-wide arbitration is permissible, Lamps Plus removed this arrow from the quiver.
In sum, Lamps Plus essentially ensures that the clause construction inquiry has a single, correct answer. But is this how judges and arbitrators have understood the opinion? The next Part tackles that question.
III. Study and Implications
This Part investigates how courts and arbitrators have interpreted Lamps Plus. It shows that although most judges saw the Court’s opinion as a blanket rule against interpreting a generic arbitration clause to allow class proceedings, a substantial minority of arbitrators reached the opposite conclusion. Finally, it explains how this finding informs discussions about arbitral lawlessness.
As the Court reeled from Bazzle to Stolt-Nielsen to Oxford Health to Lamps Plus, lower courts struggled mightily with whether clause construction is presumptively a matter for judges or for arbitrators.
Despite the Bazzle plurality’s determination that arbitrators enjoy jurisdiction over the topic, most federal appellate courts have now decided that the high stakes of class arbitration tip the scales toward keeping the issue in the court system.
Yet it is easy for parties to inadvertently draft around this default principle. For example, in most circuits, simply stating that arbitration should be conducted under the rules of the AAA—which, as noted, contemplate that arbitrators will perform clause construction—passes the baton to arbitrators.
In turn, this means that both judges and arbitrators are resolving the issue, and the factors that control where a matter ends up are essentially random.
Accordingly, research for this Piece was conducted by searching Westlaw and Bloomberg for judicial clause construction opinions from between April 24, 2019—when the Court decided Lamps Plus—and December 31, 2024. Research for this Piece was also conducted by scouring the AAA’s class arbitration docket for clause construction awards from the same period. This yielded sixty-eight documents: forty-two judicial opinions and twenty-six arbitral awards.
Table 1 reveals that, even after Lamps Plus, some arbitrators held that a generic arbitration clause allows class procedures. Zero of the forty-two courts (0%) but seven of twenty-six arbitrators (27%) determined that class arbitration was permissible. This difference is statistically significant (p < 0.001).
Table 1. Clause Construction Outcomes After Lamps Plus
|
N |
Class Actions Allowed |
| Court |
42 |
0 (0%) |
| Arbitration |
26 |
7 (27%)***
(p=0.0002) |
| Total |
48 |
10% |
Notes:
* p < 0.05, ** p < 0.01, *** p < 0.001
|
Most courts read Lamps Plus as a “magic words” test that requires the parties to expressly agree to class arbitration.
This reduced the interpretive task to skimming a generic arbitration clause to confirm that the words “class action” did not appear. For instance, this is the entirety of the clause construction discussion in the Ninth Circuit’s 2020 opinion in Shivkov v. Artex Risk Solutions, Inc.:
The final issue that we must decide on class arbitration is straightforward. “Neither silence nor ambiguity provides a sufficient basis for concluding that parties to an arbitration agreement agreed to undermine the central benefits of arbitration itself,” . . . namely, “the individualized form of arbitration envisioned by the FAA[]” . . . . As the district court concluded, because the Agreements are silent on class arbitration, they do not permit it. Thus, the court properly compelled individual arbitration pursuant to the Agreements.
Likewise, in 2021, the First Circuit tackled clause construction in American Institute for Foreign Study, Inc. v. Fernandez-Jimenezwith a two-sentence analysis: “The [a]greement does not provide an affirmative basis to conclude that the parties agreed to class arbitration. The arbitration clause is silent about class arbitration.”
Indeed, the lack of a textual reference to class procedures was considered dispositive even when the parties had broadly agreed to arbitrate “any dispute, claim, or controversy.”
Thus, the majority of judges believed that “this issue has been definitively resolved by the United States Supreme Court[].”
Admittedly, a few courts did not read Lamps Plus as a bright line rule.
For example, the Second Circuit took this position in November 2019 when it decided Jock v. Sterling Jewelers Inc.
A class of female retail sales agents alleged that Sterling Jewelers had violated Title VII by paying them less than their male peers.
The parties’ contract featured an arbitration provision that incorporated the AAA Rules and allowed the arbitrator “to award any types of legal or equitable relief that would be available in a court of competent jurisdiction.”
An arbitrator had cited this robust remedial power as a sign that she could aggregate claims and certified a class of about 44,000 women.
In the Second Circuit, Sterling asserted that the arbitrator had no right to bind class members who had not opted in to arbitration.
The Second Circuit disagreed.
The appellate panel distinguished Lamps Plus on the grounds that while the parties in Lamps Plus agreed that a judge would perform clause construction, which subjected that decision to de novo review, in Jock, the AAA rules had delegated the interpretive task to the arbitrator, which triggered Oxford Health’s extraordinarily “deferential standard of review” of the arbitrator’s award.
This was dispositive because, as noted above, under Oxford Health, an “arbitral decision ‘even arguably construing or applying the contract’ must stand, regardless of a court’s view of its (de)merits.”
Unfortunately, the Second Circuit did not stop there. Instead, it muddied the waters by adding a throwaway sentence hinting that the arbitrator had interpreted the contract correctly: “Lamps Plus leaves undisturbed the proposition, affirmed in Stolt-Nielsen, that an arbitration agreement may be interpreted to include implicit consent to class procedures.”
Yet Jock’s take on Lamps Plus never gained traction in the judicial system. This was likely because that portion of Jock is dictum. As mentioned, because the Sterling Jewelers contract incorporated the AAA Rules, the propriety of the clause construction award was not under the Second Circuit’s microscope; rather, under Oxford Health, the key was that the arbitrator simply tried to interpret the agreement.
Thus, courts engaging in clause construction have not cited Jock for the proposition that courts may construe generic arbitration provisions to implicitly permit class arbitrations—not even a 2023 opinion from the Southern District of New York (for which Jock is binding authority).
Conversely, arbitration awards tended to fall into one of three rough camps. First, about 20% were indistinguishable from judicial rulings. These private judges echoed their public counterparts by opining that, “[i]n Lamps Plus, the majority of the Supreme Court . . . firmly closed the door on the availability of class arbitration in the face of ambiguous contract language.”
For example, in early 2019, in Veliz v. University of Southern California, an arbitrator had interpreted an employment contract to permit class arbitration.
But when the Supreme Court published Lamps Plus two weeks later, the arbitrator executed a screeching U-turn and reconsidered her prior determination:
Regardless of what one thinks about the majority’s understanding of either arbitration or class actions, especially in the context of employment contracts, arbitrators do not sit to dispense their own brand of justice. The majority decision in Lamps Plus is the law of the land and must be adhered to.
Second, approximately half of the awards also ruled in favor of businesses but offered more nuanced analysis. These arbitrators read Lamps Plus as tilting the scales sharply—but not all the way—towards individual arbitration:
“[C]lasswide arbitration” [need not] be mentioned in the arbitration agreement in order to conclude that the parties intended to arbitrate on a classwide basis, although its absence makes it harder to show that the parties consented to it. Therefore, if there is a failure to mention class arbitration, it is very likely that, at best, the [‘]any disputes[‘] language would be ambiguous [and] . . . such language would not be sufficient to find that there is an affirmative contractual basis for concluding that the parties agreed to classwide arbitration.
The awards then parsed the arbitration clause and held that it did not overcome this strong presumption. The logic varied with the provision’s text and the parties’ assertions. For example, several plaintiffs cited the absence of a class waiver or the vast scope of an arbitration clause as evidence of intent to allow class procedures.
But arbitrators were unpersuaded.
They observed that because these features are common to many arbitration clauses, the natural “extension of this argument is that most arbitration agreements would then also allow for class arbitration—a conclusion clearly at odds with current U.S. Supreme Court arbitration jurisprudence.”
Alternatively, some arbitrators found that contract excluded class claims by referring to the relevant signatories in individualized terms—such as “I and Employer both agree” to resolve disputes in the arbitral forum.
In turn, this winnowing language meant that the transaction “contemplates a two-party arbitration.”
Third, and most importantly, the remaining awards allowed class arbitration. These rulings seemed to be heroic efforts to reach a predetermined outcome. Some barely engaged with the Supreme Court’s clause construction jurisprudence. For instance, in McCoy v. Road Runner Sports, Inc., the arbitrator relied heavily on the fact that the parties had agreed to arbitrate “all disputes”
—the type of broad clause that failed to authorize class arbitration in Lamps Plus.
The award cited Lamps Plus just once, in the last substantive line of the decision, at the end of the following sentence: “Considering all the provisions of the arbitration agreement, I find there is an ‘affirmative “contractual basis for concluding that [the parties] agreed to”’ classwide arbitration.”
This reduced the Court’s opinion, which should have been the star of the show, to the role of a faceless extra.
Smith v. STK Bellevue LLC involved a variation on the same theme.
A poorly drafted employment agreement appeared both to allow and to bar class arbitration.
It mandated arbitration for “claims[] on an individual or class basis” (which the arbitrator termed the “Scope Provision”) but also specified that “[c]laims pertaining to different employees will be heard in separate proceedings” (the “Separate Proceedings Provision”).
The arbitrator noted that ambiguity is not a “sufficient basis”
for class arbitration under Lamps Plus and that the scope and separate proceedings terms were “at cross-purposes.”
Yet the arbitrator held that the parties had clearly agreed to arbitrate class claims because the scope provision was “broad and all-encompassing” and the separate proceedings clause did not expressly prohibit class actions.
Strikingly, several arbitrators did something that no court surveyed did: they leaned into Jock. One cited the Second Circuit’s dictum about clause construction as proof that “an arbitration agreement may be interpreted to include implicit consent to class procedures.”
Another mentioned Jock several times, distinguished more than a dozen contrary judicial opinions, and faulted these courts for not “consider[ing] the point made by the Second Circuit in Jock that the legal sufficiency of implicit consent survives Lamps Plus.”
One final point deserves emphasis: Litigants appealed five of the awards that allowed class arbitration and lost every time.
Under Oxford Health, if the arbitrator at least pretended to interpret the contract, their analysis survived, “however good, bad, or ugly.”
This reinforces that there is almost no chance of escaping a rogue arbitral decision.
In sum, at least in the context of clause construction, arbitral lawlessness seems to be a bona fide—albeit not rampant—phenomenon. Defenders of arbitral decisionmaking have observed that there is “little . . . evidence that arbitrators definitively differ from judges in their attitudes and practices toward legal issues.”
But this Piece finds that courts and arbitrators did not see Lamps Plus the same way. Judges almost always interpreted Chief Justice Roberts’s opinion as a per se bar on discovering an implicit agreement for class arbitration. But about 80% of arbitrators read the decision as merely clarifying that neither ambiguity nor contra proferentem can “provide the necessary ‘contractual basis’ for compelling class arbitration.”
These dueling conceptions of the law led more than a quarter of arbitrators to reach a result that no court did.
To be sure, this discrepancy only proves so much. For starters, many awards seemed like good faith attempts to apply Lamps Plus. In addition, arbitrators have become less likely to authorize class arbitration as the Court has made it harder to do so: The rate of awards permitting class claims has declined from 70% (before Stolt-Nielsen) to 55% (between Stolt-Nielsen and Oxford Health) to 27% (after Lamps Plus).
Finally, clause construction could be unique. Arbitrators might deviate less from judges when they confront, say, a consumer protection or employment discrimination complaint.
Nevertheless, it is unsettling that the lawless arbitrators in this study skewed their rulings in a way that furthered their financial self-interest. Whether plaintiffs or defendants benefit, this potential for bias raises questions about the wisdom of outsourcing an ever-expanding swath of the civil justice system to private judges.
Conclusion
Out of the disputes studied, one case perfectly illustrates this Piece’s key findings. In 2020, Bally Corporation, which makes gambling equipment, faced two class actions: one in Illinois and one in New York.
The plaintiffs, tribal casino owners, alleged that Bally had suppressed competition by obtaining bogus patents for its technology, forcing them to pay more to rent its automated playing card shufflers.
The lease between the parties contained a generic arbitration clause.
The full text of that provision read:
The parties agree that any and all controversies, disputes or claims of any nature arising directly or indirectly out of or in connection with this Agreement (including without limitation claims relating to the validity, performance, breach, and/or termination of this Agreement) shall be submitted to binding arbitration for final resolution. The arbitration shall follow the Commercial Arbitration Rules of the American Arbitration Association (“AAA”) or other mutually agreed-upon procedures and shall be conducted in a mutually agreeable location.
In the Illinois matter, a federal district court held that each plaintiff needed to pursue their own arbitrations against Bally.
Quoting Lamps Plus twice, the judge devoted just a few sentences to analyzing the issue: “[I]t is undisputed that the parties’ arbitration agreements do not provide for class arbitration; the agreements are silent on the topic . . . . Based on this lack of relevant language, the agreements do not provide a ‘sufficient basis to conclude that [the parties] agreed to . . . arbitrate on a class-wide basis.’”
Yet the New York case went down a starkly different path. An arbitrator, not a judge, considered whether Bally’s arbitration provision allowed class actions.
And it was as though this inquiry took place in an alternate dimension. The arbitrator cited Oxford Health for the proposition that “[a]n arbitrator may interpret a contract as permitting class arbitration even where the parties[’] contract does not specifically mention class arbitration.”
Then, in an eleven-page award, the arbitrator held that the breadth of the Bally clause, which covered “any and all controversies, disputes[,] or claims of any nature,” demonstrated that it “unambiguously permits class arbitration.”
The matter featured similar plaintiffs and the same defendant making the same arguments about the same contractual text.
Only the decisionmaker was different.