When a three-arbitrator tribunal issues a unanimous award but one of the arbitrators is found to have had actual or apparent bias, must the award be vacated?
As seen in Canada, the U.S. and public international law (under the Convention on the Settlement of Investment Disputes (ICSID Convention)), the answer to this question depends on the assumption about “how the sausage [award] is made.” Where it is assumed that each arbitrator necessarily influenced the others during tribunal deliberations, the award has been vacated. By contrast, where it is assumed that each arbitrator reached the decision independently, the award has been upheld.
This article comparatively analyzes the above approaches. First, it reviews the standard for partiality under Canadian law, U.S. law, and the ICSID Convention. Second, it briefly reviews decisions addressing the issue of unanimous award vacatur where one arbitrator was or appeared to be biased. Third, it provides practice tips for arbitrators to avoid the risk that their award may be vacated if their co-arbitrator was or appeared to be biased.
In Canada, each province and territory has its own separate arbitration legislation. Federally, the Commercial Arbitration Act applies to all commercial arbitrations where at least one of the parties is a federal department or Crown corporation or in relation to admiralty or maritime matters. The Supreme Court of Canada established the test for arbitrator bias as a reasonable apprehension of bias, not actual bias. A number of provincial domestic arbitration statutes, including Ontario’s and Alberta’s, codify the same test. Under the British Columbia legislation, the test for challenging an arbitrator on the basis of impartiality is arguably stricter, requiring “justifiable doubts” as to impartiality. “Justifiable doubts” are established only if there is a “real danger of bias” on the part of the arbitrator.
Whether a reasonable apprehension of bias exists is a factual inquiry. The test has been applied to arbitrators in the same manner as it applies to courts. The threshold for finding reasonable apprehension of bias is high, and the inquiry is objective. There are two elements of objectivity: (1) the hypothetical person is informed and reasonable; and (2) their apprehension of bias must be reasonable. The inquiry has been formulated as “what would an informed person, viewing the matter realistically and practically – and having thought the matter through – conclude? Would the person think that it is more likely than not that the decision-maker, whether consciously or unconsciously, would not decide fairly?”
Section 10(2) of the Federal Arbitration Act provides that a court “may” vacate an award “where there was evident partiality or corruption in the arbitrators, or either of them.” The sole Supreme Court case interpreting this provision, Commonwealth Coatings Corp. v. Continental Casualty Co., provided unclear guidance. A plurality of the Court – four out of nine justices – held that “any tribunal . . . not only must be unbiased but also must avoid the appearance of bias.” However, the concurring two justices stated that “[t]he Court does not decide that arbitrators are to be held to the standards of . . . judges,” which, in turn, prohibits the appearance of bias.
The result has been a circuit split. The Ninth and Eleventh Circuits have adopted the plurality’s articulation: “evident partiality” in an arbitrator exists where “undisclosed facts show a reasonable impression of partiality.” The First, Second, Third, Fourth, Fifth, and Sixth Circuits, meanwhile, adopt a higher standard: “a reasonable person would have to conclude that an arbitrator was partial.” While it must be shown that the arbitrator actually was partial, this can be proven inferentially through surrounding facts rather than through direct evidence of the arbitrator’s subjective mindset. The Eighth Circuit leaves the question open.
Article 52(1) of the ICSID Convention provides that an award may be annulled if “(a) the Tribunal was not properly constituted” or “(d) there has been a serious departure from a fundamental rule of procedure.” An undisclosed ground to disqualify a tribunal member may implicate these bases for annulment. Article 14(1) of the ICSID Convention, in its various language versions, requires arbitrators to be “independent” or “impartial[].” This does “not require proof of actual dependence or bias; rather it is sufficient to establish the appearance of dependence or bias.” This showing may be made, for example, where the arbitrator: (a) is partner in a law firm that is adverse to or representing a party in a simultaneous proceeding; (b) has already ruled on the same factual issues; (c) works as counsel with a party’s expert in other cases; or (d) criticizes a party in the media. Many ICSID and other international arbitral tribunals take guidance from the illustrative list of conflict situations provided in the International Bar Association’s (IBA) Guidelines on Conflicts of Interest in International Arbitration.
The provincial domestic arbitration statutes provide that an arbitral award may be set aside on the grounds of an arbitrator’s impartiality. There is a dearth of case law dealing with setting aside of unanimous arbitral awards where one of the arbitrators is alleged to be biased. This indicates that a finding of reasonable apprehension of bias in respect of one of the arbitrators does not necessarily mean that the award is set aside.
In the 2023 Vento decision, the Ontario Court refused to set aside a unanimous arbitral award despite finding a reasonable apprehension of bias in relation to one of arbitrators. The essential consideration was whether the procedural error produced real unfairness or real practical injustice. The Court considered factors such as the seriousness of the breach, the potential impact of the breach on the result, and the potential prejudice flowing from the need to redo the arbitration were the award to be set aside. Evidence showed that the impugned arbitrator spent significantly more time on the case than the other two arbitrators. Nevertheless, the Court found that given the strong presumption of impartiality and independence that applies to arbitrators, it could not assume that the other two arbitrators “passively accepted [the impugned arbitrator’s] views.” They likely reached their decision independently, and therefore one arbitrator’s bias did not taint the unanimous award.
The appeal of the Vento decision was heard on November 4, 2024. The Ontario Court of Appeal reserved its decision.
No national rule requires vacatur of a unanimous award because of one arbitrator’s evident partiality. Although the Federal Arbitration Act (§ 10(2)) provides that evident partiality “in the arbitrators, or either of them” is a ground for vacatur, vacatur under Section 10 is always optional. In Commonwealth Coatings, the Supreme Court required vacatur of a unanimous award because the presiding arbitrator regularly provided significant business services to the arbitral respondent. However, this was a plurality opinion, and it did not acknowledge or analyze the fact that the award was unanimous – a criticism noted in the dissent. The concurrence, meanwhile, said it was “enough” to hold that a significant business relationship “must be disclosed”; it did not specify whether the penalty for nondisclosure should necessarily be vacatur.
There is therefore a split among lower courts. Only the Ninth Circuit has analyzed the question in depth. It concluded: “A finding of evident partiality in one arbitrator generally requires vacatur of the arbitration award,” because arbitrators “are not isolated from each other; they hear and decide the case as a panel after joint discussion, debate and deliberation. Each panel member has an opportunity to persuade the others.” Furthermore, because the Ninth Circuit adopts the lower “reasonable impression of partiality” standard, the focus is on the “integrity of the process,” not the outcome of the arbitration. Likewise in the Eighth Circuit, “it is appropriate to assume that the concealed partiality [of one arbitrator] prejudicially tainted the award.”
The First Circuit leaves the question open. The Eleventh Circuit is internally divided. The appellate court once found that an arbitrator’s partiality was “prima facie evidence that the [unanimous] award should be vacated,” without further analysis, but lower courts have opined in dictum that even if one arbitrator had been partial, this would not have warranted vacatur because the award was unanimous.
Courts in the remaining circuits have not had occasion to find evident partiality in only one member of a tribunal. But they have cited the unanimity of the award in support of the conclusion that a given arbitrator must have decided objectively.
At ICSID, only the Eiser v. Spain annulment committee has addressed this question. It found that an arbitrator appeared to be biased to the claimant because he simultaneously worked as counsel and in two ongoing arbitrations hired the same damages expert as the claimant had in Eiser. The committee found that it had to annul the award notwithstanding the fact that the award was unanimous. Because it was impossible to “pierce the veil of a tribunal’s deliberations. . . each member of the Tribunal, including [the impugned arbitrator], is expected to have influenced the other two with his views and analysis, during the course of deliberations. It is in the very nature of deliberations that arbitrators exchange opinions and are persuaded or influenced by the opinions of their colleagues.”. The Committee reiterated this assumption: “[I]n the ordinary course, the views of each arbitrator influence and are expected to influence the views of the others, during deliberations.” Yet “[t]he influence of [the impugned arbitrator] on his co-arbitrators would have been perceived differently in every material respect, had they known the full facts and extent of his” relationship with the expert, and the co-arbitrators may not have accepted the expert’s conclusions had they been more skeptical of the impugned arbitrator’s endorsement of the expert’s conclusions.
The standard of partiality in Canada, the U.S. Ninth and Eleventh Circuits, and under the ICSID Convention is the reasonable apprehension of bias. The relatively higher standards in the U.S. First, Second, Third, Fourth, Fifth, and Sixth Circuits require a reasonable person to conclude actual bias.
Reviewing courts and tribunals have had limited opportunities to consider whether one arbitrator’s actual or reasonably perceived bias warrants vacatur of a unanimous award. Where they have done so, their key consideration has been whether the compromised arbitrator likely influenced the co-arbitrators during deliberations and, in turn, the outcome of the arbitration. Since the tribunal’s deliberations are confidential, the result boils down to one’s starting assumption. If the starting assumption is that each arbitrator reached its decision independently, vacatur is not warranted unless there is evidence to the contrary. But if the starting assumption is that the arbitrators necessarily influenced one another, then one arbitrator’s actual or perceived bias affected the outcome in the award, which must then be vacated.
Where the award is upheld despite a finding of actual or apparent bias, such a decision arguably impacts the parties’ contractual bargain. A contractual right to arbitration typically also sets out the arbitration process, including the composition of a panel consisting of independent arbitrators. Even if one arbitrator’s bias does not affect the outcome, it derogates from the parties’ bargained-for process and therefore arguably requires vacatur on that basis alone.
Arbitrators can protect themselves against the risk of vacatur if a co-arbitrator was or appeared partial. The following approaches may mitigate this risk:
These methods will strike the necessary balance between achieving the benefits of the deliberative process while maintaining arbitrator independence and in turn the enforceability of the award.
