Back to Arbitration Committee publications

Maxi Scherer
Queen Mary University of London, School of International Arbitration; WilmerHale, London
maxi.scherer@wilmerhale.com

Natalie Hall
hall.natalielouise@gmail.com

Introduction

The UK Supreme Court has handed down its long-awaited judgment in Halliburton Company v Chubb Bermuda Insurance Ltd (formerly known as Ace Bermuda Insurance Ltd), which clarifies the English law on arbitrator bias and is now the leading case on this issue. Perhaps most significantly, the Supreme Court established that an arbitrator has a duty of disclosure, and that breach of this duty could be evidence that he or she was not acting fairly and impartially, justifying removal under the English Arbitration Act 1996. The recognition of a legal duty of disclosure has been welcomed by arbitral institutions and organisations which made submissions to the Supreme Court in favour of it.

Another interesting aspect of the decision was the way in which the Supreme Court dealt with the issue of unconscious bias. Throughout the decision, the Supreme Court emphasised the critical importance that an arbitrator act fairly and impartially, referring to it as a ‘cardinal duty’.¹ It referenced the findings of a recent Queen Mary University of London arbitration survey, which shows that one of the main reasons why parties choose to arbitrate in England is that the English legal system guarantees neutrality and impartiality.² However, as this article explains, from a psychological point of view, there is evidence to suggest that an arbitrator’s decision-making processes, like all human decision making, can be easily influenced by unconscious bias. The Supreme Court acknowledged the difficulty of identifying unconscious bias,³ and cautioned against attempting to open ‘windows into [people’s] souls’, therefore favouring an objective assessment of apparent bias.⁴ Given these limitations, it is important that the international arbitration community is open-minded to other ways of protecting against unconscious bias and this article considers some of the creative solutions which have been suggested to date.

The Supreme Court’s decision

The case arises out of the Deepwater Horizon disaster in the Gulf of Mexico. Halliburton had provided cementing and well-monitoring services to the lessee of the Deepwater Horizon drilling rig for the temporary abandonment and plugging of the well. Following the blowout of the well, Halliburton commenced arbitration proceedings against its insurer, Chubb, over Chubb’s refusal to pay Halliburton’s claim under a Bermuda Form insurance policy. The policy was governed by New York law and provided for ad hoc arbitration seated in London.

The parties failed to agree on the selection of the presiding arbitrator and therefore the English High Court appointed Kenneth Rokinson QC, who was one of the arbitrators that Chubb had previously proposed and Halliburton had objected to, for a number of reasons, including that insurers had a practice of repeatedly appointing retired judges or QCs known to them, such as Rokinson.

Before Rokinson’s appointment, he disclosed he had been an arbitrator in previous arbitrations involving Chubb, including arbitrations where he had been appointed by Chubb. He also disclosed that he was acting as an arbitrator in relation to two ongoing arbitrations involving Chubb. After Rokinson took up his appointment, he accepted two further appointments which he did not disclose in the Halliburton matter: (1) an appointment by Chubb relating to a claim under a Bermuda Form insurance policy with another insured party, Transocean; and (2) an appointment by Transocean in relation to a claim against a different insurer.

After learning of these appointments, Halliburton asked Rokinson to resign, which Rokinson refused to do. Halliburton then made an application to the English court for his removal under section 24(1)(a) of the Arbitration Act 1996 (the 1996 Act), which was unsuccessful. Halliburton appealed to the Court of Appeal, which also rejected the challenge.

The issues raised in the appeal to the Supreme Court were: (1) whether and to what extent an arbitrator may accept appointments in multiple references concerning the same or overlapping subject matter(s) with only one common party without giving rise to an appearance of bias; and (2) if the arbitrator may do so without disclosure.

The Supreme Court first of all emphasised the fundamental duty of any arbitrator to act impartially, which is enshrined in section 33 of the 1996 Act. It then went on to provide a thorough judgment on the issue of arbitrator bias and clarified the English law position. The relevant test is whether a ‘fair-minded and informed observer, having considered the facts, would conclude that there was a real possibility that the tribunal was biased.’⁵ This assessment involves a ‘balanced approach to any information she is given, she will take the trouble to inform herself on all matters that are relevant… She is fair-minded, so she will appreciate that the context forms an important part of the material which she must consider before passing judgment.’⁶

The Supreme Court explained that the distinct characteristics of arbitration must be taken into account in applying the fair-minded observer test, including: its private nature, the limited right of appeal, the appointment process, differing understandings of the role and obligations of the party-appointed arbitrator, the reputation and experience of an arbitrator (although this is unlikely to be given much weight) and the possibility of opportunistic or tactical challenges.⁷ The Court explained that there may be circumstances in which the acceptance of appointments in multiple references concerning the same or overlapping subject matter(s) with only one common party might reasonably cause the objective observer to conclude that there is a real possibility of bias, although this will depend on the facts of the case and custom and practice in the relevant field of arbitration.⁸ Significantly, owing to the language in section 24(1)(a) of the 1996 Act, the Court found that compliance with this duty should be assessed with regard to the facts and circumstances known at the date of the hearing to remove the arbitrator, rather than being restricted to those known only at the time the duty disclose was engaged.⁹

The Supreme Court also clarified the English law position on the duty of disclosure, which was previously in a state of uncertainty. In agreement with the Court of Appeal, the Supreme Court found that an arbitrator is under a legal duty to disclose the facts and circumstances which would or might reasonably cause an objective observer to conclude that there was a real possibility of bias – even if at a later date with the benefit of further information the objective observer would conclude that there was not such a real possibility.¹⁰ The duty of disclosure is a continuing obligation and, in contrast with the test for apparent bias, compliance should be assessed with regard to the circumstances at the time the disclosure obligation arose.¹¹ The Supreme Court explained that this duty is derived from section 33 of the 1996 Act, which gives rise to an implied term in the contract between the parties and the arbitrator that the arbitrator will act fairly and impartially. This term will be breached if an arbitrator has knowledge of undisclosed circumstances which would warrant his or her removal under section 24 of the 1996 Act.¹²

The Supreme Court also stated that the duty of disclosure will be subject to an arbitrator’s privacy and confidentiality obligations. If such obligations apply, the parties to whom those obligations are owed must provide express or implied consent before the disclosure can be made.¹³ Finally, the Court clarified that a failure to make a disclosure is a factor which will be taken into account in the assessment of whether there is a real possibility of bias.¹⁴

Applying these principles to the facts of the case, the Supreme Court found that Rokinson was under a legal duty to disclose his subsequent appointment by Chubb because at the time of his subsequent appointment, the existence of potentially overlapping appointments with only one common party was a circumstance which might reasonably give rise to the real possibility of bias.

However, by the time of the removal hearing, Rokinson had explained that his disclosure failure was an oversight, for which he apologised, and this was not challenged by Halliburton. Rokinson also stated that he had not learned anything which was not public knowledge about the Deepwater drilling incident in the undisclosed arbitrations but was prepared to resign from these arbitrations if they were not decided shortly through preliminary determinations.

In the circumstances, the Court found that a fair-minded and impartial observer would not have concluded that there was a real possibility of bias at the time of the removal hearing for the following reasons:

• there was a lack of clarity in English law as to whether there was a duty of disclosure and whether disclosure was required;
• the timings of the arbitrations provided some explanation as to why Rokinson did not recognise the need for disclosure;
• it was likely that the other arbitrations would be resolved by the preliminary issue (or else Rokinson offered to resign) and therefore it was unlikely that Chubb would be able to benefit from the overlapping arbitrations;
• Rokinson had not received any secret financial benefit from the appointment; and
• there was no basis for inferring unconscious bias in the form of subconscious ill-will.¹⁵

The appeal was therefore dismissed.

Unconscious bias from a psychological perspective

The Supreme Court’s judgment has provided helpful clarification on how arbitrator bias will be assessed by the English courts and will no doubt enhance the reputation of the English legal system as a guarantor of neutrality and impartiality in international arbitration. The duty of disclosure under English law will be welcomed by many, including international arbitration institutions, such as the International Chamber of Commerce (ICC), London Court of International Arbitration(LCIA) and the Chartered Institute of Arbitrators (CIArb), who argued in favour of the recognition of such a duty.¹⁶

The importance of impartiality was repeatedly emphasised throughout the judgment. For example, Lord Hodge opened his judgment: ‘[i]t is axiomatic that a judge or an arbitrator must be impartial’ and referred to it as a ‘cardinal duty’ and ‘core principle’.¹⁷ Lady Arden referred to it as a ‘key principle of our arbitration law’ and noted that English law is ‘rigorous in its approach to arbitrator bias and conflicts of interest.’¹⁸

However, English law has traditionally taken a restrictive approach to assessing the influence of unconscious bias. As the Court of Appeal in Locabail (UK) Ltd v Bayfield Properties put it: ‘the law does not countenance the questioning of a judge about extraneous influences affecting his [or her] mind.’¹⁹ This was the fundamental issue raised by Halliburton before the Court of Appeal. Halliburton argued that in assessing whether there was a real possibility that the tribunal was biased, regard should be had to the risk of unconscious bias. The Court of Appeal recognised that the assessment of the fair-minded and informed observer could include the risk of unconscious bias. However, the Court was dismissive of the possibility of Rokinson’s unconscious bias, citing Dyson LJ’s observations in AMEC Capital Projects that ‘arbitrators are assumed to be trustworthy and to understand that they should approach every case with an open mind.’²⁰ The Court also found it was unlikely that Rokinson would have been unconsciously biased as Rokinson was a ‘well-known and highly respected international arbitrator with very extensive experience.’²¹

The Supreme Court affirmed that unconscious bias can be part of the objective assessment of impartiality and overlapping undisclosed references could, in theory, lead to a finding that the tribunal was biased. However, the Supreme Court acknowledged the difficulty of identifying unconscious bias:

‘[t]he possibility of unconscious bias on the part of the decision-maker is known, but its occurrence in a particular case is not. The allegation, which is advanced in this case, of apparent unconscious bias is difficult to establish and to refute.’²²

It further cautioned: ‘we are [not] required to ‘make windows into men’s souls’ in search of an animus against a party or any other actual bias, whether conscious or unconscious… we are only concerned with how things appear objectively.’²³

As explained above, the Supreme Court found that the failure to disclose the further appointments to Halliburton could have led a fair-minded and informed observer to conclude that there was a real possibility of bias, but by the time of the removal hearing, this was no longer the case, and therefore a fair-minded and informed observer would no longer assess that there was a real possibility of bias.²⁴ Critics of the Supreme Court’s judgment will argue that too much of the decision hinged on the Court’s view that the relevant time to make the assessment was at the hearing to remove the arbitrator, rather than the time he accepted the subsequent appointments and the duty to disclose arose.

The Supreme Court did not provide a detailed explanation of how the change in circumstances between the subsequent appointments and the removal hearing militated against the likelihood of unconscious bias, and some may doubt that there was sufficient evidence to conclude that a real possibility of unconscious bias ceased to exist.

Professor William Park, who was appointed by Halliburton, refused to join in the award as a result of his ‘profound disquiet about the arbitration’s fairness’, explaining: ‘[t]he lack of disclosure, which causes special concern in the present fact pattern, cannot be squared with the parties’ shared ex ante expectations about impartiality and even-handedness.’²⁵

Evidence from psychological research suggests that arbitrator decision making, like all human decision making, is prone to bias. This possibility of bias stems from the dual process capability of the human brain, which has intuitive and deliberate processing functions (systems one and two).²⁶ Broadly speaking, the intuitive (system one) component acts quickly and controls our instinctive skills and reactions, while the slower, deliberate (system two) part allows us to perform actions which require conscious thought.²⁷ It is not uncommon for system one to make an error, which system two may fail to cross-check, resulting in an unconscious bias. Consider this example:

‘A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost?²⁸

The intuitive answer is $0.10; however, the actual answer is $0.05 – the bat costs $1.05 and the ball costs $0.05. In a recent study, around 80 per cent of arbitrators failed to answer a small selection of cognitive reasoning tests like the bat and ball example correctly,²⁹ indicating that arbitrators are not immune from unconscious bias. Significantly, as one practitioner has explained, where the evidence or law is ambiguous, there is ‘ample room for non-evidentiary and extra-legal considerations to influence outcomes.’³⁰

There are different ways in which decision-making processes can be affected by unconscious bias. For example, arbitrators may be said to have their own frame of reference, influenced by their culture, background and prior experiences, which may lead an arbitrator to unfairly view one party’s case more favourably than another’s. Haggard and Mentschikoff conducted an experiment involving 20 different panels of arbitrators from the American Arbitration Association – half brokers and half manufacturers. The arbitrators heard the same contractual dispute concerning whether the broker defendant had the right to cancel a sale and goods contract with a manufacturer. A comparison between the decisions showed that the arbitrators from the broker background were far more likely to favour the broker defendant than the manufacturers, which supports the view that the arbitrators were subconsciously influenced by their background and experience.³¹

Arbitrator decision making may also be influenced by confirmation bias, meaning an arbitrator may form an initial view on an issue and then interpret all subsequent information to support this initial assessment. Judge Joan Donoghue of the ICJ has described how adjudicators and arbitrators can suffer from an ‘intuitive-override’ manner of judging, where they form an initial conclusion on an issue and then test that conclusion by scrutinising the relevant material.³² Tellingly, in an arbitration survey, nearly 88 per cent of arbitrators said that they sometimes reach a preliminary view after receiving the prehearing submissions.³³ If this early assessment is wrong, it may be difficult for an arbitrator to change their view as they are more likely to engage with the submissions and evidence which supports their preliminary conclusion and pay less attention to the material which does not.

Finally, much of the Supreme Court’s decision focused on whether Rokinson had breached a duty of disclosure by not informing Halliburton of his appointments in the overlapping references. The Court stated that an arbitrator’s failure to comply with the disclosure obligations could be a significant factor in assessing whether or not the arbitrator was biased. However, from a psychological perspective, there is evidence to suggest that this issue could actually cut the other way. In other words, a person can feel less responsible for making a biased decision if they have disclosed their conflict of interest. In one study, participants were asked to imagine that they were giving advice to a person who could make money by accurately estimating the number of jellybeans in a jar. The participants were told that the jar contained fewer than 2,900 jellybeans and that they would be paid $50 if the person overestimated the number. The participants were asked to rate the morality of suggesting a number above 2,900 in two conditions – when the estimator was aware of the incentive and when they were not. The participants reported that it would be less unethical to overstate the number when the estimator was aware of the conflict. In other words, disclosure can act as a ‘moral licence’ to make a biased decision.³⁴

Is there a better way to achieve impartiality?

It is possible that English law may adopt a more stringent approach to unconscious bias. Lady Arden drew attention to what she called the ‘wisdom of Parliament’ when it enacted the 1996 Act to leave certain issues to judicial development rather than codifying them in legislation.³⁵ She explained the law in this area can therefore ‘keep pace with change. It can take account of developing standards and expectations in international commercial arbitration in particular.’³⁶ However, while the Supreme Court identified some issues for judicial development (for example the question of what enquiries an arbitrator should make about potential conflicts of interest), it is difficult to imagine that the judiciary will make real inroads to tackling the complex and pervasive influence of unconscious bias.

Given the natural limitations of the law in guarding against the influence of unconscious bias, other more innovative solutions are sometimes suggested in the international arbitration community. One such solution, which the Supreme Court described as a lively debated topic, is the abolition of party-appointed arbitrators, with appointments instead being made exclusively by arbitral institutions. Advocates of this position argue that each party tries to appoint an arbitrator who is most likely to help them win the case. The parties’ expectations in this regard might subconsciously affect an arbitrator’s mindset when deciding the dispute. Others have taken a stronger position, advocating that party-appointed arbitrators are ‘ill-conceived’, unprincipled and a ‘moral hazard’ which should be dispensed with, or rigorously policed.³⁷

However, the majority view remains that party autonomy to select an arbitrator is an essential feature of international arbitration and many parties typically try to avoid institutional appointments – even of the presiding arbitrator. According to this view, institutional appointments could therefore result in decreased user confidence and satisfaction, ultimately reducing the popularity of arbitration.

Some arbitral institutions have introduced an unconventional solution. In 2014, the International Institute for Conflict Prevention and Resolution introduced a ‘blind’ appointment process in its 2014 Rules, whereby the parties inform the institution who they wish to appoint, and the institution then appoints the arbitrators without revealing which party has nominated them.³⁸ Similarly, in 2018, the Mauritius Chamber of Commerce and Industry Arbitration and Mediation Centre included a blind appointment process into its arbitration rules.³⁹

Blind appointment processes could be a sensible mechanism to reduce the risk of bias, while preserving the parties’ right to choose their own arbitrators based on their qualifications, experience and any other objective considerations. Blind appointments might also reduce the likelihood that an arbitrator’s appointment will be challenged, given it is more difficult to demonstrate partiality when an arbitrator is unaware of which party appointed them.

However, blind appointment processes have their limitations. Some critics are sceptical whether the parties would both agree to opt into a blind arbitrator appointment process after a dispute has already begun (when reaching agreements on procedural matters is often difficult), especially as such an appointment process would prevent the parties from conducting interviews with prospective candidates before nominating them (as this would reveal the identity of the parties).⁴⁰ Others are concerned that in practice arbitrators will often be able to deduce who has appointed them, either through the parties’ conduct or through educated guesswork.

Finally, users of international arbitration should endeavour to increase the arbitrator pool to include more diverse candidates. At present, parties tend to select an arbitrator who they know, or at the very least is a trusted name within their law firm. As a result, the same individuals are repeatedly appointed, making it harder for new candidates to be considered. However, if detailed and trusted information was used by the parties during the selection process, parties would have the knowledge and confidence to make a selection outside of the limited pool.⁴¹ This would have the advantage of making repeat appointments like Rokinson’s less likely. It would also hopefully introduce a greater diversity of thought, or ‘cognitive diversity’ into the arbitrator pool. It is considered that members of a cognitively diverse group are more likely to be able to overcome the perils of ‘group think’ (where members of a cohesive group seek unanimity rather than appraising alternative courses of action)⁴² and more likely to challenge each other’s perspectives and assessments, providing an effective antidote to unconscious bias.

Conclusion

The Supreme Court’s decision has provided a welcome clarification on the English law position on arbitrator bias, establishing a legal duty of disclosure in fairly broad circumstances. In this case, the appointed arbitrator was not found to have acted unfairly or partially. However, the Court stated that the lack of clarity on the duty of disclosure in English law was a material factor in its assessment, which means that other arbitrators who fail to disclose multiple appointments with one common party may be found in breach of their fairness and impartiality obligations and therefore may be removed under the 1996 Act. The Supreme Court’s decision has therefore achieved an appropriate balance between party autonomy (on which the 1996 Act is based) and establishing a firm approach to arbitrator fairness and impartiality.

However, perhaps the most interesting aspect of the case was the way in which the Supreme Court handled the slippery issue of unconscious bias. While the Court affirmed that unconscious bias could be part of the objective assessment of impartiality, it acknowledged the difficulty of trying to establish and defend its influence. In line with previous decisions, the Court cautioned against too detailed an enquiry into whether or not an arbitrator had been biased and instead maintained that the legal test must remain objective. Critics of the judgment will argue that the Court did not fully explain why, in the case at hand, the real possibility of bias ceased to exist by the time of the arbitrator’s removal hearing. However, now that the duty of disclosure has been established more clearly, similar actions could lead to an arbitrator’s removal in the future. The Supreme Court’s decision has highlighted the fact that the law cannot offer sufficient protection against the pervasive influence of unconscious bias without compromising legal certainty. The international arbitration community will therefore need to be open minded to alternative solutions to safeguard against the influence of unconscious bias.

[1][2020] UKSC 48, at para 49. On the topic more generally, see J. van Haersolte-van Hof, 35^th Annual Queen Mary Freshfields Arbitration Lecture 2020, which can be accessed at https://play.freshfields.com/35th-annual-freshfields-arbitration.

[2][2020] UKSC 48, at para 63.

[3][2020] UKSC 48, at para 70.

[4][2020] UKSC 48, at para 52.

[6][2020] UKSC 48, at para 52.

[7][2020] UKSC 48, at paras 56 – 68.

[8][2020] UKSC 48, at para 131.

[9][2020] UKSC 48, at para 121.

[10][2020] UKSC 48, at para 132.

[11][2020] UKSC 48, at para 119.

[12][2020] UKSC 48, at para 76.

[13][2020] UKSC 48, at paras 88 – 89.

[14][2020] UKSC 48, at para 117.

[15][2020] UKSC 48, at para 149.

[16][2020] UKSC 48, at para 80.

[17][2020] UKSC 48, at paras 1, 49, 151.

[18][2020] UKSC 48, at para 163.

[19]Locabail v Bayfield Properties [2000] QB 451, at 471.

[20]Halliburton Company v Chubb Bermuda Insurance Ltd [2017] EWHC 137 (Comm), at para 51.

[21]ibid, at para 98.

[22][2020] UKSC 48, at para 70.

[23][2020] UKSC 48, at para 52.

[24]Therefore, there was a real possibility of unconscious bias at the date the disclosure obligation was breached (December 2015), but not by the time of the removal hearing in January 2017.

[25][2020] UKSC 48, at para 26.

[26]E. Sussman, ‘Arbitrator Decision-Making: Unconscious Psychological Influences and What You Can Do About Them’, American Review of International Arbitration (2013) Vol. 24, No. 3, 487, 488.

[27]D. Kahneman, Thinking Fast and Slow (2011) Farrar, Straus and Giroux: New York, at p. 22.

[28]D. Kahneman, Thinking Fast and Slow, (2011) Farrar, Straus and Giroux: New York, at p. 44.

[29]C. Guthrie et al. Inside the Arbitrator’s Mind (2017) 66 Emory L.J. 1115, 1137.

[30]S. Diamond, ‘Psychological Aspects of Dispute Resolution: Issues for International Arbitration’, in A. van den Berg (ed), International Commercial Arbitration: Important Contemporary Questions (2003) Kluwer International B.V., at p. 339.

[31]Ibid, p. 336.

[32]An ICJ Judge on Bias in International Adjudication, 22 August 2019, available at https://globalarbitrationreview.com/Article/1196307/an-icj-judge-on-bias-in-international-adjudication

[33]E. Sussman, ‘The Arbitrator Survey – Practices, Preferences and Changes on the Horizon’, (2015) American Review of International Arbitration, Vol 26 No 4, at p. 524. 3.5 per cent of arbitrators said they always reached a preliminary view, 14.1 per cent said they usually did (i.e.: 75 per cent of the time), 19.3 per cent said that they often did (i.e.: around 50 per cent of the time) and 50.8 per cent said that they sometimes did (i.e.: around 25 per cent of the time).

[34]G. Helleringer and P. Ayton, ‘Bias, Vested Interests and Self-Deception in Judgment and Decision-Making: Challenges to Arbitrator Impartiality’, in T. Cole, The Roles of Psychology in International Arbitration (2017) International Arbitration Law Library, Vol. 40, at pp. 37-38.

[35][2020] UKSC 48, at para 162.

[36][2020] UKSC 48, at para 162.

[37]J. Paulsson, Moral Hazard in International Dispute Resolution (2010) 25 ICSID Rev F Investment LJ 339, at p. 340.

[38]CPR Rules for Administered Arbitration of International Disputes, Rule 5.4,

[39]MARC Arbitration Rules 2018, Article 8.1(d) and Appendix 5.

[40]E. Krishna and M. Brown, ‘Blind Arbitrator Appointment Procedures – A Welcome Sight in Institutional Rules?’, 1 May 2019, available at www.ibanet.org/Article/NewDetail?ArticleUid=B4D1E884-52C7-44A3-805A-7FE8CCBA5183#Footnote%204

[41]There are tools which can be used to allow parties to make a more informed selection, including: GAR Arbitrator Research Tool or Arbitration Intelligence Reports, which contain data and feedback about arbitrators and arbitrations.

[42]I. Janis, Victims of Groupthink (Houghton Mifflin Company, 1978), at p. 78.