As globalisation/economic liberalisation has fuelled international arbitration's status as the preferred adjudicatory process between corporate entities in cross-border disputes, arbitration has likewise proliferated in domestic US contracts - affecting not just corporate, but also public life. What are routine contractual clauses in the framework of corporate deals now regularly appear when one rents a car, gives birth, is fired or enters into credit agreements. In light of such proliferation and the size, scope and gravity of the matters determined, decision-maker bias in the arbitration process is a pressing concern for the arbitral community and its users.
The idea that a given arbitrator will always have either an element of inherent personal bias (affiliation bias), or have been selected so as to be favourable to one party (selection bias), becomes particularly unsettling when investor-state arbitration has empowered arbitrators to influence public policy. The power of arbitration tribunals to find states in violation of their treaty or contractual commitments, specifically evidenced in Philip Morris' high-profile actions against Australia and Uruguay's antismoking legislation, is a power few entities have. Insulated from judicial oversight, and unencumbered by precedent, the question of arbitrators' biases (from both aforementioned angles) is then something that must be taken seriously.
The selection and affiliation effects
Academics identify two common biases: the "selection effect" and the "affiliation effect". The "selection effect" or, rather, the strategic decision of a litigant appointing, unilaterally, a member of the arbitration panel, is the first side of the two-sided issue of arbitral bias. A given party to a set of proceedings is naturally predisposed to choosing an arbitrator that might view their side of the argument more favourably, or has historically followed a particular decision-making philosophy. This is of course, prima facie, biased for self-explanatory reasons. However, proponents of the use of the party-appointed system argue that it gives litigants some "ownership" over the arbitration process. They contend that to dispense with this feature of selection would be to encourage parties to choose conflict over rule-based dispute resolution. They moreover argue that as only two of the panel of three arbitrators are chosen by the respective parties, any pre-existing "selection bias" a given arbitrator may have is cancelled out by the other; with the final, neutral (often the chair) arbitrator balancing proceedings. Whatever the merits, the fact is that parties appointing friendly arbitrators is systemic within arbitration.
The other side of arbitral bias among decision makers is the less well-understood phenomenon of the "affiliation effect". Even if the selection effect as described in the above is discounted, arbitrators may still find it difficult to maintain impartiality as a result of implicit psychological preferences for their appointing party. This effect has been documented in many litigation settings. For instance, it's commonly accepted that when a litigant prepares an expert witness he or she will be pushed to identify with the lawyers on the litigant's side, effectively becoming a partisan member of the team. Becoming "affiliated" with one side is not, however, a conscious cognitive process. An arbitrator may, despite his or her best intentions to remain unbiased, still end up implicitly favouring one party's argument simply by knowing that he or she was selected by that party.
Understanding the extent of implicit biases – the affiliation effect – among arbitrators is frustrated by the dearth of available empirical data. In order to exclusively identify and separate this effect from the selection, Sergio Puig (University of Arizona) and Anton Strezhnev (Harvard University) designed an experiment intended to separate out observed voting correlations among arbitrators from those driven by the selection effect, and those exhibiting a true affiliation effect. Within the data, a statistically significant affiliation effect was observed among those who responded to the experiment's false scenario. The scenario was built around how the arbitrators who responded would award costs following a case decision. The winner's appointee was more likely to punish the losing party by having them reimburse all of the winner's costs. Conversely, the loser's appointed arbitrator was more likely to protect their appointed side by awarding only some of the costs (and not all) to the winner. In pure statistical terms, the experiment identified that an arbitrator appointed by the loser had only a 36 per cent chance of assigning all costs against the losing party – this jumped to around 55 per cent when appointed by the winner. While this displays a clear affiliation effect, the experiment also included results from arbitrators who were chosen "blind": that is, not told whom their appointing party was. Of these "blind" participants, appointees struck a balance when awarding costs under the fake scenario. In effect, the study evidenced that while arbitrators do not completely advance their appointing party's interests, when room for discretion arises, they're more likely to choose outcomes that favour the side that appointed them – hence the percentile results above. Finally, the study raised an important point to be considered if the issue of bias among arbitrators is to be addressed: that of blind selection.
Addressing bias: blind selection of arbitrators
While appointing decision makers on a "blind" basis would solve the issue of the affiliation effect (as the given arbitrator would not know his appointer and could therefore not possibly exhibit implicit bias), it would do little to ameliorate the selection effect. Nevertheless, such an approach would allow parties to maintain a level of ownership over the process of selecting decision makers, while keeping the latter in the dark and thus avoiding implicit biases occurring. Ultimately, the limitations with implementing this approach are defined by the context. In relation to investor-state arbitrations, many take place inside a closed community of legal professionals who will appear in multiple related actions; in this context blind appointing might not work because of market size – certain individuals' reputations will precede them. Alternatively, arbitral institutions and large providers of arbitration services could build blind appointments into their codes of best practice. In any event, to eliminate the types of bias described in the above, cooperation and implementation of change will be needed. The recent case of H v L and others  EWHC 137 (Comm) evidences why it still remains to be seen whether the impetus exists within the profession to enact such change. In dismissing an application to remove a chairman appointed in a Bermuda form application, Mr Justice Popplewell, in the English Commercial Court, found as follows:
"The duty to act independently and impartially involves arbitrators owing no allegiance to the party appointing them. Once appointed they are entirely independent of their appointing party and bound to conduct and decide the case fairly and impartially. They are not in any sense, as may sometimes be misunderstood by those in other jurisdictions, a representative of the appointing party or in some way responsible for protecting or promoting that party's interests. This independence is enshrined in s.33 of the Act, which requires the arbitrator to act fairly and impartially irrespective of who appointed him or her. This is fundamental and well known to all involved in London international arbitration. The fair-minded and informed observer would expect M, with his extensive experience and high reputation, to treat as second nature the fact that his duty of impartiality was entirely unaffected by the identity of the party appointing him, and would expect such independence to inform his entire approach to the subject reference."
While Popplewell J's comments reflect the orthodox thinking (and necessarily so for the integrity of the arbitral system) they rather conflict with the findings of Puig and Strezhnev. The courts however will likely return to this subject as on 15 February 2017 Popplewell J granted permission to appeal his decision, noting that the question of bias in the arbitration process was one "on which various arbitration communities" might welcome authoritative guidance going beyond the facts of the case.