Three recent cases have clarified the law to be applied in this area, one in the UK and two in the US. ASM Shipping Limited v Bruce Harris & ors [2007] EWHC 1513 concerned an application under section 24 of the 1996 Act for the removal of the two remaining arbitrators in circumstances where the third arbitrator stood down for apparent bias. The judge did not accept that there is an invariable rule that one member of a tribunal tainted with bias will affect the rest of the tribunal and dismissed the application.

The parties had appointed Mr Bruce Harris and Mr A G Scott as arbitrators and the Tribunal made an award in favour of ASM (the ship owners) in respect of its freight claim. The two arbitrators continued to conduct the arbitration alone for two years until Mr Harris and Mr Scott appointed Mr Duncan Matthews QC as the third arbitrator. Following a hearing in October 2004, the tribunal issued an award substantially in favour of TTMI's (the charterers) original claims for damages due to the vessel's late arrival.

ASM challenged the award under section 68 of the 1996 Act on the ground of serious irregularity. It alleged that Duncan Matthews QC should have recused himself from the proceedings because he had previously represented ASM's principal witness, a broker, in separate proceedings, on an application for disclosure of documents by that witness. Mr Matthews had disclosed this connection after the witness had given evidence at the earlier hearing, but stated that he saw no basis for him to recuse himself. Mr Justice Morison rejected the section 68 application, but recommended that Mr Matthews stand down, which he duly did. ASM then applied under section 24 for the removal of Mr Scott and Mr Harris.

Mr Justice Smith rejected the application under section 24, which allows a party to apply to the court to remove an arbitrator on the grounds "that circumstances exist that give rise to justifiable doubts as to his impartiality…and that substantial injustice has been or will be caused to the applicant." He did not accept that one arbitrator's bias will necessarily taint the rest of the tribunal. Whether they will be tainted depends upon the particular facts of the case.

In the US, the test for bias of arbitrators has been attracting significant attention in recent months. Earlier this year, the Fifth Circuit considered whether an award should be set aside due to non-disclosure by the arbitrator of a previous working relationship with counsel for one of the parties. In that case, Positive Software Solutions v. New Century Mortgage, 2007 WL 111343 (5th Cir., 18 Jan. 2007), the claimant alleged that the award should be vacated due to the appearance of bias, as the arbitrator had failed to disclose that, seven years earlier, he and the counsel for one of the parties had been amongst the 34 lawyers that represented a company in certain unrelated litigation. The arbitrator had not worked with the particular counsel on the case, although their names appeared together on some pleadings.

The Fifth Circuit held that the relationship was too trivial to warrant vacating the award and that the fact of non-disclosure was not, in itself, evidence of partiality. Rather, arbitral awards must be upheld unless there is a concrete, not speculative, impression of bias. Vacating an award for non-disclosure by an arbitrator is only warranted when it involves a significant compromising relationship, not a trivial or insubstantial one.

This case has prompted much discussion about the appropriate test for bias and the dilemma of finding a balance between promoting disclosure by arbitrators, whilst not requiring disclosure of every past relationship, no matter how trivial or minimal. The precedent on this issue in the US was set by the Supreme Court ruling in Commonwealth Coatings Corp. v. Continental Casualty Co., 393 U.S. 145 (1968). In that case, failure to disclose a previous business relationship with one of the parties led to the award being vacated. The relationship in that case was more significant than that in Positive Software Solutions, but the Court held that even the "impression of possible bias" must be avoided and that standards for arbitrator disclosure should be, if anything, more rigorous than those for judges. The Fifth Circuit in Positive Software Solutions distinguished Commonwealth Coatings on the facts, and also concluded that Justice Black's opinion regarding the "appearance of possible bias" was not binding on lower courts. That standard should be "interpreted practically rather than with utmost rigor". The Supreme Court has denied certiorari, meaning that it has decided not to review the decision of the Fifth Circuit on appeal.

The more recent Second Circuit case of Applied Industrial Materials Corporation v. Ovalar Makine Ticaret ve Sanayi, 2007 WL 1964955 (2nd Cir., 9 July 2007) appears to set a stricter standard for a finding of bias, adhering more closely to the Commonweath Coatings standard. The court here held that "when an arbitrator knows of a potential conflict, a failure to either investigate or disclose an intention not to investigate is indicative of evident partiality." This imposes a duty on an arbitrator to investigate if he has reason to believe that a non-trivial conflict of interest might exist.

Finding the "appearance of bias" standard (for removing an arbitrator due to bias) to be too low and the "proof of actual bias" standard to be too high, the court held that "evident partiality" will be found where a reasonable person would have to conclude that an arbitrator was partial to one party to the arbitration. This would be the case, for example, where an arbitrator knows of a "material relationship" with a party but fails to disclose it. Furthermore, arbitrators must actively ensure that the parties are not misled into believing that no non-trivial conflict exists. Where an arbitrator has reason to believe that a non-trivial conflict of interest might exist, he must either investigate the conflict, or disclose his reasons for believing there might be a conflict and his intention not to investigate.

These developments demonstrate that the test for a finding of bias is not fixed in stone and that close attention will be paid to the facts in any given case.