In Enka Insaat ve Sanayi AS v OOO “Insurance Company Chubb” [2020] UKSC 38 the Supreme Court considered what law applies to arbitration agreements.
The claim concerned a fire at a power plant in Russia. Chubb, the insurer, paid out to the owner, subrogated to its rights, and brought a claim in the Moscow court against a subcontractor, Enka, claiming its allegedly defective work during the construction of the plant was responsible for the fire. The arbitration agreement was part of Enka's subcontract. The subcontract was governed by Russian law. The arbitration agreement itself was silent on choice of law but provided for arbitration under the auspices of the International Chamber of Commerce, with its seat in London. Enka applied for an anti-suit injunction in London to force the dispute out of the Russian court and into arbitration. The governing law of the arbitration agreement was relevant because it would affect how widely the arbitration agreement would be construed, and whether Chubb's claim would fall into its scope, triggering injunctive relief.
The 1980 Rome Convention on the Law Applicable to Contractual Obligations and the Rome I Regulation (593/2008/EC) were not drafted to cover the applicable law for arbitration agreements. This was not for particularly convincing reasons, and against UK objections during negotiations on the Convention: see M. Giuliano and P. Lagarde, OJ C 282/2, 31 October 1980, 11–12. The UK should take advantage of leaving the EU and amend the “on-shored” Rome I Regulation to deal with arbitration agreements. But until any such reform, the matter is left to the common law, which takes the following approach:
(1) First, the court should first look at the terms of the whole contract (applying English law as the law of the forum) to determine whether there is an express or implied choice of law for the arbitration agreement within it.
(2) Second, as a fallback, if there is no such choice on the terms of the contract, the system of law most closely connected to the arbitration agreement should be applied.
Before Enka, there were two views about the approach to take where (as is often the case) no choice of law had been expressed by the parties with specific reference to the arbitration agreement. Some courts preferred to apply the law of the main contract; others preferred to apply the law of seat of the arbitration. There were also differing views as to whether either approach reflected implied party choice or the closest connection test. As Popplewell L.J. noted in Enka [2020] EWCA Civ 574, [2020] 3 All E.R. 577, at [89], the lack of clarity did “no credit to English commercial law which seeks to serve the business community by providing certainty”.
The cause for this mess was the weight placed on the principle that an arbitration agreement is separate from rest of the wider contract to which it relates. As Andrew Baker J. put it at first instance in Enka, “[t]he separability of an arbitration agreement makes it a natural candidate for at least the possibility that it might be governed by a system of law different to that which governs the contract generally”: [2019] EWHC 3568 (Comm), [2020] Bus LR 463, at [48]. The Supreme Court in Enka rightly recognised that this weight on separability is misplaced. The separability principle is normally unknown to commercial parties when contracting; they instead reasonably operate on the footing that a single system of law applies to the whole contract, including the arbitration agreement. It follows that separability usually should have no bearing on the construction of the contract. Nor should it be relevant as a matter of principle. The separability principle exists to ensure the validity and enforceability of the arbitration agreement when the main contract is challenged as void or unenforceable: see section 7 of the Arbitration Act 1996. As the Supreme Court recognised, it is of limited, if any, relevance for other purposes, such as applicable law.
The majority in the Supreme Court (Lords Hamblen and Leggatt, Lord Kerr agreeing) held that the subcontract was governed by Russian law not by choice but instead because of Article 4(3) of Rome I (the European version of the closest connection test). As such, there was no choice of law from the main contract to be read across to the arbitration agreement. Instead, the common law's closest connection test was to be applied to the arbitration agreement. The selection of London as the seat of the arbitration meant that English law should apply as the governing law. The majority made clear that this choice of arbitral seat could not, without more, be taken as an implied choice of law (at [116]–[117]). Instead, the choice of seat provides the answer to the question of what system of law is most closely connected with the arbitration agreement (at [118]–[146]). This move away from an artificially expansive approach to implied choice is to be welcomed. Where party autonomy has not been exercised, the correct solution is to apply the fallback rule to identify the applicable law – not to second-guess what the parties could have agreed as the applicable law, had they turned their minds to the issue.
Lords Burrows and Sales dissented. They considered that the main contract was governed by Russian law as a matter of implied choice under Article 3 of Rome I. This must be the better view, having regard to the examples of implied choice given by Giuliano and Lagarde (p. 17) in their authoritative report on the Rome Convention. Lords Burrows and Sales said it was “natural, rational and realistic” to treat that choice as extending to the arbitration agreement too (at [228]). But, significantly, the dissenting judges went further and said that, even if Russian law was applied to the main contract not by choice but by some other rule (as the majority had concluded), there should still be a general rule that the governing law of arbitration agreement was the same as the main contract. While Lord Burrows said that this rule of consistency between the main contract and the arbitration agreement should apply at “any” stage of the enquiry (at [257]), if there is a relevant express choice of law, it is difficult to see what role there would be for this rule. The better view is that it could only be relevant for implied choice of law or the closest connection test.
The difference between the Justices arose because in Enka the majority concluded that there was no party choice of law for the main contract that could be read across to the arbitration agreement; the issue which vexed them will not therefore arise for many commercial contracts. But where, as in Enka, there is no party choice of law for the main contract, the minority's conclusion is to be preferred to that of Lords Hamblen and Leggatt. If (as otherwise held in Enka) the contract is treated as a whole, rather than as separable parts, it must be wrong to say that the legal system with the closest connection to the arbitration agreement is the law of the seat not the law applying to the rest of the contract. The majority approach essentially closed the front door on separability only to let it in through the back door.
One other point merits mention. The Supreme Court agreed that the purposive principle to “validate if possible” has a role to play when looking for the parties’ choice of law at the first stage of the analysis. A court should hesitate to apply a law potentially identified by the terms if it would impair the arbitration agreement. The degree of impairment necessary to engage the principle was left open by the court (see [95]–[109]). However, Lords Hamblen and Leggatt appeared to be willing to give an extraordinarily broad effect to this principle, extending beyond proper questions of validity to other matters. In particular, they suggested that, if a potentially applicable law would give narrow scope to the arbitration agreement, that might be a reason for applying another law instead. This muscular pro-arbitration stance may be born of a pragmatic desire to make English courts as attractive as possible for international commercial actors. But, as Lords Burrows and Sales noted (at [199], [277]) its basis in principle is dubious. And, if construction is a relevant factor for the validating tendency, might other parts of a prima facie applicable law be relevant too? For example, what about if the arbitration agreement is varied orally to include a new party but the applicable law gives effect to a no oral modification (NOM) clause in the main contract? Giving effect to the NOM clause impairs the arbitration agreement by preventing it from extending to the new party. That very issue arose recently in Kabab-Ji SAL v Kout Food Group [2020] EWCA Civ 6, [2020] 1 CLC 90. The Supreme Court has given permission to appeal in Kabab-Ji, so it may not be too long before we are given a better sense of what this muscular validating tendency means for arbitration in English law.
