What approach should be taken to the proper law of an arbitration agreement?

We each provide brief thoughts in the lee of the Supreme Court’s hearings of the appeals in Enka and Kabab-Ji[1] on various potential approaches, or emphases of approach, to the issue of determining the proper law of an arbitration agreement.

The validation principle

Steven Lim

It is time for the English common law, including Singapore, to reassess the Sulamérica test, and its confusing progeny, and realign with the NY Convention.[2]

The Sulamérica test does not conform with the NY Convention. The third stage of the Sulamérica test looks to the law with the closest and most real connection. The NY Convention has a similar three-stage test. It first looks to the express or implied law. In the absence of any indication of this, the Convention provides for the law of the seat to apply.

This may not in itself make much practical difference as in most cases the choice of law, if not express, is very likely resolved by an implied law, at the second stage. And in any event, the law with the closest and most real connection is quite often the law of the seat (as seen in Sulamérica and other cases).

However, as Kabab-Ji shows, the English courts’ reliance on English contract precedent for determination of the proper law of the arbitration can potentially lead to greater divergence. The Court of Appeal questioned, but did not decide, whether the requirement for business efficacy in Marks & Spencer v BNP Paribas [2015] UKSC 72; [2016] AC 742 (“Marks & Spencer”) could be satisfied where the Sulamérica test or the NY Convention provided a fallback default choice of either the law of the country with the closest connection or the law of the place where the award was made. In other words, was there a necessity for an implied choice if there was a fallback default choice? Going down that route would be a significant departure from the NY Convention and English contract law principles on implied terms should not be applied to the determination of the proper law.

English authority has vacillated between giving primacy to the substantive law of the contract and the law of the seat. Instead of laying down a presumptive implied law, it makes more sense, and is more transparent, to apply the validation principle as required under Articles II and V(1)(a) of the NY Convention.

The validation principle gives effect to the parties’ commercial intention to agree an effective and workable international dispute resolution mechanism. It provides that if an international arbitration agreement is substantively valid under any of the laws that may be potentially applicable to it, then its validity will be upheld, even if it is not valid under any of the other potentially applicable laws. This is mandated under Article II of the NY Convention[3] which requires presumptive validity of arbitration agreements, subject only to defined generally applicable exceptions of contract law[4] without reference to national rules including special, discriminatory or idiosyncratic burdens or treatment.

Further, Article V(1)(a) NY Convention[5] provides a choice of law rule, looking to the law to which the parties have subjected the arbitration agreement, or failing indication thereon, under the law of the country where the award is made. While there is debate about this, there is a strong case that for consistency in the recognition of arbitration agreements and awards, the choice of law rules in Article V(1)(a) should also be applied in Article II. Therefore, Article II requires States to recognise and give effect to parties’ agreement on the proper law of the arbitration agreement, whether express or implied. And the implied choice would be the law that gives effect to the parties’ agreement to arbitrate.

However, the validation principle is unlikely to find favour with the Supreme Court. The validation principle does not get much mention in English common law cases. There is either an indifference or even antipathy towards it. This may be because it is not understood.

In BNA, the Singapore High Court expressly considered and rejected the validation principle on the grounds that it:

  1. was impermissibly instrumental;
  1. could be inconsistent with the parties’ intentions;
  2. was unnecessary because Singapore already endorsed the principle verba ita sunt intelligenda ut res magis valeat quam pereat i.e. words are to be understood in a manner that the subject matter be preserved rather than destroyed; and
  3. could create problems at the enforcement stage because article V(1)(a) of the NY Convention contains choice of law provisions starting with the parties’ intentions, whereas the validation principle seeks to validate an arbitration agreement without necessary regard to the parties’ choice of law.

Even though the High Court rejected the validation principle, it applied a validation approach (reading “arbitration in Shanghai” as designating venue only and not seat). The validation principle is not inconsistent with the parties’ intentions; it gives effect to the parties’ agreement to arbitrate. And it is instrumental only in giving effect to the parties’ agreement, which English contract law also strives to do. There is no conflict between the validation principle and Article V(1)(a) (and Article II) of the NY Convention as the validation principle is derived from the choice of law principles and pro-enforcement policy in both Articles II and V(1)(a).

It appears the Singapore High Court did not properly understand the validation principle and was resistant to it because it does not flow from English authority but is otherwise in sympathy with and acted in accordance with its aims. It may take some time yet before the Singapore or English courts expressly accept the validation principle.

Presumptive law of seat approach

Philippe Kuhn

This suggested approach is closely aligned with the reasoning of the Court of Appeal in Enka, albeit with a difference in emphasis and one more substantive departure from Sulamérica, namely abandoning the third stage (i.e. the closest and most real connection stage). On this approach, the English courts would (1) search for an express choice of law and, failing that, (2) apply a rebuttable presumption in favour of the law of the seat, as a matter of implication (by general imputed intention) or as a general English policy rule. The arguments in favour of this approach may be summarised as follows.

First, it is consistent with the “arbitration package” concept. The idea is that considerations of neutrality, certainty and the powers of the law of the seat (especially in relation to anti-suit injunctive relief and challenges to awards) would weigh heavily with international commercial parties. In particular if they are contracting with counter-parties from unfamiliar or multiple jurisdictions. Popplewell LJ in Enka went as far as considering that the choice of a seat was arguably analogous to an exclusive jurisdiction agreement in the litigation context.[6] While it is correct to say that the proper law of the arbitration agreement would not itself impact the power of the seat court to issue injunctions or perform its supervisory role,[7] the parties’ choice of the seat, with the requisite curial powers, is a strong ground for implication. If parties have chosen the English court’s machinery, why would they not also want English law to govern questions like validity?

Secondly, it is sensitive to the principle of separability. That is of course now well-established in English arbitration law, given Fiona Trust and section 7 of the 1996 Act. This argument was emphasised by Popplewell LJ in Enka.[8] However, in my view, separability is a secondary factor reinforcing the arbitration package argument. There are notable discussions of this point in Sulamérica[9] and in BCY[10]. The reasoning in Sulamérica as to why parties would presume the arbitration agreement to be governed by the same law as the substantive contract is quite thin and reliant on pre-Fiona Trust cases and older commentary.[11] The NY Convention and negotiation-style reasoning in BCY is stronger. However, in an ever more international arbitration climate, a clearly chosen seat is generally more revealing than the substantive law of the contract, especially when dealing with international parties given neutrality considerations.

Thirdly, a clear presumption in favour of the law of the seat can enhance legal certainty without frustrating the parties’ intentions, provided the presumption is truly rebuttable. A good example where the presumption was quite properly rebutted is Arsanovia Ltd, Burley Holdings Ltd, Unitech Ltd v Cruz City Mauritius Holdings [2012] EWHC 3702.[12] Such a presumption is particularly appropriate when dealing with ad hoc submission agreements entered into after a dispute has arisen or when dealing with other standalone arbitration agreements.[13]

It is accepted that a presumption in favour of the law of the seat does not come without difficulties. The main concern, acknowledged in Enka,[14] is that English law does not recognise the concept of a ‘floating proper law’ of the arbitration agreement. Consequently, there may be a circularity issue with the proposed presumption in cases where the parties have not clearly chosen a seat for the arbitration. For example, an arbitration clause may conflate the venue and the seat of the arbitration (as in BNA). That said, there are mechanisms for identifying the seat in most institutional rules.[15] Particularly where a seat is determined through such rules it is fair to concede that the law of seat presumption is based on general imputation.

The other valid criticism of this approach is consistency with the NY Convention. It is accepted that the more natural interpretation of that convention is that it applies the law of the seat by default only if there is no express or implied choice: see Article V(1)(a). However, the application of the NY Convention approach by the English courts is less straightforward given its indirect implementation and the long-standing approach of treating the proper law of the arbitration agreement as a contract law question. The Supreme Court is unlikely to depart radically from that paradigm.

A common law approach to interpretation and implied terms

Niraj Modha

The Supreme Court may decide that the proper law of an arbitration agreement is a matter of contractual interpretation and implication rather than rebuttable presumptions. At stage one, a court must consider whether there is an express selection of the applicable law. If not, at stage two, the court must consider which law should be implied in order to give effect to the parties’ intention to arbitrate.

The rationale for this approach is straightforward. Whether it is comprised within a single clause in a lengthy document or a submission agreement, an arbitration agreement is a contract. It ought to be treated like any other contract. This approach respects the principle of party autonomy. It also avoids the knotty “closest and most real connection” test, on which there is little guidance, and for which there is no proper authority.

The processes of interpretation and implication are fundamentally distinct, even if they share similar features.[16] At the first stage of the suggested test, the court must interpret the arbitration agreement. In a series of decisions over the past decade, the Supreme Court has confirmed that both the contextual approach (giving significance to commercial considerations) and the textual approach (looking most carefully at the words used) are equally valid when interpreting a contract. In the context of arbitration agreements, the courts have adopted a purposive approach to interpretation.[17] This is uncontroversial.

It may be that the Court in Kabab-Ji adopted an impermissibly wide approach to interpretation at stage one of the Sulamérica test. Then again, if the arbitration agreement in Kabab-Ji was not binding because the respondent did not later become a party to it, then that does not necessarily demonstrate a defect in the exercise of interpretation. Not every arbitration agreement will be valid and binding in every circumstance.

At the second stage of the suggested test, the Court must consider the implication of a term. A term may only be implied into a contract in order to give that contract business efficacy.[18] At the time of contracting, the parties could not have sensibly decided to include an arbitration agreement which would be scuppered because there is no reference to the law which governs it, or the wrong law is applied to it. The implication of the appropriate law upholds the principle of party autonomy. Where there is a choice of several laws, a court might find that the parties must have intended for the arbitration agreement to result in a binding award which was enforceable at the seat, and therefore the law of the seat applies. Alternatively the principle of separability may lead to the same result. There is no need for a presumptive law. The question of which one of several national laws should be implied should be left to the court.

As with each of the other suggested approaches, there are criticisms which may be levelled. First, on what basis can the “closest and most real connection” test be discarded? If one is applying common law principles, and conflict of laws rules ordinarily apply where the substantive law is uncertain, there is an argument that these rules should apply to the arbitration agreement.[19] However, that is only a weak justification. In Sulamérica it was “common ground” that conflict of laws rules applied, but there is no higher authority which supports this.[20] Neither the NY Convention, nor the 1996 Act, nor the Rome I Regulation (which expressly does not apply to arbitration agreements) indicates that a conflict of laws test is appropriate.[21] Stages one and two should be exhaustive.

Second, is it artificial to require a court to imply or impute a term into an arbitration agreement? On this analysis the court’s hands are tied: it must find a law, in order to save the arbitration agreement. Arguably, that is not an intellectually honest approach. Yet, one should not ignore the practical reality. Plainly an arbitration agreement must be governed by a national law. The court is not re-writing the bargain or improving it for one party. Invariably, the court simply is choosing between a law which maximises the workability of the arbitration agreement, as the parties must have intended, or one which undermines it.

Third, if the aim is to validate the arbitration agreement then why not adopt the validation principle? One of the criticisms of Sulamérica is that the Court of Appeal obliquely referred to the law which would maintain the “effectiveness” of the arbitration agreement but did not decide the case on that basis. There is a difficulty here, albeit possibly only one of nomenclature. The courts do apply a validation principle where there is ambiguity in the interpretation of an agreement. The principle provides that, where there are competing meanings of a clause, the court should prefer an interpretation which upholds its validity rather than invalidates it.[22] This pre-supposes that there is an express term which has more than one meaning. Arguably, at present there is no scope for deploying the validation principle as a tool of interpretation.

A final plea: do not be hesitant to find an express choice

Ben Olbourne

On any of the normative approaches that have been considered here, the starting point in every case remains that effect must first and foremost be given to a choice of law expressly adopted by the parties, if one can properly be discerned: this is the first stage of any two-part or three-part approach to ascertaining the proper law of an arbitration agreement. If the result of that express choice of law, after whatever sympathetic approach to interpretation may be appropriate in the particular case, is that the parties’ arbitration agreement is void or incapable of being given effect to, that may be disappointing but that is the result of poor drafting and not any lacuna in the law of international arbitration, and it is not for a tribunal or a court to offer up a recourse to arbitration that the parties were unable to provide for themselves.

There ought not to be any tension here conceptually with the validation approach as analysed above: in each case, the aim is to give effect to the parties’ intentions to the greatest extent possible, both as regards the proper law to be applied to the arbitration agreement and as regards their evident intention to submit their disputes to arbitration. This aim ought to be achievable in most cases, but it may not be so in every case. But in seeking to give effect to what is construed to be an agreement to arbitrate, one should not be too quick to forget that ultimately the task at hand is one of contractual interpretation and the case for a different set of rules to be applied to arbitration agreements or choice of law provisions as opposed to all other contract provisions is not necessarily made out.

When applying the first stage of the analysis, there are good grounds for contending that a court or tribunal should not be hesitant to discern an express choice of law for the purposes of an arbitration agreement from express stipulations in its main contract. First, it is not clear that the principle of separability extends further than the minimum we identified earlier. It certainly is not the case that an arbitration agreement can be interpreted wholly in isolation from its main contract. Why, then, draw a sharp line between a choice of law provision in a main contract and the other provisions of the main contract when it comes to construing the terms of an arbitration agreement?  Secondly, arbitration clauses, just like the main contracts containing them, are often drafted by commercial people or by lawyers without specialised arbitration law knowledge. In those circumstances, there is no reason to presume that, generally, the drafters were even alive to the issue that an arbitration agreement might have a proper law differing from that of the main contract. Accordingly there is no reason to approach an express choice of law provision in the main contract with the suspicion that it was not meant also to extend to the arbitration agreement. There is, plainly, some conceptual overlap here with the second-stage implied law or rebuttable presumption analysis. However, the plea here is to consider giving express words in the main contract their ordinary meaning when it comes to construing the terms of an arbitration agreement.

Like with other approaches, this ‘gloss’ is no panacea by itself. It does not assist at all where the main contract does not contain an express choice of law provision, and it does of course leave open the possibility of reasonable minds reaching different conclusions as to issues of contractual construction, particularly in cases where there are competing or ambiguous provisions. As to this last point, specifically, it would appear that the English Court of Appeal in Kabab-Ji had little difficulty in construing the particular provisions in the contract before it as amounting to an express choice of law for the purposes of the arbitration agreement. However, a number of commentators have taken the view that the court pressed the boundaries of contractual interpretation too far in that case and/or that the decision in that case represents the outer limit of the bounds for finding an express choice of law. Others may see nothing at all surprising in the result. Finally, this gloss of course does not assist at all where, on a proper construction, the express provisions of the main contract do not identify a choice of law for the purposes of the arbitration agreement. In those circumstances, one returns squarely to the “muddle” embodied in the cases considered above and the differences of approach that have been considered.

Concluding remarks

We expect that the Supreme Court will clear away at least some of the muddle in its consideration of the appeals in Enka and Kabab-Ji. The Supreme Court’s decisions ought, at the very least, provide some certainty as to the approach to be adopted. That would be welcomed by potential parties to arbitrations and those advising them, but it remains to be seen whether the Court will also resolve the conceptual debates. We eagerly anticipate those decisions and intend to publish a follow-up piece (or pieces) considering them shortly after they have been handed down.

[1]      Permission to appeal in Kabab-Ji was granted by the Supreme Court on 8 July 2020, but a hearing is yet to be fixed.

[2]      This is also discussed in Steven Lim’s Kluwer Arbitration Blog post, Time to Re-Evaluate the Common Law Approach to the Proper Law of the Arbitration Agreement (http://arbitrationblog.kluwerarbitration.com/2020/07/05/time-to-re-evaluate-the-common-law-approach-to-the-proper-law-of-the-arbitration-agreement/?doing_wp_cron=1594541212.3820850849151611328125).

[3]      Reflected in Article 8 Model Law and section 9(4) English Arbitration Act.

[4]      It is null, void, inoperative or incapable of being performed.

[5]      Reflected in Article 34(2)(a)(i) and Article 36(1)(a)(i) of the Model law, and section 103(2)(b) of the 1996 Act.

[6]      [47]–[53].

[7]      [63].

[8]      [92]–[95].

[9]      [10]–[12], [18].

[10]     [60]–[61].

[11]     Moore-Bick LJ relied heavily on dicta in the Channel Tunnel case and comments in Mustill & Boyd, The Law and Practice of Commercial Arbitration in England (2nd ed., 1989).

[12]     In that case, Indian law was found to be the proper law of the arbitration agreement despite a London seat because of two combined factors: (1) Indian law as the substantive law of shareholders’ agreement, together with (2) the express exclusion of certain provisions of the Indian Arbitration and Conciliation Act.

[13]     See further Russell on Arbitration (24th ed.) at [2-120]; Sulamérica at [26]; BCY at [66]–[67].

[14]     [103].

[15]     E.g. ICC Rules, article 18.1; LCIA Rules, article 16; UNCITRAL Rules, article 16. For discussion see Russell on Arbitration (24th ed.) at [2-127].

[16]     Marks & Spencer, per Lord Neuberger at [26].

[17]     See Fiona Trust, per Lord Hoffmann at [8].

[18]     Marks & Spencer, per Lord Neuberger at [21].

[19]     Section 46 of the 1996 Act.

[20]     [9].

[21]     Article 1(2)(e) Regulation (EC) No. 593/2008 (Rome I).

[22]     Tillman v Egon Zehnder [2019] UKSC 32; [2020] AC 154 per Lord Wilson JSC at [38].

Steven Lim, Ben Olbourne, Niraj Modha, Philippe Kuhn

Author: Steven Lim
Author: Ben Olbourne
Author: Niraj Modha
Author: Philippe Kuhn