The post below was written by Alex Mills, who is Professor of Public and Private International Law at University College London. It is the first contribution to the EAPIL online symposium on the English Law Commission’s proposed reform of the law governing arbitration agreements. The other posts are written by Manuel Penades, George Bermann, Sylvain Bollée, Matthias Lehmann and Giuditta Cordero-Moss.
Readers are encouraged to participate in the discussion by commenting on the posts.
The Law Commission of England and Wales has produced a deeply thoughtful and well-researched Report, which proposes a number of very welcome reforms to the Arbitration Act 1996. Regretfully, however, I have significant reservations about the proposal which is the subject of this Symposium – the adoption of a new choice of law rule for arbitration agreements. This proposal is based on the Second Consultation Paper produced by the Law Commission in March 2023, and this comment draws on my Submission which responded to that Consultation Paper.
The rules for identifying the law applicable to an arbitration agreement have long been the subject of debate. The issue was prominently addressed by the UK Supreme Court in Enka v Chubb  UKSC 38, which acknowledged (at ) that it had “long divided courts and commentators, both in this country and internationally”. The decision in Enka v Chubb has, however, strikingly failed to end the division among commentators. I understand why the Law Commission considered it desirable to address this question, because of the importance of the issue and the policy considerations it presents, and because it has been suggested that there is a lack of clarity in the Supreme Court’s judgment in Enka v Chubb. This issue is complex and reasonable arguments can certainly be made on both sides, as indeed acknowledged in the impressive Report and Second Consultation Paper. I am, however, not convinced of the proposal set out in the Report, which is that “the Arbitration Act 1996 be amended to provide that the arbitration agreement is governed by the law of the seat, unless the parties expressly agree otherwise” [Report, 12.77]. In this post I set out what I understand to be the relevant principles, and explain how these broadly support the rule adopted by the Supreme Court in Enka v Chubb, which has also been followed in other common law jurisdictions (such as Singapore and Hong Kong).
A first and well-known key principle is that the law governing the arbitration agreement need not be the same as that governing the remainder of the contract, sometimes referred to as the ‘matrix contract’. This is because of the principle of separability, which allows for a distinct analysis of the arbitration clause’s applicable law.
A second key principle is party autonomy, which is the starting point for analysis of any contractual choice of law issue, and particularly important in arbitration because of its contractual foundations. An agreement as to the law which governs a contract or a clause of a contract must generally be given effect, absent considerations of public policy. Traditionally, a choice of law may be express or implied – if the latter, the search is for factors which demonstrate a real (but undocumented) choice, not a choice which is imputed to the parties as one which they ought to have made.
In the absence of a real choice, it is necessary to consider not the intentions of the parties but the objective factors linking the contract to a particular system of law. Arbitration clauses remain subject to the common law choice of law rule, under which the objective test is sometimes described as a search for the system of law with ‘the closest and most real connection’ to the contract or contractual clause. An arbitration clause will generally be most closely connected to the place where it is to be performed, which is the seat of the arbitration (see further Enka v Chubb, at  et seq).
In the law of arbitration, another principle is that of efficiency, but this principle is secondary to that of party autonomy. While the fact that efficiency is generally a goal for parties and for arbitration can assist in interpreting arbitration clauses (see eg Fiona Trust v Privalov  UKHL 40), parties may choose to have their agreements resolved according to inefficient arbitral procedures should they so wish. The law should not interfere with their choices merely because they are thought unwise or undesirable.
Choice of Law Rule in Enka v Chubb
On the basis of these clear principles, the law applicable to an arbitration agreement should be governed by the following rule, comprised of three parts in hierarchical order. This is, in essence although not form, the rule set out by the Supreme Court in Enka v Chubb.
Subject to considerations of public policy, an arbitration agreement is governed by:
(i) The law expressly chosen to govern it;
(ii) The law implicitly chosen to govern it;
(iii) The law with which it has its closest and most real connection, which will ordinarily be the law of the seat of the arbitration.
This rule is simple in appearance, although its application may be complex in particular circumstances, as explained further below. The analysis below does not consider the application of public policy, but it remains an important limitation.
Choice of Law Rule in the Report
The Report proposes to amend the Arbitration Act 1996, to insert the following choice of law rule:
(1) The law applicable to an arbitration agreement is—
(a) the law that the parties expressly agree applies to the arbitration agreement, or
(b) where no such agreement is made, the law of the seat of the arbitration in question.
(2) For the purposes of subsection (1), agreement between the parties that a particular law applies to an agreement of which the arbitration agreement forms a part does not, of itself, constitute express agreement that that law also applies to the arbitration agreement.
This rule differs from the previous rule in three respects. First, for an express choice to be made, it is necessary that the parties expressly agree that it applies to the arbitration agreement. Second, there is no possibility for an implied choice. Third, in default of a choice, the law of the seat is automatically applied, rather than being the ordinary outcome of the rule.
Under existing law a choice of law for an arbitration agreement may arise in one of three ways.
First, the contract may contain a specific express choice of law agreement for the arbitration clause. In this case, the application of this law to the arbitration clause is self-evidently based on the principles of party autonomy and separability, and is not controversial. This position is maintained in the Law Commission’s proposal.
Second, there may be an implied choice of law for the arbitration agreement. This could arise, for example, where the parties have indicated an understanding that certain statutory provisions which are specific to a governing law will apply to the validity of the arbitration agreement. In this case, the application of the chosen law to the arbitration agreement once again follows straightforwardly as a matter of party autonomy and the principle of separability. One important question in this context is whether a choice of arbitral seat should give rise to an implied choice of law for the arbitration clause. This would certainly be a factor indicating a possible choice of the law of the seat, but it is not generally considered to be a decisive one on its own, as the inquiry is concerned with identifying a real choice made (but not documented) by the parties, and must be attentive to the terms of the contract and other relevant circumstances. This rule would thus in many cases lead to the same outcome as the proposed rule 1(b) in the Report, but would do so not because of a fixed rule of law but because of an implied agreement of the parties. This possibility is rejected in the Law Commission’s proposal.
Third, the matrix contract may contain an express or implied choice of law which should, unless the contrary is agreed, be interpreted to extend to the arbitration clause. This is understood to follow from party autonomy, in combination with the common sense presumption that if parties have made a choice of law for their entire contract, and have not specified a different applicable law for any particular clause of the contract, their choice extends to all of the terms of their contract – including any arbitration agreement (see eg Enka v Chubb, at ). This presumption is, however, rebuttable, if there are indications that the parties would not have wanted their choice to cover the arbitration agreement. It is important, however, to understand that this question is about the correct interpretation of the scope of a choice which has been made by the parties. (Here I depart slightly from the reasoning in Enka v Chubb, as I take the view that an express choice of law in the matrix contract which also applies to the arbitration clause is an express, not implied, choice of law for the arbitration clause – see also Report, at [12.34] et seq.) The issue is whether there is evidence which might rebut the common sense presumption. The rule proposed in the Report abolishes the presumption and indeed the possibility of a choice of law in the matrix contract extending to the arbitration agreement, unless it does so specifically and expressly.
There are two main justifications offered for the changes in the Report. The first is that they align with the principle of separability (Report, [12.72]). The analysis of the law applicable to the arbitration agreement is treated as an issue which is entirely unrelated to the contract of which it forms part. It is submitted, however, that this takes separability too far (see eg Enka v Chubb, at  and  et seq). Separability as a principle rightly ensures that the validity of an arbitration clause is analysed separately from the matrix contract, so that challenges to the validity of the matrix contract do not necessarily undermine the validity of the arbitration clause. This does not, however, require that the arbitration clause be treated as an entirely free-floating agreement, ignoring the context in which it was formed. Indeed, if a choice of law clause in the matrix contract is (as proposed in the Report) deemed to be irrelevant to the arbitration clause, this raises the question whether other clauses in the matrix contract are similarly irrelevant. What if the matrix contract contains an ‘entire agreement’ clause, or a ‘no oral modification’ clause? Are they also irrelevant to the arbitration clause? If not, why is the choice of law singled out, particularly as it may also have interpretive effect?
The second is that the rule proposed in the Report would be more desirable for various policy reasons. The rule would, for example, undoubtedly be clearer and easier to apply than the current position (Report, [12.74]). Applying the rule would also strongly favour the selection of English law to determine the validity of an arbitration agreement with the seat of arbitration in England, which the Report considers to be desirable on various grounds, such as the alignment of the law governing the arbitration agreement and the law governing the arbitration process, and the favourable approach of English law toward arbitration agreements (see Report, [12.16] et seq). It is submitted, however, that these justifications are also not persuasive, as they elevate efficiency and other similar policy considerations above party autonomy. In the absence of an express choice of law specific to the arbitration clause, the fixed rule in the Report in favour of the law of the seat no longer requires but rather excludes an inquiry into what the parties have actually agreed. Contrary to the analysis in the Report (eg, at [12.53], [12.73]), this is a significant constraint on party autonomy. Where parties have chosen a seat for their arbitration, but have (expressly or impliedly) chosen a different governing law for their arbitration clause, the fact that they have thereby chosen different laws for the law governing the arbitration process and the law governing the arbitration agreement may be considered undesirable, and it may be inefficient, but it is submitted that this is not a sufficient reason for the law to disrespect their choice, which is the very foundation of arbitration. The proposed rule also has the undesirable effect that the arbitration agreement and the matrix contract are more likely to be governed by different laws, which raises difficult questions concerning their consistent interpretation and validity (see, eg, Enka v Chubb, at  and  et seq). There is also a concern that arbitrators will be faced with a difficult choice between applying a law chosen by the parties (for example, through a matrix choice of law agreement, or through an implied choice), which they may consider themselves to be required to do as a matter of their contractual mandate, and applying the law that will be applied by the English courts if their award is challenged.
The Law Commission’s Report is overall an excellent example of law reform, offering carefully crafted and well-reasoned proposals for improvement. On this issue, it makes its case well, and there would undoubtedly be some benefits to the reforms which it proposes. Ultimately, however, I am not persuaded that they are consistent with the core principles that should be guiding the law. A simple and clear rule is often desirable, but in this case it is my view that the complexities of the existing rule simply reflect the complexities of arbitration, which cannot and should not be legislated away.