Choice of Law – Angus Macinnis and Gianluca Rossi

December 2nd, 2020

If the parties to an arbitration agreement have not chosen a law to govern it, how should Australian courts select the proper law?

Angus Macinnis
Director of Dispute Resolution, StevensVuaran Lawyers
View Profile

Gianluca Rossi
LLB (Hons) student at the Australian Catholic University
View Profile


Australian courts (and many Australian lawyers) are yet to grapple with an issue affecting the resolution of disputes by international commercial arbitration; namely, what is the governing law of the arbitration agreement when the parties do not expressly choose one? Is it the law chosen by the parties for the main contract or is it the law of the seat of arbitration chosen by the parties? This article first considers how the issue arises, and then considers how Australian lawyers can resolve it (or, if they do not resolve it, the contexts in which the issue may need to be resolved for them by an Australian court).  The article then proposes that the law of the matrix contract should prevail, because that result is likely to better reflect the commercial reality of the way in which commercial parties and their lawyers approach choice of law issues.


It is often no easy feat to direct the attention of contracting parties to choice of law issues at all, let alone to command this attention for a sufficient period of time to answer not just one choice of law question, but two. However, where international contracts contain agreements to arbitrate, a series of cases from around the world have recently demonstrated that there are two distinct choice of law questions and if both are not given attention, this can lead to disputation in later court proceedings.

The first choice of law question is, what is the law of the main or “matrix” contract.  The second choice of law question is, what is the law governing the arbitration agreement.  Where there is no express choice of law governing the arbitration agreement, complications arise in determining which governing law should be applied. Competing schools of thought propose two solutions; first, that the law of the arbitration agreement should follow the express choice of law of the matrix contract, alternatively, that the law of the arbitration agreement should follow the law of the seat.

Although the question has been considered in a number of jurisdictions, the question has not yet been considered by an Australian court.  This article considers the competing arguments and suggests that Australian courts should hold that the unexpressed law of the arbitration agreement should follow the law of the matrix contract where the parties have expressly chosen the latter.[1]

Determination of the governing law of the arbitration agreement is not merely of idle academic interest. It has practical significance in that it is the law that determines the parties to the arbitration agreement and how the arbitration agreement is to be interpreted.

Identifying the governing law of the arbitration agreement

Australian law provides the parties to a contract with the freedom to choose the law that will govern their contract.  In the absence of a choice (either express or inferred), then, by the operation of choice of law rules, “the law will itself select a proper law”.[2]  Although a contract must have at least one governing law, the principle of dépeçage (the French word meaning literally the dismemberment or butchering of an animal) permits different parts of the same contract to be governed by different governing laws. The fact that the law permits dépeçage does not, of course, mean that it is necessarily a good idea.  In most cases, one governing law is desirable.

However, in the context of contracts that contain agreements to arbitrate, an invitation to sharpen the butchers’ knives may arise from the fact that the arbitration agreement is “separable” from the main contract.  The doctrine of separability, which is recognised in Article 16(1) of the UNCITRAL Model Law on International Commercial Arbitration (‘Model Law’) is primarily relevant (some authorities would say, only relevant[3]) as a means of preserving agreements to arbitrate in circumstances where the matrix contract has been impugned. However, the doctrine can raise additional questions, including whether an arbitration clause can be “separated” from the matrix contract’s governing law by a contrary indication in the arbitration clause. The most common example of such a contrary indication is the choice of an arbitral seat, and thus a choice of lex arbitri (being the procedural law of the arbitration[4]), which differs from the law governing the matrix contract.

In the absence of a choice of law in the matrix contract, a choice of seat can supply an indication as to the law applicable to the entire contract.  As the High Court said in Akai (at 442), albeit not in an arbitration context:

A submission, in the contract, to the exclusive jurisdiction of the tribunals of a particular country, may be taken as an indication of the intention of the parties that the law of that country is to be the proper law of the contract.[5]

In Habas Sinai Ve Tibbi Gazlar Istihsal Endustrisi AS v VSC Steel Company Ltd, [6] a case in which the parties had not chosen a governing law for the contract, Hamblen J referred to Cie Tunisienne and said (at [102]):

[T]he terms of the arbitration clause may themselves connote an implied choice of law. It is recognised that they may operate as an implied choice of law for the matrix contract itself  . . . In such cases they must surely equally operate as an implied choice of law for the arbitration agreement.

So, the question becomes how to resolve two potential choices, where they pull in different directions? This question was considered in Sulamerica CIA Nacional De Seguros SA v Enesa Engenharia SA (‘Sulamerica’)[7]. In that case, Moore-Bick LJ proposed a three-stage test for determining the law of the arbitration agreement. The test can be summarised as follows:

  • If the parties have made an express choice of law to govern the arbitration agreement, the choice would be effective, regardless of the law applicable to the contract as a whole;
  • If there is no express choice of law, the next question is whether an implied choice of law be identified. One basis for such an implication would be a conclusion that the choice of law provision in the matrix contract was also intended to govern the arbitration agreement; and
  • If an implied choice of law cannot be identified, the governing law will be the law that has the ‘closest and most real connection’ to the arbitration agreement.[8]

As a practical matter, if the third stage is reached, this will generally point to the law of seat, because ‘the supporting and supervisory jurisdiction necessary to ensure that the arbitral procedure is effective’[9] will provide a strong connecting factor with the seat, and connecting factors relevant to determining the governing law of the matrix contract (for example, the place of performance of contractual obligations) will be less relevant.

It may be noted that the three-stage approach laid down in Sulamerica, viewed through Australian eyes, looks suspiciously like the “three tiered” approach which was rejected in Akai as ‘serv[ing] needlessly to complicate matters’.[10] However, the test proposed in Sulamerica is now considered to be a widely accepted test (at least from a common law perspective) for resolving the governing law of the arbitration agreement.[11] Numerous academic commentaries, including Born,[12] Lew[13] and Redfern and Hunter,[14] have also endorsed this test.

Accordingly, even in the context of what was said in Akai, the Sulamerica test provides a starting point for an Australian court to resolve the choice of law in respect of an arbitration agreement. Before considering what an Australian court might do, however, we shall first consider the means available to parties to avoid ending up in court to debate the question.

Solving the problem by avoiding it in the first place

The first part of avoiding the problem is recognising that there is a problem if the governing law of the arbitration is not specified.  Once the problem has been recognised, it can be avoided by well-advised drafting, or by the selection of appropriate arbitral rules, or both.

The simplest solution to the problem is to include (within the arbitration clause) express language specifying the governing law of the arbitration clause.  That language should be in plain terms (and it is difficult to improve on the language suggested in the HKIAC Model Clause discussed below, namely “The law of this arbitration clause shall be (the selected law)”).

However, it is also possible to draft a choice of law provision in the matrix contract which extends to the arbitration agreement.  In Klöckner Pentaplast GMBH & Co KG v. Advance Technology (HK) Co Ltd,[15] the “Governing Law and Jurisdiction” clause in the matrix contract provided that ‘This MOU and all of the obligations contained herein shall be governed in its entirety by the laws of the Federal Republic of Germany’. [emphasis supplied].  Saunders J held that ‘The two expressions, “all of the obligations”, and “in its entirety”, are both expressions which on their face plainly refer to the whole of the contract, including the arbitration clause’. [16]

To prompt consideration of the question by parties who may not be aware of the issue, some arbitral institutions include the issue in their model clauses.  For example:

  1. in August 2014, the HKIAC Model Clause was amended to include the optional provision specifying the law which is to govern the arbitration agreement (set out above); and
  2. the model clause proposed by VIAC[17] for parties wishing to use the 2018 VIAC Arbitration Rules includes the question of “the substantive law applicable to the arbitration agreement” as one of a number of possible topics to be dealt with by “Optional supplementary agreements”.

For other arbitral institutions, the question is resolved by providing for a default provision in the rules where the parties have not specified the law to govern the arbitration agreement.  For example:

  1. Rule 23.5 of the 2016 ACICA Arbitration Rules (which came into effect on 1 January 2016) provides as follows:The law of the seat shall be the governing law of the arbitration agreement, unless the parties have expressly agreed otherwise and that agreement is not prohibited by an applicable law; and
  1. Rule 16.4 of the 2014 LCIA[18] Rules (which came into effect on 1 October 2014) provides as follows:The law applicable to the Arbitration Agreement and the arbitration shall be the law applicable at the seat of the arbitration, unless and to the extent that the parties have agreed in writing on the application of other laws or rules of law and such agreement is not prohibited by the law applicable at the arbitral seat.

If the parties are unable to reach agreement on the law to govern the arbitration agreement, but they have agreed on the ACICA Rules, then they have indirectly reached agreement on the law governing the arbitration agreement (namely, the law of the seat). The existence of default provisions specifying the governing law of the arbitration agreement are thus relevant to the multi-faceted decision of which arbitral rules to select.

We acknowledge that the default provision in the ACICA Rules, which points to the law of the seat, differs from what we contend to be the correct Australian approach; namely, that a choice of law of the matrix contract should also apply to the arbitration agreement.  However, the purpose of a default provision is to create certainty where the parties themselves have not provided it. Rule 23.1 of the ACICA Rules also provides for Sydney to be the default seat if the parties have not chosen a seat, so the combination of the two parts of Rule 23 provides what should be a foolproof default provision.

Accordingly, the fact that the law of the seat may be a preferred solution where the parties have made no choice does not, we respectfully submit, determine the question which we are considering here.  The question here is “which of the two choices which the parties have made should prevail?”; not (as is the case for the question answered by Rule 23), “how should a choice be supplied when the parties have failed to make any choice at all?”.

In what circumstances is an Australian court likely to encounter this problem?

If the parties have not managed to keep themselves out of court using the means suggested above, and the contest between the two choices (either the governing law of the matrix contract or the law of the seat) must be resolved, the question of which law applies will only arise in a case where that question is determinative of some real issue between the parties.  However, this was not so in Singapore; in BCY v BCZ,[19] Chong J noted (at [5]) that:

The contest between New York law [being the governing law of the putative matrix contract] and Singapore law [being the law of the seat] was misplaced simply because both parties acknowledged during the hearing that there was, in real and practical terms, no material difference between the two systems of law insofar as they relate to the only substantive issue before this court [being what is the governing law of the putative arbitration agreement].

Few Australian judges would have much patience for debate about a choice of law issue if the resolution of that dispute is of no material importance to the overall outcome of the case.  However, assuming that the competing contenders for the governing law of the arbitration agreement did affect some real issue between the parties, it is likely that the issue might arise in the context of either:

  1. an application by a party to stay court proceedings (and for the referral of the matters in those proceedings to arbitration) under section 7(2) of the International Arbitration Act 1974 (Cth). In that context, a party seeking to oppose the stay may wish to impugn the arbitration agreement as being null and void, inoperative or incapable of being performed.  That was the factual scenario which gave rise to Hong Kong proceedings in Klöckner Pentaplast;[20]
  2. an anti-suit injunction by an Australian court seeking to restrain proceedings in a foreign court where those proceedings are inconsistent with an arbitration agreement providing for arbitration in Australia (a like factual scenario gave rise to the proceedings in Sulamerica); or
  3. a proceeding in which the recognition and enforcement of a foreign arbitral award is resisted on the basis that, in accordance with Article V(1)(a) of the New York Convention, the arbitration “agreement is not valid under the law to which the parties have subjected it”.

Many cases in which the issue is likely to arise will be cases in which one party is impugning the validity of the arbitration agreement (i.e. the first situation).  In that context, some might argue that the answer is to be obtained by selecting the law which will uphold the validity of the arbitration agreement, regardless of whether that involves choosing the law of the matrix contract or the law of the seat.  After all, it would be reasonable to assume that, having concluded an arbitration agreement, the parties shared an intention that the agreement to arbitrate is valid rather than invalid. Born refers to this as the “validation principle”.[21]

However, this argument, although superficially attractive, does not (despite its support by noted arbitration law scholars such as Gary Born[22]) provide a tenable solution under Australian law.  Amongst other problems, if the basis for the invalidity arises after the time of contracting, the consequence will be that the determination of the “validating law” as the governing law will be made not at the time of the conclusion of the contract, but at some later time.  This is tantamount to an impermissible “floating choice of law”.[23] Indeed, the Singapore courts have recently determined that the so-called “validation principle” is not part of Singapore law.[24]

In the absence of an express choice, should the “law of the matrix contract” or the “law of seat” prevail as the law of the arbitration agreement?

When the best arguments for each competing school of thought are surveyed, the arguments supporting the choice of the law of the matrix contract are, in our view, superior. The primary reason for this is one of commercial practicality. From a commercial perspective, contracting parties are doing well if they can agree on one choice of law (let alone two). Accordingly, it should not lightly be assumed that where parties have made a choice about a seat (which may be driven principally by practical considerations) they have also intended to make a legal choice about the governing law of the arbitration agreement.

Where there is an express choice of law of the matrix contract, this arguably meets the first-step of Moore-Bick LJ’s test.[25] This is because such express choice provides a contractual choice of law applicable to all parts of the contract, which the parties have accepted by making their contract. The High Court of Singapore articulated the reason for this:

Where the arbitration agreement is a clause forming part of a main contract, it is reasonable to assume the contracting parties intend their entire relationship to be governed by the same system of law.[26]

If the first step is not satisfied and it is necessary to move to the second step, the identification of an implied choice requires the consideration of the specific factors from which that choice is to be implied.  An express choice of law of the matrix contract may (even if it is not regarded as an express choice) be a relevant factor; in considering the second step, Moore-Bick LJ said:

[A]lthough one may start from the assumption that the parties intended the same law to govern the whole of the contract, including the arbitration agreement, specific factors may lead to the conclusion that that cannot in fact have been their intention. [27]

The question that then arises is whether the choice of the seat is one of those ‘specific factors’ that might displace the starting assumption.  Those who argue that it does rely principally on the doctrine of separability and the availability of dépeçage. However, the doctrine of separability does not provide a strong argument in favour of an implication of the law of the seat as the governing law of the arbitration agreement. The better view of the doctrine of separability is that it should not prevent the express choice of law in the matrix contract extending to the arbitration agreement. Rather, the doctrine assists by ensuring that that the arbitration agreement will still be valid even if the matrix contract is impugned.[28] The doctrine should not be deployed in support of an argument that that the arbitration clause will necessarily have a “separate” existence such that the governing law clause in the matrix contract has no application to it.

Moore-Bick LJ also considered that where the third-step is satisfied, the reason for the law of the seat having the ‘closest and most real connection to the arbitration agreement’ is a result of that seat exercising supervisory jurisdiction over the arbitration.[29]  It is plainly correct to say that an arbitration agreement, standing alone, has a closer connection with the law of the seat than with any other law. However, in the circumstance which we are considering here, the arbitration agreement does not stand alone.  In particular, it is becoming increasingly common for parties to adopt “staged” or “multi-tiered” dispute resolution clauses in which a non-binding process (for example, negotiation or mediation) is required as a precondition to arbitration.  It would seem a surprising and unhelpful result if that part of a multi-tiered dispute resolution clause dealing with arbitration was covered by the law of the seat when everything else in the clause was governed by the law of the matrix contract.

Perhaps the best argument for the law of the seat as the governing law of the arbitration agreement is that when parties opt for dispute resolution to occur in a neutral country, they do so in order to ‘insulate the dispute resolution mechanism from the national law of either party’.[30] Applying the law of the seat as the governing law of the arbitration agreement achieves this objective by ‘insulating’ the arbitration clause from the matrix contract, and in some circumstances, this is a worthy aim.  However, if this is what the parties want, it is a simple thing for them to say so expressly. That a desire for “insulation” must necessarily carry with it a desire for a second choice of law debate is not a result, which, in our view, is consistent with the commercial reality of choice of law in the context of arbitration clauses (sometimes referred to as ‘midnight clauses’).


Given the complexity of the debate about this question, and the differing responses which courts around the world have applied, it is desirable that Australian practitioners serve their clients well by selecting the governing law of their arbitration agreements rather than leaving the matter to be determined by an Australian court.  If the matter is to be determined by a court, the solution which we propose accords with the United Kingdom and Singapore courts, and does not require dismemberment of parties’ contractual intentions.

[1] Absent an express choice of the governing law of the matrix contract, Article 28 of the UNCITRAL Model Law on International Commercial Arbitration provides that the Arbitral Tribunal may choose the governing law that it considers to be appropriate.

[2] Akai Pty Ltd v People’s Insurance Co Ltd (1996) 188 CLR 418 at 442 per Toohey, Gaudron and Gummow JJ (‘Akai’).

[3] BCY v BCZ [2016] SGHC 249 at [60].

[4] Invariably the applicable arbitration statute in force at the seat.

[5] Citing Cie Tunisienne v Cie d’Armement [1971] AC 572 (‘Cie Tunisienne’).

[6] [2013] EWHC 4071 (Comm) (unreported Hamblen J, 19 December 2013).

[7] [2012] EWCA Civ 638.

[8] Sulamerica CIA Nacional De Seguros SA v Enesa Engenharia SA [2012] EWCA Civ 638 (‘Sulamerica), at [25].

[9] Sulamerica at [32].

[10]Akai, 442.

[11] The test has been endorsed by the High Court of Singapore in the case of BCY v BCZ [2016] SGHC 249.

[12] Gary Born, International Commercial Arbitration (Wolters Kluwer, 2nd ed, 2014), 520.

[13] Julian D. M. Lew, The Law applicable to the Form and Substance of the Arbitration Clause in ICCA Congress Series no. 9, p. 136.

[14] Nigel Blackaby and Martin Hunter, Redfern and Hunter on International Commercial Arbitration (Oxford, Oxford University Press, 6th ed, 2015), 159-160.

[15] [2011] HKCFI 458; [2011] 4 HKLRD 262; HCA 1526/2010 (unreported, Court of First Instance, Saunders J, 14 July 2011).

[16] Ibid, at [29].

[17] Vienna International Arbitral Center.

[18] London Court of International Arbitration.

[19] [2016] SGHC 249.

[20] Above n 12.

[21] Gary Born, International Commercial Arbitration (Kluwer Law International, 2nd Ed, 2014) at pp 24-27

[22] In particular, Gary Born; see for example, Gary Born, above n 7, 547-542 and Gary Born, ‘The Law Governing International Arbitration Agreements: An International Perspective’ 26 Sing. Acad. of LJ. 814 at 834-843.

[23] Armar Shipping Co Ltd v. Caisse Algerienne [1981] 1 WLR 207; The Iran Vojden [1984] 2 Lloyd’s Rep 380.

[24] BNA v BNB [2019] SGHC 142 at [52].

[25] Sulamerica at [25].

[26] BCY v BCZ [2016] SGHC 249 at [59].

[27] Sulamerica at [31].

[28] BCY v BCZ [2016] SGHC 249, at [79].

[29] Sulamerica at [32].

[30]Ian Glick QC and Niranjan Venkatesan, ‘Choosing the Law Governing the Arbitration Agreement’, Jurisdiction, Admissibility and Choice of Law in International Arbitration: Liber Amicorum Michael Pryles (Wolters Kluwer, 2018), 144.