Can EU Member States Set Conflict-of-Laws Rules for Substantive Matters in Arbitration?
The relation between EU private international law and arbitration is notoriously complex. While the arbitration exemption of the Brussels I bis Regulation has rendered both much case law as well as legal debate, less attention has been paid to whether and how the EU conflict of law rules in the Rome I Regulation apply to the substance of disputes resolved by arbitration.
A revealing example comes from Sweden.
In 2018, the Swedish Arbitration Act (officially: lag [1999:116] om skiljeförfarande) was revised. One of the amendments was the introduction of a choice of law rule for arbitral tribunals in Section 27 a.
Mirroring Article 27 of the 2017 SCC Arbitration Rules (which has remained substantially unchanged in the revised 2023 SCC Arbitration Rules), the provision affirms that arbitrators primarily should apply the law chosen by the parties. In the absence of such a choice, the provision lets the arbitrators determine the applicable law. It is further clarified that the arbitrators “may base the award on ex aequo et bono considerations only if the parties have authorized them to do so.
When adopting this conflict of law provision, the preparatory works simply stated, with reference to Article 1(1)(e) of the Rome I Regulation, that EU private international law is not applicable to arbitration (SOU 2015:37 p. 92). However, this interpretation of Article 1(1)(e) of the Rome I Regulation deserves scrutiny.
Unlike the general exemption for “arbitration” in article 1(2)(d) of the Brussels I bis Regulation, the exclusion in Article 1(1)(e) of the Rome I Regulation encompasses “arbitration agreements and agreements on the choice of court”.
Given the doctrine of separability, it seems clear that the intention of the exemption in Rome I Regulation is simply to exclude dispute resolution clauses from the rest of the contract. It is not explicitly stated that a substantive dispute settled in an arbitration is exempted.
From a legislative point of view, it is a difference to include choice of law rules in institutional arbitration rules such as the SCC rules. Technically, institutional arbitration rules are contractual and subject to party autonomy. Introducing choice of law rules in a national arbitration act is something completely different. First, it is prohibited under EU law as EU regulations shall apply (see Article 288 of the Treaty on the Functioning of the European Union). Second, it may put arbitrators in an awkward position where they must either rely on the Swedish Arbitration Act or the EU Regulation.
In practice, this may not make much difference in most commercial cases. Both Section 27a of the Swedish Arbitration Act and the Rome I Regulation prioritize party autonomy. But the divergence becomes critical in cases involving mandatory rules, such as employment disputes. Under article 8 of the Rome I Regulation, the law chosen by the parties must be related to the law that would have been applicable if no choice of law was done. Must a Swedish arbitral tribunal dealing with an international employment dispute take the Rome I Regulation into consideration or can it apply Section 27 a of the Swedish Arbitration Act?
In my opinion, there is no doubt that the exemption of the Rome I Regulation to arbitration agreements cannot be extended to cover also substantive matters in an arbitration. Hence, the Rome I Regulation shall be applied. That must also be the case when an arbitral tribunal rules in a regular matter.
Very interesting question. The inapplicability of the Rome I Regulation to substantive issues in arbitration is not based, I believe, on the exclusions from its scope of application, but on the determination of the very addressees of the rules. The Regulation, which is anchored in the scope of the European judicial area and judicial cooperation in civil matters, obliges the judicial or even non-judicial authorities of the Member States, but not arbitral tribunals, in my opinion. Arbitration is not covered by the guarantee of access to justice [Article 67(4) TFEU)], constitutes an alternative to jurisdiction and implies a renunciation of jurisdiction within the framework of the will of the parties. Moreover, the possibility of choice of applicable law is not the same in both cases, since in arbitration it is generally possible to choose a non-domestic body of law (the Unidroit Principles, for example), which is not possible in the Regulation, rather than as an incorporation by reference. In the absence of choice, arbitral tribunals may also opt for a non-domestic law, so that the application of, for example, Article 4 of the Rules would not make sense in such a case. In sum, I believe that the discarding of the Rome I Regulation is not only justified, but also advisable.
As regards your first point. An arbitral award can still be challenged in court. In such a case, EU law clearly applies (see e.g. Mostaza Claro, C-168/05, EU:C:2006:675. To say that only the court hearing the challenge, but not the arbitral tribunal hearing the case is obliged to apply EU law seems a little artificial. Why should not the same principle apply for rights under the Rome I Regulation? As an example, an arbitral tribunal seated in Sweden may have to deal with an employment dispute for an employee who has carried out his work in France. Under his employment contract, the law of Florida shall apply. If Section 27 a of the Swedish Arbitration Act is applied, the law of Florida will be applied. If article 8 of the Rome I Regulation is applied, the law of Florida will also be applied, but French law protective provisions that cannot be derogated from will break-through the law chosen by the parties. What shall the arbitral tribunal do? Apply Section 27 a Swedish Arbitration Act or Article 8 of the Rome I Regulation?
As regards your second point, the difference between applying a non-domestic body of law as an incorporation by reference or in arbitration can also be seen in the light of the possibilities to challenge the award. In my opinion, the limits for the arbitral tribunal’s margin of applying such a non-domestic body of law are set by the possibilities to challenge the award under the law of the seat.
I agree with Sixto that the issue should be resolved by exploring the general scope of EU instruments (ie whether they only apply to authorities of the Member States) rather than the scope of particular instruments.
I am less convinced by the argument that choice of law rules for arbitration are different from those of the Rome I Regulation. If the Rome I Regulation is binding on arbitral tribunals, it should displace any contrary rule, including rules allowing for choice of non state law.
All this being said, decisions of arbitral tribunals on the applicable law are essentially unreviewable, except if they fail to apply certain overriding mandatory provisions, so I think that the issue is largely academic.
Of course, Gilles, if the Rome I Regulation is binding any choice of non-domestic law would be unavailable and this is not an advisable consequence.
Arbitral awards are not unreviewable and the risk that the conflict-of-laws issue leads to a wrongful application of mandatory provisions is very real. See e.g. the scenario in my reply to Sixto’s comment above. In many member states, arbitration in employment disputes is common. Sweden is such an example. Last year, it was the third most common contract type to be arbitrated under the SCC rules (see statistics https://sccarbitrationinstitute.se/en/statistics-2024/).
In international commercial arbitration, arbitral awards are unreviewable on the merits. This means that, in principle, an ‘error’ of the determination of the applicable law is unreviewable. The only exceptions are 1) a clear agreement of the parties which would be ignored by arbitrators, and 2) violation of the substantive public policy of the forum. So, if an arbitral tribunal applies non state law instead of a national law, I am not sure on which ground that could be reviewed.
I am not sure about employment arbitration. I come from a jurisdiction where this is simply forbidden. Maybe certain jurisdictions allow it with a full review of the merits by courts. In any case, I think it is unwise to draw general conclusions on the applicability of the Rome I regulation in arbitration based on the protections offered by Article 8.
Even if your point on limited review is correct in most cases, I wonder if one can so generally state that there are only two exceptions for review. Isn’t this fully dependent on the law of the seat of the tribunal?
It is probably hard to think of a case where the application of non-state law instead of national law as such is the problem. In the case of employment arbitration, the problem arises when protective provisions that cannot be derogated from have been disregarded.
I do not understand your point on not drawing general conclusions for the application of the Rome I based on article 8. As I mention in the post, applying the Swedish conflict of laws rule or the Rome I Regulation will not make a difference in most commercial disputes as both rely on party autonomy. It is when the issue is really put to the test (like in employment arbitration) that its true nature is revealed.
An arbitral award may indeed be annulled for reasons of public policy linked to the non-application of European mandatory rules, including those relating to the protection of workers or consumers, irrespective of which law has been applied and by what means that law has been established. As noted by the International Chamber of Commerce, acting as amicus curiae in the ‘Mitsubishi’ case, or by the CJEU since ‘Eco-Swiss’, arbitrators are obliged to apply or consider such mandatory rules. Arbitration in the field of employment or in the field of consumer contracts also raises a question of capabilty or settlement by arbitration in the sense of article V.2 of the New York Convention, currently submitted to the lex fori. However, it is not conceivable, nor do I know of any case, of control of the law applied by the arbitral tribunal, and even less so for the mere fact of not applying the Rome I Regulation, unless the arbitral tribunal, as Gilles emphasises, has not respected the agreement or mission entrusted by the parties to apply a certain law, or because the arbitral tribunal has made its award in equity when arbitration in law was appropriate by agreement of the parties or by requirement of the lex arbitri.
Nor am I aware of any cases annulling or setting aside an arbitral award for not having applied the Rome I Regulation. Still, I think that it could happen in an employment dispute like the Florida/France example I outlined in a reply to one of your comments yesterday.
I agree with you that an arbitral tribunal must consider mandatory rules. There are however many and conflicting such rules out there. This stresses of course the importance of taking private international law seriously in arbitration. I think there is no room for “ex aequo et bono” decisions in such issues without consideration of the hard law consequences (annulment, setting aside, non-recognition et.c.).
As you rightly point out, arbitration in international employment disputes raises also the issue of conflicting rules on arbitrability. That issue I have written an article on in the 2023 Hague Academy anthology “Applicable Law Issues in International Arbitration”. Also in that contribution I draw general conclusions from employment cases. If you are interested in reading my article, you are more than welcome to send me an e-mail and I’ll send you a pdf.
Many thanks, Erik. Your contribution is available in our Library.
If Gazprom (C-536/13) was decided as it was because the Brussels I Regulation was not aimed at arbitral tribunals, which could do as they otherwise saw fit, why would Rome I be materially different ? The technical difference in the langauge of the two arbi-exclusions cannot justify it, and I would have thought that there was little more to be said.
Why cannot the language of the two exemption clauses justify differences between the application of the two regulations?
The view that the Rome I Regulation binds arbitral tribunals was also held by the late Professor Mankowski, who published on the issue several times (eg RIW 2011, 30 and RIW 2018, 1). The arguments raised in support of the position of applicability are substantial and cannot be brushed aside. I completely agree that it is not as simple as referring to Article 1(2)(e).
However, the Giuliano-Lagarde Report does contain certain hints that the “otherness” of arbitration for the purpose of the Rome Convention was recognized more broadly than is now evident from the text (“Having regard to the fact that the solutions which can and have been considered generally for arbitration are very complex and show great disparity (…)”, Article 1 para 5 of the Report). On the whole, I believe that the view of non-applicability is more convincing.
The citation from the Guiliano/Lagarde report that you refer to relates to the exclusion of arbitration agreements.
Here is the Giuliano/Lagarde argument in its entirety:
“Having regard to the fact that the solutions which can and have been considered generally for arbitration are very complex and show great disparity, a delegate proposed that this matter should be studied separately and any results embodied in a Protocol. The Group adopted this proposal and consequently excluded arbitration agreements from the scope of the uniform rules, subject to returning to an examination of these problems and of agreements on the choice of court once the Convention has been finally drawn up.
The exclusion of arbitration agreements does not relate solely to the procedural aspects, but also to the formation, validity and effects of such agreements. Where the arbitration clause forms an integral part of a contract, the exclusion relates only to the clause itself and not to the contract as a whole. This exclusion does not prevent such clauses being taken into consideration for the purposes of Article 3 (1).”
It relates to the exclusion of arbitration agreements, but is – in my view – an argument on the peculiarity of arbitration itself, not of arbitration agreements (“solutions … for arbitration”, not “solutions … for arbitration agreements”).
So is this the proposition ? That whenever an arbitration has its seat within the borders of the European Union, the tribunal may not exercise – and the parties may not ask or direct the tribunal to exercise – the power to choose the law (or other rules, principles, or quasi-laws) to be applied by them to the substance of the dispute which currently exists in most developed arbitral systems. Instead, the tribunal is required to apply the law of a nation state under the direction of Articles 3 to 12 (or thereabouts) of the Rome I Regulation ? Including mandatory rules of the forum (which presumably means of the state of the seat) ? If that is so, one would think that it would be even more discouraging or damaging to arbitration on the territory of the Member States than was the decision in The Prestige. From a vantage point outside the European Union, it is surprising to see this possibility being ventilated within it. But one assumes that those who are paid to notice these things and who assess risks will notice this one also.
No, that is not the proposition. As I see it, it is perfectly fine for an arbitral tribunal to exercise the power to choose the applicable law if it does that on a contractual mandate from the parties. This is also the point that makes it troublesome to include conflict-of-laws rules in a national arbitration act compared to in institutional arbitration rules. Whereas institutional arbitration rules are applied on a contractual basis, the national arbitration act is supposed to be applied as the fall-back rule in e.g. ad hoc arbitration. In the example of an employment dispute there might be a situation of conflicting conflict-of-laws rules in the national arbitration act and in the Rome I Regulation. Therefore, I mean that an EU member state cannot legislate in conflict with the Rome I Regulation.
In most commercial disputes, the application of the Rome I Regulation will not make a difference as it also relies on party autonomy.
Thank you. But how would a tribunal be allowed to apply the laws or principles of a religious commnity, or abstract principles of good faith, or the lex mercatoria, ot the general principles of international law, or anything like that ? The parties may want, choose, such a basis for the resolution of their disputes, but if the Regulation tells the tribunal whether/what the parties may choose, surely it will exclude all these from the menu?
First recital 11 states that “[t]he parties’ freedom to choose the applicable law should be one of the cornerstones of the system of conflict-of-law rules in matters of contractual obligations”. Recital 13 further clarifies that “[the Rome I Regulation] does not preclude parties from incorporating by reference into their contract a non-State body of law or an international convention”.
With support of the recitals above, I cannot see that the widely accepted principle of party autonomy in article 3 would limit arbitrators’ application of such frameworks in general if they have a contractual mandate from the parties.
Another thing is though that the application of such frameworks cannot be done in a vacuum, but that it needs to be related to the hard law reality. Without knowing for sure, I bet that even an English arbitral award would be hard to enforce in the UK if it is incompatible with e.g. UK sanctions against Russia?
At least here, the general view is, as it is put in Dicey, ‘It is clear that the reference in Art 3(1) of the Rome Regulation to the parties’ choice of “the law” to govern a contract is a reference to the law of a country. It does not sanction the choice or application of a non-national system of law, such as the lex merctoria, or general principles of law, or the law of a religious community…’. There was a half-hearted attempt, in the work leading up to the Regulation, to allow for a choice of UNIDROIT principles, worded in such a way as to make it clear, without saying so, that choice of a religious ‘law’ would still not be effective, but the proposal was withdrawn.
I am sure you are right about the impact of UK sanctions against Russia and those who facilitate its invasive tendencies. But that will be because the English court is bound to apply UK sanctions law, and cannot, therefore, make an order, say to give leave to enforce the award as a judgment, which conflicts with it. I have not really reflected on it, and so say this hesitantly, but I do not think that it follows from that that arbitrators in London have to apply these sanction rules.
In my opinion, the exclusion of the applicability of the Rome I Regulation to international commercial arbitration does not derive from Article 1.1.e, but rather from legal provisions that create an autonomous rule for determining the law applicable to international commercial arbitration. I refer, on the one hand, to Article 28 of the UNCITRAL Model Law on International Commercial Arbitration (transposed, for example, into Article 34.2 of the Spanish Arbitration Law). On the other hand, I refer to Article 7.1 of the 1961 Geneva Convention on International Commercial Arbitration. These provisions apply to international commercial arbitration and exclude the applicability of the Rome I Regulation because they are earlier and more specific.
The argument that special choice of law rules for arbitration derogate from general rules of PIL is generally a good one, except in a European context, where EU law prevails over national laws.
Indeed, Gilles. EU law does not prevail over national laws in general, but within its scope of application. As Prof. Sánchez Lorenzo previously explained, the scope of application of Rome I Regulation is civil judicial cooperation. Therefore, in my view, international commercial arbitral tribunals are not bound by Rome I Regulation.