Autonomy and Opinion 1/17 – a matter of coherence?

By Francisco de Abreu Duarte

On the 30th of April this year, the CJEU handed down its highly anticipated Opinion 1/17 on the compatibility of the CETA agreement with EU law. As Ankersmit details in his blogpost, the request for an opinion had been part of a widely known quarrel within Belgian internal politics, with Wallonia demanding the Belgium government to expressly consult the Court of Justice of the European Union (CJEU) on the legal merits of that agreement. Respecting that decision from its regional parliament, Belgium asked the CJEU, among other things, whether such an agreement was compatible with the principle of autonomy of the EU.

I will circumscribe this post to the analysis of the precise question of autonomy and leave out many of the other troubling questions such as the ones raised by Schepel’s in his previous post. The argument I put forth is as simple as it is controversial: autonomy, due to its abstract characteristics, is often subject to power injections leading to incoherent interpretations depending on the subject-matter at hand.

Let us see how autonomy has been interpreted before Opinion 1/17 and then analyze it in that light.

  1. Decades of autonomy for nothing?

To begin with, it is important to clarify that the principle of autonomy is a fundamental core principle of EU law that goes much beyond the questions of external action. I defend that autonomy was deeply embedded in the process of European integration, many years before external action took the importance it has today and should be read in that light. As Odermatt suggests, I find it prudent to begin by distinguishing two dimensions of the principle for the sake of conceptual clarity: the first is the notion of internal autonomy; the second the notion of external autonomy.

When a new international organization is born out of international law (even such a different and unique one as the EU), the first tension that immediately arises is the one between member states and the organization itself. This struggle is normally dominated by a discourse of independence of the latter, of the autonomy of its essential characteristics, one of the creation of a new legal persona. This initial clash is not irrelevant for our discussions on external action or on how CETA is regarded by the CJEU. In fact, in every decision where autonomy played a part, the CJEU resorted to Van Gend en Loos to anchor the principle to the essential characteristics of EU law. This means that the CJEU resorted to one of its most powerful cases to claim that the fight is always about the same: the control or monopoly of jurisdiction of the CJEU in order to ensure the protection of the essential characteristics of the EU. The second dimension of autonomy appears only at a later stage in the life of an international organization. Once the organization has settled the initial clash with member states, freed from the chains of its master in a Frankenstein scenario, it starts facing a different and complex second fight. In external autonomy, the organization must protect the same essential characteristics, but this time must do it from other international legal orders and the realm of international law. It is this second dimension that is normally discussed when one speaks of the CETA opinion, of Opinion 2/13 or of the decision in Achmea. But one should not forget that the Court sees both notions of autonomy as a continuum or two faces of the same coin: it was used in the early times to justify internal integration, fostering a growing equilibrium between member states and a Union of law (with the principles of direct effect, primacy, implied competences or human rights, to name just a few); now it is used to protect it from other potential unbalances caused by international legal orders. I think that it is important to keep this in mind when analyzing Opinion 1/17. I do not think the CJEU is psychologically obsessed with protecting its monopoly of jurisdiction against every bit of international law, but one should never forget that over 40 years of legal constructions aiming at granting legal independence to the EU must leave its scars. Even if done in a healthy way with the aid of the domestic courts of the member states.

Focusing now on external autonomy I would like to put forth three fundamental criteria extracted from the jurisprudence of the CJEU on this matter:

  • ‘The allocation of competences between member states and the Union’ The prohibition of external courts to adjudicate upon the division of competences/powers or the allocation of responsibilities between the EU and member states;
  • ‘The respect for the mechanism of preliminary ruling’ – The duty not to break the existential link between the CJEU and the domestic courts of the member states;
  • ‘The control of EU Law’ – The existence of some degree of control by the CJEU in case there is a disrespect of EU law.

a) Allocation of competences/division of powers

The first dimension of this external autonomy is connected to a simple problem. The division of competences between the Union and member states was a fundamental part of the initial struggle of the Union for independence. Indeed, when one speaks of competences, it is all about the allocation of sovereign rights. It is in such articles that we find the core social contract between the Union and its members, where one defines what is yours and what is mine, how far you can go and how far am I willing to give up on my original power. Hence it is normal that, if one wishes to maintain such a balance between the Union and its creators, whenever facing a problem of autonomy, the CJEU must above all maintain this division of competences intact. In the jurisprudence of the CJEU, this dimension is often present: in Opinion 1/91, for example, the Court claimed that if the EEA Court could be called upon to interpret the expression ‘Contracting Party’ (C-1/91, European Economic Area [1991], ¶ 34), then that would constitute a breach of autonomy as it ‘is likely adversely to affect the allocation of responsibilities defined in the Treaties and, hence, the autonomy of the Community legal order, respect for which must be assured by the Court of Justice pursuant to Article 164 of the EEC Treaty ‘ (C-1/91, European Economic Area, [1991] ¶ 35). The same line of reasoning was followed in many other cases, such as the Mox Plant case where the CJEU claimed ‘The act of submitting a dispute of this nature to a judicial forum such as the Arbitral Tribunal involves the risk that a judicial forum other than the Court will rule on the scope of obligations imposed on the Member states pursuant to Community law’ (C-459/03, Comission v. Ireland, [2006],¶ 177).

Another example is Opinion 2/13 on the co-respondent mechanism ‘However, the fact remains that, in carrying out that review, the ECtHR would be required to assess the rules of EU law governing the division of powers between the EU and its Member states as well as the criteria for the attribution of their acts or omissions, in order to adopt a final decision in that regard which would be binding both on the Member states and on the EU. […]  Such a review would be liable to interfere with the division of powers between the EU and its Member states’(C-2/13, Accession to the European Convention of Human Rights, [2014],¶ 224-225).

The jurisprudence demonstrates the connection between both autonomies: the CJEU has long fought for independence of the EU from member states and is not ready to give it up now to let other international courts or tribunals decide in its place.

b) The mechanism of preliminary ruling

A second fundamental criterion is the respect for the mechanism of preliminary ruling. This criterion basically protects the EU from breaking its fundamental link between domestic courts and the CJEU and maintains the balance achieved in the early times of integration. Autonomy is perceived as crucial, normally referred to as the ‘keystone’ of judicial integration, used as the main argument in several cases on this topic.

It was, for example, the main argument in Opinion 1/09, concerning the establishment of a European and Community Patents Court. The CJEU could not allow such a court to bypass domestic courts and obtain exclusive jurisdiction over that part of EU law. This would mean the destruction of the fundamental link, even if the Patents’ Court was imbedded with a mechanism of previous engagement with the CJEU itself. In the words of the CJEU ‘[…] the tasks attributed to the national courts and to the Court of Justice respectively are indispensable to the preservation of the very nature of the law established by the Treaties’ (C-1/09, European and Community Patents Court, [2011], ¶85).  As was also one of the arguments advanced in Opinion 2/13 regarding the relationship between the new mechanism of Protocol 16 to the ECHR and the CJEU, as the CJEU feared that such previous engagement with the CJEU could eventually render preliminary rulings unnecessary and hence destroy the “keystone of the judicial system”, meaning the judicial cooperation between the CJEU and member states’ courts (C-2/13, Accession to the European Convention of Human Rights, [2014], ¶196-200).

In 2018, in the Achmea case, the importance of this criterion grew even larger. According to the CJEU, because arbitral courts could not be seen as courts in the sense of article 267 TFEU, they stood outside the EU’s legal system and could not interpret EU law. The destruction of intra-EU BIT’s through autonomy was received with much surprise, and fears of ‘contamination’ of external investment agreements (such as CETA) or commercial arbitration rose immediately. The CJEU affirmed there with great confidence that ‘ the judicial system as thus conceived has as its keystone the preliminary ruling procedure provided for in Article 267 TFEU, which, by setting up a dialogue between one court and another, specifically between the Court of Justice and the courts and tribunals of the Member states, has the object of securing uniform interpretation of EU law, thereby serving to ensure its consistency, its full effect and its autonomy as well as, ultimately, the particular nature of the law established by the Treaties’ ( C- 284/16, Achmea, [2018], ¶ 37).

c) The control of EU Law

Finally, I advance a final criterion. I think this one is often ignored when one speaks of autonomy but I give it great importance. Autonomy means also control when things go wrong, meaning when some dispute settlement body misinterprets EU law. This is the final weapon that the CJEU has, even with domestic courts: if a certain Portuguese court fails to uphold EU law by not requesting a mandatory preliminary ruling or if an Irish court fails to comply with a certain interpretation given by the CJEU in an answer to a preliminary ruling, the CJEU has always the possibility of sanctions. Member states are subject to infringement proceedings and the Commission makes good use of them. At the same time, through years of jurisprudence such as Köbler scenarios, the CJEU built an idea of responsibility of judges themselves for failing to comply with EU law, hence granting individuals a claim for legality on those cases (see for example C-416/17, Commission v France, [2018]).

This is not possible when we speak of an outsider court or tribunal. It was part of the problem in Opinion 1/09: if the Patents’ Court failed to comply with the answer of the consultation to the CJEU, or if it simply refused to ask a preliminary question, what could the CJEU do? Not much evidently, as the relevant court is not attached to a member state which could be held responsible for the breach (C-1/09, European and Community Patents Court, [2011], ¶84). Likewise, in Opinion 2/13, the CJEU feared that interstate disputes under article 33 ECHR could lead to an absolute lack of control over good or bad interpretations of EU law (C-2/13, Accession to the European Convention of Human Rights, [2014], ¶205-210) and was not even satisfied with the Advocate General’s solutions of having a deference mechanism (similar to the one found in UNCLOS/ITLOS) or simply sanctioning one of the member states who had brought the action (through infringement proceedings).

  1. The surprise: Opinion 1/17 and the incoherent move

Now that we have briefly outlined how the CJEU developed autonomy up until Opinion 1/17, the decision must come across as a surprise. Taking into consideration the enormous emphasis put on the respect for the mechanism of preliminary ruling, one could almost immediately rule out any ISDS body which disregarded such a ‘keystone’ relationship between the CJEU and domestic courts. Apparently, however, autonomy can swing more than some of us expected, even if some structural problems of coherence could be raised.

Let use the same criteria set forth before and apply them to the ICS.

The first criterion is not a controversial one in my opinion and it rather shows that the Commission did learn from past confrontations with the CJEU. If one takes a look at the division of competences or allocation of powers in the CETA agreement, the project is quite protective of the CJEU’s monopoly of jurisdiction by attributing the judges in Luxembourg the final word on who is responsible in casu. Through article 8.21, the CETA agreement clearly defers to the CJEU the decision on who should become the respondent and accepts its decision as the sole interpreter of the division of competencies as set out in the treaties. Whoever the Court choses to face legal action before the ICS, either member states or the Union, the ICS must respect with no room for arguing. In Opinion 1/17, the CJEU was very happy with this – ‘The fact that there is no jurisdiction to interpret the rules of EU law other than the provisions of the CETA is also reflected in Article 8.21 of that agreement, which confers not on the CETA Tribunal, but on the Union, the power to determine, when a Canadian investor seeks to challenge measures adopted by a Member State and/or by the Union, whether the dispute is, in the light of the rules on the division of powers between the Union and its Member States, to be brought against that Member State or against the Union’.

But autonomy is (or was!) much more than giving the CJEU the chance to adjudicate on competencies. It was about ‘protecting’ the essential characteristics of EU law and, above all, to protect the keystone link of preliminary ruling between domestic courts and the Court of Justice. And this is where the reasoning gets strangely incoherent with the previous jurisprudence.

In fact, the ICS system, as set out in CETA, completely ignores the existence of a similar mechanism of preliminary ruling (as tribunals are prevented from actually using that mechanism as described in 267 TFEU). There is no preventive interaction with the CJEU like the one we had seen carefully designed in the project of accession to the ECHR; no prior involvement; no mimicking of preliminary ruling as in the Patent’s Court project. Nothing. While the accession to the ECHR would even demand an exhaustion of legal remedies, hence ensuring the preventive control of the CJEU, in the ICS there is a complete bypass. Why would the CJEU not demand a similar system for the ICS. On the contrary, the CJEU remains fully apart from the decisions of the ICS which are not even reviewable before domestic courts. A pure external system outside the CJEU’s control.

How could the CJEU accept all of this when in Opinion 2/13 it had put so much emphasis on the mechanism of preliminary ruling? The reasoning is a departure from a long tradition and it is based on an interesting paradox. While in most of the older decisions concerning autonomy the CJEU complained about the risks of having outside jurisdictions interpreting EU law, precisely because they stood outside the system (and hence the control) of the Court, it is now such a profound distance that dictates the validity of the ICS. How? Through an innovation brought by the Commission that apparently saves the day – EU law as “a matter of fact”.

There is so much on these six words that hardly any blogpost could conduct a full analysis of its meaning. However, I truly hope that more and more colleagues can debate this in order to reach a sensible conclusion on the distinction between law as a fact and law as a set of norms. I believe this is essential as it seems to be the core justification for the validity of the ICS in Opinion 1/17. According to the CJEU:

  • There is no need for any preliminary ruling mechanism as the ICS will never interpret EU law but only take it as a matter of fact;
  • Even if it does need to take into consideration a domestic measure (which can very well be executing EU law or be EU law) article 8.31.2 obliges the ICS to follow the CJEU’s prevailing interpretation (as a matter of fact) hence respecting the acquis.
  • Given the nature of arbitration and the need to have speedy judgements, it is absolutely understandable that there is no interim moment of consultation as that would delay the process (already very arguable as studies show quite the opposite with ICSID cases taking now 3 years and 8 months – see here and here). It also justifies the lack of domestic review as it was precisely to avoid such courts that the process was created.

But what if the ICS fails to take into account EU law correctly? What if it departs from the mainstream jurisprudence (which might not be as simple as mere ‘facts’)? What if the appeals court, when reviewing a decision by ICS’ first instance, actually interprets EU law to judge on whether an error of interpretation of domestic law exists? There is no risk there?

To all of this the CJEU answers in a surprising way:

While Article 8.28.2(b) of the CETA adds that the Appellate Tribunal may also identify ‘manifest errors in the appreciation of the facts, including the appreciation of relevant domestic law’, it is nonetheless clear from the preceding provisions that it was in no way the intention of the Parties to confer on the Appellate Tribunal jurisdiction to interpret domestic law (C-1/17, Accord ECG UE-Canada, [2019], ¶133).

The will of the parties, for the first time ever in the jurisprudence of autonomy, plays a fundamental role on the reasoning of the CJEU. One should not forget that these parties were, on the EU side, the same parties that agreed that the accession to the European Convention of Human Rights was absolutely necessary for the advancement of the EU and that the project would never jeopardize the monopoly of jurisdiction of the CJEU. And the same that so desperately tried to construct mechanisms of exchange between both Courts to precisely avoid such scenario. Apparently, all the safeguards put forth were never enough, even if manifestly stronger than those in the ICS. They lacked the precise wording: law as a matter of fact.

Finally, a last question: how could the CJEU control such a situation? In Opinion 1/09, the problem was precisely this. If the Patents’ Court interprets EU law wrongly how could the Union control the cohesion of its system? The CJEU did not allow the project to carry on as such, not only because the Patents Court would bypass domestic courts but also because, should things go wrong, the CJEU could not sanction anyone (as it would do with the member state with an infringement procedure for example). But in the ICS system there is absolutely no control should things go wrong. And, as we have seen, it is very likely to happen.

To this bold accusation the CJEU replies in Opinion 1/17 with a striking difference: even if such a ‘grey scenario’ would happen, one where the ICS does interpret EU law in a wrong way, there is absolutely no problem as member states or the EU are not in any way bound by such interpretations and it does not create case-law (C-1/17, Accord ECG UE-Canada, [2019], ¶131). Which is to say, CETA has no direct effect nor do the decisions of the ICS, even if they get it wrong, are to constitute case law. However, there is someone that, in the middle of all things, is bound in casu by such an interpretation – member states. They will have to pay huge sums to investors in case the ICS makes a mistake interpreting law as a fact, even if that does not change the nature of EU law or the CJEU’s case-law. And they will be held responsible for acts which could even have been enacted by imposition of EU law (such as measures implementing Directives or simply domestic legislation in accordance with the jurisprudence of the CJEU). It does not appear fair to place member states in a situation of impossible conflicting duties, whereby the CJEU could even (ad absurdum) sanction the member state for having changed its legislation to comply with rulings by the ICS triggered by Canadian investors.

  1. Autonomy is (not) all the same: from human rights to investment arbitration

I believe that, in the end, there is more to autonomy than meets the eye. Such a fundamental ‘principle’ with such an abstract nature should require a consistent and technical approach. But by resorting always to some conceptual mysticism, the CJEU is actually able to forge new solutions and reject or accept draft agreements depending on many factors. When De Witte writes that there is selfishness involved or Gragl speaks of jealousy, they might not be entirely fair. But I am sure that those feelings play their part on the big game of autonomy. In the end, the principle of autonomy is about how much international integration we want for our Union and how diminished would the CJEU be in face of conflicting jurisdictions and disorganized legal orders such as those we find in international law. The fragmentation and lack of cohesion feeds the fears of a CJEU that has fought hard to maintain its independence from member states and now from other courts and tribunals. Alongside these fears we find other factors that place pressure on the judges: economic lobbies; political alliances; the risk of being the one to blame for failing to move forward in a globalized world.

I find it difficult to believe that any Supreme or Constitutional Court can claim total immunity from external pressures. Nor is it fair to ask people to act like gods even if, like Weiler said long time ago (Weiler, The Transformation of Europe, p. 2428), we give them this ‘deep-seated legitimacy that derives from the mythical neutrality and religious-like authority with which we invest our supreme courts.’ Judges are people and people are subject to pressures. I believe this can help to explain the strong divergence between Opinion 2/13 and Opinion 1/17. From the pure legal point of view, the reason does not seem absolutely coherent nor respectful of previous decisions. Other factors might have played a part and those should be understood and studied just the way we address the legal aspects.

But we should ask for coherence nonetheless. I believe autonomy should be all the same, regardless of the subject-matter at hand, whether we speak of integrating human rights’ orders or arbitration tribunals. We can not hide in formal legal arguments when autonomy is not treated as a strict norm (certainly not a rule and hardly as a principle) but rather as an open-ended proposition, subject to power considerations. We must strive for a better dogmatics that will ensure legal certainty for future draft agreements. Otherwise we seriously risk critiques of legitimacy, of a CJEU that strikes down the biggest leap towards human rights interaction in the history of the EU but allows for the ICS to thrive.

One comment

  1. Szilárd Gáspár-Szilágyi

    My thoughts exactly. I think the CJEU dug itself so deep into this autonomy mess in Opinion 2-13 (repeated summarily in Achmea) that it had to find a way out. I am happy you touched upon the exact issue I pin-pointed in late 2015/early 2016, before Opinion 1/17 even arose, when the ICS was included in TTIP. In my analysis I said that the interpretation of EU law will occur de facto, even if the text says that EU law shall be used as a matter of fact. I am happy we share this view.
    My article can be found here:

    One more thing I would add after Achmea and 1/17 is what President Lenaerts mentioned 2 months ago in London at a conference. Mutual trust. It now all of a sudden became part of the autonomy toolkit. In Achmea, he said, mutual trust was the issue, but then the CJEU says in 1/17 that mutual trust is not an issue anymore. Plus in 1/17 the CJEU completely forgets about Article 344.

Leave a Reply