Thursday, 7 April 2016

Is the EU-Turkey refugee and migration deal a treaty?

Maarten den Heijer*, Thomas Spijkerboer**

*Assistant professor of international law at the Universiteit van Amsterdam
**Professor of migration law at the Vrije Universiteit Amsterdam.

In the European Parliament, questions were asked about the legal nature of the EU-Turkey Statement of 18 March, pursuant to which Greece has started to return asylum seekers to Turkey this week. Apparently, the EU’s procedure for negotiating and concluding treaties with third countries, laid down in in Art. 218 TFEU, has not been followed. The European Parliament wants to know whether the Council nonetheless considers the Statement to be a treaty, and, if not, whether Turkey has been informed about the non-binding nature. Importantly, for treaties “covering fields to which the ordinary legislative procedure applies” (asylum and immigration is such a field), the Council may only conclude a treaty with a third country after obtaining consent of the European Parliament (Art. 218(6)(a)(v) TFEU).

It seems that legal experts of the Commission and the Council have identified the issue. Shortly after the EU-Turkey Statement, the Commission proposed to amend the Relocation Decisions relating to Italy and Greece, in order to transfer some of the relocation commitments concerning asylum seekers arriving in Italy and Greece to Syrians in Turkey. The proposal appears to contradict the view that the EU-Turkey Statement of 18 March did not intend to produce legal effects. However, in consideration 4 of the proposal’s preamble, the presented rationale for the amendment is the Statement of the EU Heads of State or Government of 7 March 2016, in which the Members of the European Council (and not Turkey) agreed to work towards the Turkish proposal of resettling, “for every Syrian readmitted by Turkey from Greek islands, another Syrian from Turkey to the Member States, within the framework of the existing commitments”. The Commission would seem to be navigating around the EU-Turkey Statement as the ground for amending the 22 September Council Decision, possibly fearing that to do otherwise may lend support to the argument that the Statement is, in fact, a treaty.

It could be argued that the statement is not a treaty in the meaning of the Vienna Convention on the Law of Treaties or an international agreement in the meaning of Article 216 TFEU, precisely because it is merely a “statement”. This is the view of Steve Peers on this blog: “Since the agreement will take the form of a ‘statement’, in my view it will not as such be legally binding. Therefore there will be no procedure to approve it at either EU or national level, besides its endorsement by the summit meeting. Nor can it be legally challenged as such. However, the individual elements of it – new new Greek, Turkish and EU laws (or their implementation), and the further implementation of the EU/Turkey readmission agreement – will have to be approved at the relevant level, or implemented in individual cases if they are already in force.” Karolína Babická appears to share this view: “The EU-Turkey statement as such is not legally binding. It is only a politically binding joint declaration. It is not challengeable as such but its implementation in practice will be possibly challenged in court.”

A further reason not to view the statement as a treaty is that it does not use terms as shall and should, which are normally used in international law to indicate obligations of result (shall) or obligations of effort (should). Instead, the more indistinct term ‘will’ is used. On the other hand, the Statement says that the EU and Turkey “have agreed on the following additional points”. Article 216 TFEU uses the term ‘agreement’ when referring to a treaty with third countries. If two parties agree to something, can the result be anything less than an “agreement”? Or is the meaning of the term agreement in Art. 216 TFEU different from its ordinary meaning?

If one would embrace the thought that the Statement of 18 March is not a treaty or agreement because it is designated as “Statement” and uses the term “will”, it would follow that the EU could neglect the constitutional safeguards of Art. 218 TFEU by changing the form or terminology of a particular text. It would be rather odd if the EP and CJEU could be sidetracked by such clever ruses. It would mean that the applicability of constitutional safeguards depends entirely on choices regarding the design instead of content made by Commission or Council.

That the form is not decisive is confirmed in the case law of the International Court of Justice. In Aegean Sea, the question was whether a joint communiqué, issued after a meeting between the Prime Ministers of Greece and Turkey, in which they agreed that a territorial dispute dividing the two countries should be resolved by the ICJ, constituted a treaty on the basis of which the ICJ had jurisdiction over the case. The Court held:

95. The Brussels Communiqué of 31 May 1975 does not bear any signature or initials, and the Court was informed by counsel for Greece that the Prime Ministers issued it directly to the press during a press conference held at the conclusion of their meeting on that date. The Turkish Government, in the observations which it transmitted to the Court on 25 August 1976, considered it "evident that a joint communiqué does not amount to an agreement under international law", adding that "If it were one, it would need to be ratified at least on the part of Turkey" (para. 15). The Greek Government, on the other hand, maintains that a joint communiqué may constitute such an agreement. To have this effect, it says, "It is necessary, and it is sufficient, for the communiqué to include-in addition to the customary forms, protestations of friendship, recital of major principles and declarations of intent-provisions of a treaty nature" (Memorial, para. 279). Counsel for Greece, moreover, referred to the issue of joint communiqués as "a modern ritual which has acquired full status in international practice".

96. On the question of form, the Court need only observe that it knows of no rule of international law which might preclude a joint communiqué from constituting an international agreement to submit a dispute to arbitration or judicial settlement (cf. Arts. 2, 3 and 11 of the Vienna Convention on the Law of Treaties). Accordingly, whether the Brussels Communiqué of 31 May 1975 does or does not constitute such an agreement essentially depends on the nature of the act or transaction to which the Communiqué gives expression; and it does not settle the question simply to refer to the form - a communiqué - in which that act or transaction is embodied. On the contrary, in determining what was indeed the nature of the act or transaction embodied in the Brussels Communiqué, the Court must have regard above all to its actual terms and to the particular circumstances in which it was drawn up.

The ICJ found that the terms of the communiqué, using terms as “decision” and “obligation” were indicative of the parties intending to bind themselves. However, it transpired from the context, namely previous and later negotiations and diplomatic exchanges between the parties, that they had not yet undertaken an unconditional commitment to submit the continental shelf dispute to the Court.
In Qatar/Bahrain, the question was whether minutes of a meeting between two Foreign Ministers constituted a treaty. The ICJ held:

24. The 1990 Minutes refer to the consultations between the two Foreign Ministers of Bahrain and Qatar, in the presence of the Foreign Minister of Saudi Arabia, and state what had been "agreed" between the Parties. In paragraph 1 the commitments previously entered into are reaffirmed (which includes, at the least, the agreement constituted by the exchanges of letters of December 1987). In paragraph 2, the Minutes provide for the good offices of the King of Saudi Arabia to continue until May 1991, and exclude the submission of the dispute to the Court prior thereto. The circumstances are addressed under which the dispute may subsequently be submitted to the Court. Qatar's acceptance of the Bahraini formula is placed on record. The Minutes provide that the Saudi good offices are to continue while the case is pending before the Court, and go on to Say that, if a compromise agreement is reached during that time, the case is to be withdrawn. 25. Thus the 1990 Minutes include a reaffirmation of obligations previously entered into; they entrust King Fahd with the task of attempting to find a solution to the dispute during a period of six months; and, lastly, they address the circumstances under which the Court could be seised after May 1991. Accordingly, and contrary to the contentions of Bahrain, the Minutes are not a simple record of a meeting, similar to those drawn up within the framework of the Tripartite Committee; they do not merely give an account of discussions and summarize points of agreement and disagreement. They enumerate the commitments to which the Parties have consented. They thus create rights and obligations in international law for the Parties. They constitute an international agreement.

On that basis, the ICJ concluded the dispute to be within its jurisdiction. It follows that the question of whether a text is a treaty does not depend on form but on whether the parties intended to bind themselves. Whether there is such intent, depends on the terms used and the context in which the text was drawn up.

There is no reason to assume that this reasoning does not apply to the EU (which is not a party to the Vienna Convention on the Law of Treaties). In interpreting agreements concluded between the EU and third countries, the CJEU consistently observes that even though the Vienna Convention does not bind either the Community or all its Member States, a series of provisions in that convention reflect the rules of customary international law which, as such, are binding upon the Community institutions and form part of the Community legal order (C-386/08, Brita, par 42). Presumably, the definition of a treaty in Art. 2(1)(a) VCLT belongs to customary international law. The 1986 Vienna Convention on the Law of Treaties between States and International Organizations or between International Organizations, which has not yet entered into force, uses the same definition and expands it to agreements concluded between international organizations or an international organization and a state.

Both the text and context of the EU-Turkey Statement support the view that it is a treaty. The parties “decided” to end the irregular migration from Turkey to the EU, and, to that purpose, they “agreed” on a number of action points. These include a commitment on the part of Turkey to accept returned migrants and a commitment on the part of the EU to accept for resettlement one Syrian for every one Syrian returned to Turkey. Further, the Statement reaffirms the joint action plan of November 2015 and mentions that it is already being implemented. Indeed, several implementation reports have been drawn up since November 2015, from which it is clear that the previous action plan has been activated (here and here). The EU-Turkey Statement now at issue is also being implemented. For example, the Greek parliament has passed a law allowing migrants arriving in the country to be returned to Turkey. On Monday 4 April 2016, Turkey accepted the first returned asylum seekers from Greece. All this indicates that the EU-Turkey Statement was meant to sort legal effects. This, in turn, indicates that both parties intended to bind themselves and that, therefore, it is a treaty.

One way to argue that the EU-Turkey statement is not an agreement in the sense of Article 216 TFEU would be to posit that it merely reconfirms already existing obligations from previous agreements (such as the EU-Turkey and Greece-Turkey Readmission Agreements). But textually as well as contextually, that argument is difficult to uphold. First, the substantive part of the agreement opens with the decision to return all irregular migrants to Turkey. Even though this sentence is followed by qualifications about compatibility with international and European law and even the explicit statement that this does not constitute collective expulsion, this is a highly novel (and legally very questionable) element, which can hardly be construed as a restatement of pre-existing obligations. The same is true for the EU commitments to resettle Syrians from Turkey and the additional funding for the Facility for Refugees in Turkey of 3 billion euro. Secondly, it is well known that the pre-existing readmission obligations (on the basis of the EU-Turkey and Greece-Turkey Readmission Agreements) were barely being applied. Therefore, the fact that Turkey agreed that, as of 20 March 2016, all irregular migrants were to be accepted is a substantively novel element. The idea that the EU-Turkey Statement merely repeats pre-existing legal obligations is not convincing.

Does the fact that the internal EU rules were possibly not followed mean that the Statement does not have legal effect? Probably not, as the Statement was agreed by the Members of the European Council, whom Turkey could have considered to have full powers to bind the EU. Article 46 VCLT provides that a party may not “invoke the fact that its consent to be bound by a treaty has been expressed in violation of a provision of its internal law regarding competence to conclude treaties as invalidating its consent unless that violation was manifest and concerned a rule of its internal law of fundamental importance”. Paragraph 2 of that provision provides that a violation is manifest if it would be objectively evident to any State conducting itself in the matter in accordance with normal practice and in good faith. In Qatar/Bahrain, the ICJ did not consider it relevant that Qatar had not followed the procedures required by its own Constitution for the conclusion of treaties: “Nor is there anything in the material before the Court which would justify deducing from any disregard by Qatar of its constitutional rules relating to the conclusion of treaties that it did not intend to conclude, and did not consider that it had concluded, an instrument of that kind; nor could any such intention, even if shown to exist, prevail over the actual terms of the instrument in question.” (par. 29).

We are therefore of the view that the EU-Turkey Statement is a treaty with legal effects, despite its name and despite internal EU rules not having been observed.

Why is the binding nature relevant?

That the Statement is a treaty implies not only that the EU and Turkey must uphold its terms; it also opens up a debate out is legal effects, including possible challenges against its legality in view of possible conflict with other rules and treaties, such as human rights. The fact that the Statement has already been concluded and is therefore no longer merely ‘envisaged’, means, however, that it is no longer possible to obtain an opinion of the CJEU “as to whether an agreement envisaged is compatible with the Treaties” (Art. 218(11) TFEU). It is still possible for one of the EU institutions or a Member State to bring an action for annulment of the act of the European Council to conclude the agreement with Turkey. Such an action was successfully brought in Commission v France (C-327/91), when the ECJ declared void the act whereby the Commission sought to conclude a competition agreement with the US, for reason of the Commission not being empowered to do so. However, this left the Agreement with the US itself intact, which is in conformity with the rule of Article 46 VCLT.

In view of the default position in international law that all treaties are equal, it further is difficult to argue that the Statement is void because of a possible conflict with human rights such as guaranteed in the ECHR or within the EU legal order, such as the right to asylum and the prohibitions of non-refoulement and collective expulsion. Only if the EU-Turkey Statement conflicts with jus cogens, is it to be considered void and may Member States not give effect to it (Art. 53 VCLT).

It is however possible for individuals (such as those being returned from Greece to Turkey) to challenge the implementation of the EU-Turkey agreement before national courts, arguing that it conflicts with fundamental rights. This in turn, may lead to a referral to the CJEU or a complaint before the ECtHR. Is the agreement in violation of human rights? As has been argued by UNHCR (here and here) and many others (eg here), the agreement may well raise issues under (at least) the prohibition of refoulement (is Turkey safe and is there a risk of expulsion from Turkey?), the right to liberty (is systematic detention in Greece allowed?) and the prohibition of collective expulsion (are the returnees able to challenge their return on individual basis, including before a court?). However, the Statement does not prescribe how, exactly, returns are to be effectuated and does not oblige Greece to systematically detain all asylum seekers who enter the country from Turkey. The Statement says that returns are totake place in full accordance with EU and international law, thus excluding any kind of collective expulsion” and that “[a]ll migrants will be protected in accordance with the relevant international standards and in respect of the principle of non-refoulement.” Further, migrants are to be “duly registered and any application for asylum will be processed individually by the Greek authorities in accordance with the Asylum Procedures Directive.” It would seem therefore that the Statement itself does not directly violate international norms – it leaves the Member States sufficient freedom to implement the obligations in harmony with human rights. It follows that the Member States (Greece) must implement the agreement in harmony with human rights: “Where a number of apparently contradictory instruments are simultaneously applicable, they must be construed in such a way as to coordinate their effects and avoid any opposition between them. Two diverging commitments must therefore be harmonised as far as possible so that they produce effects that are fully in accordance with existing law.” (ECtHR Nada v Switzerland, par 170).


This brings us to two concluding observations. First, the devil of implementing the EU-Turkey deal is in the detail. Although its effectiveness in terms of stopping irregular migration by creating a deterrent effect may depend on returning all persons arriving in Greece as quickly as possible, fundamental rights may well halt returns in individual cases or result in lengthy procedures. It is indeed the question whether the appropriate human rights framework is in place in Greece (as is observed by UNHCR).  Second, the EP is right in asking critical questions about the Council not following the rules for concluding a treaty (also see earlier questions about the EU-Turkey deal of 29 November 2015). Although one could take the view that time did not allow to await an Opinion of the CJEU, the agreement was not concluded with Turkey overnight and there would at least seem to have been opportunity to ask consent from European Parliament (Art. 218(6) says that, in an “urgent situation”, EP and Council may agree on a time-limit for consent). That the institutional role of the EP has been neglected confirms the worrying trend that intergovernmental decision-making is taking over in the Union, and that national interests increasingly often prevail over the common values of the Union. This is bad for European democracy.

Barnard & Peers: chapter 24, chapter 26
JHA4: chapter I:5

Photo credit:


  1. Are your sure that the Statement is only a one side declaration not representing also the Turkey position? This is not what arises from the President Tusk declaration after the summit (see here : ). That having been said I could agree that even if it is presented as en EU-Turkey agreement this is not an... EU Agreement (it reminds me the Magritte painting "this is not a pipe"). However I agree with you that it is an agreement from an international law perspective and this explain why Greece has immediately taken some legislative and operational initiatives by accepting also on its territory some Turkish representatives to check the implementation of the agreement on the Greek side. But this agreement has also an impact on the EU legal framework? I think so because (as you highlight) in your article the Head of State and Government of the EU (and behind them the Commission) have taken an international obligation in domain falling in the EU competence by so disregarding the principle of sincere cooperation which should frame the relations between the Member States and the EU institutions as well as the implementation of the democratic principle in the EU because they (and the Commission) have sidelined the only directly elected institution. All in all this is not a good example for an EU which pretend to be trasparent and committed to the better law making principles... Emilio De Capitani

  2. I consider that the "Statement" is both illegal and illegitimate, and is more dangerous than Trump's recent ban on immigration ban since is a covert one, and not open like in US. But US is democracy still because the Court can rule out the executive order by the President himself. While ECJ and ECtHR are silent. Turkey is autocracy (how "irregular migrants"can be resettled in the situation and country which derogated against the minimum human rights guarantees indeterminately and oppresses and represses its own citizens by violation right to life, prohibition on torture and imposing severe restrictions on the right to movement, speech and expression). Therefore, the "statement" is illegal per se since sends back the people to the unsafe country like Turkey.

    1. The ECJ and ECtHR have not yet spoken on the deal, however. They can certainly rule on the implementation of the statement, even if it is soft law that cannot as such be challenged.