Muddy Waters: A Guide to the Legal Questions surrounding ‘Pushbacks’ at the External Borders at Sea and at Land
By Daniel Thym, Professor of European and International Law and Director of the Research Centre Immigration & Asylum Law, University of Konstanz, Germany.
15 years ago, the term ‘Aegean Sea’ was shorthand for crystalline beaches. Nowadays, we associate it with a different set of images. Islands such as Lesvos stand for the partial failure of the EU asylum policy at ensuring adequate reception conditions and fair and efficient asylum procedures. Frontex is under fire for having tolerated or supported pushbacks practices of the Greek coastguard. More recently, the Spanish exclave of Ceuta entered the limelight once again when border guards returned several thousand migrants to Morocco, seemingly without even rudimentary procedural guarantees. These ‘pushbacks’ at land and sea borders raise important challenges of institutional governance, operational reporting, and external monitoring. Moreover, it can be difficult to identify the correct legal standards, which shall be at the heart of this blog post.
The spectrum of opinions on the legality of the pushbacks practices appears to be irreconcilable. On the one hand, the term ‘pushbacks’ is often associated with automatic illegality. On the other hand, Article 6 of the Sea Borders Regulation (EU) No 656/2014 explicitly authorises national border guards to ‘order’ vessels ‘engaged in the smuggling of migrants by sea’ to change course. Earlier last year, the Spanish policy of summary returns was famously found to comply with human rights.
Comments below have the objective of introducing readers to the veritably labyrinth of legal issues involved. They are deliberately not meant to take a definite stance, let alone assess specific incidents. Rather, the objective is to provide readers with a mental map allowing them to chart a legal territory, which, unfortunately, is full of shallow and muddy waters where one can easily get lost. We shall see that the debate should pay more attention to the higher level of protection in secondary legislation instead of abstract human rights guarantees.
Human Rights: N.D. & N.T judgment on the Spanish-Moroccan Border
It is widely known that the Grand Chamber of the European Court of Human Rights (ECHR) delivered a landmark judgment last year, which is often cited in defence of the Greek practices. Judges in Strasbourg famously found the Spanish policy of ‘hot returns’ to be compatible with human rights, although border guards did not undertake even basic screening of the individual situation before they sent the applicants back to Morocco. The absence of any procedure did not prevent judges from concluding that the Spanish practice respected the prohibition of collective expulsion. For more detail, I refer to my earlier blog post on the judgment.
For our purposes, it is notoriously difficult to determine whether the argument can be extended to the Aegean Sea. A response may be to distinguish land borders from sea borders (here, p. 3), thus giving greater leeway to Croatian and Spanish border guards than to the Greek coastguard. Yet, there is nothing in the Court’s reasoning indicating such differentiation. Rather, judges relied on what they called the ‘own conduct’ of individuals (here, paras 200, 231). Unfortunately, this criterion does not resolve the question either, since the judgment contains two overlapping explanations of why the ‘own conduct’ excluded the need for basic screening, as I explain in a recent article in the International Journal of Refugee Law (here, pp. 576-579).
On the one hand, judges emphasised that the applicants had tried to cross the border illegally despite legal pathways being available at a very abstract level and without any guarantee that they would have worked for the applicants (here, paras 155, 201, 209-232). This argument could possibly be applied to the Aegean Sea, since individuals can theoretically apply for asylum at regular border crossing points, as the ECHR noted in a follow-up judgment on Poland (here, paras 207-209). Moreover, there is some resettlement from Turkey to the EU. To be sure, there would be no guarantee whatsoever that individuals may enter the EU legally—as in the N.D. & N.T. judgment where judges held abstract entry options to be sufficient.
On the other hand, the own conduct was described as the use of force and the storming of border fences (here, para 201). If that was the decisive aspect, many boats in the Aegean Sea would not be covered by the exclusion, and the prohibition of collective expulsion would apply. The same could even be argued for the recent mass arrival in Ceuta, which witnessed much less violence than previously after the Moroccan police had effectively opened the gate. In short, there is no straightforward answer whether the finding of legality extends to contemporary practices.
Distinguishing Non-Refoulement and the Right to Asylum
Protection against refoulement is the gold standard of international refugee law, which Article 78(1) TFEU and Article 19 of the Charter uphold as a constitutional necessity for EU asylum law. For our purposes, however, that commitment entails less than many think intuitively. Why? The prohibition of refoulement must be distinguished from the right to asylum. The right to asylum essentially entails that anyone making an application must be allowed to enter the territory. She receives an asylum procedure and can be deported only if the outcome is negative. Implications of the right to asylum for contemporary pushback practices are discussed below.
By contrast, the prohibition of refoulement obliges states not to send anyone back to a situation where he would be persecuted. The Italian or Maltese coastguard would almost certainly violate that obligation if it actively returned migrants to Libya (whether operational support or the provision of information to the Libyan coast guard entails the same legal consequence is a different matter). It is far from certain, however, whether the same conclusion applies to Turkey. More than three million refugees and migrants live there more or less peacefully. Afghans and Syrians are not generally persecuted there, they have access to status determination, and living conditions are difficult but not dreadful. As a result, returning them will not automatically amount to illegal refoulement.
That is relevant for the assessment of the pushbacks, since any invocation on the prohibition of refoulement presupposes a ‘real risk’ of inhumane or degrading treatment (here, para 79), and the right to an effective remedy presupposes an ‘arguable complaint’ that return would violate human rights (here, para 78). Both standards are not automatically met in the case of Turkey, also considering that both the Refugee Convention and the ECHR prohibit refoulement but do not grant a right to asylum—in the habitual reading at least (here, paras 200, 231; here, pp. 358-369). In other words, human rights do not generally require states to admit anyone saying the word ‘asylum’ to the territory, for as long as returning the person does not amount to refoulement. Of course, other rules can still result in the verdict of illegality in the absence of a danger of refoulement.
Procedural Conundrum: Inherent Limits of Non-Refoulement
A major doctrinal conundrum of refugee law are the procedural guarantees the protection against refoulement entails in relations with countries like Turkey or Morocco, which are—unlike Libya—not generally unsafe. The starting point is straightforward in line with previous comments: whether there is a ‘real risk’ of illegal refoulement must be assessed in light of information such as country reports, which states must scrutinise vigorously; additional information may be brought forward by individuals. Below the threshold of an arguable claim, however, the prohibition of refoulement seems not entail any procedural obligations (here, paras 85-98).
The problem is that this standard is a logical dead end from a practical perspective, since it is notoriously difficult, if not impossible, to know whether someone has an arguable claim or not without basic screening. Judges recognised the problem in N.D. & N.T.: generous interpretation of the prohibition of collective expulsion was meant to fill the procedural gap by requiring rudimentary screening of anyone rejected at the border irrespective of whether there was a danger of refoulement or not (here, paras 171-181). If rudimentary screening identifies the danger of a real risk, more advanced procedural guarantees under the prohibition of refoulement apply. If not, the person can be returned directly. Four our purposes, these thoughts bring us back to where we started, since judges consciously decided not to extend that conclusion to the Spanish ‘hot returns’. In some scenarios, the ECHR considers blanket returns to be legal.
One step further, the Grand Chamber maintained in Ilias & Ahmed, delivered a few months earlier, that returning ‘asylum seekers’ to a third state (as opposed to the deportation of nationals to home states) requires access to procedural guarantees against direct or indirect refoulement in that country (here, paras 131, 134). Judges insisted on a ‘thorough examination’ and stipulated that doing so would usually require a ‘legal procedure resulting in a legal decision’, also for migrants without protection needs (here, paras 135-138). At the same time, however, they introduced a backdoor by undermining the seemingly clear-cut insistence on mandatory procedures. Article 3 ECHR is being complied with in case of a thorough assessment of the situation in the third state even ‘(i)n the absence of such a (individual procedure)’ in the country pushing someone back (here, para 137). That seemingly abstract assessment is to be done on the basis of existing reports, without the need for individualised hearings or screenings (here, paras 137-141).
In the case of Turkey, it would not be particularly hard for the Greek authorities to claim that it has complied with these abstract requirements given that basic status determination seems to be available in Turkey. In the Spanish case, most recent arrivals in Ceuta would not be able to rely on that standard anyway, since many of them were Moroccan nationals that were returned to their home state (as opposed to Congolese nationals being returned to Morocco). It is frustrating that there are so many legal uncertainties surrounding the human rights case law of the Strasbourg Court. The procedural conundrum of how to disentangle the right to asylum from the prohibition of refoulement awaits being resolved.
Directives and Regulations as a Safety Net
There is often a mismatch between a widespread focus on abstract human rights in the public debate and the level of protection, which is often higher under EU legislation. Directives and regulations also have the advantage of usually being more precise, thus providing for more reliable answers. The situation in the Aegean Sea and in Ceuta is no exception. It seems to me that a major shortcoming of the legal debate so far has been the relative neglect of statutory prescriptions in the Asylum Procedures Directive, the Schengen Borders Code, and the Sea Borders Regulation.
The remainder of this blog post will concentrate on these legislative instruments, although they cannot provide definite answers either. The instruments do not define important terminology, and we have few judgments of the European Court of Justice. Unfortunately, Greek courts are highly reluctant to make preliminary references to judges in Luxembourg. While the Commission could bring infringement proceedings, doing so would be procedurally risky, since it bears the burden of proof for administrative practices on the ground (here, paras 48-49).
Asylum Procedures Directive: Individual Right to Asylum
Article 3 of the Asylum Procedures Directive 2013/32/EU goes beyond human rights requirements by laying down an individual right to apply for asylum ‘in the territorial waters’. That option may even be mandatory under Article 18 of the Charter, although we do not know for sure what that article requires, since it was consciously formulated ambiguously; it does not clarify to what extent it goes beyond the prohibition of refoulement and the level of protection in secondary legislation to which the wording seems to defer (here, para 63). However, these uncertainties are not directly relevant for our purposes, since legislation is precise enough: Member States must permit asylum applications ‘in the territorial waters’.
We know from recent Court judgments that it is not necessary to use the word ‘asylum’ and that applications can be lodged with any state authority, including with border guards (here, paras 93-94; here, paras 94-102). Judges also emphasised that Article 8(1) and Recital 28 of the Asylum Procedures Directive require states to actively provide information on at border crossing points the possibility to make an application and to ensure basic communication via unspecified interpretation arrangements. All this is to be done by border guards. Specifically trained personnel and the presence of a translator are foreseen at a later stage within the territory: during the personal interview in line with Article 15.
Thus, the degree of procedural safeguards depends of the stage of the procedure: the personal interview is not the same as applications at border crossing points. For our purposes, another set of rules applies, since pushbacks usually take place during ‘border surveillance’ in between crossing points. Article 6(1)(3) and Recital 26 demand less in this respect: neither proactive provision of information nor interpretation, just training for border guards so that they can inform individuals about what to do next once they have made an application. During border surveillance, secondary legislation puts more responsibilities on individuals.
What would this entail, if migrants and refugees started printing ‘ασύλου’ (asylum) on the boat or wave flags with that word? From a purely doctrinal perspective, one might argue that Articles 3 and 6 require direct physical contact with state authorities—in the same vein as an application ‘at the border’ does probably not oblige border guards to cut a hole in the fence in Ceuta if third country nationals shout ‘asilo’ (asylum) across the border ‘at’ which they are present. Be it as it may, the Greek coast guard would practically have to allow the vessel to reach the next Greek harbour, since enforcing an obligation to leave the territorial waters would create the jurisdictional link Article 3 could be interpreted to require.
The same applies to those actively applying for asylum once they have crossed the border fence in Ceuta and stand on Spanish territory: they must be given a procedure in accordance with the Asylum Procedures Directive. Secondary legislation does not contain any backdoor that would allow from an exit from procedural obligations provided that third country nationals have applied for asylum. The example of Ceuta illustrates that this is not always the case. Many Moroccan nationals appear uninterested in entering the asylum system, since the likelihood of a negative outcome is great and might result in entry bans. That being said, the answer to the question whether someone applied for asylum or not is a crucial factual condition to conclude that pushbacks were illegal.
Schengen Borders Code: Stricter Standards for ‘Border Surveillance‘
Some readers will be confused by these highly technical doctrinal comments on individual articles in secondary legislation. Yet, these provisions are what the European Parliament and the Council agreed upon during lengthy negotiations, and they offer the degree of detail abstract human rights guarantees often lack. The same holds for Articles 13-14 of the Schengen Borders Code Regulation (EU) 2016/399 on refusals of entry.
Their contents reiterate our earlier finding about differentiated standards. Article 14 requires a written ‘refusal of entry’ with basic reasons in accordance with Annex V at border crossing points only, while Article 13 refers to weaker guarantees under the Return Directive for scenarios of border surveillance. The Commission elaborated on the distinction in some detail in its recent informal guidance note (here, p. 5).
For our purposes, the reference in the Schengen Borders Code Regulation to the Return Directive 2008/115/EC results in few obligations, since Article 2(2)(a) authorises Member States ‘not to apply this Directive’ to irregular crossings (see also here, paras 71-72). Some guarantees, listed in Article 4(4), apply nonetheless, but they are not directly relevant for ongoing debates, since they do not oblige Greek or Spanish authorities to adopt return decisions subject to legal remedies under Articles 6 and 13. If domestic laws use this option, the Return Directive is largely irrelevant for our analysis. The right to an effective remedy in Article 47 of the Charter cannot be used to overturn that outcome, since it builds upon the potential violation of an individual right the Return Directive does not establish.
Sea Borders Regulation: Rudimentary Screening
All of the above leaves us, finally, with the only piece of legislation dealing with border surveillance in detail: the Sea Borders Regulation (EU) No 656/2014. It had always aimed at accommodating human rights with the political desire for ‘enhanced measures to combat illegal immigration’ in line with the United Nations Convention against Transnational Organised Crime. As stated previously, Article 6(2)(b) explicitly authorises border guards to order any vessel ‘engaged in the smuggling of migrants by sea’ ‘to alter its course’ outside of the territorial sea. The legislature assumed, in other words, that pushbacks are not always illegal. The same conclusion applies, as we have seen, to land border where written refusals of entry are not required in scenarios of border surveillance—provided that third country nationals do not apply for asylum with border guards.
However, this is not the end of the argument, since Article 4(3) Sea Borders Regulation lays down rudimentary procedural standards in case of interception or rescue. The text refers to identification, assessment of personal circumstances, and the option to bring forward arguments about a danger of refoulement. Reference to information about the destination, mentioned explicitly in Article 4(3), reminds us of the factual scenario the legislature had in mind: disembarkation in third states by border guards, mirroring the facts of the ECHR’s Hirsi judgment about former Italian practices towards Libya. The Directive was meant to specify the meaning of Hirsi.
Article 4(3) underlines once again that the degree of procedural guarantees may vary. The level of protection is comparatively low, mirroring the case law of the ECHR on the prohibition of collective expulsion. Whenever the prohibition applies (unlike in the N.D. & N.T. judgment), judges in Strasbourg similarly require states to foresee basic identification and the option to bring forward arguments, not, however, personal interviews with trained personnel other than border guards or legal remedies with suspensive effect (here, paras 237-242, 276-277; here, paras 57-61). In that respects, the case law mirrors the wording of the Asylum Procedures Directive, which does not oblige border guards to actively support third country nationals in making an asylum application during border surveillance.
‘Interception’: Essential Legal Threshold
Unfortunately, the Sea Borders Regulation refrains from defining the term ‘interception’, although that notion is highly relevant, since it triggers the applicability of the procedural safeguards in Article 4(3), which apply, as we have seen, irrespective of whether someone applied for asylum. Any interpretation of the term cannot build on a uniformly recognised meaning in public international law. Moreover, it is settled case law that provisions in secondary legislation often have an autonomous meaning of their own independent of how they are construed in international law.
There are at least two possibilities what ‘interception’ involves. On the one hand, one could interpret the term broadly covering any prevention of onward travel. If that was the case, ordering a ship to leave the territorial waters would amount to interception. On the other hand, one might require a certain level of control of border guards over the vessel. Doing so coincides with the general scheme of the Sea Borders Regulation: Article 6 is primarily concerned with operational measures, including ‘boarding and searching’ the vessel, and the procedural safeguards in Article 4(3) effectively presuppose the presence of migrants on border guard ships (like in the case of ‘rescue’ to which this provision refers as well).
It is, once again, somewhat frustrating that these questions ultimately have to remain open in the absence of a Court judgment from Luxembourg. As stated previously, the legislature had the facts of the Hirsi judgment in mind when drafting the Sea Borders Regulation. The term ‘interception’ may have been assumed, therefore, to mirror the notion of ‘jurisdiction’ under human rights law, which judges in Strasbourg have not applied, so far at least, to a scenario not involving physical presence on a coast guard ship (here, paras 70-82). ‘Pushbacks’ below the threshold of jurisdiction, let alone ‘pullbacks’ by the coast guard of third states, were not the focus of attention of the legislative debate at the time and have not been scrutinised in light of the ECHR so far. It presents a formidable challenge, therefore, to apply the Sea Borders Regulation to these scenarios.
Conclusion: Pitfalls of Legal Uncertainties and the Danger of Formalism
This contribution set out to provide readers with a mental map how to navigate the muddy legal waters closer inspection of human rights law and secondary legislation brings to the fore. It deliberately refrained from suggesting the ‘correct’ interpretation, let alone assess specific scenarios. Of course, it would be ideal if the legislature agreed to amend secondary legislation to overcome these uncertainties, although there is little hope that this could realistically be achieved in practice. Defining the conditions under which pushbacks are (il)legal would be at least as controversial as the protracted debate about the reform of the Dublin III Regulation.
At a practical level, it is essential, therefore, to improve reporting requirements of Frontex missions and national border guards about the factual situation on the ground. Such facts-oriented regular reporting on the basis of precise questions could be done without a consensus how the various provisions are to be interpreted legally. To do so would prepare the ground for other institutions to engage in ex post assessments from a political and legal perspective, thus increasing the chances that we get judgments from Luxembourg authoritatively interpreting EU law.
In the absence of such rulings, Member States will continue using legal uncertainties as a cover for the defence of restrictive practices. They reiterate the basic lesson that a functioning asylum system requires more than compliance with the letter of the law, which, as we have seen, often leaves room for states to justify their behaviour. A functioning asylum system requires a political consensus about the way forward, which, unfortunately, does not seem to exist within Europe at this juncture. In the absence of a basic consensus, we are left with ultimately unsatisfactory debates about the meaning of complex rules that do not allow for clear-cut verdicts.