May states select among refugees?

ABSTRACT This article argues that there is no general ‘yes’ or ‘no’ answer to the question of whether states may permissibly use group-based criteria to permanently resettle refugees from certain contexts. Indeed, I argue that this question only makes sense with respect to a range of narrowly circumscribed scenarios. And, even though it may appear that states have wide discretion with respect to refugees in these scenarios, the way that these scenarios have arisen and continue to be maintained casts doubt upon even this limited conclusion Nevertheless, consulting both the history and practice of international refugee law can help us understand why some particular forms of group-based prioritizations are foreclosed. Such attention to practice based details also sensitizes us to the solution structure (and historic interest constellation) behind refugee law, as well as those institutional tweaks and patches that could actually stand a chance of making a marked difference.


Introduction: clarifying the question
The way that the questions of our symposium are framed causes me some concern. At a general level this is for three reasons. First there is a principle of statutory interpretation -Expressio unius est exclusio alterius (the express mention of one thing excludes all others)which effectively moots some of the questions of our symposium. For, in asking after 'acceptable and unacceptable criteria for accepting/rejecting refugees', it is surely relevant that the Refugee Convention already explicitly lists the grounds upon which refugee status can be denied to those who would otherwise qualify. 1 There are no other grounds of exclusion. Second, it is controversial to frame accepting and rejecting as two sides of the same coin. In some circumstances a decision to accept or prioritize one applicant may have downstream effects on another applicant's chances without being appropriately described as an act of 'rejection'. Lastly, I worry the timing of our symposium could suggest that there is legitimate debate about certain criteria when there is notwhen in fact certain matters are 'settled'. The conference behind our symposium arose at a time when a few world leaders made public remarks about prioritizing refugees based on religion. Yet, far from identifying an 'open' 2 question in liberal democracies or amongst refugee scholars, these remarks merely reveal how politicians make insincere appeals to dead-letter concepts to garner votes.
All this makes my engagement quite fraught. Nevertheless, I think fruitful engagement is possible. At a general level I hope to dispel a common way of seeing things, highlight just how few situations our questions apply to, and to reveal very clear but consistently overlooked resources from practice that can inform our theory. More specifically, I think the main question to engage is this: Are there standards of group identity that states may use to prioritize the resettlement of some refugees? From the outset I should make it clear that it is a strength of my account that it does not give a general 'yes' or 'no' answer to this question. Instead, I simply explain why some prioritizations are foreclosed and then sketch the broad contours of a more positive answer. However, even my more pared-down question requires greater clarity along three dimensions: (i) the standard we rely on in asking what states may do (introduction), (ii) the institutional and legal grey areas that can make our question an open one (sections 1 and 2), (iii) how states use those grey areas to behave badly towards likely refugees, and why this matters for how we perceive the current state of affairs and conceptualize our obligations (section 3).
The first two clarifications seem to indicate that, in certain non-ideal contexts, some groupbased prioritizations may be used since states are not obliged to resettle anyone from those contexts. The third clarification challenges this: the impression that these contexts are free from obligations is mostly illusory since these contexts are themselves largely the creation of states. These facts provide strong countervailing reasons against using group-based prioritizations. Thus, given our present non-ideal circumstances, even prioritizations that could be defended in theory should be approached with great care. I start by explaining my general approach, why human rights are the right standard to rely on, and how I understand human rights. Philosophers of migration often write in a genre of applied ethics: moral commitments or principles alone supposedly lead us to policy limitations or prescriptions. Moral theory is seen to guide political and legal practice (Rossi and Sleat 2014). Here, I adopt a different tack since I think the standard lens is inadequate for refugee issues in particular. Refugees are 'a distinctly modern political artifact of the international order of states' (Owen 2016, 269). If we construe our duties to them in terms of pre-political 2 See Cristina Lafont (2020) 38-39 on the two senses for a political question to be 'settled'. In one sense a question is settled (at least for the time being) insofar as it is resolved by proper procedures. For instance, domestic acceptance of the refugee convention makes it impermissible to resettle refugees based on religion (see below). Here 'we speak of a settled decision, whether or not … participants have a settled view on the matter'. In another sense a question can be settled, 'insofar as a shared view on its proper answer has been reached'. Here 'we speak of a settled view on the proper answer (or range of answers) that would lead citizens to settle a decision accordingly.' In this second sense the nature of the issue at hand informs the range of permissible answers and the mechanism of settlement (pure majority rule, lottery, bargaining, etc. moral rights or apolitical humanitarianism then we overlook how they are interstitial and exceptional: they exist in spaces between and outside of states and are therefore an exception to the normal structure of our international order. Moreover, an applied ethics approach is apt to miss how the complex problem of refugees involves dynamics that are fruitfully analysed with the tools and internal 'logics' of political science or economics. My approach thus rests on several assumptions: overemphasizing humanitarianism has its own problems, 3 a proper analysis of refugee issues must stray into domains beyond moral philosophy, and that a better understanding of practice can inform our theory. The first relevant practice-based fact is that refugee rights are legal rights distinct from yet connected to international human rights. 4 Many theorists confuse refugee and human rights. They are closely related but not the same. 'Contrary to conventional wisdom, the Geneva [Refugee] Convention is not a human rights treaty in the orthodox sense, for both historical and legal reasons' (Chetail 2014, 22). Refugee rights can be quite different. For example, Article 19 of the Convention helps refugees in the 'liberal professions' continue their career in a new state. 5 Much like humanitarian law refugee law is its own domain.
Nevertheless, human rights are the correct standard for determining what states may do in this domain. Despite any differences human and refugee rights both inhabit the territory between morality and law. While human and refugee rights ought not be confused, it is still true that 'human rights law has radically informed and transformed' refugee law (Chetail 2014, 22). Our question must take the relationship between refugee and human rights seriously. Yet, since I also think a strictly moral and apolitical approach is inadequate for refugee issues I understand our question in terms of institutional political morality or the morality of institutions (Scanlon 2016). Approaching the question in this way recommends a political or 'practical' conception of human rights.
The basic claim of this approach is that 'human rights are the constitutive norms of a global practice whose aim is to protect individuals against threats to their most important interests arising from the acts and omissions of their governments' (Beitz 2011, 197). Another account reasons that since international law is structured around sovereign nation-states which exercise authority over literally everyone this 'requires the international legal order to attend to pathologies of its own making' (Macklem 2015, 1;Owen 2016, 275). In short human (and refugee) rights mitigate the problems created by the modern state system and how it allocates rights, duties, and responsibility. This approach to human rights focuses on the emergent practice of human rights instead of on inherent moral rights or what relevant actors can agree to call 'human rights'. We proceed by looking at the descriptive facts of the practice and its underlying normative aims. By working at these two levels we can produce a critical reconstruction of the practice that picks out instances when 'the practice's norms are ill-suited to advance its aims' (Beitz 2011, 105) and which guides us in assessing the lower limits of what governments may do. Now that my approach is on the table I shall turn to the parts of both human and refugee rights that bear on our specific question: where refugees are located 6 and non-discrimination. I argue it only makes sense to ask our question with respect to refugees in legal limbos like camps or non-signatories (as most such prioritizations are prohibited for territorially present refugees) and that the refugee right to non-discrimination is more helpful than the human right.

Location
The most obvious reason why location matters is definitional. Refugees have a well-founded fear of persecution in their country of origin and are outside it. The 'persecution' and 'alienage' conditions are constitutive of refugee status. As for alienage, the Convention was crafted to give refugees 'legal status and protection … until they … acquire new or renewed national protection' (Hathaway and Foster 2014, 17). It is not a tool for migration generally or all forced migration. The substantive rights of the Convention are 'directly related to the predicament of being outside one's own country' and are 'the sort of rights required to facilitate integration' (Hathaway and Foster 2014, 21-22, fn. 34). Refugee rights are not 'entitlements of generic utility' for anyone, but rights tailored to a specific population. Accordingly, there must be a 'match between the beneficiary class and the remedy' (Hathaway and Foster 2014, 21-22). Ignoring this fact squanders resources, unfairly displaces costs, and treats persons arbitrarily (Cherem 2016, 195).
Two further location-based distinctions are relevant. First, the Convention has five 'levels of attachment' between a refugee claimant and an asylum state. 7 Second, there is a distinction between refugees in camps or other limbos and those who spontaneously show up at borders. The distinctions are related. While the second used to be produced by chance and showed up quite differently than it does today, nowadays states can 'externalize' their borders to manipulate the levels of attachment, lessen the number of refugees arriving spontaneously, and increase those in camps or other limbos. Let's examine the distinctions and their relationship.
The Refugee Convention sets out a 'five-part assimilative path' defining a 'continuum of legal attachment to the asylum state' (Hathaway 2005, 156-157). A refugee who has left their country and presses their claim against another state may be: (i) present in a state's jurisdictioni.e. 'under its control or authority' (Hathaway 2005, 156), (ii) physically present within that state's territory, (iii) lawfully present in the state (while their claim is being processed), (iv) lawfully staying within the state (after recognition), (v) a durable resident in the state (after all residency requirements have been met).
The levels of attachment 8 were born out of a dilemma: refugee status is both declaratory and potentially verified by a formal status assessment. Anyone who has left their country is 6 I develop an extended argument about location in Why Location Matters in Refugee Debates (unpublished ms.) 7 I can't talk about the groups that attachments are indexed to, nor how this creates a 'dual contingency' for many rights. 8 Rights were framed in terms of attachments without explicitly dubbing it a 5-step process. Here I cannot discuss how some scholars seem to imply (without argument) that there are 3-4 attachments. Above I mention two reasons adopting the levels of attachment made sense but don't mean to imply these were the only or primary reasons. The fact that the 'philosophy of migration' literature consistently overlooks this fundamental aspect of refugee law should give pause.
a refugee if they fit the definition. But, since not everyone who claims to be a refugee is, a state may verify each applicant's claim (Hathaway and Foster,25 fn. 51). 9 The dual nature of refugee status raised the question of when refugee rights apply. The answer that drafters came up with used the levels of attachment: 'pending status verification' states should 'provisionally … honor basic refugee rights' that correspond to the first three levels of attachment (Hathaway and Foster 26, Hathaway 2005, 159). More rights kick-in after verification. Full rights accrue after residency requirements. This variegated approach to rights is unlike standard human rights treatieswhere all rights either apply or don't. For instance, states who sign the ICCPR cannot 'pick and choose' which rights inhabitants enjoy. 10 Yet, this variegated approach makes sense for refugee law. Refugee issues arise not only between an individual and at least one state but also between states. Refugee rights are often calibrated to what can be expected of a state based on the level of attachment. Thus, refugees who are merely under a state's jurisdiction, like those interdicted on the high seas, sensibly have rights to rationing (Art. 20), against discrimination (Art. 3), and against 'refoulement' (Art. 33)being unilaterally returned. But, only refugees who are lawfully staying enjoy the right to wage-earning employment (Art. 17).
The levels of attachment created their own problems. For historical reasons the drafters neglected scenarios where someone had fled their country but not yet established an attachment. 11 Such a person is marooned between the declaratory and verified senses of refugee status. The ambiguity created by this scenario makes our question an open one.
States have learned that refugee rights are variegated and that the first level of attachment has the least clarity. Many states now use 'extraterritorial migration controls': tactics to stop or divert migrants en route. These include setting up refugee camps or offshore processing centres, coast guard interdictions to either return or divert people to a third state (both often undertaken without status assessment or adequate procedural rights), 'readmission agreements' that quickly return migrants to a (possibly non-signatory) state they transited through, posting immigration officers in foreign ports to screen documents and deny onward travel, or levying fines on transport companies like airlines for each irregularly documented arrival. In theory each tactic could be administered to respect refugee rightslikely with multilateral treaties, supranational regulation, and an independent body to arbitrate disputes. In practice these tactics are unilaterally designed and enforced by individual states in ways that violate rights (Silverman 2014). These practices created a new manifestation of the distinction between asylum seekers who show up at borders and refugees resettled from a temporary safe haven. Let's look at this distinction.
Spontaneously arriving asylum seekers are nearly always encountered at the second level of attachment and lodge their claim against a specific state. Their effective enjoyment of protection against refoulement and new membership hinges on that state's assessment. But, those resettled from a temporary safe havene.g. a non-signatory that permits residencemay have no level of attachment whatsoever against the resettling state. This gives the resettling state huge discretion. If refugees are resettled from this second scenario they bypass the first levels of attachment, are lawfully staying on arrival, and slowly establish durable residence and membership (maybe citizenship). This distinction has long been a part of refugee law. In drafting the Convention the French representative noted, 'the situation of island countries could not be compared with that of the continental countries, which were in direct physical contact with the source from which the refugees came'. 12 The Canadian representative noted his country 'was separated by an ocean from the refugee zones. Thanks to that situation, all refugees immigrating to Canada were ipso facto legally admitted and enjoyed the recognized rights granted to foreigners admitted for residence' (Hathaway 2005, 157 fn. 12). In the 1950s 'overseas asylum states' like Canada 'mainly … receive[d] refugees preselected for resettlement'. This difference is another reason it made sense to adopt the levels of attachment so that 'any unauthorized refugee … would benefit from protections … [but] would not … immediately acquire all the rights of "regularly admitted" refugees … preauthorized to enter and reside' (Hathaway 2005, 157). Otherwise, states that happened to have refugee-producing neighbours would be unfairly burdened.
While the distinction is old it was a product of chance factors that might impact any state like regional instability or 'accessibility'. Yet, over time, powerful states turned themselves into 'islands' by both hardening and externalizing their borders. The extraterritorial migration controls noted above try to stop any and all irregular migrants (including refugees) before they come in a state's jurisdiction or territory. This allows states to sidestep legal obligations and select the refugees they want to resettle from safe havens.
We can now see how these distinctions bear on our question. It is simply false to assert that spontaneous arrivals could be preferentially treated if they fit into the host society while others could be unilaterally turned away, rejected, or deprioritized if they were not expected to fit-in. People either show up at your borders or they don't. 13 Those who are refugees must all be treated accordingly. But refugees living in temporary safe havens could be selected for resettlement, perhaps with group-based criteria. This is because, absent any level of attachment, there is no particularized obligation between these refugees and any resettlement state. There is a hortatory 'collective obligation' of all signatories towards refugees in general but this can be discharged in many ways (e.g. donations to UNHCR).
Moreover, this distinction tracks a real difference insofar as, 'there is a morally relevant difference between imposing risk and failing to protect against it' (Miller 2016, 84). If a state unilaterally turns away spontaneous arrivals (without adequate alternative arrangements) then it would be actively imposing risk and violating international obligations. If a state fails to resettle refugees from a temporary safe haven, such as a camp or a non-signatory state, then it fails to protect against risks associated with those precarious scenarios. But, if the state is unconnected to their original flight or the fact that they're now in a temporary safe haven then the duty is like general humanitarian duties in that there is no particularized reason to discharge one's duty towards specific refugees, let alone via resettlement.
Another way we could understand this difference is in terms of being 'obliged as a matter of justice to right a wrong' versus having 'a reason of justice' to right a wrong: "As a general matter, we are not required as a matter of justice to correct the injustice that others perpetrate, although we may have reason to do so. If Bert steals Anne's money, justice does not require Charles to right this wrong, although if Charles happens to be so placed that he can direct the money back to Anne, this would very likely be the right thing for him to do. Charles commits no injustice if he fails to secure the return of Anne's money, because responsibility for Anne's loss rests entirely with Bert … we might say that Charles has a reason of justice to direct money back to Anne. But this is different from saying that he is obliged as a matter of justice to right the wrong Bert has caused. How weighty this reason will be depends on questions such as whether Charles is the only person other than Bert who is able to see that Anne gets her money back, how much it will cost him to do so, etc." (Miller 2013, 218 and fn. 11) Camps or other limbos may seem analogous. Actors administering such situations might be like Charles helping Anne so she will not suffer the full effects of Bert's wronging. Criticizing Charles seems misplaced as he is taking care of someone that he is not, strictly speaking, obliged to help. Criticism would also seem misplaced if Charles were to help Ayn because of some group-based criteria (e.g. language) but refrain from helping similarly situated Anne.
I will argue this is the wrong way to see things (and, given how widespread I think this view is I take this to be one of my paper's main contributions). Such a view makes it seem as if these are contexts of sheer discretion and no obligation. This account construes the permissibility of state action as exhausted by the internal protection of rights and the (relatively easily achieved) external respect for both other states' sovereignty and outsiders' human rights. But, as extraterritorial migration controls show, states can exploit the interstitial spaces of the international order in morally suspect ways that violate the spirit if not the letter of the law. These are the spaces that matter for our question.

Non-discrimination
It may seem like the human right against discrimination could resolve our question: if this right is universal and typically violated by the use of group-based prioritizations then we have an answer. Unfortunately, human rights practice is not that neat and tidy. But, the refugee right to non-discrimination can help. This underscores the importance of differentiating between refugee and human rights.
Discrimination is hard to define. It is easiest to spot when it is direct and intentional but may occur unintentionally or via indirect effects. Theorists locate the wrong-making feature(s) of discrimination in the unequal distributions it engenders, the motivations it presupposes, and the messages it conveys. Some think it is bad since it harms those who are discriminated against and others because it wrongs or disrespects them.
Here I steer clear of articulating precisely what makes discrimination bad. It suffices to say that it is bad, and that it works by imposing 'disadvantage on persons based on their membership in a salient social group … relative to some appropriate comparison social group' (Altman 2016). I use this broad, anodyne definition since I only aim to show that, in our context, identifying the salient comparison group is difficult. The international human right against discrimination cannot give clear guidance as to whether the differential treatment in our question amounts to discrimination.
Apart from specialized non-discrimination treaties (CEDAW, CERD) there is a human right against discrimination in Article 26 of the ICCPR and Article 14 of the ECHR. The former prohibits discrimination due to 'race, color, sex, language, religion, political or other opinion, national or social origin, property, birth or other status'. The latter adds 'association with a national minority, property, birth or other status'. Art. 26 is an autonomous right prohibiting discrimination in laws of signatories whereas Art. 14 is an accessory right prohibiting discrimination in other ECHR rights.
Let's assume the scope of this right is universal. Do states then violate nondiscrimination when they use any of the above-mentioned group-based metrics (e.g. language) to prioritize refugees from safe havens for resettlement? That conclusion seems too quick. Our question concerns who states may prefer from such contexts not who they disprefer. In domestic contexts discriminatory effects are almost always the flipside of a preference for one group. But in an international scenario like ours it is possible for preferences to come apart from disadvantage or disrespect. So, if Canada gives French-speaking refugees in temporary safe havens some priority for initial resettlement to Quebec then I don't think it is violating the human right to non-discrimination. Certainly, prioritizations may be illiberal, unwise, or un-virtuous. But in this Quebec example it seems clear that prioritizations need not harm or wrong either other refuges or non-French-speaking Canadians. Indeed, prioritizations may serve salutary goals like revitalizing minority communities in a larger nation.
In the domestic context the comparison group against which relative disadvantage is assessed is easily identified. For instance, between 1999 and 2002 several Roma who gave birth in Slovakia's Krompachy public hospital were non-consensually sterilized. The clear comparison classall women giving birth at that hospitallets us determine this was discriminatory by showing, say, that this happened only to Roma, that they had to use segregated rooms, that this is a vestigial practice of discrimination from the erstwhile communist regime, etc (Kinsella 2012). 14 It would still be discrimination if the state preferentially treated Slovaks yet deprioritized, neglected, or refused to treat all others. Slovakia is equally responsible for all Slovakians and must protect the basic rights of all in Slovakia.
Yet in our context it is hard to tell precisely who the relevant comparison class is. If a state were to mainly resettle refugees of a certain group it is hard to identify the disadvantaged comparison class. All refugees? Those in temporary safe havens? All those (from the same country? Fleeing the same basis of persecution?) left behind in a specific safe haven? Not only do states lack particularized resettlement responsibility towards those in safe havens, those remaining need not be harmed, disadvantaged, or disrespected. For example, if there are fewer people in a camp then those remaining may have a greater share of resources or a better chance of having their application accepted by another state. It is also unclear if the refugees left behind would be disrespected. For instance, the UNHCR triages applicants into 'emergency', 'urgent', and 'normal' resettlement priority based on their vulnerability to further harm in safe haven and this sometimes involves the use of group-based metrics (e.g. gender, age, sexuality, ethnicity). 15 We cannot read-off disrespect from the mere fact that a refugee is not prioritized and thereby selected.
In international contexts it is often hard to determine the comparison class. Partly for this reason it is hard to prove violations of the human right to non-discrimination. International courts often permit differential treatment if states cite 'objective and reasonable' bases and 'proportionally' undertaken measures in pursuit of a 'legitimate aim'. While suspect criteria exist like race, sex, language, religion, political opinion, or nationality even differential treatment based on these grounds can be permitted if the state's aim is deemed 'very weighty' (Morawa 2001, 176-196). Courts use the 'margin of appreciation' doctrine in such casesa weaker standard of review favouring statesbased on the thought that 'states are usually in the best position to make legislative policy choices and decisions' and because of 'differences between national traditions and perceptions as well as cultural variation' (Gerards 2018, 30). International bodies have developed the 'objective and reasonable' standard so as to make this human right a weak tool for contexts like ours (Hathaway 2005, 129-47).
In short, the international human right of non-discrimination doesn't provide a lot of clarity on issues involving interstate dynamics. The decision will favour states. Human rights practice on non-discrimination at the international level seems to be at odds with (or at least does not resolve tension between) its normative aims. This is precisely the kind of case that causes trouble for the practical approach to human rights.
Fortunately, refugee law interacts with human rights law in a helpful way. Article 3 of the Convention states, 'the Contracting States shall apply the provisions of this Convention to refugees without discrimination as to race, religion, or country of origin'. While early drafts referred to refugees in the territory of a state this language was debated and deleted (Marx and Staff 2011, 647). For, if this right were indexed to the second (territorial) level of attachment states could discriminate at or just beyond their borders to avoid obligations towards entire groups of refugees. Since this would fatally undermine the regime this right was given the broadest interpretation possible. This means non-discrimination in refugee law applies to any refugees in the jurisdictione.g. under the (effective) control or authorityof a state. Moreover, the relevant drafting history of Art. 3 is explicit about the comparison class with respect to which discrimination is to be assessed: discrimination cannot take place between or among refugees on any of these three bases (Marx and Staff 2011, 650;Hathaway 2005, 245).
It is true that Art. 3 is an accessory right, that coming under jurisdiction only triggers limited rights, and that it is contingent whether a refugee counts as in the 'jurisdiction' of a resettling state (Hathaway and Gammeltoft-Hansen 2014). Nevertheless, the drafting history reveals Art. 3 was clearly meant to apply extraterritorially. This gives us quite a bit of guidance since presence in jurisdiction and/or territory is clearly established in many of the scenarios where states want to use prioritization: camps in a signatory, island 'reception centers', airport transit zones, high-seas interdictions, etc. Even in less clear scenarios jurisdiction may still be relevant. 16 Finally, the limited rights triggered by jurisdiction are expansive enough that prioritizations based on the prohibited grounds would impact other rights in a way prohibited by refugee law. This is not simply a point about what happens to be legally prohibited. Given my approach to human and refuge rights the fact that this prohibition was not only explicitly inserted but that it has stood the test of time grounds a (defeasible) presumption that it upholds the overall normative aims of the refugee regime. So, I think we can say that the authority immanent to the practice of refugee rights buttresses an intuitive conclusion: group-based resettlement prioritizations are not the norm; they can only be used in the comparatively rare case of resettlement from safe havens and, even then, prioritizations based on race, religion, or country of origin are simply off the table. This has value in that we cannot simply 'back-engineer' arguments to validate our pretheoretical intuitions. Moreover, these prohibited grounds are precisely the ones that politicians have bandied about in recent years. That these prioritizations are legal 'deadends' clues us in to what is really going on when leaders appeal to them. Finally, while there are other unsavoury grounds of prioritization I don't need to exhaustively treat them. If institutional realities and the distinctiveness of a refugee are properly understood this counsels against most group-based prioritizations.

States behaving badly
Above I speculated that many people see refugee resettlement from camps or other limbos as wholly discretionary because they see these spaces as analogous to Charles' act of charity. If Bert steal's Anne's money then it is a good thing if Charles gives her some of his extra. But, he is not obliged to do so since he didn't commit the initial wrong of theft. This is especially true if other agents -Dave, Edwin, etc.could also help better or more easily. I also said this was the wrong way to see things. Why?
Notice how the status and relationship of these actors matters. At least in safe havens like refugee camps Bert is a refugee-producing state, Anne is an individual refugee, and Charles is an international organization (e.g. UNHCR, FRONTEX, IOM) or non-state actor (e.g. INGO, private security company) acting on behalf of a state or the international community (or both). So, it is more like this: Bert wrongs Anne. Given prior practice 17 and explicit agreements among Bert, Dave, Edwin and nearly everyone else, the way Bert wronged Anne can make Dave and Edwin obliged to grant or otherwise secure her membership if she shows up on their doorstep (territory) or otherwise comes under their effective control (jurisdiction). Dave and Edwin don't want to honour their obligations, so they use Charles to stop Anne before she can clearly trigger such obligations. They give Charles money so that they can avoid responsibilities they had signed up to shoulder. If they are unhappy with Charles' decisions or advocacy they withhold funding. They tell Charles to ensure Anne ekes out an existence. They periodically review applications from Anne to be allowed to enjoy what they would have been obliged to do for her if her access to the clear triggering condition (territorial presence) had not been blocked by mechanisms of their agent, Charles.
Scaling the analogy up in this way is stylized. Much will depend on the organizations and activities lumped together under 'Charles' (we might be less sceptical of UNHCR acting on behalf of most states and refugees than of IOM or a private security firm). But here I'm only interested in how 'Charles' is a non-state actor partly acting on behalf of states. I simply wish to describe a general dynamic: powerful states increasingly use international organizations, non-state actors, side-agreements, or quiet diplomacy with non-signatories (e.g. Jordan, Lebanon) to indefinitely contain refugees. States use such practices to pre-empt territorial presence that would clearly trigger obligations. This gives states discretion over those for whom they must secure membership. Moreover, relative public ignorance about such mechanisms affords states reputational benefits for the refugees they do resettle, and states may also quietly use admissions as leverage in international negotiations.
The problem is not simply that those who de facto govern such limbosusually a mix of UNHCR, INGOs, private security firms, and host state security forceshappen to be unable to offer state membership. The fact that they lack this capacity but are charged with administering such spaces is by design. Over the past few decades, 'the institutionalization of detention centers abroad through funding and operating externalized processing centers and refugee camps is becoming increasingly permanent'. These tactics hold 'asylum seekers and irregular immigrants outside the borders of their destination states at the behest of those states' (Silverman 2014, 610). A policy of 'containment' has arisen that 'warehouses' refugees.
If this broad characterization is correct then states are using intermediaries to act badly in one of two ways: by illicitly shirking their obligations or by knowingly benefitting by sustaining a wrong.
The first characterization is straightforward. Most states have signed the refugee convention. This agreement was designed to address a certain situationthe presence of individuals who are outside their country and made stateless by persecution. The agreement begins to trigger obligations when refugees come in a signatory's jurisdiction. On this understanding states have made an agreement among themselves and refugees are third-party beneficiaries. When states violate the spirit and letter of that agreement by using extraterritorial migration controls to shunt refugees into camps or non-signatories they wrong the other states and the third party beneficiaries suffer. 18 Unlike the 'Charles' of Miller's original scenario the powerful states that use these tactics cannot plausibly maintain that they don't have a hand in Anne's predicament. While responsibility for Anne's initial loss rests 'entirely with Bert' these states have undoubtedly helped block her from accessing the remedy that she should have been able to help herself to. Camps (and other such limbos) are never explicitly mentioned in the Refugee Convention. They often prevent refugees from triggering the law of host states (Wilde 1998, 110) and are essentially extralegal artefacts accreted by state policy rather than an intentional part of international refugee law (this is so even though an administrative framework has arisen for many UNHCR camps). Similarly, many of the extraterritorial migration control mechanisms surveyed above rest on shakyor simply nobases of international law. Thus, Dave and Edwin cannot appeal to Anne's lack of territorial presence to justify why they should have complete discretion in choosing to resettle her (or not) with whatever group-based metric they please. Here the thought is something like the prevention doctrine in contract law: you cannot get out of your agreement by appealing to the failure to meet a certain condition in the contract when you yourself have acted to prevent the fulfilment of that condition. 19 18 Thanks to Nicholas Cornell for clarifying my thinking here, the prevention doctrine point, and Infra fn. 20. 19 If refugees are third-party beneficiaries the analogy only works for cases when a party prevents its own liability.
The second characterization is more contentious: when states block access to their territory through the creation and maintenance of camps and other such limbos this effectively amounts to knowingly benefitting by sustaining a wrong (Barry and Wiens 2016). States that shunt refugees into spaces outside the scope of refugee convention collectively contribute to the persistence of a wrong that the refugee's country of origin unjustifiably inflicted upon her. Since this characterization seems to attribute special salience to the particular wrong of being made into a refugee my thought here relies on something akin to Macklem's conception of human rights: international human (and refugee) rights arose to mitigate the wrongs created by the specific organizational principle of modern state sovereignty, and being forced to flee one's country and rendered statelessness by persecution is an especially predictable, salient or even 'exemplary' pathology of our international order (Owen 2016, 269). As this wrong is structural it stands in need of a unified remedy (already offered by refugee law) as opposed to many situations of statelessness, the predicament of so-called 'climate refugees', and so forth.
But is this second characterization plausible? Aren't states just avoiding an obligation created by an agreement rather than sustaining a particularly egregious political wrong? For, if we characterize cooperative deterrence as sustaining a wrong then this seems to imply that many states (those powerful enough to turn themselves into 'islands') are judging themselves to be better off in a world where persecution continues, refugees exist, and they (mostly) have discretion in resettlement compared to a world where all such persecution and refugee flows stopped. 20 A cynic or hardcore political realist might think that this is true in some casesfor example, Italy and Libya (under Gaddafi) could perhaps each gain more power by using refugees as bargaining chips (Libya against Italy, Italy against the EU) than they could if the flows between them did not exist. But it seems like a 'conspiracy theory' of refugee law to think that it is generally true that powerful states judge themselves better off in a world with refugees and the discretion to relieve refugee flows (e.g. if they effect allies) or not (e.g. if they effect rivals) than in a world without refugees. Even if refugees are sometimes purely used as leverage (Stedman and Tanner 2003;Greenhill 2011) that doesn't seem like a generally accurate description.
I don't need to decide between these characterizations. Each can likely illuminate different refugee crises or aspects of the same crisis. All I have sought to show is that refugees in camps or other limbos would have no resettlement claim against particular states if they simply happened to end up there, but that this description is not plausible. Nowadays refugees are shunted into camps or non-signatories by a thick web of extraterritorial migration control tactics. If these tactics amount to either illicitly avoiding an obligation or benefitting by sustaining a wrong then they make the states that most use them at least partly responsible to the refugees in those spaces.
Since location outside the refugee regime is now often an engineered rather than chance occurrence states cannot point to location as a sufficient justification for why they may choose refugees for resettlement in any which way (and using whatever group based metrics) they like. Moreover, once we understand institutional realities and the distinctiveness of a refugee it makes it clear that many of the contexts where our question even makes sense should themselves probably not even existsurely not in their present form. Refugee law did not envision mechanisms of management and aid that offer no guarantee of or timeline for a durable solution. Convention framers assumed, like many political philosophers, that people are 'either subject to the full force of the state's authority, or they are not' (Walzer 1981, 31). Yet camps and other limbos are increasingly used to create places where the full force of a state's authority is strategically withdrawn and partly transferred to international or private groups that administer goods and services without a definite end.
At the start I noted that a proper analysis of a complex problem like that of refugees must go beyond moral philosophy to include the tools of political science or economics. This is where that observation becomes relevant. An approach of 'applied ethics' or 'political moralism' (Rossi and Sleat 2014) is apt to miss crucially important facets of the global refugee problem. As Wendy Brown has observed in questioning the supposedly 'apolitical' nature of human rights, 'there is no such thing as mere reduction of suffering or protection from abuse … Just as abuse itself is never generic but always has particular social and subjective content, so the matter of how it is relieved is consequential. Yes, the abuse must be stopped but by whom, with what techniques, with what unintended effects, and above all, unfolding what possible futures?' (Brown 2011, 143 my emphasis). 21 In other work I have shown why the category of a refugee is distinctive and inassimilable to generic suffering or harm (Cherem 2016). Here I shall close by emphasizing that in theorizing about refugees we cannot sensibly restrict ourselves to those who need help and those who can offer it. 22 Any assessment of the permissibility of group-based prioritizations needs to go beyond a dyadic analysis of the refugee and the state of resettlement. This is because the fact that many states are behaving badly indirectly impacts whether a particular state may use (non-prohibited) group-based metrics to prioritize certain refugees for resettlement. For if too many actors shrug their share of the collective burden then this threatens the efficacy of the regime, and such prioritizations can only take place in the context of the regime continuing to function. While such a characterization is oversimplified, I think we can nevertheless fruitfully think about the refugee regime as a prior response to a collective action problema response that is now subject to all the dynamics of free-riding, defection, and so forth.
The refugee regime is not simply rooted in human rights. It was also 'conceived out of enlightened self interest'a 'mutual self-interest' among signatories. The international community learned from institutional responses to refugees over the decades until finally 'the prewar understanding of assimilation as a source of internal stability' was supplemented by 'concerns to promote burden sharing' and 'maintaining the 21 As Rossi and Sleat note, the phrase 'political moralism' was initially suggested by Bernard Williams. Brown discusses 'political moralism' (and its perils) in more detail in Brown (2001), chapter 2. 22 The philosophy of migration literature would be well-served to move from thinking of refugee scenarios as a two party-relation (e.g. refugees and resettlement states) to thinking of them as a relation between three or more parties (e.g. adding the country of origin, 'temporary' host states, the UNHCR, the refugee regime itself, etc.) One error in treating a multi-party relation as a two-party relation is that (for example) by failing to pursue remedies that solve refugee crises by incorporating the state of origin into the distribution of costs, illiberal and authoritarian states have an incentive to create or otherwise intensify an 'externality' borne by other states that can be leveraged in negotiations on other issues (Greenhill 2011). viability of overseas resettlement' (Hathaway 2005, 85, 93). From this perspective, the modern refugee system was not simply designed to benefit of refugees, but also to benefit states (though, not optimally so) by acting like a floodgate to relieve migratory pressure or perhaps even an insurance policy. As for the floodgate: 'so long as the admission of refugees was understood to be formally sanctioned by states, their arrival would cease to be legally destabilizing' (Hathaway 2005, 85). As for insurance: the overseas resettlement model that had been used by the International Refugee Organization could guard against any one state being unfairly burdened by the bad luck of having a crisis on one's borders. For, while it became increasingly apparent that liberal states could not simply send people back (because of the norm of nonrefoulement) it was also not always feasible to automatically or fully integrate everyone either (the current predicament of several non-signatories in the Middle East illustrates this risk). States needed to both absorb and distribute the flows. Sadly, while the 'control and assimilate' aspect of the refugee regime developed the later notion of burden sharing via overseas resettlement was never concretely fleshed out (though, there have been attemptse.g. the European Refugee Fund).
Expanding the aperture of our analysis and taking such concerns of mutual selfinterest alongside those of human rights offers some very limited guidance insofar as we can reconstuct the "problem structure" and "solution structure" of an international regime formulated in response to a specific problem (Raustiala and Slaughter 2002, 545). In short, resettlement prioritizations using metrics of group-identity that both squarely support refugee rights and equitably distribute burdens between states would seem to be positively recommended. The UNHCR's prioritization of refugees who are vulnerable to further harm in a safe haven (e.g. often women, children, and sexual minorities) would appear to fit this bill, as would the prioritization of family reunification for refugees (and family is surely a metric of group identity). Conversely, groupbased prioritizations that would not support refugee rights or that would undermine burden sharing by optimally benefiting some states at the expense of others seem quite fraught. For instance, while the letter of Art. 19 might 23 technically allow France to prioritize resettling hundreds of thousands of highly-educated French-speaking refugees from the 'liberal professions' (on condition that they live in overseas departments), 24 this seems terribly unjust. While such an odd proposal might realize short-term political gains or 'development' goals it would undermine refugee rights and (in more ways than one) inequitably distribute resettlement burdens.
More realistically, if a state wants to prioritize resettling refugees from safe havens by using group-based criteria that optimally benefit itjob skills or languagewe need to look at factors of individual responsibility and collective action. Is that state using extraterritorial migration controls? Is it also giving priority to refugees at risk of further harm in safe-havens and adequate financial support to UNHCR? Does the resettlement state already have remedial duties towards some refugees that it should discharge given that it had a hand in 23 I mention this as a hypothetical only. I suspect this is legally foreclosed and thus a settled question in both senses; Supra fn. 2. Much depends on if the arrangement was (as per Art. 19) 'consistent … with the laws and constitution of France'. I suspect it is not even though there would be a great deal of discretion in the initial selection from overseas camps and limbos since Art. 19 is only required to apply to "lawfully staying" refugees. The socio-political dynamics with and in departments also make this a pure hypothetical. 24 Supra fn. 5. creating their situation? Is the refugee regime itself stretched to its limit? Whether a state's pursuit of its own interest in this regard is compatible with the commitment to refugee rights and the mutual self-interest of states is highly contingent. To answer these questions in a particular case or say anything more specific about generally acceptable actions in the existing regime we will need perspectives from political science and economics.
The conference behind our symposium began with a set of questions. One asked after 'the right principles of admitting and rejecting refugees when asylum … is undersupplied in a non-ideal world'. I'll end with a question: Why take this undersupply as a given? Given that UNHCR receives only 2% of its funding from the UN General Budget, that around 98% of its budget comes from voluntary (?!) governmental donations, and that (mainly earmarked) contributions from a small group of highly developed states (those most likely and able to engage in resettlement) make up nearly all of that 98% (Sazak 2015, 309;Vayrynen 2001, 150), why look any further for the source of our undersupply? The literature in refugee studies, political science, and economics already contains a variety of proposals to create institutional 'patches' (Hathaway and Neve 1997;Sazak 2015;Blocher and Gulati 2016) or ways of 'matching' refugees (Jones and Teytelboym 2017) 25 with states that would address the current undersupply. In short, instead of focusing on group-based resettlement prioritizations that could serially be undertaken by individual states I think we need to target the structures of our background institutions so as to both incentivize good behaviour and deal with the collective action problems that have come to beset a regime not designed for the modern era. Making headway on a complex problem involves asking the right questions.