Sanctuary, firewalls, regularisation: three inclusive responses to the presence of irregular migrants

ABSTRACT Liberal democratic states can respond to the presence of irregular migrants in three basic ways: by attempting to deport all migrants without legal status, by ignoring their irregular presence without providing them with rights or legal status, or by attempting to include them in the host society. This epilogue to a special issue on sanctuary discusses three inclusive responses to unauthorised immigration: sanctuaries, firewalls and regularizations. We describe their characteristic features by examining their specific benefits for migrants (protection from deportation, access to public services, and pathways to membership), the types of actors promoting or providing these responses (subnational or national governments, civil society actors), and the challenges to national immigration control they raise (contestations over jurisdiction, division of competencies, and determination of legal status). We acknowledge overlaps and ambiguities between the three responses and discuss whether asylum can be considered as sanctuary within the international state system and whether not only regularizations, but also sanctuaries and firewalls can promote inclusive membership. Finally, we lay the ground for empirical and normative analyses of justifications for each of the three responses by showing that these invoke specific claims about contestation rights, benefits for migrants, and benefits for the wider society.


Introduction
All independent states claim for themselves the power to control immigration into their territory. Yet such control can never be perfect and its effectiveness varies depending on resources that states are able and willing to devote to this task, technologies they can employ to achieve it, and legal and political constraints that impede their control efforts. As a result, all destination countries of forced or voluntary migration have to face the presence of undocumented and irregular migrants in their territory. Some of these have crossed borders without the required visas and passports; others have held valid documents at the time of entry but have stayed beyond the permitted period; and still others have filed asylum applications that have been rejected.
Political authorities have basically three options how to respond. They can try, first, to punish past and deter future attempts through deporting irregular migrants from their territory. Yet governments' efforts to do so are often hampered by non-cooperation of transit and origin states who are unwilling to take migrants back and by the considerable costs of staging raids, running detention centres, and hiring charter flights (Ellermann 2008). In liberal democracies, deportations are also often stopped by court injunctions invoking the rights of children or a more general right to private and family life that applies to undocumented long-term residents (Thym 2008;Ronen 2012). Immigration authorities also frequently face political pressure from employers where undocumented migrants provide an essential low-cost workforce (for example in fruit picking or domestic care work). In these cases, employers and consumers have an interest that migrants are not deported but also in avoiding taxation and regulation of migrant labour that would make it more costly and thus in maintaining migrants' precarious irregular status (de Haas, Castles, and Miller 2020, 56;Piore 1980). Finally, governments may be confronted with grassroots mobilizations defending the right to stay of those who are seen as belonging to the local community (Rosenberger, Stern, and Merhaut 2018).
Faced with these obstacles, governments may, second, decide not to decide and grudgingly tolerate the presence of migrants without legal status. Sometimes legislators create a formal status of temporary toleration, such as that of 'Duldung' of asylum seekers in Germany who fail to qualify for a protection status but cannot be deported (Schütze 2022). More often, informal toleration of irregular migrants simply is an implication of the above-mentioned constraints on deportation (Rosenberger 2019). Yet while enforcing deportations risks jeopardising human rights of migrants as well as mobilising resistance by domestic interest groups and civil society actors, turning a blind eye undermines the effectiveness of immigration control and the rule of law. It can also feed a growing shadow economy that diminishes the state's tax revenues and flouts its efforts of regulating labour markets and securing social minimum standards. For the migrants themselves, a policy of formal or informal toleration creates permanent insecurity over their future prospects in the country and blocks also their mobility, since leaving temporarily comes with the risk of being caught and rejected at the next attempt to re-enter (Stavilă 2015;De Genova 2002;Villegas 2014).
In this paper we will not discuss further exclusionary and toleration responses to irregular migration. Instead, we will focus on a third kind of reaction that can be considered inclusionary in the sense that it regards migrants as bearers of rights and aims to block or overcome the threat of deportation. We argue that an inclusionary approach consists itself of three distinct types of policies: sanctuaries, firewalls and regularizations. Sanctuaries are spaces created to protect irregular migrants from deportation. In the introduction to this special issue we distinguish between territorial, societal and discursive spheres of sanctuary (Mourão Permoser and Bauböck 2023). Firewalls are internal policies of subnational administrations (cities, counties, provinces) or organizatons interacting with undocumented migrants not to enquire about their legal status and not share such information with immigration authorities. Regularizations refer to laws and policies that transform irregular statuses of migrants into regular ones by issuing work authorizations or residence permits to them. Our aim in this epilogue is to discuss the differences and relations between these three inclusive responses to irregular migration.
The literature on sanctuaries has become wider and deeper in recent years. It has become, first, wider by expanding an initial focus on sanctuary cities in the U.S. and Canada to Western European cases and by starting to compare systematically these cases within and across countries (Lippert and Rehaag 2012;Bauder 2017;Bauder and Gonzalez 2018;Lasch et al. 2018;Darling and Bauder 2019;Bazurli and de Graauw 2023), making us thereby aware that the phenomenon has a wider geographic scope and also of the variety of national contexts in which sanctuary responses to irregular migration have emerged. In this respect, there is still much ground to be covered, especially in countries of transit and new immigration countries outside the Global North Bauder 2022a, 2022b).
Second, the literature has become deeper in the sense of covering multiple levels and spheres of sanctuary. Territorial sanctuary is now studied not only at municipal level but also that of autonomous regions (Wyn Edwards and Wisthaler 2023), and research has covered social sanctuary movements that operate across borders (Boudou 2023) or through local networks of private individuals (Elsrud, Lundberg, and Söderman 2023). Finally, the discursive aspects of sanctuary have attracted more attention in empirically-based analyses of the norms and values invoked by activists (Boudou 2023) or the symbolic repertoires of political authorities (Humphris 2023). In the introduction to this special issue, we have tried to provide a conceptual grid for analysing this broader variety of phenomena that are now covered in an emergent field of 'sanctuary studies' (Mourão Permoser and Bauböck 2023).
In this epilogue we propose a further step that should help to connect this field to the flourishing literature on the 'ethics of migration' in normative political theory. As stated above, our proposal is to consider sanctuary as one of three inclusive responses to the long-term presence of irregular migrants in a society. We analyse firewalls as a second response that is, on the one hand, a necessary component of effective protection of migrants in sanctuary spaces, but can also function, on the other hand, independently as an organisational guideline of non-cooperation with immigration authorities in the provision of public services. Finally, we add regularisation as a third response that has been widely studied in the migration policy literature but has been almost completely ignored in the literature on sanctuary although it offers the most durable solution to the problem to which sanctuary providers respond.
The next section discusses the particular features of sanctuaries, firewalls and regularizations. Section 3 considers relations and partial overlaps between the three responses and Section 4 discusses which normative justifications may be invoked by political actors when promoting one or the other policy. A brief concluding summary suggests implications for future research.

Sanctuary
As discussed more extensively in the introduction to this special issue, the core feature of sanctuary is to create a safe space where migrants will be at least temporarily protected from political authorities whose aim is to remove them from their territory (Mourão Permoser and Bauböck 2023). In modern states, central governments claim a monopoly over determination of legal status and deportation from the national territory (Gibney and Hansen 2003). Sanctuary involves therefore carving out safe spaces at subnational levels by agents that act independently of, or in opposition to national governments. These agents can be governments of sanctuary cities or regions that use their territorial autonomy to defy a central government's attempts at enforcing deportations (Humphris; Bazurli and de Graauw 2023;Wyn Edwards and Wisthaler 2023). They can also be national or subnational providers of public services in specialised institutions, such as hospitals or schools, who turn these institutions into safe spaces that immigration authorities cannot enter. Finally, such spaces may also be created by organisations, groups and individuals in civil society who give shelter to undocumented migrants, including churches (Mourão Permoser 2022a, 2022b), social initiatives and networks, and private individuals (Elsrud, Lundberg, and Söderman 2023;Boudou 2023).
In our introduction we have added 'discursive sanctuary' to these territorial and social modes of protection. While purely discursive declarations of sanctuary cannot themselves create spaces of effective protection, they are important in signalling a commitment on which institutional and social actors can follow up by providing effective sanctuary themselves or pressuring governments to do so (Mourão Permoser and Bauböck 2023). In any case, the action of creating territorial, social or discursive sanctuary disputes the deportation powers of national authorities either in all or in a limited set of cases. It raises thus a challenge to a basic feature of territorial jurisdiction that has developed with the consolidation of modern states: the monopoly of national governments to determine the legal rules for immigration and to control their enforcement throughout the national territory (Torpey 2000). This is why, in Table 1, we suggest that sanctuaries involve a contestation over jurisdiction.

Firewalls
Firewalls are created by internal rules for public administrations and public service providers that prevent them from sharing information about their clients' legal status with immigration authorities. They differ from sanctuaries mainly in their more limited purpose. Firewalls aim at providing irregular migrants with access to public services by removing deportation threats from the interaction between them and service providers (Carens 2013, 130-147). They do so by creating barriers of information (don't ask), communication (don't tell) and cooperation (don't assist) between clients, service providers and branches of public administrations charged with law enforcement (Lasch et al. 2018;Hermansson et al. 2022).
The builders of firewalls are motivated by two distinct concerns: a migrant-centred and an organisation-centred motive. The former aims to ensure that migrants have access to fundamental human rights to which they are entitled independently of legal status, such as emergency health care or primary education (Crépeau and Hastie 2015). The second concern is effective delivery of public services to a target population, which would be jeopardised by collaboration with immigration authorities (Hermansson et al. 2022). Often humanitarian and organisational goals will mutually reinforce each other as they do in the case of medical and educational service providers. In an epidemic, deterring irregular migrants from using public health services enhances the risk of infection for the general population. Similarly, denying undocumented children public education is likely to enhance risks for the wider population emerging from a fully marginalised underclass. Yet the two motivations for firewalls can also operate independently of each other. Organisations whose services are not essential for human rights protection may still adopt ethics codes that instruct their employees not to ask about migrants' irregular status. Conversely, other organisations may build firewalls without any humanitarian motivation merely for the sake of securing effective service delivery. A good example are police investigations into crimes where migrant witnesses with irregular status are unlikely to come forward if they have to fear deportation as a consequence (Timmerman et al. 2020;Spencer and Delvino 2019).
Yet the building of firewall policies can also be constrained by public administrative tasks. For example, labour inspections that aim to fight exploitative working conditions cannot ignore the irregular status of migrants whom they detect when carrying out workplace raids. Sanctioning employers for hiring irregular migrants inevitably puts the migrants themselves at risk. Instead of being shielded from immigration enforcement by a firewall, they can then only be protected temporarily as witnesses in trials against their employer, or permanently through regularisation offers (Fox-Ruhs and Ruhs 2022).
When we compare firewalls to sanctuaries, the former appear to be a more limited response to irregular migration for several reasons. First, firewalls adopted as an internal policy of organisations will also be limited by the purposes of these organisations. Medical firewalls do not protect healthy migrants and educational firewalls cannot protect migrant children outside the school environment. Sanctuary as a safe space requires building a comprehensive multipurpose firewall around a space within which irregular migrants can live most of their everyday lives and such a space is unlikely to emerge from the uncoordinated efforts of distinct service providers. Second, firewalls do not challenge immigration law enforcement as comprehensively as sanctuary policies do. In contrast with sanctuaries, firewall policies do not only protect migrants but also service providers against interference by immigration authorities with their publicly recognised or mandated tasks. Unlike sanctuaries, firewalls may therefore even be legally approved or required by national law or be covered by lawful self-regulation powers and internal ethics codes of service providers.
Nevertheless, firewalls still create conflicts over competences (Armacost 2016). As the above-mentioned example of crime reporting illustrates, such conflicts can arise even within the state's security apparatus. Information-sharing and cooperation between different branches of the public administration and the providers of public services is a basic requirement for the efficiency of modern states. Immigration authorities cannot implement the law if all other public administrations refuse to cooperate. Governments keen on improving their deportation enforcement record will therefore be tempted to interfere with the primary goals of public service providers by forcing them to report irregular migrants to immigration authorities.

Regularisation
The third inclusive response to the presence of undocumented migrants is regularisation. Unlike the other two, it addresses the root cause of the problem by moving people out of irregular status. This can be done through 'statute of limitation' laws that provide individuals with automatic or optional access to regular status after several years of residence and no criminal record other than violations of immigration law. Alternatively, governments or legislatures may implement collective 'amnesties', i.e. allow undocumented migrants a transfer into regular status within a limited time period until a certain deadline. Individual regularizations through statute of limitations for transgressions of immigration law implicitly acknowledge a right to domicile that emerges from de facto residence and social membership (Carens 2013). By contrast, collective amnesties are mainly motivated by concerns about effective immigration law enforcement where numbers of irregular migrants have grown so large that deportation of most is no longer a feasible goal. Amnesties allow immigration authorities a fresh start with a 'clean slate', which explains why they are usually accompanied by new and tighter sanctions for irregular migrants and their employers in order to dampen a pull effect from expectations of future new amnesties.
For migrants, the effect of regularisation is not just temporary protection as in sanctuaries and access to public services through firewalls but a recognition of their membership in the destination society, which provides them with fuller access to social rights than firewalls can do (Kraler 2019) and normally also opens a pathway to full citizenship. Some authors have proposed to make regularisation more acceptable for current citizens by excluding those who have a record of irregular status from access to citizenship for some time (by adding more years to a legal residence requirement) or forever (Maas 2022).
Such policies of delayed or excluded access to citizenship have been justified as a way of penalising past unlawful behaviour and maintaining the fair value of citizenship for regular immigrants. These arguments raise, however, questions about the underlying conception of membership. Should citizenship be granted as a reward for good behaviour or based on factual social membership? While the absence of a serious criminal record is often a condition for access to permanent residence or citizenship status, the case for such a conditionality is much weaker when it refers to trespassing immigration law. In many countries, there are virtually no channels of legal entry for those seeking asylum or for unskilled migrants. Penalising migrants indefinitely for crossing the border irregularly where this is the only way to cross it would be disproportionate and normatively unjustifiedparticularly so in the case of those who were good faith asylum seekers, even if their application was ultimately rejected or of those arriving as minors. Arguments about penalising migrants for irregular entry by delaying or rejecting citizenship can also be challenged as freezing a non-citizen status that is problematic from a perspective of democratic inclusion and equality in societies of immigration. We believe therefore that regularisation should mean access to the same legal status as that enjoyed by regular immigrants, including the same opportunities to apply for citizenship after a period of legal residencewhich in the case of irregular migrants would anyhow add up to longer de facto residence if periods of irregular presence are not counted. The practices of democratic states seem to follow in most cases a membership logic of regularisation. A well-known exception is the 2012 U.S. American DACA law (Deferred Action for Childhood Arrivals) that provided persons who arrived in the US as minor children without authorisation with a renewable residence permit and travel documents as well as eligibility for work permits without giving them access to permanent residence and thus eventually to naturalisation (Gonzales, Terriquez, and Ruszczyk 2014). The reason for this half-way regularisation was (and still is) a divided Congress where there is no bipartisan consensus on immigration reform (Johnson 2018).
In contrast with the other responses, regularisation is naturally a competence of national level legislatures and administrations. If subnational governments could carry out their own regularisation programmes, the other jurisdictions may not be bound to recognise beneficiaries as having legal status also in their territory. Regularisations are politically often hotly contested. The issue is, however, not which authorities ought to have the relevant powers and competencies, but whether national governments adopting regularisations create additional pull factors for future irregular migration and undermine the rule of law. However, as Sarah Song and Irene Bloemraad (2022) have recently suggested, rule of law arguments can be also mobilised in favour of regularisation. The authors support this with five arguments. First, legalising undocumented migrants on the basis of precisely defined criteria may enhance predictability of immigration law that diminishes administrative discretion. Second, statute of limitation provisions are a regular feature of criminal law that should also be applied to immigration law. Third, regularisation enhances 'publicity' by bringing migrants out of the shadows and underground economy. Fourth, it acknowledges limited enforcement capacities and improves the like treatment of like cases, instead of selectively detaining and deporting small numbers of migrants. Fifth, legalisation could also be justified as 'a corrective measure for the government's past failure to ensure due process in asylum and deportation cases (Song and Bloemraad 2022, 32). Table 1 summarises the characteristic features of the three inclusive responses to the presence of irregular migrants.

Ambiguous contexts and cases
As most empirical typologies, ours of three inclusive responses to irregular migration should not be interpreted as a rigid classification, but as describing predominant patterns. Some of the most interesting cases are those that defy easy categorisation. We have already pointed out above that sanctuaries may also be described as spaces where irregular migrants are protected by a comprehensive firewall around a certain jurisdiction (or a space that for practical purposes is treated as if it were a separate jurisdiction, such as a church building). In this section we discuss instances where apparently defining characteristics of one response seem to apply also to others. The point of this exercise is not to cast doubt on the usefulness of our conceptual distinctions but to clarify further in which respects the three responses differ even if their characteristics seem to overlap.

Sanctuary at national level
In a way, modern asylum law is a functional equivalent of sanctuary offered by states to individuals fleeing persecution or violence elsewhere. In effect, the origins of state asylum can be traced back to a nationalisation of earlier forms of sanctuary (Marfleet 2011;Mourão Permoser 2022b). For the historian Philip Marfleet, when King Charles II of England offered asylum to the persecuted Huguenot minority of France in 1681, this represented a secularisation of the institution of sanctuary which now 'was no longer associated with specific local sites in which protection was an expression of Godly or saintly authority but with the entire territory of the nation state' (Marfleet 2011, 440).
With the emergence of international refugee law, asylum has been upgraded into an internationally recognised right to seek protection in other states and a duty of nonrefoulement that prevents states from sending refugees to countries where their fundamental rights would be in jeopardy. At least in terms of the underlying legal norms, asylum has thus become a stronger form of protection than can be offered by contemporary forms of sanctuary at sub-state level. At the same time, however, contemporary refugee protection has been split into different levels of protection with those fleeing from international and civil wars often receiving merely temporary protection instead of the more secure status of asylum according to the Geneva Refugee Convention. This creates another similarity between national-level refugee protection and temporary protection provided by sanctuaries arrangements within states. David Owen has recently reintroduced the notion of sanctuary into the literature on refugee protection by distinguishing between asylum as a response of the international state system to the persecution of individuals in or by their state of origin, sanctuary as protection for those displaced by generalised violence and the breakdown of public order, and refuge as the sheltering by another state of persons displaced by events like famines and droughts (Owen 2020, 54-65).
In all of these ways, the protection of refugees and internationally displaced persons can be compared with the historic institution and contemporary practices of sanctuary. The decisive difference between international asylum and within-state sanctuary lies, however, in what migrants are protected from: violence in their country of origin or deportation to that country. These two contexts can hardly be lumped together under the same concept of sanctuary. It makes therefore sense to apply this term today specifically to the latter case and only in a more metaphorical sense also to the former. Asylum law is a form of international protection enacted by states against other states. It is offered to individuals outside their country of nationality whose lives and fundamental liberties are threatened by that state or on whose protection they can no longer rely. By contrast, sanctuary practices of civil society and local governments create spaces of protection against immigration authorities within this very same state and by doing so put up resistance to the norms and practices of national state authorities.

Membership effects of sanctuaries and firewalls
Sanctuary can turn into a de facto regularisation that is territorially circumscribed if irregular migrants are granted voting rights at the local level, employability, and access to all goods and services provided by the local administration. 1 There are some sanctuary cities in the US that try to do just that. In such cases membership norms, which characterise regularisations, strongly drive policies at the local level and thus transcend the core features of sanctuary as a space of protection. 2 However, besides being territorially limited, this kind of proto-regularisation through sanctuary is also weaker than 'real' regularisation because the former is not anchored in a formal legal status. It is a matter of practice and its viability depends on the current policies of particular administrations at local and national levels. It is therefore (at least in principle) also only temporary. If a new government pursues a hardline policy towards irregular migrants, the locally regularised may again become irregular.
Even if they cannot regularise undocumented migrants, some firewall and sanctuary policies impact also on migrant membership by providing partial substitutes for legal status and by guaranteeing or attributing rights that are jeopardised by irregular status. Amongst other cities, New York, San Francisco, Los Angeles, Washington D.C., Barcelona, Madrid, Zürich and Paris have issued city ID cards that provide access to some municipal services and are also available to migrants without residence permits (de Graauw 2014(de Graauw , 2021Kaufmann and Strebel 2021). Although these services are often quite limited and city ID cards do not protect holders from deportation, issuing identity documents to undocumented migrants carries a strong symbolic message that their holders are considered local citizens by virtue of their residence in the municipality, even if they do not enjoy residence rights or citizenship status at the national level (Bauböck 2003;Bauböck and Orgad 2020).
Sanctuaries generate, however, citizenship effects also in a quite different sense. Private providers of sanctuary are enacting their own citizenship by protesting or defying perceived unjust exclusion. Irregular migrants participating in active forms of protest can similarly be regarded as enacting, performing and claiming citizenship (Isin and Nielsen 2008), but this should not be confused with citizenship as a status of membership recognised by political institutions. Performative citizenship in the context of sanctuary may be either productive for claims to full citizenship status or counter-productive when sanctuary consolidates into a substitute of semi-citizenship (Cohen 2010), i.e. a protracted situation where immigration laws are not enforced and migrants enjoy de facto access to many rights, but that remains overshadowed by the threat of forced removal from the territory (Bagelman 2016).

Justifications for sanctuary, firewalls and regularisation
Having examined characteristic features of the three inclusive responses to irregular migration, we now turn to reasons that can be invoked to justify them. We neither present an empirical account of normative arguments used by political and social actors in defence of their contestation of the state's immigration powers, nor do we aim to evaluate these arguments from the perspective of a comprehensive theory of justice in migration policies. Our goal is more modest. We want to lay the grounds for both kinds of research by mapping arguments that seem prima facie the most plausible justifications. These arguments still have to be tested in empirical research to see whether they correspond to the motivations articulated by those advocating one or the other of the three responses and also would have to be tested by normative theories for their internal coherence and correspondence with conceptions of justice or democratic legitimacy. The ideas presented in this section should therefore not be interpreted as our own normative theory or inductive findings from research, but as initial material for empirical hypotheses and normative propositions.
What we hope to show is that although justifications for the three responses partly overlap because they are driven by the same impulse to secure rights and membership for irregular migrants, there are also distinct arguments that can be used to support each of the responses. Humanitarian justifications can be provided for all of them. In the following table we want to highlight differences by listing normative reasons that can be more specifically used to justify a particular type of response.
In Table 2 and the discussion below we distinguish (in the rows) between justifications of the contestatory aspect of these responses, i.e. their resistance against or deviation from general legal norms applying to irregular migrants, and justifications referring to their beneficial aspects, which consist of benefits for migrants (e.g. protection from deportation, access to services, security of residence) and for the wider society (e.g. reducing public health risks, shrinking the shadow economy, restoring the rule of law). What we call (in the first row) 'contestation rights' refers to claims of actors that they have a right to contest immigration authorities' policies grounded in their political powers, organisational autonomy or in the perception that they share with migrants a common membership in the society. Justifications are partly actor-specific; some cannot be invoked by social actors (civil society organisations or private individuals) in the same way as by public administrations and government authorities. We distinguish therefore between territorial and social sanctuary. The former refers to territorial jurisdictions that are turned in spaces of sanctuary by state actors through laws and regulations; the latter to physical spaces transformed into sanctuary spaces by non-state actors (Mourão Permoser and Bauböck 2023). A similar distinction could be made between firewalls erected by public administrations and non-governmental organisations, but we assume that the justificatory reasons for these two types of firewalls are broadly the same and therefore lump them together. We leave out here the third type of discursive sanctuary, as it is normally a supplementary aspect of territorial and Shrinking shadow economy, restoring rule of law social sanctuary that does not require distinct justification. Where discursive sanctuary remains just a symbolic proclamation without effective protection measures linked to it, it can be justified only weakly on grounds of potential impact on policy reform (Mourão Permoser and Bauböck 2023). When fully enacted, sanctuary is a quite radical practice in the sense that it brings about a de facto suspension of state authorityalbeit for a limited period of time within a limited space. Sanctuary activists have different motivations for engaging in these practices. Some providers of sanctuary justify their actions on the basis of a normative ideal of open borders. Others raise more modest claims that certain kinds of immigrants should not be deported and have claims to regular status (e.g. long-term residents and their dependent children, rejected asylum seekers with non-refoulement claims). Yet others do not question the authority of the state over immigration decisions per se, but rather argue that the way in which this authority is exercised is illegitimate because current laws are immoral and unjust. From this perspective, sanctuary does not entail an overall rejection of state authority over irregularised migrants and refugees. It is rather a form of resistance against more particular injustice. Nevertheless, sanctuary still has the function of making visible the existence of potential alternatives to existing border regimes and maybe even of enacting at the micro-level a utopia in which borders are open and crossing them does not create distinctions between individuals' rights. In this respect, sanctuary remains, however, caught in a paradox: Such utopias can only be enacted by creating another type of boundary that serves to keep state authorities out. Migrants continue to be at risk when they cross it by leaving the protected space.
We suggest that all these motivations allow to characterise the provision of sanctuary by actors in civil society as acts of civil disobedience. This would entail that normative justifications of and constraints on civil disobedience also apply to social sanctuary. In this view, acting against the law by providing sanctuary is justified if providers sincerely and reasonably believe that enforcing immigration laws would entail severe injustice, and if they use non-violent means to defend sanctuary spaces against state authorities. Another condition for justified civil disobedience emphasised by theorists like John Rawls (1999) 3 and Jürgen Habermas (1985) is that actors perform their actions publicly and aim to convince other citizens of the injustice they are protesting against, with the goal of achieving a policy change. As we have discussed in the introduction to this special issue, this is the case where sanctuary is also enacted discursively (Mourão Permoser and Bauböck 2023). Unlike in many other instances of civil disobedience, however, sanctuary activists do not tend to communicate what they do as an illegal action, that is, as a conscious and purposeful act of disobedience, for which they are willing to accept legal punishment. Rather, activists often argue that what they are doing is 'obedience to the law', only that the law in question is the higher law of human rights that is infringed by national or European immigration law. Cédric Herrou's challenging of French law criminalising private sanctuary provision at the Conseil d'État illustrates this well (Boudou 2023), as do the narratives of church activists and search-and-rescue NGOs (Mourão Permoser 2022a; Mann and Permoser 2022). By contrast, the network of private individuals sheltering irregular immigrants in Sweden do not make their actions visible to the public in this way as this could jeopardise the effective provision of sanctuary (Elsrud, Lundberg, and Söderman 2023). Our conclusion is that justifications focusing on benefits for migrants may potentially clash with normative conditions for the justifiability of civil disobedience.
We have presented the somewhat ambiguous case of a civil disobedience defence for social sanctuary first, in order to highlight the contrast with territorial sanctuary, firewalls and regularisation where this justification seems less applicable. It is not clear that subnational governments and branches of the public administration can claim a civil disobedience defence against the state of which they are a part. Individual civil servants may invoke it when refusing to carry out orders that risk to expose migrants to unjust deportations, but mayors or heads of public offices acting in their official role cannot. In this case, a different type of argument emerges that invokes a distribution of competences between branches of government. The core of the argument here is that a sufficiently robust form of territorial autonomy allows sub-national political actors to refuse cooperation and provide sanctuary against national-level immigration law enforcement. This does not remove the contestation aspect from their actions but places them within a different justificatory framework where contestation is over the vertical distribution of legal powers and policy competencies in multilevel democracies.
With regard to benefits for migrants territorial and social sanctuary closely resemble each other. Both aim at protecting them from detention and deportation. Sanctuary cities and regions can moreover ensure that irregularised migrants enjoy access to public services within their territory and can lead a more or less normal and full life, which is unlikely to be the case for migrants sheltered in churches, private homes or on search and rescue vessels operated by NGOs. The providers of sanctuary are less likely to foreground benefits for the wider society than those erecting firewalls or advocating for regularisation. However, we can still hypothesise about what kind of defences might be available to them when challenged on such grounds. Subnational governments could claim that in multilevel democracies, immigration enforcement powers should not be an exclusive domain of the central government. While some tasks, such as policing state borders or issuing visa must remain under central government control in order to preserve internal freedom of movement within the country, in other policy areas local governments should have a say as they are responsible for providing public services to local communities that include irregular migrants. Such immigration regime pluralism would be arguably beneficial for democracy in terms of an additional element of checks and balances between levels of government and in strengthening a residence-based conception of local citizenship (Bauböck 2018, 75--80). Territorial sanctuary could thus also signal the need for a constitutional reform that provides subnational governments with additional legal powers in arrangements that have been labelled 'immigration federalism' (Bauböck 2001;Spiro 2001;Varsanyi et al. 2012). Social sanctuary is clearly different in this respect. Insofar as it can be justified as legitimate civil disobedience, this entails a general appeal to citizens' sense of justice. Activists can also argue that their activities contribute to a more robust civil society that is ready to confront state authorities when they abuse their powers. Discursive proclamations of sanctuary strengthen both kinds of claims. They help to mobilise local citizenries for autonomous immigration policies of subnational governments and they articulate in a performative manner the autonomy of civil society vis-à-vis the state.
Turning now to firewalls, organisational autonomy rather than civil disobedience is more likely to be invoked as a justification for their contestatory features. Branches of the public administration, such as a national health service, as well as non-governmental organisations providing essential services to migrants can point to their specific organisational tasks that could be jeopardised if they collaborated with immigration authorities. Based on qualitative research, Spencer and Delvino have identified five frames adopted by policy makers to justify providing social services to irregular migrants: increasing security for all city residents (security frame), improving the health and well-being of vulnerable individuals (humanitarian frame), protecting the human rights of the undocumented (human rights frame), treating migrant workers as future citizens (deserving workers frame), improving the socioeconomic situations of the city by ensuring the well-being of all inhabitants (socioeconomic frame), and finally, improving the service provision of the city (efficiency frame) (Spencer and Delvino 2019).
Despite these multiple justifications for expanding service provision to irregular migrants, the substantive justification for contestation of national powers is organisational autonomy. This is quite different from the justification for supporting territorial sanctuary. Mayors institutionalising sanctuary may point to their local political mandate and constitutional powers, while service providers need to invoke other justifications for creating immigration firewalls. An internal one could be professional ethics that do not allow them to do the work of police forces. Civil society organisations (but not public administrations) may also claim a right to non-interference by the state with their activities derived from freedom of association or private property rights. Such internal reasons generally need to be backed up by others addressed to the broader public and the beneficial aspects of firewalls. Next to humanitarian reasons, it is likely that organisations shielding their operations through firewalls will argue that this is beneficial not only for irregularised migrants but also for regular ones and resident citizens who receive their services (Hermansson et al. 2022).
Regularisation differs more fundamentally from the other two responses. As it involves a reform carried out by government authorities on the basis of a regular change of laws and policies, a civil disobedience justification is obviously inapplicable. Moreover, as discussed above, since regularisations normally occur at the national level, the implementing institutions cannot and do not need to invoke their autonomy vis-à-vis the central government. However, most regularisations are still strongly contested in the arena of democratic politics and their proponents have to provide justifications for overturning an existing immigration law that irregularises large numbers of migrants. We suggest that in all cases of regularisation the core justification will be that their beneficiaries meet conditions for membership in the host society.
In the case of individual regularisation, a claim to regular status needs to be justified on the basis of a sufficiently strong relation to the host society, i.e. in terms of social membership (Carens 2013). Collective amnesties that transfer large numbers of irregular migrants into regular status within a short period of time or until a certain deadline often cast the net wider by including also migrants who do not yet qualify as members. Their justification needs therefore additional reasons referring to benefits for the wider society. These include the shrinking of a shadow economy in which labour laws and tax duties are routinely infringed. The other main rationale for collective regularisations is to reboot immigration law implementation by getting rid of large numbers of cases where deportations cannot be enforced. As argued by Song and Bloemraad (2022), regularizations of both types can therefore be regarded as strengthening rather than undermining the rule of law.

Conclusions
Irregular migration is a pervasive and persistent phenomenon in contemporary destination countries of major migration movements. In this article we have examined three policy responses that can be described as inclusive in the sense that they neither try to reverse irregular migration through deportation nor ignore it through toleration. We have labelled these three responses 'sanctuaries', 'firewalls' and 'regularisations'. They have been covered so far by rather separate literatures. Our goal was to systematically analyse the differences, similarities and overlaps between the three responses. This is mainly a mapping exercise. It has required conceptual analysis that is inductively informed by empirical examples discussed in the literature. We have described the characteristic features of sanctuaries, firewalls and regularisations by examining their specific benefits for migrants (protection from deportation, access to public services, and pathways to membership), the types of actors promoting or providing these responses (subnational or national governments, civil society actors), and the challenges to national immigration control they raise (contestations over jurisdiction, division of competencies, and determination of legal status). We acknowledge overlaps and ambiguities between the three responses and discuss whether asylum can be considered as sanctuary within the international state system and whether not only regularisations, but also sanctuaries and firewalls can promote inclusive membership. Finally, we lay the ground for empirical and normative analyses of justifications for each of the three responses by showing that these invoke specific claims about contestation rights, benefits for migrants, and benefits for the wider society.
Empirical studies using tools such as discourse analysis could try to find out how often organisations, officials and activists engaged with one of the three responses invoke these justifications and how they evaluate the alternative responses. 4 For normative theorists our analysis might be helpful to move beyond questions of what justice or democracy require and to engage more closely with policy alternatives and dilemmas where each option seems to be endorsed by the same liberal values.
For example, firewalls prevent political authorities from distinguishing between migrants with regular and irregular status. However, regularisation (both individual and collective amnesties) presupposes this distinction when bringing about a transfer from one status to the other (Schmid forthcoming). Is this merely a practical dilemma that can be easily resolved or also an ethical one between the goals of access to public services and to full citizenship in immigration societies?
Posing such questions requires a broader view of the field that is not limited to the perspectives of migrants and actors in civil society supporting them but considers ethical dilemmas of migration policies also from the perspective of democratic legislators and executives (Bauböck, Permoser, and Ruhs 2022). We hope that our conceptual reflections on inclusive responses to irregular migration will contribute to both empirical and normative research agendas, and to the much-needed dialogue between scholars and policy actors.