Non‐governmental/civil society organisations and the European Union‐externalisation of migration management in Tunisia and Egypt

Funding information Nederlandse Organisatie voor Wetenschappelijk Onderzoek, Grant/Award Number: 453-12-004 Abstract This paper sheds light on the role of non-governmental/civil society organisations, as well as the United Nations High Commissioner for Refugees, in the externalisation of migration management in Tunisia and Egypt. These organisations are involved in migration-related activities, which, from the externalisation perspective, should prevent people from attempting the sea-crossing to Europe, thus immobilising them in North Africa. The paper is an ethnographic border regime analysis drawing on extensive fieldwork. It shows that European Union-externalisation is not a univocal and smooth process. Instead, externalisation is entangled in a complex network of actors and dynamics. Measures meant to support externalisation may produce effects contrary to those envisaged, whereas practices accomplishing externalisation goals may be in fact the result of internalisation. By analysing the various actors in their diverse practices and in their—sometimes conflictual—relationships, the paper also tries to minimise state-centrism and Euro-centrism in the study of externalisation.

dynamics in North Africa are mostly studied from the Euro-centric (and state-centric) perspective of EU-externalisation.
Externalisation is the process through which destination countries promote, support, delegate, impose, or directly carry out activities related to migration and border management outside their territories (i.e., in the territories of countries of transit and origin or in international waters) in order to prevent unwanted arrivals at their territorial borders (Bialasiewicz, 2012;Boswell, 2003;Burridge, Gill, Kocher, & Martin, 2017;Cobarrubias et al., 2015;Menjívar, 2014;Zaiotti, 2016;Zolberg, 2003). Externalisation largely relies on the active cooperation from countries of transit and origin, which are pushed and incentivised to adopt specific legislative and policy measures.
However, destination countries also carry out activities in foreign territories directly, for example through visa policies (Bigo & Guild, 2005) or information campaigns (Watkins, 2017), and they outsource controls to private companies acting there, such as carriers (Rodenhäuser, 2014). Similarly, migration-related activities in countries of transit and origin can be delegated to IOs (Lavenex, 2016;Wunderlich, 2012) and NGOs/CSOs. More specifically, the EU and its member states have identified NGOs/CSOs in the Southern Mediterranean neighbourhoodespecially the more professionalised and westernised ones (Bürkner & Scott, 2018)-"as natural partners" (Bürkner, 2018, p. 180) in the process of region-building of which the externalisation of migration control is a part: European "funding instruments for migration-related projects in third countries [show] a strong geographical focus on […] the southern Mediterranean" (Den Hertog, 2016, p. 1), and "[i]n many cases, […] the recipients are actually NGOs" (El Qadim, 2019, p. 353; see also Irrera, 2019).
However, studies on the relationship between NGOs/CSOs and externalisation are still extremely rare and limited in scope. Andersson (2014) notes that NGOs in West African countries, attracted by the funding opportunities offered by European state donors, have "to obey the rules of the deterrence game" (pp. 44-45) on which the funding is conditioned. Case studies only focus on information campaigns aimed at discouraging people from leaving: Rodriguez (2019) questions the effectiveness of such schemes carried out by NGOs in Senegal on behalf of EU states; Poutignat and Streiff-Fénart (2010) note that some Mauritanian organisations adopt a clearer externalisation agenda, whereas others are more concerned about the migrants' rights, and aim to provide independent information.
The paper adds to these studies in addressing this research gap (see also Dini & Giusa, forthcoming). In analysing the NGOs/CSOs' role in externalisation, it focuses on those NGOs/CSOs that are either funded by states or IOs or are engaged in activities, which, regardless of the funding sources, typically lend themselves to serve externalisation purposes. 3 Besides NGOs/CSOs, the paper also examines UNHCR. Indeed, the work of NGOs/CSOs is strongly connected with that of IOs, to which NGOs/CSOs are often subcontractors. Moreover, migration and border studies have paid scant attention, if at all, to IOs in Tunisia and Egypt. Although both of the two major IOs specifically devoted to migration and asylum issues often work in (formal as well as informal) cooperation with NGOs/CSOs, in Tunisia and Egypt UNHCR relied more than IOM on NGOs/CSOs in the period under research (2011)(2012)(2013)(2014)(2015)(2016)(2017). This period, indeed, saw an increase in the number of people fleeing Libya and Syria, as well as in "UNHCR's […] relevance and effectiveness with respect to mixed migration in North Africa" (Eyster & Paoletti, 2015, p. 150). This is why UNHCR was the most suitable case study.
Indeed, EU-externalisation "is not a smooth top-down process" (Bartels, 2018, p. 64) and "cannot be understood entirely in terms of an old style geopolitics of dominance" (Collyer, 2016, p. 610).
Governments of third countries are not passive recipients of EU inputs (Reslow, 2012), their responsiveness being "shaped by their respective domestic and regional concerns" (Cassarino, 2018, p. 405).
Some states resist to externalisation or turn it into a leverage to rebalance power relations (Cassarino, 2005;El Qadim, 2015;Paoletti, 2010). IOs, in turn, can be, to some extent, autonomous and independent actors (Barnett & Finnemore, 1999;Keohane & Nye, 1974) and play at least the role of counterweights (Lavenex, 2016). Similarly, NGOs/CSOs are both objects and subjects of government (Sending & Neumann, 2006); their positions vis-à-vis externalised migration management are diversified, and the outcomes of their activities in this field unpredictable. Moreover, the agency of individuals on the move, towards which externalisation is directed, can adapt and react to the latter's moves (Papadopoulos, Stephenson, & Tsianos, 2008). Finally, each actor can mobilise different narratives and discourses (human rights, humanitarianism, security, terrorism, development, etc.), which, again, engage in a dynamic and sometimes conflictual interaction with one another.
The concept of "migration (and/or border) regime" (Hess, Kasparek, & Schwertl, 2018;Horvath, Amelina, & Peters, 2017;Kasparek, De Genova, & Hess, 2015;Pott, Rass, & Wolff, 2018) and, more specifically, the 'ethnographic border regime analysis' approach (Hess, 2012) help make sense of such complexity. A regime is the result of "multiple entanglements of discourses, power relations, and subjectivities" (Horvath et al., 2017, p. 305). The analysis of practices plays a crucial role in revealing such entanglements, whereby the different practices of the various actors involved "relate to each other but are not ordered in the form of a central logic or rationality" (Hess, 2012, p. 430); rather, the concept of "regime" implies "a multidimensional multi-scalar space of conflict and negotiation" (Kasparek et al., 2015, p. 70) whose structure is always fluctuating and subjected to changes. 4 By focusing on the practices of NGOs/CSOs and UNHCR, and by analysing these actors in their diversity and in their dialogic and sometimes conflictual relationship with the various state and nonstate actors in the field, against the background of the Tunisian/Egyptian context, this paper aims to contribute a "multiperspectival" (Rumford, 2012) gaze on this Mediterranean border regime and to minimise the risks of methodological nationalism/Europeanism. Clearly, it is impossible to completely avoid Euro-centrism and state-centrism when studying EU-externalisation. However, it is arguably possible to both acknowledge the relevance of the externalisation process and relativise it as just one (if possibly the most powerful) of the drivers determining the dynamics of the North African border regime.
The ethnographic border regime analysis presented here draws on 50 interviews, as well as a number of informal conversations, carried out with representatives from NGOs/CSOs, IOs, state authorities, and migrants during fieldwork in Tunisia (2016) 5 and Egypt (2017). 6 Another three interviews were carried out on Skype or on the phone shortly after the Tunisian fieldwork. Given the severe limitations imposed by the Egyptian government on the freedom of association, and on human rights in general, NGOs/CSOs operating in Egypt are anonymised when quoted, and their names are never mentioned in association with (potentially) sensitive information, unless the latter is retrieved from publicly available sources.
The paper proceeds as follows. Section 2 provides an overview of IOs and NGOs/CSOs in the Tunisian and Egyptian contexts after the 2011 uprisings. Section 3 analyses UNHCR as an example of a major IO in the field and unpacks its role in migration management from the externalisation perspective. Section 4 is divided into four subsections, each exploring a different category of activities carried out by various NGOs/CSOs. Finally, each subsection briefly connects externalisation through NGOs/CSOs with the concepts of migration management, delocalization, civic solidarity, and humanitarian border, respectively.

| IOS AND NGOS/CSOS IN TUNISIA AND EGYPT
In both Tunisia and Egypt, the number of organisations involved in activities related to migration has increased in recent years. However, there are some differences between the two national contexts.
In Tunisia, until the 2011 revolution, the few existing NGOs/CSOs were poignantly defined "truly governmental organizations" (Boubakri, 2013), because they were controlled by the Ben Ali regime (the others had been outlawed, or their lives had been made almost impossible). Migration and asylum were managed mainly by With the revolution came the freedom of association, which resulted in the flourishing of civil society movements. Thousands of local and dozens of international NGOs/CSOs were established in the country (Bartels, 2015;Eyster & Paoletti, 2015). At the same time, the Libya war caused hundreds of thousands people to flee and seek refuge in Tunisia. Many NGOs were involved in managing the crisis, mostly under the coordination of UNHCR, whose importance grew exponentially. In the Choucha camp alone, where most refugees were hosted, nearly 20 NGOs worked under UNHCR coordination in the first months of the crisis. Some of them kept working in Tunisia even after the crisis ended and were still there at the time of fieldwork.
In Egypt, instead, and despite restrictions, there were already large numbers of NGOs before the 2011 uprising. Contrary to what happened in Tunisia, "the revolution prompted a crackdown on NGOs which […] had a chilling effect on all civil society organisations" (Jones, 2012, p. 16). The situation even worsened under Morsi's successor Al Sisi (Norman, 2016), who took the power in 2013. Since then, many Egyptian NGOs have been shut down by the government, and new laws and regulations have granted authorities more leeway for denying the registration of new associations while also imposing limitations regarding funding from abroad. This restrictive turn is directed towards NGOs/CSOs (e.g., human rights organisations) that may be regarded as opponents of the regime, and thus towards Egyptian citizens in the first place. However, restrictions inevitably affect NGOs/CSOs supporting migrants' human rights as well.
Also, the two migration-related IOs have a longer history of rootedness in Egypt. IOM opened its first office in the country in 1991 (with Cairo also hosting the Regional Office for the Middle East and North Africa), whereas the UNHCR Egypt office was opened as early as 1954. Whereas the Libyan crisis had a more limited impact on Egypt, and refugees fleeing the neighbouring country did not remain for long on Egyptian territory, the Syrian crisis had significant and long-lasting consequences for this country. As in Tunisia, this resulted in an increasing importance of UNHCR Egypt and in the growing number of its subcontracting NGOs/CSOs.

| EXTERNALISATION THROUGH IOS? THE CASE OF UNHCR
Scholarship has analysed the role of IOs in the externalisation of migration management, considering (and often criticising) them as essential components in such processes (on IOM and UNHCR, see Lavenex, 2016;Wunderlich, 2012; on IOM, see Ashutosh & Mountz, 2011;Bartels, 2017;Brachet, 2015;Nieuwenhuys & Pécoud, 2007;Pécoud, 2010; on UNHCR, see Fine, 2018;Scheel & Ratfisch, 2014;Valluy, 2007; on the International Centre for Migration Policy Development, see Cobarrubias, 2019;Georgi, 2007;Hess, 2010). Indeed, whereas international relations scholars are divided about whether and in how far IOs are autonomous and independent actors (Mearsheimer, 1994), migration scholars tend to see migration-related IOs as instruments of the most powerful states.
However, several authors (Geiger, 2010;Hess, 2010;Lavenex, 2016;Pécoud, 2017) also pointed out that UNHCR and other IOs are not mere executors of EU plans: They can also play a role as counterweights and influence policymaking in destination countries.
Many NGOs/CSOs have some kind of formal cooperation with IOs, which mostly means implementing their projects. Other NGOs/CSOs cannot help having at least informal cooperation with IOs, especially in contexts in which the gaps created by the inaction of local state governments are particularly wide, and humanitarian emergencies leave no other option than cooperation between all actors. With regard to migration, this is the case of both Tunisia and Egypt. Consequently, understanding the role of IOs is key to the analysis of NGOs/CSOs. This section sheds some light on the role of UNHCR, which was indicated by many interviewees as the most powerful player in Egypt, and clearly still had a dominant position in Tunisia over 2 years after the closure of the Choucha camp. In both Tunisia and Egypt, like in other transit countries, the agency is faced with the need to strike a balance between its own mandate, the interests of EU destination countries, which are among the main UNHCR donors, and the interests of the host countries, allowing UNHCR to work there. According to its mandate, UNHCR should provide protection for refugees and seek durable solutions for them. The interests of EU destination countries are preventing asylum seekers from reaching Europe and establishing national protection systems in transit countries. Indeed, if the latter could guarantee sufficient protection standards, they would fall into the category of "safe third countries" (Achermann & Gattiker, 1995), and asylum seekers could be returned there. The interests of host countries are keeping the number of refugees in their territories as low as possible, by: a) limiting access to formal protection mechanisms, b) preventing effective asylum systems from being established on their territories, in order to avoid attracting refugees from countries in crisis and being sent back those who do manage to reach Europe, c) deporting as many people as possible.
UNHCR's "diplomatic" attitude of balancing different and conflicting interests (Loescher, Betts, & Milner, 2008) is reflected in a series of practices. In Egypt, for example, the refugee status determination procedure imposes on asylum seekers heavy and unusual limitations of their rights, exposing them to abuses from state authorities. Refugees contacting UNHCR to express their wish to apply for asylum are given an appointment for registration up to 4 months later (the average waiting time being 2 months). During this period, in which they are not officially registered with UNHCR, they can be detained and expelled by state authorities. Moreover, among those who eventually manage to register with UNHCR as asylum seekers, only those who have a valid identity document are given a "yellow card," which grants a residence permit during the asylum procedure. The others only receive a "white card," which, again, does not protect from arbitrary arrest and deportation. By doing this, "UNHCR is creating a protection gap. They do this because they were pressured by the Egyptian government", the director of an NGO told me. The reason provided by UNHCR-Egypt Officer 1, instead, was that "here in Egypt, we have people who come for reasons other than asylum, and sometimes they have documents but conceal them, so we try to regularise asylum in Egypt". These words show that UNHCR makes the security narrative of state authorities its own. Similarly, UNHCR is scanning the irises of all asylum seekers: "So we are sure we are advocating for the right people, and we can check that they don't sell their refugee documents," UNHCR-Egypt Officer 2 told me. Furthermore, NGO representatives told me that when UNHCR is allowed to access Egyptian detention centres to register asylum seekers there, it "indiscriminately scans the irises of all foreign detainees," not just of refugees. This confirms the agency's tendency to expand the boundaries of its operational activities, as well as of its policy interests (Betts, 2012;Crisp, 2009;Lavenex, 2007) and, thus, its increasing involvement in the global policing of mobility (Franke, 2009;Scheel & Ratfisch, 2014).
More importantly, this shows that the process of externalisation in which UNHCR is involved is not a straightforward and univocal one. From the EU perspective, UNHCR should foster the externalisation of asylum through the establishment of effective protection systems in third countries. This would discourage beneficiaries from attempting the sea-crossing (given the satisfactory living conditions in the EU-neighbouring countries) and allow returns from Europe to "safe" countries. Instead, UNHCR bows to the pressure of the local state authorities aimed at reducing the obstacles for the Egyptian deportation regime. On the one hand, this may to some extent be still in line with the objectives of EU externalisation, insofar as incarcerations and deportations arguably contribute to keeping unwanted migrants away from Europe. On the other hand, the extremely low level of protection in Egypt, the difficult living conditions, and the high degree of arbitrariness of state authorities are likely to push increasing numbers of people to attempt the crossing to Europe, in search for better conditions. Even people with a regular residence status live in a permanent state of precariousness, which may push them to travel further to Europe. This is even more so for the undocumented onesincluding those falling into the "protection gap" created by UNHCR to please Egyptian authorities. Furthermore, specific categories of refugees with resident status, such as Syrians, may suffer from particularly harsh conditions imposed on them by authorities and tolerated by UNHCR. Sea-crossings of Syrians from Egypt to Italy only started in 2013, after Al Sisi seized power and launched his anti-Syrian campaign (Denaro, 2016)  which also exceeds its mandate by merging humanitarian activities with security-related ones. Moreover, some practices of the Tunisian UNHCR contribute to reducing the level of refugee protection. An example is provided by the rescapés de la mer (French for "those rescued at sea"). Some of the people attempting the sea-crossing from Libya to Italy end up being rescued by the Tunisian authorities and brought to Zarzis or other Tunisian ports next to the Libyan border, in the Médenine governorate. When asked about what they do when they are informed about rescapés arriving at the port, a UNHCR officer explains: "We don't go to the port. We don't advertise it, we don't want people to ask protection only because they see the UNHCR logo: 90% are migrants, not refugees!" (see also Garelli & Tazzioli, 2017, p. 25). The screening of potential asylum seekers-one of UNHCR's core activities-is thus delegated to RC. As a result, according to IOM-Tunisia Officer 1, "only 15% of the rescapés are selected as potential refugees and transferred from RC to UNHCR. If they were disembarked in Europe rather than here, many more would be allowed to apply for asylum." Not only are most people prevented from applying, but those whose applications are rejected cannot challenge the decision. According to my UNHCR respondent, people "can file a first and a second appeal". However, as I learnt from rejected asylum seekers and NGO volunteers, no legal advice is provided in the Médenine governorate, so people are not even aware of this opportunity, and, more generally, of their rights.
Like in Egypt, UNHCR is not only involved in migration control with the Tunisian authorities, but it also allows for protection gaps to be created. Like in Egypt, this ends up pushing people to attempt the sea-crossing to Europe (Boubakri, 2015;Garelli & Tazzioli, 2017), mostly from neighbouring Libya. In 2013, in order to dismantle the Choucha camp at the end of the refugee crisis, UNHCR even offered the refugees who still lived there 1,500 dinars if they accepted to be relocated to Tunisian towns within an "urban integration" scheme, knowing that this was exactly the amount of money needed to go to the next Libyan port and take a boat to Italy, which is what most beneficiaries did according to many of my interviewees (see also Garelli & Tazzioli, 2017, pp. 23-24). The result was not externalisation but rather the opposite.
Finally, in both countries, UNHCR was unable to achieve one of its main aims: the adoption of an asylum law. For the 2015-2018 period alone, UNHCR Tunisia received 900,000 Euro from the EU for expanding and formalising the country's protection space, with particular stress being put on the adoption of an asylum law (Délégation de l'Union européenne en Tunisie, 2016). Indeed, the revolution had sparked hope "that Tunisia might eventually adopt an asylum law" (Ambroso, 2012, p. 7). In 2016, the law was ready but the Tunisian government had "other priorities," as both IOM and UNHCR officers told me. Four years later, the law has not been passed yet. In Egypt, instead, the situation never appeared likely to change (Ambroso, 2012), and the authorities do not even simulate an interest: "we don't need an asylum law, we already ratified the 1951 convention and the 1967 protocol", the Chairperson of the National

Coordinating Committee on Preventing and Combating Illegal
Migration and Trafficking in Persons told me. The scarce advancements made in setting up national protection systems in Tunisia and Egypt may partly be due to UNHCR's "lacking resources," as diplomatically argued by UNHCR-Egypt Officer 2. To a large extent, however, they are clearly the result of insufficient cooperation-or even opposition-from the local governments, fearing the potential effects of EU-externalisation of asylum to North Africa.
The accommodative attitude of UNHCR towards host countries is explained by the fear that a deterioration of the relations with these may result in the closure of the UNHCR country office, which may arguably be of even more harm to refugees. In principle, host countries (even those less inclined to host refugees) have an interest in being released from their responsibilities (i.e., assessing asylum claims and providing services for refugees) through UNHCR. This, however, only results in a limited bargaining power for the UN agency. Externalisation through UNHCR is only tolerated as long as it does not result in a transfer of responsibilities and burdens to the host government.
While the activities carried out by UNHCR may be, in theory, part of EU-externalisation efforts, their outcomes are in fact the result of complex dynamics between different actors, including UNHCR, donor states, asylum seekers, and authorities of the host country. These dynamics may end up producing effects contrary to those envisaged by externalising actors. According to Lemberg-Pedersen (2017), externalisation, besides being a reaction to migration and displacements that are already in place, and an attempt to prevent or confine them geographically, can also be "a cause of transnational displacement and forced migration in itself" (p. 31), which Lemberg-Pedersen calls "border-induced displacement." Here, ironically, the "displacement"

| Economic integration
Livelihood projects aim at integrating refugees into the local work market, and thus into the social fabric of the host country. Typically, these projects provide professional training and micro-credit opportunities for the beneficiaries to start an income-generating activity. This is not easy in Tunisia and Egypt because both countries impose limitations on foreign citizens (also including refugees) to obtain a work permit, with local employers required to demonstrate that no nationals of the host country could be employed instead of the foreign beneficiary.
In Tunisia, Caritas, Adventist Development and Relief Agency (ADRA) and Islamic Relief Tunisia (IRT) have all carried out projects in this field, but not all have been successful. A microcredit project by Caritas, funded by the French branch of a religious organisation, had been going on for one and a half year when I interviewed a Caritas representative, who claimed that the beneficiaries "are not reliable. As soon as they earn something, they disappear", that is, they use the money to organise the sea-crossing to Italy, rather than to integrate in the Tunisian society. This is also the case for most of those who receive contributions in cash (the alternative kind of financial support provided by Caritas Tunis): "After we gave them the money, we didn't see them anymore: they used it to go to Europe". Caritas' initiatives are aimed at making people stay, not at facilitating irregularised border crossings: "When they tell us that they are thinking about trying to cross, we try to make them reflect about it, but with little success." Caritas Tunis is not easily categorisable as an instrument of EU externalisation strategies, because its donors are not the EU or its member states. Therefore, it could be seen as a relatively autonomous actor of humanitarian containment. However, its activities produce, at least in part, an opposite, anti-externalisation effect.
Such a side effect was also produced by a UNHCR livelihood scheme run by IRT. People were trained and equipped to start an economic activity, but then "most of them sold the equipment and used the money to take a boat from Libya, so we had to stop the program," two IRT representatives explained.
A more "successful" project seems to be the UNHCR-funded one managed by ADRA, which had just completed its first year of life when I met an ADRA's representative. "We see the potential that people sell the goods, but we haven't seen this happening so far," he said.
Similarly, two NGOs/CSOs involved in livelihood schemes in Egypt (a UNHCR implementing partner and an organisation running projects funded by a EU country and a non-European country of the Global North) both present their projects as "successful." As an NGO/CSO representative explained: "Despite the risk that people may use the money for travelling to Europe, the success rate is high." For all interlocutors mentioned in this subsection, "success" means that people remain where they are. Their idea of integration is intimately connected with rootedness in the host country. The idea that economic integration may legitimately result in a higher degree of selfdetermination and increased opportunities for beneficiaries to fulfil their migratory projects by travelling further to Europe never emerged during the interviews. Despite their sympathetic attitude towards migrants, and regardless of whether their activities are funded by the EU or other sources, these people seem to perfectly embody the spirit of EU-externalisation. More broadly, their attitude seems to reflect the narrative of "migration management." Indeed, the multiplication and diversification of the actors involved in migration-related activities is an essential component of what has been defined as a "shift from control to management" (Kalm, 2012). This shift materialises in a technocratic and depoliticised approach to migration (Andrijasevic & Walters, 2010;Geiger & Pécoud, 2010Pécoud, 2015) that presents itself as pragmatic, realistic, humane, and benefiting all parties involved-countries of origin, countries of destination, and migrants alike.
The migration management approach has been criticised (Düvell, 2003;Overbeek, 2002) as a move aimed to serve "the interests of both developed states and the corporate sector" (Boucher, 2008(Boucher, , p. 1469. Arguably, a managerial approach to migration specifically facilitates the "voluntary" adaptation of countries of origin and transit, as well as (would-be) migrants, to the externalisation policies designed by destination countries of the Global North (Zaiotti, 2016). Thus, there is clearly a correlation between migration management and "externalisation." However, the perspective of migration management is less state-centric than the externalisation perspective. The migration management discourse more broadly supports the worldwide disciplining of mobility (Geiger & Pécoud, 2013). Analysing its logics allows to go beyond the observation of inter-state dynamics to address other relationships, such as that between the frontiers of capital and territorial borders (Mezzadra & Neilson, 2013), which also include, but are not limited to, state-driven externalisation.
This subsection suggests that, even in contexts dominated by the paradigm of migration management, externalisation attempts may fail, and prove counter-productive, because not everything can be "managed" "successfully." Some of the next subsections also suggest that there may still be some leeway for more critical NGOs/CSOs to act in a less depoliticised, "managerial" way. Research could shed light on such processes of repoliticisation by looking not only at destination countries of the Global North (Sinatti, 2019) or international waters (Cuttitta, 2018a) but also at countries of transit and origin, against the background of externalisation.

| Awareness campaigns
One of the NGOs/CSOs involved in livelihood activities in Egypt also carries out awareness campaigns for (would-be) migrants about the risks of irregular migration. Again, the words of my interviewee abide by the politics of containment and deterrence: "I don't think the sea is a good option anyway. It's a bad way of leaving the country. Our aim is to either do resettlements or keep people in Egypt until the crisis ends and they can return home". In this case, success may also mean returning to one's home country, or being resettled, whereby it is well known that resettlement is only a remote chance. 8 Other NGOs carrying out information campaigns have a more nuanced attitude. Some are even critical of the current restrictive border regime and of externalisation policies. In Zarzis, in the South-East of Tunisia, the members of the local fishermen's association Le Pêcheur account for almost 90% of the local fishermen. I had a collective conversation with some of them. As seamen, they are well aware of the risks of a sea-crossing, but they are also sympathetic with those who wish to cross, and of course, they would not let anyone drown or go adrift. On several occasions, they have provided food, water, or fuel to people they met on the high seas-these people did not want to be "rescued" and brought to Tunisia, because their boats were still able to continue the journey. In 2018, six members of the association, including the president, were even arrested by the Italian authorities-and released only after 3 weeks-for towing a migrant boat in distress towards Italian waters. However, normally, "if we find people in distress in international waters we call the international radio rescue channel, or we call MSF [Médecins Sans Frontières]: they have rescue ships operating in the area; they can rescue people and bring them to Europe", one of the fishermen explained. Since 2014, despite its sympathetic attitude towards the migrants' aims to travel to Europe, the NGO has explained young fishermen and unemployed people from Zarzis the dangers of attempting the sea-crossing. This is done within World Bank-funded projects that also aim at combating sea pollution and illegal fishery, as well as providing training in fishing techniques and other related activitieswhich shows that issue-linkage transcends the limits of state-state relations to penetrate IOs-funded schemes implemented by NGOs. In sum, Le Pêcheur simultaneously supports mobility from North Africa to Europe through its practices at sea and externalised containment through information campaigns.
Some of the NGOs/CSOs carrying out information campaigns are more openly critical towards the current restrictive border regime and towards externalisation policies. One is the Tunisian association La terre pour tous, whose president told me that "the responsibility for migrant deaths lies on the policies of the EU and its member states in the first place" (see also Tazzioli, 2018, p. 806). He is also very critical towards non-independent organisations: "I don't talk with those who only take money from the states", he said when I mentioned a large, government-friendly organisation and asked him if they have ever cooperated with them. Therefore, and although it cooperated with IOM by carrying out information campaigns within a large-scale threeyear (2012-2015) project, La terre pour tous is now carrying out such activities only as a voluntary service, outside any formal context. Its president thinks that his critical attitude is not in contradiction with information campaigns: "It is correct to provide people with accurate information. Then, it's up to them to decide whether to remain or to leave".
In Egypt, the head of an NGO/CSO has a particularly critical stance towards UNHCR and its subcontracting NGO partners, and thinks that the organisation's different funders-including the EU, private foundations, faith-based organisations and foreign diplomatic missions of European governments and other countries of the Global North-grant much more autonomy in the project implementation.
This shows that the "humanitarian marketplace" "within which states are able to select between different agencies competing for funding and contracts" (Betts, 2012, p. 57) also works the other way round, allowing for NGOs/CSOs to look for funding sources that are more appropriate for their aims. In its awareness campaigns the organisation "explain[s] people how long the crossing is, what the risks are, but also their entitlements, trying to give them the full picture, so that they can take an autonomous decision".
Interestingly, besides information campaigns, there are also other, informal ways of persuading migrants to give up their migratory projects. As I learnt from NGO/CSO representatives, some UNHCR's implementing partners working in Egyptian detention centres try to convince detainees to accept repatriation, even if this is not their job: "they tell them: why don't you go back home? They use exactly this phrase!". Given the repressive attitude of Egyptian authorities towards both migrants and NGOs/CSOs, the motivations for humanitarian workers to abide by the logics of containment by pushing people to accept repatriation may have less to do with the externalisation of the EU border than with the internalisation of the Egyptian one. Border scholars have pointed out that borders have become potentially ubiquitous (Balibar, 2009), and processes of externalisation go along with ones of internalisation (Menjívar, 2014). The outcomes of internalisation practices enacted by a 'transit' country may well be in line with those envisaged by externalisation practices of destination countries (as in the case of detained people accepting repatriation from Egypt), but internalisation is often enacted by local authorities independently from EU-externalisation. This suggests that the perspective of externalisation is not exhaustive of the dynamic potential of borders, which might be more comprehensively grasped through less univocal notions, such as that of "delocalization." The term delocalization (Cuttitta, 2018b) does not only fully explain the "mobile" (Amilhat Szary & Giraut, 2015) and elusive nature of borders, the ability of the latter to perform their functions away from their official demarcation lines. It is also more neutral than "externalisation" and "internalisation," insofar as it does not take a univocal perspective implying a clear-cut spatial dynamic (i.e., outward or inward) and leaves the question of "who borders?" (Rumford, 2006) open. This seems particularly productive in a context in which the multiplication of both state and non-state actors makes it increasingly difficult to clearly identify the bordering agent(s), and to assess in how far each of the actors involved contributes to determining the dynamics producing border shifts and changes.

| Cultural integration
In Egypt, NGOs/CSOs considering themselves "critical" and "independent" differentiate themselves from the others for advocacy or a political commitment inspired by-as a representative of an NGO told me-"the belief that those people have the right to cross the sea". Some have strong sentiments against UNHCR, which is considered complicit with both EU-externalisation and Egyptian repressive policies. However, not all of them refuse cooperation with the UN agency. This is particularly visible in the field of cultural integration.
The social theatre group Outa Hamra promotes societal change through its work with marginalised groups in Alexandria and Cairo. Its interactive theatre allows for artists and audiences from different nationalities to share their stories and experiences. The aim of its UNHCR-funded projects is, according to the UN agency, to stimulate social interaction and mutual understanding between refugees and host communities. Outa Hamra tries to go beyond this narrow objective of mere pacification between different national groups by stimulating its migrant and Egyptian audiences to put their relationship with power into question, thus raising political awareness of the context (Mercier, 2017).
Similarly, an NGO/CSO managed a 3-month project funded by UNHCR and Caritas to facilitate the integration of Syrian refugees through handicraft and music courses, as well as technology workshops and theatre shows: "We want them to feel that they are part of a community here. And then maybe they will be able to travel regularly, and go to Europe for tourism, or even for work, and maybe come back one day", a representative explained. The organisation wants to make sure that Syrian refugees are neither forced to leave nor prevented from leaving Egypt. Like Outa Hamra, its aim is to empower them, so they can make their own choice.
From the externalisation perspective, such activities should be seen as part of the efforts aimed at preventing people from continuing their journey towards Europe. However, the motivation of these NGOs/CSOs is different: they want to support migrants so that they can take by themselves any decision regarding their future. Consequently, the meaning of success is more flexible: successful integration activities may also produce effects contrary to those envisaged by the externalisation policies of the donors, putting migrants in the condition to travel further to Europe, rather than making them remain where they are. Contrary to the unintended effects of "unsuccessful" economic integration schemes or other UNHCR initiatives, (potential) anti-externalisation effects of these activities are consciously taken into account and may be seen as part of an effort to defuse typical externalisation devices such as UNHCR-funded integration schemes.
This attitude of challenging exclusionary policies can be read as an example of what Agustín and Jørgensen (2019) call "civic solidarity" (p. 41). Whereas solidarity, according to these authors, is both "relational and spatial" (p. 25), this arguably does not only mean that "alliances and practices of solidarity are constituted at different scales" (p. 36) but also that solidarity may produce spatial effects by facilitating movement across space. Some accept these conditions more passively, whereas others do not spare efforts to make all detention facilities accessible, or try to combine health care provision with some sort of advocacy. Similarly, some are supposed to assist only refugees and asylum seekers, and stick to this limitation, whereas others refuse it and assist people regardless of their legal status. 9

| Provision of health care and legal advice
Health care provision through NGOs relieves local state authorities from responsibilities in this field. Insofar, as it is (directly or indirectly) funded by European sources, medical aid through NGOs/CSOs allows for European authorities to provide care at a distance to those very persons that they are confining away from their territories. Thus, while being, more broadly, part of the humanitarian apparatus in charge of "managing the undesirable" (Agier, 2010) worldwide, health care provision through NGOs also shows a specific relationship with externalisation. Similar considerations are also applicable to the provision of legal advice (counselling) and support (representation in court).
Human rights are part of the humanitarian discourse that has become increasingly central in the management of migration across the Mediterranean (Cuttitta, 2018b). The role played by humanitarianism in externalisation processes, combining the logics of care and control vis-à-vis people on the move, is exemplarily reflected in the attempts to externalise asylum. The provision of legal aid through activities funded by the EU (e.g., a project managed by Terre d'asile in Tunisia, another one managed by an Egyptian NGO from 2013 to 2016) or UNHCR (two NGOs in Egypt were working for the UN agency at the time of fieldwork) may thus be seen as part of the process of externalisation of asylum. However, legal advice through NGOs/CSOs often goes beyond the externalisation of asylum to include issues that are not (or not directly) related to asylum: helping people out of detention, preventing or delaying deportations, obtaining certificates or permits, assisting in criminal proceedings, and so forth.
In sum, migrants' human rights in countries of transit and origin are supported by (EU) destination countries in a range of fields from health care to legal aid. From the perspective of externalisation, the relevant activities might be understood as part of integration measures aimed at "keeping them there." However, the idea that people whose most basic needs are met and whose fundamental rights are protected may be less motivated to travel further to Europe is not necessarily more convincing than its opposite, namely that healthier people enjoying more rights may meet less obstacles in their migratory projects and be more likely to attempt the sea-crossing. Therefore, these initiatives can be seen, more broadly, as part of the process of humanitarianisation of the border that builds on the respect of human rights and humanitarian standards to enhance migration management as a whole.
The growing body of work on the "humanitarian border" (Walters, 2011), analysing how logics of care and control coexist and support each other in the management of migration (Agier, 2010;Fassin, 2005;Franko Aas & Gundhus, 2015;Hyndman, 2000;_ Işleyen, 2018a;Little & Vaughan-Williams, 2017;Pallister-Wilkins, 2015;Perkowski, 2018;Ticktin, 2006;Williams, 2015), has prompted some authors to also pay specific attention to the role of NGOs/CSOs (Cuttitta, 2018a;Gerard & Weber, 2019;Lopez-Sala & Godenau, 2019;Pallister-Wilkins, 2017;Prokkola, 2018;Vandevoordt, 2017). However, most of the studies on NGOs/CSOs at the humanitarian border look at the territories of destination countries of the Global North, and-even in the few looking at countries of origin and transit (Sunata & Tosun, 2019)the relationship between NGOs/CSOs, humanitarian migration management, and externalisation remains largely under-researched. This is a promising avenue for further research, not lastly because addressing the role of human rights and humanitarianisation allows to include an analysis of Europe's normative power (Manners, 2002;Pace, 2007), and of its ability to legitimise European hegemonic neighbourhood policies (Del Sarto, 2016), in the analysis of externalisation. processes. This surely reflects the states' attempt to outsource to non-state, "neutral" actors activities they would not be allowed to carry out directly in foreign territories for sovereignty reasons.

| CONCLUSIONS
However, the paper also shows that EU-externalisation through NGOs/CSOs (and UNHCR) is not a univocal and smooth process.
Instead, externalisation is entangled in a complex network of actors and dynamics in the North African "border regime." It meets resistance from the authorities of third countries, as well as from the migrants themselves, who are able to turn externalised initiatives aimed at their immobilisation into instruments facilitating their mobility towards Europe. Motivations and practices of NGOs/CSOs may also contribute to undermining the effectiveness of externalisation.
Indeed, EU-Mediterranean regions are constituted through "interactive and dialogic processes" (Scott, 2018, p. 175), and these produce "varied and sometimes even chaotic results" (Collyer, 2016, p. 608) in the field of migration, revealing the "scattered nature of externalised border management" (Gaibazzi, Bellagamba, & Dünnwald, 2017a, pp. 9-10). The fact that NGOs/CSOs' donors include the governments of non-European countries (among others: USA, Australia, and Japan) further questions the picture of a uniform linear connection between externalising Europe and externalised North African countries. Moreover, NGOs/CSOs' donors may themselves be neither state actors nor IOs. Therefore, while confirming the relevance of externalisation as both a governmental policy and a heuristic perspective, this paper also tries to relativise it by pointing out the limits of univocal, Euro-centric and state-centric visions of externalisation.
First, the paper shows that state authorities in countries of transit and origin are not passive objects of Europeanisation: "the agency of the Europeanised" (Celata & Coletti, 2016, p. 19) is also important.
Moreover, the "Europeanised" do not only react to initiatives of their European counterparts, for example, by delaying ad libitum the adoption of asylum laws to oppose the externalisation of asylum. They also act autonomously, responding to domestic concerns, in ways that impact directly or indirectly on the mobility of foreign citizens (e.g., by discriminating against certain nationalities), with consequences that may be in line or not with the aims but are not a result of externalisation. European actors may then simply take advantage of restrictive policies adopted by third countries and rely on them (Badalič, 2019) to fulfil their externalisation objectives. In sum, a narrow focus on externalisation, when studying migration dynamics in the North African region, risks to give in to Euro-centrism and miss important parts of the picture.
Second, while states are arguably the main actors in migration and border regimes, and by definition the main actors of externalisation, the focus on NGOs/CSOs allows to look at EU-externalisation from a less state-centric perspective. Rumford (2008)  Although these findings suggest that it might often be impossible to identify clear patterns, further research may help shedding light on different specific relationships, for example, between the source of funding and the likelihood to work closely with the local governments.
Whatever the motivations and funding sources of NGOs/CSOs, the paper also shows that measures meant to support externalisation (e.g., livelihood projects) can produce effects contrary to those envisaged by externalising actors, insofar as they end up facilitate the seacrossing to Europe. Similarly, UNHCR practices (and, more generally, UNHCR's failure in externalising asylum in transit countries) may end up pushing people to leave for Europe. On the other hand, some practices of UNHCR and NGOs/CSOs may appear to support the host country's deportation regime and thus accomplish the externalisation goal of keeping people away from European borders. However, this may be regarded as the outcome of internalised borders of the host country, rather than EU-externalisation. This speaks to the flexible and ubiquitous nature of borders and suggests that more neutral concepts, such as "delocalization," may prove useful to grasp the complexity of regional border regimes' dynamics-a complexity that often "confounds the distinction between internal and external" (Perera, 2007, p. 208).
More broadly, the paper suggests that NGOs/CSOs, as well as UNHCR, can be seen as components of the process of humanitarianisation and managerialisation of the border-a process that sees humanitarian and human rights arguments and logics becoming more central in the global disciplining of mobility, and migration as a phenomenon that can be successfully managed through technocratic expertise. The perspective of "humanitarian migration management," while not in contradiction with that of state-led externalisation, provides a less state-centric view of the NGOs/CSOs' and IOs' roles.
The perspective of externalisation is, in itself, inevitably statecentric and Euro-centric. Seeing the centre-margin axis (Parker, 2008) within the broader context of the border regime, as this paper has tried to do, may help building alternatives to concentric (Casas-Cortés & Cobarrubias, 2019; _ Işleyen, 2018b) and state-centric views.

ACKNOWLEDGEMENTS
Research for this paper was funded by the Nederlandse Organisatie Regeni, who was killed for carrying out fieldwork in Egypt.

CONFLICTS OF INTEREST
I declare that I have no conflicts of interest.

ENDNOTES
1 The high degree of diversity among the non-state actors other than IOs begs the terminological question of how these should be defined. Indeed, "trade unions or professional associations […] often do not self-identify as NGOs, but rather as CSOs" (Tomlinson, 2013, p. 214), and some regard NGOs "as a subset of CSOs involved in development cooperation, albeit often one with no clear boundaries" (Tomlinson, 2013, p. 214). According to Bürkner and Scott (2018), "'CSO' […] denominates all forms of civil selforganisation that may assume formalised structures but often remain informal and temporary," as opposed to those "'non-governmental organisations' operating either at a national level within a formalised organisational structure […] or at a global scale as part of a larger organisation" (p. 11). This paper uses both terms, either jointly ("CSOs/NGOs") or separately. In the latter case, it follows Bürkner and Scott. 2 The concept of transit is contested (Düvell, 2012), for example, because it lends itself to obscuring the variety of migration patterns in countries that are not just of transit and to creating categories ("transit migration," "transit countries," etc.) used by policymakers to support their agendas. This explains the inverted commas.