“Biting our tongues”: Policy Legacies and Memories in the Making of the Syrian Refugee Response in Jordan

This article analyses the signiﬁcance of policy legacies and policy memories for refugee policy in conﬂict-neighbouring countries, where most of the world’s displaced live. Drawing on insights from critical policy analysis, it views refugee policy as co-produced by national and international agencies on the basis of previous dynamics that are already the product of an intense history of interaction and translation. This approach is illustrated by analysing two different aspects of refugee policy in Jordan: the process of counting Syrians in the country and the partial integration of Syrians into the formal labour market. Both examples reveal an overarching legacy of accommodation that ties international and host government actors together. Despite sometimes differing over preferred outcomes, the main goals for the various actors involved have been to strike compromises, safeguard organisational interests, and create outward policy success. In order to meet these goals, the agencies involved have learned to tolerate unresolved ambiguities and disregard other inconvenient legacies and memories that would only complicate policy negotiations. Acknowledging this intertwinement of agencies, technologies, and rationales of government is essential for rethinking policy change and responsibility in contexts of mass displacement.


INTRODUCTION
This article discusses the significance of the past for the making of current refugee policy. To do so, it makes a case for more fully appreciating the complexities of both past and present interaction between the variety of agencies involved in shaping such policies in practice. While conceptual discussions on refugee policy and governance in Western countries have become much more fine-grained in this regard, 1 they remain comparatively coarse regarding those parts of the world where most of the world's displaced live.
Explicit conceptual contributions on refugee policy in countries neighbouring conflict zones, which often depend on international aid and the involvement of international organisations, still tend to be framed in the language of "levels of analysis". They neatly distinguish between host state and international actors or policies when it comes to determining legacies that have shaped refugee policy. I show in this article that empirical observations belie such simple analytical distinctions, and argue for a different approach to analysing the significance of the past for refugee policies in countries of first reception. I argue that such policies are shaped by a number of policy legacies and policy memories, which are translated as they move across different arenas, policy fields, and time periods by actors that often cannot be neatly described as individual, national, or international. While the term "policy legacies" highlights the structural dimension, "policy memories" stresses the role of actors in making sense of the past, and in mobilising different narratives about it with a view to the present. Such a re-conceptualisation is not just of analytical value but also practically relevant. It changes the way we look at the production of refugee policies, as well as at the prospects of policy change.
The article develops this approach by exploring the making of the Syrian refugee response in Jordan, which many policymakers currently see as the most promising laboratory for putting in place new policies due to its leadership's ostensible willingness to cooperate with the international humanitarian and development community. Yet I argue that such changes cannot be engineered on the drawing board, as refugee policies in the country are shaped by a variety of legacies and memories inside and outside of refugee governance, which have created their own materialities. As will become apparent, the practices of international humanitarian and development agencies as well as donors are not external to these legacies but fundamentally part of them.
The article begins by discussing the state of the art of scholarship on refugee policy in most of the world with a view to policy legacies and memories, and discusses how the repertoire of critical policy studies can productively contribute to their analysis. It then briefly lays out fundamental policy legacies and memories in the Syrian refugee response in Jordan. Subsequently, it explores two instances of refugee policymaking in which such legacies and memories have played important roles: practices of counting refugees, which on the surface show relative continuity; and the ostensible recent policy change towards opening up the labour market for Syrian refugees. In both cases, I show that intertwined assemblages of local, national, and transnational agencies, ideas, technologies, and practices have created a variety of legacies and memories that have, in turn, shaped the prospects for policy continuity and change. While the first example demonstrates ambivalent learning processes of both Jordanian governmental and humanitarian actors based particularly on memories of the preceding Iraqi refugee response, the second one rather points to a neglect of policy legacies and memories in other relevant policy fields, which has limited the prospect for meaningful policy changes. Both examples reveal what I identify as an overarching politics of accommodation tying together international and host government actors, whose main goal has been to strike policy compromises, safeguard organisational interests, and create outward policy success. For the sake of achieving these goals, they have either been disregarding difficult legacies that would have complicated things or "biting their tongues", as a United Nations (UN) manager described as one of their main learning processes.
The article is based on extensive documentary analysis of policy documents, as well as around 75 interviews and informal conversations conducted with a variety of policy-shapers in English and partly in Arabic in four different research stays in Jordan of 1-4 months each between May 2015 and September 2017. These include middle and upper management of different UN agencies, international non-governmental organization ((I)NGO) representatives and staff, donor agency staff, project managers, journalists, researchers, and consultants, as well as staff of various government agencies involved in the refugee response in Jordan. Due to being generally perceived as a white, female researcher based at a European/UK University, my access to perspectives of international organisations and donors was comparatively easy, yet my familiarity with social and political dynamics in the country, and knowledge of the local Arabic dialect, also repeatedly opened up possibilities for in-depth conversations with staff of Jordanian government agencies and Jordanian NGO workers.

DETERMINANTS OF REFUGEE POLICY ACROSS
THE GLOBAL-NATIONAL DIVIDE When trying to conceptually grasp the significance of the past for refugee policy particularly in host countries bordering conflict zones, we can observe a bifurcation in much of the literature. One strand particularly deals with the legacies of humanitarian intervention as a global endeavour led by an international aid industry, which produces distinct types of governmentality, 2 or with the development of refugee policy at a global level. 3 The other primarily looks at determinants of host government policies, 4 and at how those relate to the global refugee regime. 5 Both have teased out important factors and dynamics that have shaped refugee policies over decades. Yet they have often adopted a macro-perspective that posits governments and international organisations as distinct and monolithic actors operating on scales that are taken for granted and seen as fixed, and have neglected the everyday dimensions of how refugees are governed in interaction between a range of relevant agencies. This has been the case even where crucial empirical findings point to the need for a different conceptualisation. In her ground-breaking article on host country policies, for example, Karen Jacobsen discusses street-level bureaucrats' motivations and their impact on government policy, as well as power struggles between different institutions and decision-makers as crucial factors shaping refugee policy that warrant empirical analysis, but still conceives of government as a unitary and confined actor in this process, which can be squarely located on the national level. 6 On the other end of the spectrum, many contributions subsumed under "global refugee policy" are in fact sophisticated empirical discussions of both host country and global regulations and their relation to each other, yet they do not explicitly problematise the conceptual implications of this. 7 There is, however, a rich body of critical analyses of refugee governance in specific spaces, which can help add a practice dimension into the discussion around legacies shaping refugee policy. Some analyses of encampment, for example, explicitly highlight how policies directed at refugees should be understood as spatially and materially embodied practices, which only become meaningful in day-to-day operations that bring together various international and national agencies, and stress the role of refugees in shaping what policies are in the end. 8 The recently burgeoning literature on refugee governance in urban, peri-urban, or rural settings is even more helpful in conceptually grasping such confluences and daily practices of governance. 9 Landau and Amit, for example, point to the common gap between legislated, official policies and actual practice. In order to understand this gap, they advocate focusing on bureaucratic autonomy on a sub-national scale, and for appreciating the variety of agencies involved and the daily dynamics of governance in a disaggregated, spatially sensitive way. 10 They also argue that refugee policy is determined by a variety of other policy fields, such as housing and labour market policies, which are often most relevant to urban refugees' lives. 11 These findings are echoed, notably, by authors who are simultaneously involved in the field as analysts and practitioners, 12 and by in-depth country analyses of refugee policies, which clearly show the variety of actors, agencies, and agendas involved. Often located in anthropology or geography, they point to broader legacies that have also left their mark on refugee policies, say the politics of inaction, or of keeping regulations or announcements ambiguous as a distinct mode of governing. 13 What is more, they make visible spatialised patterns of governance that further question the notion of fixed levels of analysis. 14 3. POLICY LEGACIES AND MEMORIES THROUGH THE LENS OF CRITICAL POLICY ANALYSIS There is a need to more squarely connect such analyses, which often conceptually draw on governance and governmentality studies, with debates about refugee policies and the role of different legacies and memories feeding into them. To do this, I draw on the repertoire of critical policy analysis, a body of approaches that questions many of the core assumptions of mainstream policy science -most notably its conceptualisation of policy as an evidence-based and linear process of creating specific measures with clearly defined stages and targets. Three of its re-conceptualisations of policy are particularly productive to grasp refugee governance in Jordan and elsewhere. First, to conceive of policy as a technology of government, which seeks to make populations legible and governable by creating specific bodies of knowledge around them, which in turn form the basis for political intervention. 15 Second, to understand policy as a struggle over contested meanings between a broader set of actors and audiences than that usually conceptualised as "decision-makers". Such "policy-shapers" may include legislators, but also various implementers, a variety of designated (supposedly technical) experts and consultants, as well as (potential) clients, who interpret policy messages in different ways for their own purposes. 16 Third, to see policy-making as a process and labour of assembling heterogeneous elements -technologies of government, actors, discourses, keywords, objects, as well as quotidian practices, bureaucratic logics, and individual desires. While these elements carry different histories with them, these are recombined when travelling from place to place, and take on new meanings during the process. 17 Based on such an understanding , the ways in which past policies affect the composition and shape of current ones is much more open than in historical institutionalist accounts and associated notions of path dependency, which have dominated traditional policy sciences. 18 Policy legacies can derive from multiple and overlapping regimes or policy fields, 19 or from the history of single elements, technologies, or paradigms that have developed a life of their own, and which policy-relevant actors draw on in processes of bricolage. 20 Finally, this framework makes it possible to highlight the subjective dimension of policy legacies, and bring in the personal histories of policy-relevant actors, or the ways in which they (selectively) remember past events or procedures when designing or implementing policies. 21 In order to underline the relevance of this dimension, and more broadly of agency and practices in shaping policy, I use the term "policy memories" in conjunction with that of policy legacies.
Regarding (forced) migration, such memories -individual as well as collective acts of social construction that are geared towards the present -often refer to the imagined community of the nation and its past. 22 Histories and memories of (forced) migrant reception are crucial in moulding the strategies of various actors and agencies involved. They may shape red lines and constraints of policy-shaping directly or indirectly. Nationalist and nativist discourses may evoke images of past inter-group homogeneity, which needs to be defended against newcomers, rather than referring to previous successful forms of integration. 23 Yet even those memories of migrant integration that may be framed as "local" (i.e. embedded in the specific country history) usually relate to policy assemblages that were always already transnational in nature. This is particularly important to acknowledge in the context of refugee-hosting states neighbouring conflict zones, which have often been shaped by international agencies' involvement in post-colonial state-and nation-building. It is also pertinent in a field like humanitarianism and, more broadly, international development, where policy blueprints circulate globally and professionals often refer to experiences made elsewhere when making sense of the context they currently find themselves in. 24 In what follows, I thus use the terms policy legacies and policy memories in order to grasp the significance of the past in policy-shaping processes, specifically with regard to refugee-related policies. I use both terms together, in order to stress the broader, structural aspects that make up the institutional setting in which current refugee policy-shaping takes place, and to highlight that these legacies, as well as specific historical instances, only ever become meaningful through individual as well as collective (institutionalised or broader discursive) reconstructions by a variety of policy-shapers across scales. I also single out how particular legacies have been remembered or (more or less strategically) forgotten, and how this has shaped the response to Syrian displacement to Jordan.

THE SYRIAN REFUGEE RESPONSE IN JORDAN:
LEGACIES AND MEMORIES The range of institutional actors involved in refugee policy addressing Syrians in Jordan is broad, and marked by institutional overlaps and competition. This goes both for the array of Jordanian governmental agencies involved, where for example, the Prime Ministry, the Royal Court, the secret service, different units in the Ministry of Interior, as well as in the Ministry of Planning, and in various line ministries, vie for decision-making authority 25 and push for different policy measures, and for the even wider range of international agencies that compete with each other in seeking to secure significant roles in the response, determine its overall focus and orientation, and occupy distinct fields of activity. 26 In the latter case, conflicts were particularly discernible between international NGOs and UN organisations about roles and performance in the protection regime, 27 as well as the capacity to deliver targeted assistance in specific fields; in the tug-of-war between The United Nations High Commissioner for Refugees (UNHCR) 28  27 Many international NGOs, for example, see UNHCR as sacrificing important protection issues for the sake of appeasing the government, as all-dominant, and as mistakenly following a one-size-fits-all script in every new context. Interlocutors also perceived the ILO as neglecting protection issues in the strive to employ as many Syrians as possible. E.g. Interview with Economic Recovery Manager, DRC, May 2017. 28 The United Nations High Commissioner for Refugees (UNHCR) is legally bound -by international law, and a Memorandum of Understanding (MoU) with the Government of Jordan (GoJ) -to determine the status of asylum-seekers and provide protection to refugees. In practice, it has taken also the institutional lead in the refugee response, and coordinates the Inter-Agency Task Forces, which seek to organise all agencies involved into sectors according to a global blueprint for a refugee response. Other agencies involved might serve as implementing partners or operational partners of UNHCR in specific sub-fields. 29 The UN office for the Coordination of Humanitarian Affairs (OCHA) is the part of the UN Secretariat responsible for bringing together humanitarian actors to ensure a coherent response to complex emergencies and disasters (whereas UNHCR is supposed to play this role in refugee crises).
between UNHCR and The UN Development Programme (UNDP) 30 on operationally leading that response, as well as on providing humanitarian assistance versus a more developmental orientation. This fractured institutional landscape and the policies enacted in it are made sense of and take shape through the selective mobilisation of past experiences of (governing) migration and displacement. After all, Jordan has a long history of hosting forced migrants. Large population influxes of Palestinian refugees (in 1948 and 1967), Palestinian-origin Jordanians who had lived in the Gulf for decades (in 1991), and Iraqis (in the 1990s as well as after the toppling of Saddam Hussein in 2003), as well as smaller influxes of Lebanese, Libyan, and other groups, have shaped society, politics, and economy in the country. In effect, Jordan is among the top two countries globally with the largest number of registered refugees in relation to its overall population. 31 Given their size and presence in the minds of policy-shapers, the Palestinian and the Iraqi influxes are most relevant when it comes to policy legacies and memories. However, this presence has different characteristics. As mass displacement from Iraq to Jordan is relatively recent, a number of policy-shapers responsible for the Syrian response in the country, at least in the first years of the response, had also been directly involved in managing the Iraqi file in Jordan or elsewhere. This includes highranking UNHCR managers, 32 government representatives, 33 as well as consultants and NGO workers. Some policy-shapers thus have personal memories of, and professional experiences with, governing Iraqi displacement. Yet stories about the Iraqi response are also told to newcomers who come to work with international NGOs or UN agencies in Jordan, and government representatives who were not then in positions of responsibility call upon received wisdom as well. 34 In comparison, the Palestinian influx is largely present in an indirect way. This does not mean that it is less relevant. One international consultant with long experience in the country aptly summarised the significance of this policy memory: "The Palestinian experience is not often brought up, but it is the narrative that lies underneath all of it." 35 According to common estimates, around half of the population of Jordan, likely more, is of Palestinian descent. The permanence of their presence in the country, and their status as "refugee-citizens", 36 have shaped political discourses and policies about refugees in the country. The experience has, for example, resulted in a profound reluctance among Jordanian government agencies to let another group of people become a permanent population, thereby further altering the precarious population balance in the country. This has played out directly in the Government of Jordan (GoJ)'s hardening stance, and eventual refusal, to let Palestinian refugees from Syria enter the country. But the experience of Palestinian displacement to Jordan has also played into other aspects of the Syrian response. A personal aspect shaping policy memories and thus approaches to the Syrian refugee presence is that many NGO and local UN staff in Jordan are of Palestinian descent. For many of them, this has created a sense of connection and duty through a shared experience of being refugees. 37 At the same time, it has also generated a dynamic of othering, which locates Palestinian Jordanians in the national collective or a collective of urban professionals dealing with a group of new outsiders. These tend to be either framed as victims or, as current governmental and media discourse would have it, as a burden on national resources. 38 Policy memories of Jordan hosting both Iraqi and Palestinian displaced, whether personal or narrated through others, thus take their current shape through a filtering process, in which the logics of the agencies people have been working with, as well as common media discourses, have shaped much of the resulting narrative.
Both the Iraqi and the Palestinian influxes have also never been purely "national" (Jordanian) crises, but have involved global or international actors, agencies, and policies. Consider the role of The United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNRWA) 39 ,which remains a permanent core feature of the humanitarian landscape in Jordan and the broader region until this day. Given not only its role in education and health care provision, but also as one of the largest employers for Palestinian refugees in Jordan, it has become a "quasistate institution", 40 which foils any attempt to neatly differentiate between global and local agencies. To a degree, this also goes for UNHCR, which has been physically present in Jordan, and assumed different roles in refugee influxes to the country since 1990. 41  endangered, 43 and the memory of almost being expelled from the country has shaped the politics of accommodation that, as I highlight further down, has characterised interactions between UNHCR management and Jordanian governmental agencies ever since. Notwithstanding their changing fortunes in varying political circumstances, the crucial role of such international agencies is very much part of the Jordanian legacy of governing refugees. This is not an exclusive Jordanian legacy though. It forms part of what is often referred to as "the great compromise" of global refugee governance. 44 According to it, governments in many conflict-neighbouring states have accepted hosting and not forcefully returning large groups of refugees in return for devolving a large part of the required funding, as well as responsibility for registration and material assistance to refugee populations, onto such international actors. 45 Refugees themselves have become relatively inured to this division of roles. 46 This aid regime has also intertwined with a neoliberal framework. In the Syrian response, the greater part of service provision, particularly in the larger refugee camps of Zaatari and Azraq, is left to a host of non-governmental agencies, and government intervention is limited. 47 Outside of the camps, government schools and health centres have played a much stronger role. However, self-reliance for Syrians has become an important catchword both inside and outside the camps, implying the need to provide for themselves in the context of dwindling humanitarian assistance, and a lot of programming and funding has shifted in this direction. 48 In fact, there was never a neat separation of roles and responsibilities between the Jordanian host government and international organisations, neither regarding security issues, which have preoccupied governmental agencies since the Palestinian influx, nor regarding the provision of infrastructure, services, and funding. This has created an assemblage of overlapping and parallel institutional structures of refugee governance, in which different agencies and technologies of care and control have intersected. In the following, I explore two examples of these confluences, highlighting different memories and legacies at play.

COUNTING SYRIANS AND SYRIAN REFUGEES: AMBIVALENT
LEARNING PROCESSES One of the most contentious issues in the process of requesting donor support, as well as for the actual programming and planning of any form of international intervention -that is, for making refugee policy -is determining how many people in need there are in the first place. As Harrell-Bond et al. already noted over two decades ago, "since the time international humanitarian agencies became involved in assisting refugees in developing countries, this seemingly reasonable requirementthe need to count refugees -has, to a significant extent, dominated policy, planning, implementation and evaluation". 49 This policy relevance has made it appealing for various agencies with differing agendas -not just humanitarian ones -to be involved in producing refugee statistics. This, in turn, has contributed to making such statistics relatively unreliable and disparate. 50 This is also the case in Jordan, where a variety of governmental and international agencies have engaged in efforts to count Syrians in the country, with very different results. The ways in which Syrians have been counted, and the ways in which the results have been subsequently used, demonstrates a fundamental lesson that policy actors learned from the Iraqi response: do not confront each other over numbers but let them co-exist in silent disagreement. This has reinforced an overarching politics of mutual accommodation that has characterised interactions between humanitarian organisations and hosting governments in much of the region since that response took shape in the mid-2000s. 51 A number of fundamental dynamics of government connected to (refugee) population counts have tied together different agencies involved in refugee governance in Jordan and elsewhere. First, counting populations constitutes one of the prime forms through which modern government makes populations legible and governable, and expands governmental reach and authority. 52 This equally applies to refugees, and concerns host governments as well as humanitarian agencies. Refugee counts make it possible not only to control their whereabouts but also to claim responsibility for (taking care of) the concerned populations, and to assert authority over them. Care and control are closely intertwined in humanitarian interventions. 53 For UNHCR, this has, for example, long meant prioritising refugee camps over refugee selfsettlement in urban contexts, as in the latter case UNHCR has to rely on government data. In Jordan, regaining control over the numbers was one of UNHCR's main motivations for setting up Zaatari camp, in contrast to the purely urban presence of displaced Iraqis. 54 Data collection does not only enhance governmental reach overall, but also the relative political weight of the specific agencies that engage in the process. applies to struggles for dominance in the refugee response between humanitarian versus governmental agencies as well as between each of these groups of policy actors. The former director of Jordan's Syrian Refugee Affairs Directorate (SRAD)a directorate of the Jordanian Ministry of Interior created in 2013 to oversee the Syrian presence in Jordan -noted, for example, that other government agencies raised their own numbers, but asserted that they were supposed to simply adopt the SRAD figures, citing a Prime Ministerial decision. 55 A corollary of this attempt to increase control and political clout through refugee counts, which is equally of relevance for humanitarian and host government efforts, is securitisation. Indeed, the Syrian response in Jordan illustrates how data collection about refugees, technological innovation, and securitisation are closely connected. The UNHCR-driven introduction of the Iris scan technology for refugee registration and recognition, as well as the Jordanian governmental "upgrade" of their identity cards for Syrians to include biometric information, are a case in point. In both cases, new technologies have intensified control and increased the relevance of the respective agencies. The SRAD, for example, was initially only responsible for camp security, but with its introduction of biometric identity cards, its jurisdiction was expanded to also include non-camp settings from 2014. 56 At the same time, the hype around new technologies such as the Iris scan has demonstrated the lack of concern for the risks it puts the screened population under, and the ways in which such technologies can violate their rights. 57 Funding is also a shared concern. It is not just governments that seek to access and justify international assistance by pointing to the numbers of refugees, as some recent work seems to suggest, 58 but also humanitarian actors. 59 As part of the established game of appealing for humanitarian funding, agencies like UNHCR have reiterated a narrative that depicts refugees as a grave burden on hosting countries and have advised hosting governments to use higher population estimates in their appeals to reinforce this message in Jordan and many other cases. 60 As Sophia Hoffmann thus summarises: "UNHCR and the [. . .] Jordanian government [. . .] share, in particular, two interests: to raise humanitarian funds and to centralise information and control over refugee populations." 61 Notwithstanding these broader shared concerns, host governments and humanitarian agencies are not monolithic entities. The various concerns or interests related to refugee counts can differ between as well as among hosting governments and international agencies, depending on the organisational rationales of these agencies. example, while management of the (Jordanian) SRAD stressed their need to know about the number of Syrians and their whereabouts for security reasons, Ministry of Planning staff rather highlighted the burden Syrians placed on the resources of the country, bolstering this claim by referring to the highest circulating numbers, and pointing out the need for more international funding. 63 UNHCR managers, in the meantime, claimed to be primarily interested in the numbers so as to be able to adequately respond to Syrians' needs and demonstrate "evidence-based programming", 64 which links up to the care and control dimension mentioned above. Yet as we have and will again see below, funding is also an important concern for humanitarian agencies. Differing figures can thus reflect as well as help with accommodating a variety of rationales present within one agency or between different ones.
Such and the census and highlighted their political nature. 69 A recent report comparing and re-calculating available figures also concluded that the census significantly overestimated the number of Syrians in the country. 70 Finally, the higher figure is put in question by the results of the urban verification exercise, a process of (re-)registering all Syrians residing in the country (i.e. not just refugees) 71 led by the SRAD in cooperation with the UNHCR from early 2015. Despite various attempts to remove fees and other bureaucratic obstacles to increase numbers, the total number of Syrians registered with the SRAD -a prerequisite for accessing government services -never exceeded that of Syrian refugees registered with UNHCR. 72 Despite these widespread doubts and counter-evidence, the government's estimate of 1.25-1.4 million Syrians in the country has become the dominant reference point in Jordanian public discussions, at least in the sense of not being openly contradicted. This is the result of a complex assemblage, in which policy legacies and memories play an important role. One decisive factor in this was that since late 2013, the Jordanian king, Abdullah II publicly declared the number of Syrians in the country to be around 1.2 million, and later even more. 73 As an interviewee involved in the negotiation processes around the figures put it: as well as those who came as seasonal labourers and stayed permanently once the armed conflicts began. When the king says "this is the number", nobody will challenge the king. And that was one of the challenges, some didn't see it as a challenge; they saw it as a convenient number that was also politically useful to then build into their work; but for others who tried to actually arrive at an accurate number, that presented a problem. 74 This observation attests to the legacy of and memories associated with institutional authority, which shape not only refugee policy but most policy fields. It points to the significance of monarchic rule in Jordan, where one can openly criticise any government but where ordinary people as well as public figures, including politicians, usually refrain from contradicting the king, and use his utterances as guidelines for permissible public speech. This is not just due to the fact that criticising the king is legally criminalised, but also due to the well-founded concern that it may result in being excluded from the distribution of offices or perks. 75 Besides the "royal factor", the widespread sensitivity among both Jordanian and international agencies about the numbers of Syrians harks back strongly to previous experiences of counting refugee populations in Jordan, most of all that of counting Iraqis, which was marked by controversy. In 2007, the GoJ requested the Norwegian research institute Fafo to conduct an assessment of Iraqis in the country. Fafo, based on a sample survey, came to the conclusion that the number of Iraqis in the country was around 161,000, but given that this conflicted strongly with the 450,000-750,000 estimate cited by the GoJ at the time, 76 it was then pressed by the GoJ to reassess its estimate based on entry and exit numbers as well as the number of Iraqi mobile phone subscriptions. In the final report, Fafo estimated there to be between 450,000 and 500,000 Iraqis in the country. 77 This precedent is relevant in several ways. Methodologically, we can see an overlap in the usage of entry and exit numbers to estimate Syrians in the country today, but also in the criticism that this method is unreliable. Analysts agree that the higher estimate reluctantly given by Fafo at the time was significantly overstated. 78 This overstatement, however, reinforced Jordan's claim regarding the immense burden it 73 E.g. Petra News Agency, "His Majesty King Abdullah II Addresses the London Donor Conference", 4 Feb. 2016, available at: http://www.jordanembassyus.org/blog/his-majesty-king-abdullah-ii-addresses-lon don-donor-conference (last visited 20 Jan. 2020). 74 Interview, summer 2015. This, my interlocutor reported, was particularly the case for the SRAD, which had articulated a clear preference for simply registering and thereby recording all Syrians in the country. I assume that in order to maintain its position of relative dominance, and in order to accommodate the King's statement as a tacit guideline, it nevertheless officially adopted the estimate-based higher figure. was shouldering, and served as grounds to request funding from the international community, which had already poured in generously shortly before but did even more so after the assessment. With the US as the most important donor, followed by the European Union and others, funding greatly exceeded the aid given to other refugee crises in the world at the time, a clear indication that it was motivated by both the political strategy to support Jordan as a major ally, and the desire to specifically assist Iraqis displaced by the 2003 US intervention. Part of the assistance was given directly to the government as institutional and infrastructural support, which led to more government concessions regarding the hitherto precarious status of Iraqis in the country. 79 Besides the government and UNHCR, whose budget also grew immensely thanks to the perceived scope of the refugee crisis, using a high number was also convenient for financially hard-pressed NGOs in the country. In view of the funding available for assisting refugees, the latter switched the focus of their efforts to providing emergency relief for Iraqi refugees, and could use those numbers in their proposals. 80 An independent consultant who had been professionally involved with both the Iraq and the Syria refugee response in Jordan tied the two experiences together by calling the Iraqi population count "the great lie", and the Syrian one "the secondgrade lie". 81 Whether "lie" or "bargaining strategy" is the more appropriate term, the experience of counting Iraqis led to different processes of policy learning: from the perspective of the Jordanian Ministry of Planning (MoPIC) and other government actors, the strategy of setting very high numbers for migrant/refugee populations was relatively effective. In the Iraqi case, it brought in a lot of funding for undeniably pressing challenges in local development, which did not receive as much attention as the plight of Iraqi refugees. 82 The same strategy is now pursued with regard to Syrians. Whether seen as a major success or as "still not enough", 83 the Iraqi experience seems to suggest to many GoJ actors that using high estimates for Syrians remains fundamental to receiving high levels of international aid.
Outside of the GoJ, other learning processes can be noted. While the inflated numbers of Iraqis in Jordan substantially increased the budgets of UNHCR and its (I)NGO partners, it also created considerable planning problems. An (I)NGO director involved in the response at the time described the intense competition the unexpectedly low number of Iraqis produced between agencies, who were supposed to include a substantial percentage of them in their programming: "People fought over Iraqis: 'that one's mine, I got him first'." 84 Another former (I)NGO worker recalled: In 2008, both UNHCR and partners were looking for these lost Iraqis. [. . .] We called them at the time "the invisible Iraqis"; we know they are here but we can't quite put our finger on where they are. And so a lot of the planning figures were based around registered Iraqis, but also the assumption of unregistered Iraqis. [. . .] Sure enough, they never found the missing 450,000 Iraqis. [. . .] As it got more and more clear that there just didn't seem to be this many Iraqis, everybody was a bit [upset] -everybody, including the donors that also bought into this. I remember the head of USAID saying at one point: "we have never paid so much for so little". 85 In consequence, UNHCR and its partners now rely exclusively on their own figures for planning purposes. However, they do not openly contest the government numbers but "bite their tongues", as a UN manager remarked. 86 The former UNHCR country representative explained: One thing which we have learned from Iraq is: don't get into a battle over something that is not going to make a significant change. So, if the government estimates their number, that's fine. We acknowledge that, and we just say how many are actually registered with UNHCR. 87 This politics of accommodation is not something unique to the case of Jordan, or the Syrian response. It is part of a common strategy of humanitarian agencies to deal with host governments in a way that seeks to avoid open confrontation over numbers or other disagreements for the sake of other potential wins. 88 Allowing for different numbers to circulate can, from the perspective of humanitarian actors, also simultaneously serve the goal of maximising funding, and that of targeted planning and control. From the perspective of GoJ actors, the existence of a differing UNHCR figure has equally made it possible to officially adhere to "their" figure and still accommodate conflicting priorities, especially related to security versus funding.
We cannot say for certain how significant the inflated numbers have been for international funding commitments to Jordan during the Syrian response. Factors such as the overall geostrategic relevance of the country, donors' assessment of the GoJ as a comparably easy government to deal with, their eagerness after the 2015 summer of migration to prevent further migration movements, 89 and the GoJ's commitment to partially opening up its labour market to Syrians have certainly also played a major role in these significant commitments. 90 What the higher numbers arguably did is provide a pretext for continued funding for those who needed one. The acceptance of different co-existing numbers by all sides, and thus the deliberate tolerance of ambivalence, has, meanwhile, made it possible to accommodate differing rationales, which are part of refugee policies more broadly, but also reflect specific learning processes. This demonstrates that refugee counts and the associated politics of aid are not just driven by host governments, but that this dimension of refugee policy is borne out of interaction between a variety of agencies and technologies of government. A broader pattern of mutual accommodation -by now a policy legacy in itself -has moulded not just the strategies adopted by humanitarian actors, but the interaction between them and host governments, and has shaped involved agencies' policy memories.
6. WORK PERMITS FOR SYRIANS: NEGLECTING POLICY LEGACIES FOR THE SAKE OF COMPROMISE While numbers underlie the articulation of policies, the question of the economic (non-)integration of Syrians has formed a core aspect of these policies to date. During its first five years, the response to Syrians' displacement to Jordan was characterised by an unofficial yet clear reluctance to grant Syrians formal access to any means to gain a living. This red line has shifted though, and since the "Jordan Compact" announced in February 2016, the GoJ is officially committed to providing formal access to the labour market to Syrian refugees in return for substantial donor support to its economy and infrastructure. This is a notable policy shift on the surface, however a number of policy legacies and memories have limited its successful implementation. While some of those stem from previous refugee responses, many come from outside of the policy field, and relate to broader contours of and everyday practices in the Jordanian political economy and labour market. These legacies and memories, which were largely neglected in the process of policy development, have both shaped the long-term refusal to let Syrians work, and limited their formal labour market integration since 2016.
To start with the legacies and memories of other/previous refugee influxes and responses: the socio-economic integration of refugees is a bone of contention across the world, as many host governments fear this would allow them to settle down more permanently. While refugees may be legally granted the right to work, they often face bureaucratic hurdles, which in effect makes it hard to impossible to actually get a work permit or permission to establish their own business. 91 This was also the case for Syrian refugees in Jordan. 92 Here, the reluctance to grant them formal labour market access was strongly mediated by the experience of Palestinian displacement. Despite the fact that Palestinians' labour market integration has actually been quite successful, 93 the overriding concern of Jordanian policymakers has been to avoid letting other population groups become permanent, further "crowding out" the Transjordanian 94 population part. As Dallal Stevens points out: "The association of "refugee" with Palestinian, and hence with permanent residency, has patently distorted the discourse in Jordan. Jordan's fear of creating semi-permanent residence for another large Arab cohort is very real". 95 In effect, a policy legacy that could have been remembered as a success story to be replicated was disregarded, and substituted by policy memories of Transjordanians being demographically as well as economically side-lined. This has resulted in a discursive and structural focus on temporariness and a reluctance to integrate newcomers socio-economically. Creating obstacles for legal employment epitomised this tendency in the first years of Syrian displacement. 96 While this was tolerated by and large by the donor community for the first five years, the substantially increased Mediterranean crossings to Europe in 2015 intensified incipient discussions between UNHCR, British academics, DFID, British and other EU donor governments about a "Jordan Compact" supposed to better support Jordan as a major hosting country for Syrian refugees, and placed more pressure on the Jordanian government to change its position. 97 The Compact announced in 2016 -in line with discussions on the Global Compact on Refugeespromised to open up the Jordanian labour market to up to 200,000 Syrian refugees in a number of sectors, in exchange for major support from the international community not just for the refugee response in Jordan, but for economic development in the country more broadly. It sought to accomplish this through a number of measures, among them the provision of a major World Bank concessionary loan usually reserved for least developed countries, an amended trade agreement between Jordan and the EU, and additional grants and loans. For the first three years, efforts at labour market integration primarily focused on providing Syrian refugees with work permits. 98 The agencies involved in creating these policies largely disregarded a number of legacies in policy areas which have been ever more strongly linked with forced migration policy through the Jordan Compact -particularly labour market policy and economic, including trade, policies. An almost wilful exclusion of policy memories regarding those legacies for the sake of creating a workable compromise between different Jordanian and international agencies, has meant that the Jordan Compact did not factor in the specificities of these fields in Jordan; a dynamic that has made its implementation much more difficult than anticipated. 99 This can be illustrated by looking at the initiative to simplify the rules of origin (RoO) that govern the EU-Jordan Trade Agreement, which came into effect in July 2016. This initiative deliberately integrated trade policy, and indirectly other economic policies with forced migration policies. 100 It did so by lowering thresholds for the percentage of local input required to export to the EU under reduced tariffs. In order to benefit from these lower tariffs, exporting companies need to be located in a number of Special Economic Zones (SEZs), and employ at least a certain percentage of Syrian refugees. The expectation shared by many involved actors was that by lowering the required percentage of local value added, existing businesses would be able to expand their operations and additional investors could be attracted to a number of these zones in Jordan, which would not only create jobs for Syrians and Jordanians but also increase Jordan's export base, particularly in manufacturing. 101 Yet despite the high profile of this initiative, take-up has been very low to date, with only 13 companies successfully registering for an authorisation to export to the EU, and employing just over 1,000 workers in total, among them 291 Syrian ones by March 2019. 102 Various legacies play into this. Special Economic Zones, for one, have a long and problematic history in Jordan, as they do globally. Since the mid-1990s, the country has experimented with different types of Special Economic Zones and preferential free-trade agreements. Much like past such attempts, the recent focus on zonal development has been pushed by a variety of agencies ranging from the World Bank, the EU and other donors, international academic experts and globally operating companies, to Jordanian governmental actors. Support was partly based on these actors' established belief in SEZs as a panacea for attracting more investments and reforming economies. 103 The scheme's limitation to SEZs was also due to political concerns among both EU negotiators, who did not want to see the agreement override the recently concluded pan-Euro-Mediterranean (PEM) Convention, and Jordanian politicians, who saw this as a convenient way to limit effects of Syrian employment on Jordanian labour, which is rare in those zones. 104 Focusing on SEZs thus appeared as a workable compromise between the various agencies involved in designing the scheme. What they did not take into account, however, was the limited success of zonal development strategies. While some of the mentioned initiatives in Jordan have attracted foreign investors and raised export levels (especially in the garment sector), they have not created positive linkages with the country's economy and labour market as a whole. 105 This policy memory, which is well-established among economic experts working in and on Jordan, was left out of the discussions leading up to the scheme.
Yet even if the temporarily revised trade rules were to be applied beyond the Special Economic Zones -a prospect highlighted in the Dec 2018 revision of the agreement -106 it is questionable whether this would help expand exports to the EU. In designing the agreement, international as well as Jordanian agencies over-assumed a readiness of Jordanian as well as international investors in the country to export to the EU and employ Syrians to this end. Thereby, they again deliberately neglected a number of policy legacies and memories that contradicted these assumptions. Various factors, such as notoriously high transportation, shipping and electricity costs, taxes and fees, have hampered export-oriented investments for decades. Local manufacturers and private sector representatives, who were excluded from negotiations around the trade agreement, 107 have also noted the uncertain regulatory environment, the lack of long-term economic policies, and the prioritisation of quick fixes as major obstacles. Moreover, they have pointed out a long-standing import-rather than an export-orientation in Jordanian economic policy making. 108 Such aspects have resulted in manufacturing exports being largely limited to garments, an industry that does not require much local input or involvement, and that has mainly flourished due to a preferential trade agreement with the US, which has attracted South Asian companies to producing in Jordan's former Qualified Industrial Zones (QIZs). 109  The amended agreement, which was a response to the limited take-up, stipulates that the scheme would be extended to the entire territory of Jordan once 60,000 refugees are lawfully employed across the country -a number that is lower than the initially envisaged 200,000 but much more realistic, given that in the end of 2018, only 42,000 work permits for Syrians were estimated active. It also maintains the lower threshold of employing 15 per cent of Syrians, rather than raising it to 25 per cent in subsequent years, and limits the employment quota to the production lines geared towards EU export, rather than the factories' labour force overall (see the economic mainstays of the country. In consequence, a British diplomat involved in the negotiations conceded that "Europe has basically opened its doors for products from a country that doesn't really produce anything" 110 -a testimony to the consequences of wilfully excluding policy memories on these long-standing production and trade dynamics during the policy design process. 111 This also concerns the long history of unequal trade relations between Jordan and the EU, which was only very partially addressed in the negotiations. Jordan's exports to Europe have been minuscule in the past not only because of the previous RoO regulations, but also due to factors such as complicated certification processes, which many Jordanian companies feel unable to comply with. 112 Despite a palpable wish to make the trade initiative create actual changes in export levels, and an awareness among EU negotiators and member states that Jordan's export levels were so low that this would hardly be threatening to member states, such dynamics were not explicitly tackled. Moreover, potentially promising product groups like agricultural and food products were excluded due to member states' reluctance to opening up this fiercely protected sector. 113 In consequence, Jordanian exports to the EU have barely increased since the conclusion of the trade agreement, 114 despite various donorsponsored projects aiming to support export oriented manufacturing. Another legacy that was firmly neglected in revising the EU-Jordan trade agreement is the highly segmented nature of its labour market, and the skills and experiences Syrian refugees in Jordan actually have. This has particularly complicated the attempt to recruit Syrians to work in the garment factories, which are located in a number of the former QIZs. 115 EU negotiators, as well as Jordanian governmental representatives expected this to be a quick and politically easy win, which would get many Syrians, particularly women, into jobs. This was because the sector is relatively labour intensive, with a labour force that overwhelmingly consists of non-Jordanian female workers, so that there would not be much competition with local labour. 116 The prospect of a jointly agreed pilot scheme between Jordanian governmental actors and international negotiators led them to neglect long-standing segmentations within the labour market, where the chances of finding work in specific sectors depend not just on whether a worker is local or foreign, but also on their nationality and gender. 117 Employers have very specific perceptions about the characteristics and qualifications of certain groups of foreign workers, which means that they are not eager to substitute one with another. 118 Migrant labourers also find themselves in particular social circumstances, which affect their capacity to work in particular jobs. South Asian women, who make up the overwhelming part of the garment factories' labour force in Jordan, tend to come to Jordan alone and live in compounds on site. In contrast, young Syrian women -the group implicitly targeted by the pilot scheme -usually live with their families and have care and domestic responsibilities, which often do not allow them to be away for the long working days common in factory work, and to face additional long commutes. Most also have little experience in the sector. Take-up has thus been very low, and this has remained the case even after the scheme was opened up to camp residents, a group that was assumed to be more prone to take up these low-paying jobs. Disregarding the social embeddedness of labour, and the needs working Syrian women have, has hindered women's labour market participation in the garment industry, in manufacturing and more broadly. 119 The legacy of labour market segmentation is compounded by long-standing bureaucratic practices of recruiting and receiving (migrant) labour, which have run counter to the general willingness of the GoJ to fulfil its obligation to provide work permits to Syrian refugees. An employee of a Cash for Work project engaging Syrians and Jordanians in road works recounted, for example, that it was difficult to recruit women. This was not because women didn't volunteer for this type of work (quite the opposite), but because Ministry of Labour (MoL) directorates refused to issue work permits to them. This, in turn, was partly due to staff questioning the appropriateness of women doing road works, but equally because the registration system only allowed for "road work" as an occupational category to be assigned to men, not women. 120 In an UNHCR meeting gathering existing knowledge and research on Syrian livelihoods, a participant working for an international organisation also recounted the reluctance among MoL directorate employees to issue work permits to Syrian women, which s/he learned was copying their established practice to deny work permits to Egyptian women, aiming to restrict Egyptian labour migration to single men and avoid entire Egyptian families from coming to Jordan permanently. 121 Even though these particular examples were eventually resolved, they nevertheless highlight how the decades-long segmentation of the Jordanian labour market has sedimented in bureaucratic practices, which have, in turn, posed unexpected obstacles to the formal recruitment of (female) Syrian workers.
These examples from the process of implementing the Jordan Compact highlight that the core elements of labour market integration policies for Syrian refugees have very much been co-produced by international and national policy actors, and based on what seemed like a workable compromise between them. Yet they have neglected a number of -equally co-produced -technologies of government, dynamics and bureaucratic procedures that have shaped and permeated labour market and trade dynamics for decades. Associated policy legacies and memories were side-lined when designing the Compact, but had to be dealt with subsequently, as they have substantially complicated the implementation of the anticipated policy changes.
Some policy legacies and memories could not be easily dismissed though, and were thus factored into the Jordan Compact's design from the start to achieve a workable compromise. This is the case with the legacy of securitisation, which has shaped much of the history of governing (forced) migration in Jordan, 122 and has come into play in various instances related to the Compact. One such instance was the reluctance of the Ministry of Interior to allow camp residents to be part of the work permits scheme. 123 Camp residents have repeatedly been singled out as a particular security risk -a narrative which has materialised in increased spatial segregation of camp-based refugees from the host population over the years, and has been tolerated if not supported by international humanitarian agencies, including NGOs. 124 It was only in mid-2017, when it became clear that the numbers of permits given to Syrians were far from the needed target, that camp residents were included in the scheme, and that employment centres were set up there to recruit Syrian workers, who now make up a substantial portion of the work permits issued. 125 In the end, the logic of making the number of work permits the yardstick for dispensing further assistance partially overrode the legacy of securitising camp residents, and associated policy memories. Yet securitisation dynamics continue to conflict with the labour market integration agenda, in the camps as well as elsewhere. 126 Finally, it should be noted that the difficulties of "making refugees work" 127 also stem from the agency of refugees, who strategise based on their own policy memories. For instance, the memory of massive and unexpected food assistance cuts in 2014/5, which prompted a significant exodus from the region, has informed refugees' suspicion towards agencies' assurances that they would not lose assistance if they took up work, and has led many to avoid formalisation. 128 Refugees' own reluctance to formalise, based on concerns about losing assistance, getting on the radar of governing authorities, and the realisation that formal work does not mean better working conditions, has been among the main factors preventing a larger take-up of the work permits drive. 129 Such inconvenient policy memories are hard to reconcile with the organisational logic of most agencies involved, as well as with the Compact's overall goal of formalisation. Policy shapers sometimes frame them as regrettable misunderstandings, 130 whereas for others they provide an occasion to blame Syrian refugees for the limited success of the Compact. 131 Yet they have only been considered to a very limited degree in further developing labour market integration as well as assistance policies.

CONCLUSION: TRANSNATIONAL LEGACIES AND MEMORIES IN
REFUGEE POLICY Case studies do not just provide a window into the complexity of the world; they can -and should -lead us to also question and revise the theoretical concepts we work with. This very much applies to the categories used to grasp the significance of the past in forced migration policy -particularly when we discuss policies in and of countries of first reception neighbouring conflict settings, where international organisations tend to play a strong role. This setting has often invited analytical distinctions that separate the domestic/nation-state from the international scale, and host government from global refugee policies. However, looking at policy legacies and memories in the day-to-day dynamics and practices of the response to Syrian displacement to Jordan demonstrates that such distinctions obfuscate crucial dynamics shaping refugee policy over time.
One such dynamic, which I have highlighted in this article, is the politics of mutual accommodation between a variety of agencies and actors associated with both scales. A joint desire to find compromises in order to demonstrate outward policy success or achieve other goals, has made this all the more virulent, particularly in the case of Jordan, which has become an important showcase for "humanitarian innovation" in recent years. I have argued that this politics of accommodation has involved tacitly accepting a number of ambiguities permeating the response, and side-lining any policy legacies and memories that might undermine such compromises. This has proven productive in reconciling conflicting interests, yet it has in effect meant prioritising operational stability and political considerations over the actual improvement of Syrians' life in the country. This dynamic of refugee responses extends beyond the Jordanian case, and should receive closer attention in scholarship on the nature of refugee policies in conflict-neighbouring host states.
I have shown in this article that such overarching dynamics, which I have termed "policy legacies and memories", have shaped past and current policies and confined the range of imaginable changes. While "legacies" relate more to the institutional, structural dimensions of policy, explicitly adding the dimension of memory highlights the importance of the way in which specific actors or agencies interpret, or ignore, such legacies in line with organisational logics and/or challenges of the present. The case of refugee counting demonstrates various learning processes among different policy-relevant actors, which are particularly based on the preceding Iraqi response. These exemplify and have reinforced the broader politics of mutual accommodation, which has characterised interactions between humanitarian, development and governmental actors in the region since the mid-2000s. In the case of the recent policy changes regarding labour market integration of Syrians, the joint prioritisation of some strategic priorities at the expense of others for the sake of a workable compromise between these different agencies has neglected a number of policy legacies that have shaped and permeated labour market and trade dynamics between Jordan and its partners for decades, and the lessons that other actors have learned from them. Their side-lining during policy design has greatly complicated actual policy changes. Both aspects of refugee policy analysed here show that policy legacies and memories are transnational in nature, i.e. that refugee policy is co-produced by both local/national and international agencies, on the basis of previous dynamics that are already the product of an intense history of interaction and translation.
The Jordanian case therefore demonstrates the usefulness of thinking about refugee policy as emerging in transnational policy assemblages, in which a variety of agencies work together based on their (often entwined) histories and experiences, drawing on partly shared agendas and technologies of government, which in their combination shape the possibilities for stability and change of refugee policy. The conceptual repertoire of critical policy analysis -acknowledging emergence, ambivalence, and thinking about policy in assemblage categories -can help grasp such transnational dynamics, legacies, and memories. It is high time to more firmly link such conceptual discussions about policy with dynamics uncovered in in-depth country studies, and with the more varied conceptual repertoire that can be found in neighbouring disciplines such as anthropology and geography.
Such a move is not just of scholarly concern. Neglecting the legacies and memories at play leads to policy designs that are often unrealistic and unhelpful for actually improving targeted populations' lives. Conceptually integrating the role of "international" (humanitarian and development) agencies and donor governments in shaping policies on the ground also means acknowledging their shared responsibility for present dynamics. An analytical separation between host country and international actors creates the illusion that international refugee policy stands above, or apart from, host government policy choices, and that shortcomings are largely due to host governments' unwillingness to reform, or to accommodate refugees. Acknowledging the intertwinement of actors, technologies, and practices, however, can facilitate more equitable forms of burden-sharing that are mindful of donor governments' and international organisations' role in co-creating policies that shape daily life for the world's displaced in conflict-neighbouring countries.