The use of pseudo-causal narratives in EU policies: the case of the European Union Emergency Trust F

The EUTF aims to address the ‘root causes of migration’ by providing development assistance to countries of origin and transit. While it is allegedly based on scientific evidence, scholarly consensus suggests that development assistance is ill-suited to address irregular migration – which is something that some of the actors who designed the EUTF were aware of. We advance a new framework for understanding the emergence and success of pseudo-causal narratives (i.e., narratives relying on unproven and/or disproven causal claims) in EU policymaking. Using frame analysis, we argue that the pseudo-causal ‘root causes’ narrative was adopted against better evidence because it was plausible, compelling and had been used in EU external migration policies before. Faced with the salience of migration and the urgency to act in late 2015, and due to the absence of any clear ideas of what other measures could work, EU actors adopted this narrative to demonstrate that they were actively responding to the ‘crisis’. The narrative met little contestation, since it met the concerns of both those who were keen to stop migration and those who wanted to preserve the core of previous EU development policy.


Introduction
At the European Union (EU)-Africa summit in Vienna in December 2018, European Parliament President Antonio Tajani proclaimed: 'We need a Marshall Plan for Africa worth €50bn […]. By creating opportunities for young Africans at home we can stop migration flows to Italy and Europe' (Banks, 2018). Framing aid as a tool to address migratory movements is not completely new, not even in the EU. Ever since the December 1992 Edinburg European Council, government and EU officials have made periodic calls to address the root causes of migration through a comprehensive approach in origin and transit countries. Yet these calls had not been translated into policies. The initiatives adopted instead had focused on exporting migration control measures and had dealt only cursorily with root causes and preventive measures in countries of origin (Castles, 2004, p. 219). In short, until recently, EU policies aiming to address the root causes of migration had suffered from a gap between rhetoric and action. The so-called 'refugee crisis' of 2015 arguably bridged this gap, leading to the creation of a policy instrument that prioritized addressing 'root causes'. Adopted at the November 2015 Valetta Summit, the European Union Emergency Trust Fund for Africa ('EUTF') indeed seeks to address 'the root causes of destabilisation, forced displacement and irregular migration, in particular by promoting resilience, economic and equal opportunities, security and development' (European Commission, 2015, p. 4). As of 10 December 2020, the EUTF has approved 254 projects worth €4.85 billion (European Commission, 2020b) implemented in 26 African countries that are 'among the most fragile and affected by instability, forced displacement and irregular migration' (European Commission, 2019, p. 1).
While the narrative of preventing migration by addressing its root causes seems compelling, it contradicts the broad academic consensus that development fuels migration rather than stops it (Angenendt et al., 2017;Fratzke & Salant, 2018;Nyberg Sørense et al., 2003). Indeed, the poorest of the poor cannot migrate, as migration involves costs. This paper seeks to elucidate the processes leading to the adoption of the 'root causes' narrative of the EUTF. More specifically, we are interested in why EU policymakers, and particularly the European Commission, put a disproven causal claim, the 'root causes' narrative, at the centre of its policy, even when some of the actors involved were aware that the narrative was flawed. The European Commission's role is particularly interesting in this regard, as it is usually considered a technocratic and expertise-led actor in EU policymakingalthough there is an increasing awareness of its politicized nature (e.g., Christiansen, 1997).
Studying the so-far under-researched EUTF, our paper contributes to research on evidence-based policymaking and policy narratives. The literature on evidence-based policymaking largely concurs on the idea of a strategic and selective use or non-use of genuine research and evidence in EU policymaking. Our paper confirms these findings, but it also adds a new dimension to the debate by demonstrating how EU actorsinstead of just using evidence selectivelyknowingly adopt and accept causal narratives that go against extant evidence, while allegedly following an evidencebased approach. Research on policy narratives (Boswell et al., 2011;Roe, 1994) has shown how narratives may be based on unproven causal claims, but has not considered the use of disproven causal claims (Boswell et al., 2011, p. 1). Further developing the framework by Boswell et al. (2011) by differentiating the conditions for the adoption of a policy narrative, we, moreover, aim to provide a better understanding of how unproven and even disproven causal narratives emerge in EU policymaking. The case of the EUTF is particularly puzzling in this regard, as the EUTF specifically claims to ground its interventions on 'an evidence-based approach in order to understand drivers, dynamics of migration, and to map out responses' (European Commission, n.d.). The disconnect between the EUTF's self-proclaimed 'evidence-based' approach and its reliance on a pseudo-causal narrative contradicting widely accepted scholarly findings provide an opportunity to theorize about the conditions under which unproven and disproven pseudo-causal narratives emerge. In so doing, we do not seek to test the theory in a rigorous way, but rather to provide an exploratory study of these conditions through the case of the EUTF.
Using the method of frame analysis (Rein & Schön, 1977, p. 1996 and drawing on theories of framing and policy narratives, we assess how the assumptions linking development aid to reduced migration flows were constructed and propose several conditions for the successful emergence of a pseudo-causal narrative, whether based on an unproven or a disproven claim. We argue that the 'root causes' frame was adopted because it was plausible, compelling and well established in EU policymaking circles. Moreover, several political and contextual factors contributed to its success, such as the prominence of the perceived 'refugee crisis' and the urgent need to respond to it, the uncertainty around what policies could effectively stop migration, and the support given to the narrative both from actors who wanted to stop migration and those who wanted to preserve the core of development aid. The 'root causes' frame allowed the European Commission to rally support and reassert its legitimacy, especially at a time when it was receiving increased criticism over its inability to resolve the internal struggles in the EU regarding solidarity in refugee protection. To conduct this frame analysis, we draw on official EU documents, press releases and project reports related to the EUTF, as well as on 23 original interviews conducted in May 2019 and February 2020 with policymakers in EU institutions, national officials in Member States' Permanent Representations, and civil society organizations (see methodological appendix).
The paper is structured as follows: following a brief review of the literature on the (non-)use of evidence in policymaking, we present our theoretical framework on the construction of pseudo-causal narratives, drawing on the literature on framing and policy narratives. Based on Rein's and Schön's version of frame analysis, we analyse the evolution of the root causes frame in the EUTF by first presenting the diagnosis of the problem to be addressed according to EU actors, then discussing the remedy or prognosis proposed by the EU, and finally engaging with the political dynamics behind the adoption of the frame. We conclude by reflecting on the role of pseudo-causal narratives in EU migration policymaking (and beyond).
The (non)use of research evidence in EU policymaking and the adoption of pseudo-causal narratives Most scholarship on evidence-based policymaking deplores the insufficient or inadequate use of scientific research in the policymaking process. Some scholars argue this is because the policymaking community and the scientific community are 'two (distinct) communities' that need to respond to very different demands (Cairney, 2016;Caplan, 1979;Parkhurst, 2017;Stoker & Evans, 2016): Whilst for researchers publishing comes first and foremost, policymakers need to meet tight deadlines and respond to the political demands of different audiences. This scholarship, therefore, focuses on suggesting ways to either make the policy process more hospitable to the use of evidence or to make research more relevant to policymaking (Bartlett, 2013;Torriti, 2010). Another strand of the literature argues that policymakers use evidence strategically, i.e., in support of policy choices that they would take regardless. Hence, evidence is often driven by political concerns rather than the other way around, with political constraints leaving little space for evidence-based policies (Bartlett, 2013;Schlaufer, 2018, pp. 93-94). Research on the use of evidence in EU policymaking largely echoes this latter perspective (Bartlett, 2013;Sharman & Holmes, 2010;Torriti, 2010).
In sum, research on evidence-based policymaking has focused on assessing the reasons for the use or non-use of genuine scientific evidence. However, comparatively little attention has been paid to the creation and mobilization of pseudo-causal narratives by policymakers, that is, narratives that appear and/or claim to be scientific and factual but whose central claim is either unproven or has been disproven by the available scientific evidence. This neglect is in part due to a focus on 'the instrumental use of knowledge', i.e., we do not talk about specific evidence but evidence more generally evidence used as a tool for the betterment of policy, which does not consider the 'symbolic functions' of knowledge, namely its capacity to lend legitimacy and credibility to certain actors and their policy positions, and to justify their actions (Boswell, 2008(Boswell, , 2009. One notable exception is Boswell (2008), who draws on Brunsson (1989) to distinguish two types of organizations: action organizations and political organizations. Action organizations derive their legitimacy from the impact of their social interventions and use research evidence instrumentally to improve their performance. In contrast, political organizations derive their legitimacy from talk (referring to certain norms and values) and from appearing 'actionary' in responding to pressing issues. Boswell (2008, pp. 473-476) expects that political organizations will use expert knowledge strategically to garner support: by demonstrating a commitment to basing their decisions on evidence, political organizations not only strengthen and legitimize their own role but also substantiate their claims in contested policy areas such as migration.
Building on Boswell's framework, we posit that 'pseudo-causal' narratives, based on unproven and/or disproven causal claims, will under certain circumstances play the same legitimating and substantiating functions as genuine scientific evidence. We illustrate this theoretical expectation empirically through the example of the 'root causes' frame in the EUTF. Some research already points to the fact that migration policy, seemingly backed up by evidence, may in fact be grounded in unproven causal claims: Baldwin-Edwards et al. (2019, p. 2148, for instance, have argued that EU migration policies post-2015 have been assumption-led rather than evidence-based, with policymakers showing themselves unwilling to recognize the complex and mixed nature of migration flows. Hence, even policies that claim to be based on evidence are often based on a clever argumentative connection of cause and effect for which there is no strong empirical evidence. Whilst Baldwin-Edwards et al. present a case of what we would call unproven causal claims, the EUTF relies on a disproven causal claim, i.e., a connection of cause and effect that is widely considered incorrect among scholars. To understand how such 'pseudo-causal narratives' can convince audiences and triumph despite their empirical inaccuracy, we turn to the literature on policy narratives. According to Roe (1994, p. 51), narratives stabilize 'the assumptions needed for decision-making in the face of what is genuinely uncertain and complex [emphasis added]. They can be representationally inaccurateand recognisably sobut still persist, indeed thrive'. The relationship between migration and development certainly qualifies as uncertain and complex, since it is not clear which interventions or policies can effectively contribute to reducing migration flows. Given this uncertainty, policymakers may be prone to rhetorically establish a causal link, regardless of whether it holds or not. We would add that they do so specifically in highly salient policy areas and in times of an increased perception of the urgency to actsuch as the 2015 EU 'refugee crisis', which put policymakers in the position of having to quickly adopt a narrative concerning how their policies were going to address the crisis. Again, bearing in mind the symbolic functions of knowledge, the need for an effective policy in terms of potential outputs and outcomes was arguably less important than demonstrating to the European public that action was being taken. Besides, Boswell et al. (2011, p. 1) argue that such narrative constructions are more likely to be successful if they are cognitively plausible, dramatically or morally compelling, and if they chime with perceived interests. We would add that such narratives are also more likely to be adopted if they have been used before and gone unchallenged. Especially in situations of high levels of uncertainty and urgency, being able to draw on existing blueprints saves time and energy for policymakers, who cannot risk proposing a narrative that will meet with broad scepticism and contestation.
These narrative constructions, in turn, act as framing devices 'that guide both analysis and action in practical situations' (Rein & Schön, 1996, p. 89). 1 Drawing on theories of framing and frame analysis, we analyse how the pseudo-causal 'root causes' frame was adopted and presented as scientifically warranted. Policy frames are strong and generic narratives that include two important elements: they identify the issue at stake (diagnosis: what needs fixing) and they prescribe solutions (prognosis: what should be done about the issue). Policy frames, however, 'do not float freely': the institutional configuration, the distribution of power and the preferences of the actors involved are all crucial to understanding the emergence and effects of a frame (Lavenex, 2001, pp. 21-22;Risse-Kappen, 1994). To account for the emergence of the frame of 'addressing the root causes of migration through development aid', we examine how the problem was defined, what policy solutions were advanced in response, and the constellation of preferences of the different actors involved. These three dimensions are, of course, interdependent: defining a problem already constrains the type of policy solutions available, meaning that prognosis and diagnosis are often two sides of the same coin. Besides, the distribution of power and the preferences of actors influence both the framing of the issue at stake and the solutions put forward. Still, disentangling these three dimensions conceptually does allow for greater clarity (even though we will see in our empirical analysis that they overlap).
We summarize the conditions for the adoption of pseudo-causal narratives by political actors in Table 1, separating between the diagnostic and prognostic dimension on the one hand, and the political dimension on the other. Conditions for the successful adoption of a narrative that lie in the narrative itself are summarized under diagnosis/prognosis; exogenous conditions relating to decision-making are subsumed under politics.
In what follows, we demonstrate how these conditions affected the emergence of the 'root causes' narrative in the diagnosis, prognosis and decisionmaking processes leading to the adoption of the EUTF. Diagnosis: the 'European refugee crisis' as a result of factors in sending and transit countries As Sandberg (2006, p. 211) notes, any framing process first involves identifying the specific problem at stake and answering the question of 'who is to blame?'. This implies that problems do not exist per se but are usually defined as such by policymakers. The actors who get to define a problem, who frame it, are in a powerful position, since the frame already precludes certain solutions from being considered. The perceived problem to be addressed has shifted at different times in the 'European refugee crisis'. Initially, the EU's definition of the problem revolved around managing and distributing those asylum-seekers who had already arrived in Europe, particularly in Italy and Greece (and to a certain extent in Hungary). The Council adopted two relocation schemes in September 2015 to support these countries and to prevent the self-relocation of asylum-seekers facing inhumane conditions (2015a, 2015b). However, the negotiations over these schemes were highly controversial, partly due to a distributive conflict between Member States: while states receiving higher asylum-seeker shares (either as border or as destination countries in North-Western Europe) wanted to see redistribution, Central Eastern European countries preferred maintaining the status quo so as not to incur additional responsibilities (Biermann et al., 2019). A subsequent initiative on long-term solidarity involving a permanent stand-alone quota system turned out to be highly controversial and was soon deadlocked (Zaun, 2018).
With no durable solution in the internal dimension in sight, EU institutions and Member States focused on the external dimension of migration policies, where Member States' positions were more congruent. The problems to be addressed were now (a) the arrival of immigrants in Europe and (b) the reasons making them leave their home countries. The arrival of immigrants in Europe was addressed by the reinforcement of Frontex with the European Border and Coast Guard Regulation to stop 'irregular migration' and 'human smuggling' in the Mediterranean, and by the adoption of an EU-Turkey Statement aiming to disincentivize migrants from entering Greece irregularly from Turkey (Slominski & Trauner, 2018). Yet EU policymakers were aware that these measures, while providing a temporary band-aid to arrivals at the EU's external borders, would do little to stop immigration in the longer term. A shift in focus towards 'upstream work' (i.e., work closer to sending countries) was therefore needed (Interview_PermRep_6). Hence the resolution to directly address the root causes of the 'problem', i.e., the reasons why migrants leave their home countries. Another 'problem' that a stronger focus on 'root causes' (rather than border or return policies) could address was the significant loss of human lives occurring in the Mediterranean, represented in the media by the death of refugee child Alan Kurdi (Adler-Nissen et al., 2020).
According to the Commission Decision of November 2015 creating the EUTF, the root causes that needed to be addressed included 'demographic pressure, environmental stress, extreme poverty, internal tensions, institutional weaknesses, weak social and economic infrastructures, and insufficient resilience to food crises', as well as internal armed conflicts, terrorist threats and a deteriorated security environment (European Commission, 2015, p. 1). This was not the first time 'root causes' such as underdevelopment in sending and transit countries had been framed as a problem to be addressed by EU policies. After the 1992 European Council, the frame was taken up again in 2005, when the Spanish Canary Islands experienced an increase in arrivals of migrants by boat. Whilst the narrative was also centred on underdevelopment as a driver of African emigration, policies continued to focus on migration control rather than development, either through the 'stick' of readmission agreements or the 'carrots' of visa facilitation and mobility partnerships (Vives, 2017). Nonetheless, the long-standing (rhetorical) commitment to the 'root causes' frame signalled to policymakers thatat least politicallythe frame was not contested. This facilitated its re-emergence after the perceived 'refugee crisis' in 2015, when the Commission had to propose solutions under severe time pressure and high levels of politicizationthis time not only in Spain but across Member Statesto show that the EU was actively addressing the 'crisis' (Interview_PermRep_6). As the problem was much more encompassing in scope and scale, a significant stepping up of resources (Interview_COM_1) was required, resulting in the creation of the EUTF.
Prognosis: 'addressing the root causes of migration' through development aid The prognosis element of a frame refers to the policy ideas that are presented as a solution to a previously defined problem. In this section, we first demonstrate how the EUTF was initially conceived in the context of the EU-Sahel Strategy, providing a ready-made blueprint which was repurposed in 2015 in response to the refugee crisis. Next, we demonstrate that the EUTF claims to be evidence-based. Finally, we show how the intuitiveness of the 'root causes frame' contributed to its adoption and, finally, discuss its morally and dramatically compelling character.
In their 'garbage can model' of policymaking, Cohen et al. (1972) have argued that policy ideas are not always developed in response to a need for policy change but are often pre-existing and reused in situations where new policies must be adopted quickly. Conventional wisdom suggests that the EUTF was developed as a response to the 2015 'refugee crisis'. But in fact, the EUTF was conceived much earlier, following the adoption of the 2011 EU-Sahel Strategy, which aimed to address key issues of security and development in a region increasingly torn by destabilization and jihadist terrorism. DG DEVCO developed the idea of a Trust Fund to allow the Strategy to be implemented more rapidly (Interview_COM_3; Interview_COM_4; Interview_-COM_5). The idea was eventually enthusiastically supported by Member States and the College of Commissioners during the 2015 crisis and was sold as a migration management instrument. Given the time constraints the Commission was facing, having a ready-made 'solution' to the 'refugee crisis' in the drawer clearly simplified its adoption. The full title 'EUTF for Stability and Addressing the Root Causes of Irregular Migration and Displaced Persons in Africa' still testifies to how two different goalscreating stability and addressing the root causes of migrationwere juxtaposed in a single instrument. According to a high-ranking Commission official, the notion of 'emergency' in the EUTF initially referred to the emergency situation in the Sahel region, such as Mali or parts of Nigeria where Boko Haram is active (Interview_COM_3). However, many observers in Member States or EU institutions who were not involved in the design of the initial instrument interpret the emergency as being the increased inflow of migrants into Europe in 2015 (Interview_EP_3; Interview_PermRep_1; Interview_PermRep_2; Interview_-PermRep_6), even though most migrants did not come from Africa, but the Middle East. The further the EUTF developed, the more exclusively it emphasized projects aimed at decreasing migration rather than creating stability (Interview_CSO_1; Interview_EP_2). In short, drawing on the narrative of using aid to 'address the root causes' of migration allowed the Commission to connect the EUTF's original solution of creating stability in the Sahel region with the diagnosis of a 'European refugee crisis'. Thus, projects adopted in the former area could be sold as contributing to migration management aims.
The importance of evidence in the design of the EUTF is manifest in the Constitutive Agreement adopted in Valetta on 12 November 2015, which suggests that the EUTF would 'be guided by a strategyunderpinned by a clear evidence basewhich will determine its objectives and priority sectors' (European Commission, 2015, p. 10). In the Strategic Board meetings, both the Commission and Member States have reiterated the importance for the EUTF of investing in research and of applying an evidence-based approach (European Commission, 2016b). If the EUTF has indeed invested in research, evidence and monitoring facilities, these facilities are mainly mobilized at the implementation and evaluation stages, e.g., to assess whether EUTF projects meet their specific development goals rather than whether they have an effect on migratory flows (Interview_COM_3; Interview_PermRep_6).
We now turn to assess why using aid to 'address root causes' then became such a powerful frame despite its disregard for established research findings suggesting that aid will not stop migration. The EUTF does not merely juxtapose development aid and migration rhetorically (as previous programmes did), but strongly focuses on development projects in practice: between 2016 and 2019 (own calculations, see appendix), traditional aid programmes promoting greater economic and employment opportunities constitute the second biggest chunk of the funds provided (24 per cent of EUTF funding). The adoption of these programmes is justified by the (once again unsubstantiated) assertion 'that short and long-term grievances arising from economic and social exclusion, marginalisation and inequality are amongst the most significant drivers of violence, forced displacement and illegal migration' (European Commission, 2020a). The biggest chunk of EUTF funding, however, still goes to migration governance and border protection (32 per cent of EUTF funding), demonstrating that the EU has not fully abandoned the migration control focus. Funding for other traditional development aims of food security, better health services and education comprises 21 per cent of the EUTF budget. Conflict prevention receives 9 per cent of the funding, while infrastructure, institutions and governance receive 12 per cent. The strong focus on traditional development aid programmes and especially 'economic opportunities' seems particularly puzzling when considering the evidence on the impact of development on migration: research suggests that development aid has a limited capacity and effectiveness to contribute to employment or economic growth (Schöfberger & Venturi, 2018, p. 3). Even more paradoxically (and problematically), there is a broad scholarly consensus that economic development leads to more rather than less migration, at least in the short term. This is often referred to as the 'migration hump' (Angenendt et al., 2017;Fratzke & Salant, 2018;Nyberg Sørense et al., 2003). Economic development, and especially income-generating projects, including temporary job creation, is expected to raise the aspirations of potential migrants and facilitate their access to the resources needed to movethereby stimulating migration (Ascencio, 1990). Only when a country reaches an upper-middle income level is the relationship reversed a threshold that is out of reach in the short and medium term for the countries targeted by the EUTF. More recently, some scholars have argued that improvements in governance and the quality of and access to public services may help reduce emigration, although the effect is small (MEDAM, 2018). Still, only a small fraction (12 per cent) of EUTF funding is dedicated to investment in governance, public services and institutions.
Despite its weakness, the 'root causes' frame is repeated almost identically in every EUTF-funded action or project and linked to the overall project-specific objectives through quasi-causal arguments. These arguments are 'quasi-causal' because they may or may not hold empirical validity (Weldes, 1996, p. 282). It is not their accuracy that matters, but that they provide 'warranting conditions' which 'make a particular action or belief more "reasonable", "justified", or "appropriate", given the desires, beliefs, and expectations of the actors' (Fay, 1975, p. 85). The absence of empirical evidence or scientific backing for such claims may reinforce their capacity of persuasion by portraying these quasi-causal relationships as self-evident. Consider the following statement by Commissioner for International Cooperation and Development Neven Mimica when announcing two new projects for Senegal under the EUTF (European Commission, 2016a): The EU will continue to support the Government of Senegal in its endeavour to improve access to basic social services, to make its economy more competitive and to strengthen food security and agricultural development. I am convinced that improved living conditions in rural areas and new job opportunities will go a long way to address root causes of irregular migration and make a real difference in the prospects of Senegalese people.
Here, the audience is urged to share the confidence of Commissioner Mimica that increased economic opportunities will contribute to tackling the underlying drivers of migration (and thus help curb migration flows to Europe). The argument seemingly needs no further explanation or justification, appealing to the audience's 'common sense'. Likewise, it is not demonstrated how and why funding under the EUTF as a form of international aid is able to remedy these root causes. And yet, almost anythingor quite literally 'everything' (Interview_ EEAS_1)becomes included within the remit of EU development cooperation: with the EUTF, the EU indeed hopes 'to address a wide range of ills' with projects focusing on employment creation, basic services such as health and education, food security and environmental protection, and measures to fight human smuggling and trafficking (Alfonso & Immenkamp, 2015, pp. 8-9). These examples confirm that scientific evidence plays a symbolic rather than an instrumental role in the design of the EUTF and its 'root causes' frame: claims of reliance on an 'evidence-based approach' legitimize and naturalize the EUTF, even when the narrative is not supported by scientific research or is, as we have shown, even disproven by research. The frame's constant repetition in every EUTF policy document, project plan or press release reifies and naturalizes it, which contributes to the frame going widely unchallenged.
Aside from being intuitive and appealing to common sense, the 'root causes' narrative is also morally and dramatically compelling. In early September 2015, the widely mediatized images of Alan Kurdi, a three-year-old Syrian boy who had drowned on his way to Europe, spurred renewed calls for a more humane immigration policy and for ending the EU's 'political inaction' (Adler-Nissen et al., 2020). Meanwhile, most Member States remained strongly opposed to hosting large numbers of refugees. Preventing migration by addressing its 'root causes' was a convenient solution to address the issue of rising deaths at Europe's borders whilst keeping borders closed. Other policies aiming at stopping emigration in sending and transit countries, such as cooperation with the Libyan coast guard and policies focusing on the prevention of human smuggling and trafficking, had received widespread criticism, especially by NGOs (Statewatch, 2020). The 'root causes' frame has received much less negative attention because, in contrast to the 'fight against smuggling', the provision of development aid is less contested, especially in circles of migrant advocacy groups. This is because the 'root causes' frame' focuses on ending the drivers of forced migration, while implying less of a use of force compared to return and readmission policies. Through frame amplification, the frame's legitimacy was thus further enhanced (Benford & Snow, 2000, p. 624). Frame amplification refers to a situation where a recognized policy aimsuch as the desirability of developmentis used to legitimize and embellish a more controversial or stigmatized policy, in this case migration control. This makes the frame of 'addressing the root causes of migration' appealing to actors who would otherwise be sceptical of it, e.g., African partner countries, European development ministries and agencies, and NGOs.

Politics: the struggle between reshaping development aid and protecting its core
Framing is an exercise of power: the frame adopted and the solutions prescribed by it reflect what is politically feasible, given the preferences of the actors involved. Understanding the political dynamics behind the EUTF's design and adoption is, therefore, crucial to explaining why evidence suggesting that this policy is flawed was disregarded. We argue that the 'root causes' frame was successful not only because it was intuitive, compelling and had been used before, but also because of the urgency to take action in late 2015, the complexity and uncertainty that existed concerning true causal claims, and the fact that the narrative fitted in with the interests of decision-makers.
Migration was a highly salient issue in late 2015 when the EUTF was adopted, with the European Commission under severe 'political and popular pressure to do something' to address the 'crisis' (Interview_Perm-Rep_6). This pressure was arguably all the more prevalent as the Commission's role in proposing internal measures on the temporary and permanent relocation of refugees in Europe had not yielded the expected results, but had instead aggravated internal divisions between Member States (Interview_PermRep_1; Interview_PermRep_6; Interview_EP_1). This favoured a concerted approach within the Commission and its different DGs. Starting in late 2015, all actors working on migration (especially in the Commission but also in the European External Action Service) met regularly to streamline their activities (Interview_COM_1; Interview_COM_2; Intervie-w_EEAS_1; Interview_EEAS_2). Likewise, especially Member States that were affected by the 'crisis' faced increased pressures to act. The 'root causes' frame hence allowed both the political level of the EU and national governments to show that they were effectively responding to the 'refugee crisis' and preventing something similar from happening again (Interview_-COM_1; Interview_EP_2; Interview_EP_3).
Our interviews suggest that those who were most aware of the 'migration hump' literature were to be found mainly among EU and national bureaucrats, particularly in DG DEVCO and in domestic Development Ministries and Divisions, while the political level of the Commission and within Member States and sometimes actors with a background in Justice and Home Affairs were often unaware, buying more easily into the 'root causes' narrative (Interview_COM_1; Interview_PermRep_4; Interview_PermRep_5; Interview_Perm_Rep_7). Those in DG DEVCO who were aware of it raised this several times in the context of the EUTF's adoption; however, they were not listened to: 'We tried to advise our people regularly that evidence shows that there is a "migration hump", but it didn't take root, it was not what they wanted to hear' (Interview_COM_1). Hence, the adoption of the narrative was partly based on the fact that not all actors at the EU level and within Member States were aware of the faultiness of the 'root causes narrative', and partly due to some actors in the Commission deliberately ignoring the evidence. Even when the European Political Strategy Centre (the Commission's in-house think tank) drew attention to the existence of the 'migration hump' by stressing that 'economic development spurs migration in the short-term' (EPSC, 2017, p. 8), those conclusions were met with heavy scepticism. Some Member States even questioned whether this was 'really true' (Interview_COM_1) and the findings of the report were eventually disregarded and discussed no further (Interview_COM_1; Interview_-COM_2). This explains why the 'root causes' frame remained largely uncriticized.
What further contributed to the adoption of the frame was the complexity and uncertainty over genuine causal links between alternative migration policies and the reduction of irregular migration flows. Whilst some policymakers were made aware that development would not prevent migration, they lacked evidence of what could stop it. There is indeed much more of a consensus on what does not work to prevent migration, such as restrictive border policies (Cornelius & Salehyan, 2007) or restrictive visa practices (Czaika & Hobolth, 2016), than on what does work. This arguably contributed to the adoption of a narrative that was factually incorrect.
A final reason behind the successful adoption of the 'root causes' frame is that it bridged the divergent concerns of those actors who wanted to adopt specific migration control policies with those who aimed to preserve the development core of the EUTF. Member States that had populist parties in power (such as Italy, Austria and the Visegrad countries), as well as countries, such as France and Germany, where moderate governments were under pressure from populists gaining electoral ground, had a strong interest to stop further migration to Europe through migration control policies (Zaun, 2018;Interview_EP_1;Interview_PermRep_1;Interview_PermRep_5). These same countries are also among the EUTF's biggest bilateral donors (European Commission, 2020b). Yet countries with a strong development aid tradition sought to protect the traditional core of development assistancenot least because two-thirds of the funding came from the European Development Fund (EDF) (European Commission, 2020b). To manage these divergent preferences, the European Commission incorporated elements of the conflicting demands of the two camps (Brunsson, 1989). The compromise consisted in partly repurposing the funding to spend it on migration control purposes while otherwise rebranding previous and ongoing development policies as economic and job creating policies that address 'the root causes' of migration.
Member States favouring the migration control approach supported the 'root causes' frame as it allowed them to show to their domestic audiences that they were acting to prevent irregular migration flows. Yet even among actors with a long-standing tradition in development aid (such as Luxembourg, Ireland, Sweden, Finland or the German Development Ministry), some were aware of the flaws in the 'root causes' frame but nevertheless promoted it, albeit for a very different reason: they wanted to ensure that an important proportion of EUTF money continued to be spent on traditional development policies. As one interviewee highlights: We and others have called on the EUTF being used for poverty reduction. It's not like Member States are not aware [of the migration hump], but it's a tricky situation for several countries, because [funding under the EUTF] came from the EDF … For us, it's still development cooperation money […]. So therefore, we take the narrative, and we take the purpose that we see in the EDF, and we apply it to the EUTF, where we say, of course, root causes should be the bulk of what we do in the EUTF, and we think that's good Development Cooperation, that's a way to spend money wisely … If that's inside the EUTF or the EDF, that's secondary … We continue to push for the root causes … because this is still development cooperation money, […] we never wanted it to be 100% going to migration control. (Interview_PermRep_7) This shows that for the European Commission, presenting aid as a solution to migratory movements was a way to respond to both the demand for more migration control and for a continuation of traditional development policies. The very inaccuracy of the 'root causes' narrative, in fact, served to reconcile the need for talking about migration control while largely acting to preserve the core of development policy (Brunsson, 1989).

Conclusion
In this paper, we have analysed how the EU successfully introduced its frame of 'addressing the root causes of migration' through development aid into the EUTF. The adoption of this frame despite its inherently flawed logic is particularly paradoxical, not least when considering the EUTF's self-declared aim to provide evidence-based policymaking. Through a frame analysis of the 'root causes' narrative, we have argued that the narrative was successfully enshrined in the EUTF because of several factors lying in the narrative itself, namely that it was cognitively plausible, morally compelling, and had been previously established and used without much contestation among the actors involved. Moreover, several contextual and political factors contributed to its adoption: the decision-making on the EUTF occurred at a time of high salience and urgency, which pushed the Commission to propose a policy swiftly. This proved particularly challenging, as no positive causal narrative of what policies could stop migration was available. At the same time, the narrative bridged the concerns of different audiences, including those actors who favoured more migration control and those who wanted to maintain the core of development policy. While previous research has focused on the (non-)use of genuine evidence in policymaking, this analysis has shown that pseudo-causal narratives (based on unproven or even disproven causal claims) can fulfil the same legitimising functions, especially when an actor (such as the European Commission) faces strong political pressure to propose a response to a perceived crisis but has few formal powers to do so.
The use of pseudo-causal narratives seems to be a wider dynamic in EU migration policy. Other examples include the narrative suggesting that harmonising asylum policies across the EU prevents secondary movements from border countries, or the narrative that restrictive border policies deter irregular migration (and reversely, that operations of rescue at sea represent a pull factor). With regard to the former, research has suggested that secondary movements are not mainly driven by policies but by factors such as colonial ties, the presence of migrant networks, or a country's reputation, including as regards its economic or human rights situation (Neumayer, 2005;Thielemann, 2006). Still, up to now policy harmonization is framed as the sole solution to address secondary movements. Likewise, the assertion that strengthening border control and ending rescue operations at sea will stop irregular migration is not supported by evidence (Cornelius & Salehyan, 2007;Cusumano & Villa, 2019). Instead, these measures have been found to divert migrants towards more dangerous routes, encouraging recourse to smuggling networks, and more generally, increasing the risk of deaths at sea. Whilst these narratives are not presented as evidence-based, and while the evidence opposing the EU's causal narratives is certainly not as strongly established as in the case of the 'root causes' narrative, it is still striking that such unwarranted narratives are so widely used in EU migration policies.
The problem of using unwarranted causal narratives has become a salient issue in politics in recent years. Our analysis has shown that such narratives may also be advanced by actors that one might least expect to use them, namely technocratic actors such as the European Commission, which are usually considered to base their work on expertise and knowledge (Rimkutė & Haverland, 2015). Yet, it seems that, given their limited capacity to act, these types of actors tend to promote policies that legitimize their role politically, especially when they are under pressure. Further research could analyse to what extent similar dynamics are present in other salient policy areas of EU policymaking in which the causal links are highly uncertain and where the conditions for the adoption of unproven or disproven causal claims identified in this paper are present. Note 1. We use 'narrative' and 'frame' interchangeably, conceiving of policy narratives as a specific kind of framing device.