Abstract
The protection of asylum seekers and refugees has become one of the most politically divisive issues in the European Union, yet there has been a lack of research on public preferences for asylum and refugee policies. This article analyzes which policies Europeans prefer and why. We advance a theoretical framework that explains how asylum and refugee policies that use limits and conditions enable individuals to resolve conflicting humanitarian and perceived national interest logics. Using an original conjoint experiment in eight countries, we demonstrate that Europeans prefer policies that provide refugee protection but also impose control through limits or conditions. In contrast to the divisive political debates between European Union member states, we find consistent public preferences across European countries.
Introduction
Asylum and refugee issues have gained importance in the politics of the European Union (EU), becoming increasingly divisive since the refugee ‘crisis’ in 2015–2016 when 1.3 million asylum seekers arrived in Europe leading to the de-facto collapse of the common European asylum system. While many core aspects of asylum policy have always remained within the realm of the nation state, the difficulty of certain EU member states to cope with the refugee emergency called for new common European solutions (see Nieman and Zaun (2017) for an overview). Yet, asylum reform proposals proved to be contentious between member states. The struggle to cooperate on asylum and refugee policy issues generated a great deal of political conflict, reinforcing long-standing fault lines between the northern ‘core’ member states, the more recent eastern members, and the southern ‘frontier’ EU countries (Thielemann, 2005; Trauner, 2016). The national politics of member states, particularly the preferences of their respective electorates, are likely to be important factors that shape the different negotiating positions EU countries take regarding asylum and refugee policy.
Yet, despite this increased public scrutiny and intense political debates, we know very little about Europeans’ preferences for asylum and refugee policy and whether they differ across EU member states. While the scholarship on attitudes to asylum seekers and refugees has been growing in recent years, it remains much smaller than the large body of work on attitudes to immigration and immigrants in general (see Hainmuller and Hopkins, 2014). A few landmark studies focus on the characteristics of asylum-seekers and refugees preferred by the public (Bansak et al., 2016; Hager and Veit, 2019). Studies that investigate public preferences for asylum and refugee policies, however, are still rare. There are a few notable exceptions that examine specific aspects such as the redistribution of asylum seekers (Bansak et al., 2017; Heizmann and Ziller, 2020), the number of asylum claims that should be granted (Andersson et al., 2018; Hercowitz-Amir and Raijman, 2020) or how changes in inflows of asylum seekers affect existing residents’ attitudes (Hangartner et al., 2019). Consequently, the current scholarly understanding of public preferences vis-à-vis asylum and refugee policies remains very limited.
Evidence from self-reported surveys of people’s views on asylum and refugee issues shows European voters have contradictory positive and negative evaluations, which suggests a tendency towards ambivalence in public policy preferences. On the one hand, Europeans appear to be strongly committed to providing humanitarian protection to people who flee violence and persecution (Connor, 2018). Yet on the other hand, Europeans are simultaneously concerned about the (mis)management of refugees and asylum policy (DIW Berlin, 2017) and about ‘bogus’ asylum-seekers (Cohen, 2011; Husbands, 1994; Sussex, 2005) who threaten state ‘control’ over immigration.
This ‘Janus-facing’ nature of public preferences for asylum and refugee policy, with its conflicting humanitarian and perceived national interest concerns, renders them remarkably difficult to study. Generally, the study of public preferences tends to conceive of preferences as lying on a unidimensional, bipolar continuum of positive to negative sentiments. However, when it comes to asylum and refugee protection, such an approach is very limiting since it obscures the full and complex nature of public policy preferences. In this case, observational survey methods are inadequate instruments for ascertaining public preferences due to the simultaneous coexistence of positive and negative evaluations of asylum and refugee issues. This duality makes the answers obtained in observational survey methods unreliable and inconsistent, since citizens are often unaware of their own ambivalence or the fact that policy-making has certain unavoidable trade-offs (Citrin and Luks, 2005). Furthermore, the moral aspects of asylum and refugee affairs make self-reported preferences through surveys more vulnerable to the possibility of social desirability bias, whereby survey respondents are less likely to give their honest opinions if they feel they are not socially acceptable.
To overcome these conceptual and methodological challenges, this article employs an original conjoint experiment to reveal which asylum and refugee policies Europeans prefer and to investigate the extent to which these preferences vary across European countries. We aim to analyze not only which policies are preferred but also why, by advancing a theoretical framework that explains how policies that use limits and conditions enable individuals to resolve conflicting logics about asylum and refugee issues. Furthermore, by employing this experimental research design, we are able to isolate the separate causal effects of particular features of asylum and refugee policy on public preferences. We conducted the experiment in 2019 with 12,000 respondents across eight different population-based samples in Europe, covering Austria, France, Germany, Hungary, Italy, Poland, Spain, and Sweden.
Our study makes several contributions. First, we move beyond the common practice of thinking about public preferences for asylum and refugee policy in binary terms (i.e. ‘support’ vs. ‘oppose’). In our theoretical framework, we conceive of public preferences in a way that allows for non-binary policy choices, whereby asylum and refugee policies do not only vary in their restrictiveness but also according to the use of limits and conditions in providing protection to asylum-seekers and refugees. Exploiting our experimental design, we are able to contribute to the literature on policy preferences by considering whether and how they are contingent on the use of such policy controls, which we argue function as a way of resolving ambiguous or internally conflicting sentiments about an issue.
Additionally, the article provides the first-ever analysis of the public’s multidimensional preferences for asylum and refugee policy, a topic that has attracted little scholarly attention to date. The few existing studies have provided a unidimensional understanding of public preferences in this policy area. In practice, asylum and refugee policies are inherently multi-dimensional, as they involve decisions on various aspects of the governance of the asylum process and the scope of the protection of refugees, as well as the different ways of cross-country cooperation and assistance. Our categorization and empirical analysis of multidimensional asylum and refugee policies consider all these policy aspects, thus making an important conceptual and empirical contribution. Finally, our analysis also provides timely and relevant findings that can inform ongoing policy debates about how to reform asylum and refugee policies in Europe.
Theoretical framework
The two ‘logics’ of public preferences for asylum and refugee policy
Theoretically, the ambiguous nature of public preferences arises from the two conflicting logics that pervade asylum and refugee policy (see Rosenblum and Sakehyan, 2004). The first is the logic of humanitarian assistance, which has its basis in a moral imperative that states should provide protection to people who are fleeing persecution or face imminent harm. In addition to this moral foundation, policies based on this logic derive legitimacy for providing protection from ‘transnational discourse and structures celebrating human rights as a world-level organizing principle’ (Soysal, 1994: 3). Anchored in the 1951 Refugee Convention, this humanitarian logic underpins the approach of the EU, its member states, and the wider international community towards asylum and refugee issues. One of the consequences of a humanitarian logic is the legitimation of the protection ‘deservingness’ of certain asylum-seekers over others, and this differentiation is a basis for public attitudes towards refugees and asylum-seekers in practice. A recent study by Bansak et al. (2016) shows that Europeans are more willing to admit asylum seekers who have been victims of torture.
In contrast, the second logic relates to the perceived national interest, which takes an instrumental perspective on providing protection to asylum seekers and refugees. According to this logic, policy preferences are socio-tropic, shaped by a citizen’s subjective calculation of the perceived costs and benefits for the host country of accepting asylum seekers and refugees. This logic of perceived national interest in asylum and refugee issues dates back to the politics of the Cold War era, where countries provided refugees protection partly for their own strategic benefit (Loescher, 1989). This logic is still pervasive today since it is often the case that ‘refugees are considered to be an ever-growing burden on economies and societies and a threat to internal stability’ (Czaika, 2009: 90). This line of reasoning appeals to the preservation of scarce material resources (Ivarsflaten, 2005; Tartovsky and Walsh, 2016) where national prosperity is framed as being threatened by asylum-seekers (Greussing and Boomgaarden, 2017; Ritter and Rhomberg, 2017). National governments are under increasing pressure to protect their asylum systems from the perceived ‘menace’ of economic migrants (or ‘bogus’ asylum seekers) who are seen as exploiting a lax protection policy (Khosravinik, 2009) and, thus, as undermining the ‘integrity’ of the broader immigration system.
We argue that public ambivalence about asylum and refugee protection arises from the opposition of these two logics, which conflict in individuals’ preference formation. In the context of our study, we are theoretically interested in how these two competing logics play out in the formation of a person’s preferences for asylum and refugee policy. We reason that while the perceived national interest logic pushes individuals towards restricting or even abolishing protection, the humanitarian logic pushes individuals towards supporting generous protection with relatively few restrictions. Rather than simply polarizing the public into two groups (e.g. support vs. oppose), we expect these competing logics to generate internal contradictions within the same individual citizen, pitting a person’s humanitarian norms versus what they perceive to be in the national interest. Certainly, some people might be more motivated by one logic or the other which may cause them to feel strongly positive or negative about the issue, but based on the existing survey evidence, we can expect the majority of people to be motivated by both of these two conflicting logics.
Scholars have still not understood or explained how the contradiction between these two logics manifests itself in public preferences for asylum and refugee policy. The existing scholarship on public preferences tends to simplify public opinion in bipolar terms, as lying on a continuum that ranges from positive to negative sentiments. Yet for issues for which the public has conflicting sentiments, such as asylum and refugee affairs, employing a bipolar continuum to measure opinion is inadequate and can result in instability (see Craig and Martinez, 2005). A unidimensional conception also obscures the multidimensional nature of asylum and refugee policy, and its intersection with public preferences.
Given the lack of existing research in this area, our study is exploratory. Yet, on the basis of our theoretical framework, we do have certain expectations about Europeans’ preferences for asylum and refugee policy. We argue that Europeans, in their formation of preferences for asylum and refugee policy, need to balance their normative concerns for providing humanitarian assistance against their subjective socio-tropic concerns about preserving the national interest. It would then follow that Europeans are more likely to prefer asylum and refugee policies that allow them to reconcile both the logics of humanitarian assistance and perceived national interest. Across the policy dimensions that we have conceptualized (see below), we expect that Europeans will opt to balance both, preferring policies that neither give unlimited assistance to refugees and asylum-seekers, nor fully restrict assistance altogether. A policy design that creates balance in this sense contains policy controls such as limits (e.g. annual limits on the number of asylum-seekers) or conditions (e.g. allowing family reunification only for refugees who are able to economically support their family members) while ensuring that certain core features of protection are provided. Our expectation is, therefore, that Europeans will prefer asylum and refugee policies that implement limits and conditions to policies that are unconditional or highly restrictive in the assistance and protection they provide.
We expect to observe this pattern of policy preferences across European countries. In our view the logic of perceived national interest is likely to exist across all countries since, in general, attitudes towards immigration issues are markedly socio-tropic (see Hainmueller and Hopkins, 2015). Certainly, some countries (e.g. Sweden or Germany) have a longer-standing tradition of humanitarian assistance to refugees and asylum seekers than others (e.g. Poland or Hungary) (Juncho, 2007). Presumably, a longer national history and experience with providing refugee protection may make humanitarian assistance more culturally ingrained and more prevalent in individuals’ preferences for asylum and refugee policy. This would mean that individuals who live in countries with longer-standing traditions of providing humanitarian assistance to refugees would have preferences more inclined towards policies that provide unconditional protection while those that live in countries with more limited experience with providing refugee protection would be more inclined towards policies that restrict provision. Yet, at the same time, we believe that this is complicated by the fact that countries such as Hungary and Poland, while not having long histories of assisting asylum seekers and refugees, may nevertheless strongly adhere to the logic of humanitarian assistance given the past experiences of compatriots who fled persecution and were resettled abroad. It should also be noted that these countries have in recent years taken initiatives to host refugees, such as extending open arms to Ukrainian refugees who fled to Poland during the Russo-Ukrainian war. Moreover, in their study of what kinds of refugees Europeans prefer, Bansak et al. (2016) find that preferences are similar across countries. For instance, in that study the preferences of Hungarians are found to be remarkably similar to those of Swedes. Based on this reasoning, we expect the tension between the logic of perceived national interest and the logic of humanitarian assistance to be reflected in public preferences for asylum and refugee policy across European countries.
Conceptualizing multi-dimensional asylum and refugee policy
Considering the range of policy decisions that need to be taken when dealing with asylum seekers and refugees, we suggest that the core dimensions of asylum and refugee policy in European countries relate to: (1) the right to apply for asylum; (2) the resettlement of already recognized refugees; (3) the return of asylum seekers whose applications for protection have been unsuccessful; (4) family reunification for recognized refugees; (5) how national governments work with supranational institutions such as the EU to govern the asylum and refugee process; and (6) the provision of financial assistance to so-called ‘first countries of asylum’, i.e. lower-income countries that host large numbers of refugees near conflict regions. Within dimensions, we differentiate between policy specifications that utilize limits or conditions and those that do not. 1
Our conceptualization builds on existing approaches to measuring asylum and refugee policies that can be found in the small existing research literature (see Hatton, 2016; Helbling et al., 2017). Similar to existing approaches, we distinguish between policy dimensions that relate to the regulation of the admission/access of asylum seekers and refugees to the country on the one hand (dimensions 1 and 2), and the treatment and rights of asylum seekers and refugees after admission on the other (dimensions 3 and 4). Our approach differs by extending the scope to two policy areas that relate to governance (dimension 5) and financial assistance to lower-income countries hosting refugees near conflict regions (dimension 6). This allows us to cover the most important aspects of asylum and refugee protection discussed in public policy debates over the past decade. We use both academic research and recent proposals for policy reform in EU member states to inform our conceptualization of asylum and refugee policies. Rather than focusing on preferences towards very specific policy proposals, we aim to focus on the basic principles underlying different refugee and asylum policies.
While our conceptualization of asylum and refugee policy applies to all high-income countries, the focus of the empirical analysis in this study is on Europe. For this reason, we elaborate the following six policy dimensions below in the specific context of recent European policies and debates.
Asylum
Countries that have ratified the 1951 Geneva Convention are obligated to examine applications for asylum from any non-citizen on their territory, without any numerical limits or conditions. However, there is no legal requirement for countries to facilitate legal travel and immigration to claim asylum. In practice, most high-income countries do not offer asylum seekers, especially those from lower-income countries, the opportunity to travel to their countries legally to apply for asylum. As a consequence, most of the world’s forced migrants who are trying to claim asylum in high-income countries must do so by engaging in unauthorized crossing of national borders that often involves long and dangerous journeys across land or sea.
A central question in debates about this policy dimension relates to whether and how the right to apply for asylum (as stipulated in the Geneva Convention and also in European asylum laws) should be limited in some ways. Following the large increase in asylum applications in the EU in 2015 and 2016 several European countries have openly debated restricting the right to asylum. For example, Germany, which received almost half of all asylum applications made in the EU during 2015–2016, began to discuss the introduction of annual limits (Obergrenze) on the number of asylum applications, while a joint ‘vision statement’ for reforming Europe’s asylum and refugee policies published by the Austrian and Danish governments in November 2018 proposed to limit the right to apply for asylum to people fleeing from countries bordering the EU.
Resettlement
A second important policy dimension relates to the scale of resettlement facilitated by the overall asylum and refugee policy. Resettlement involves the transfer of refugees, whose refugee status has been determined by the United Nations High Commissioner for Refugees (UNHCR), from a ‘country of first asylum’ (i.e. a host country in or near the region of conflict) to another state that has agreed to admit them. With 85% of the world’s recognized refugees hosted by lower-income countries, resettlement to higher-income countries can, in theory, be an important tool for reducing global inequalities in the protection of refugees.
However, states decide about the scope and hence, the limits of resettlement. Currently, such limits are widely implemented: the global numbers of resettled people have typically been very small compared to the total number of refugees in the world. Resettled people constitute less than 5% of the global refugee population. Among EU member states, resettlement has increased in recent years – from 8000 in 2015 to 27,000 in 2019 – but it remains at a relatively low level. There is considerable between-country variation in the EU: while a small number of countries (the UK, Sweden, France, and Germany) admitted most resettled refugees in the EU over the past few years, some EU countries such as Austria, Poland, Hungary, and Denmark did not allow any resettlement in certain years.
Asylum seekers whose applications for protection have been unsuccessful
A fundamental question in asylum and refugee policies is how to deal with asylum-seekers whose applications have been unsuccessful. Under what conditions can and should they be sent back to their home countries? This is a critical and highly contested issue in Europe as more than half of applications for asylum in EU countries are unsuccessful (Eurostat, 2018).
According to the international legal framework for refugees, states are bound to the principle of non-refoulement, which is a core principle enshrined in the Geneva Convention and forbids states to return rejected asylum-seekers to countries where they are likely to face serious harm and danger. There is an ongoing debate about the extent to which some of the EU’s current policies adhere to the principle of non-refoulement. For example, aspects of the implementation of the EU’s migration deal with Turkey (European Council, 2016) have been criticized for violating the principle of non-refoulement (for example, Carrera et al., 2017; Council of Europe, 2020). Similarly, non-refoulement has been at the centre of controversies about ‘search-and-rescue’ missions carried out by European boats in the Mediterranean (Cogolati et al., 2015).
Family reunification
Next, we consider the rights that recognized refugees are given to bring their family members into the European country providing protection. While states are not legally bound to guarantee family reunification to recognized refugees, family reunification is widely considered a de facto human right, the underlying humanitarian principle being that the family unit is deserving of protection, not simply the individual (Cholewinski, 2002). Over the past few years, the regulation, limits, and conditions associated with family reunification for refugees have been key issues in public debates about immigration in many EU countries. EU laws, especially the EU Directive on family reunification, provide a framework for the regulation of this issue but they still leave EU member states considerable room for manoeuvre. In practice, national regulations vary across countries. Some EU member states have started to introduce conditions such as requiring that recognized refugees demonstrate that they can financially support their family members (see Council of Europe, 2017, 2018). Minimum income requirements are common in the regulation of family reunification of labour migrants, but international norms and EU laws encourage countries to waive such material requirements for refugees.
Governance
A central question in the governance of the asylum process and refugee protection in Europe relates to the role of the EU. Currently, the asylum process is mostly under the control of member states. The Common European Asylum System specifies common minimum standards that asylum procedures must fulfil at the European level but the examination of asylum applications and the decision whether to grant or reject an application rests with member states. There has been great variation across EU member states in the number of asylum applications received, the length of the asylum process, and the acceptance rates of applications made by asylum seekers from the same countries (Winn, 2021). In response, it has been proposed that some aspects of policymaking on asylum and refugees should be shifted from the national to the EU level (see Lücke, 2018). Those in favour argue that a strong centralized European agency would ensure more uniformity and fairness in the asylum process by standardizing asylum procedures and the criteria for obtaining protection. Advocates of greater centralization of European asylum processes (supported by greater joint financing) have also suggested that it may help ease the burden on the countries that currently receive most asylum seekers.
Financial assistance to low income countries hosting refugees near conflict regions
A sixth important dimension of a country’s overall asylum and refugee policy relates to how much and what types of economic and financial assistance it provides to ‘first countries of asylum’ in or near conflict regions. Providing financial assistance to non-EU countries hosting large numbers of refugees is an important policy tool. A number of European political leaders have advocated a radical paradigm shift in Europe’s asylum and refugee policies, away from protecting refugees in Europe to providing much more assistance to first countries of asylum near conflict regions to protect refugees and facilitate their economic and social integration in the host country. The justifications for this approach typically include the argument that those refugees who make it to Europe are not among the most vulnerable, and that it would be significantly cheaper to help protect and integrate refugees in lower-income countries near conflict regions rather than in Europe.
In addition to the question of how much financial assistance should be provided to first countries of asylum, a key policy question relates to whether and how this assistance should be conditional on these countries’ efforts to help control migration to Europe. For example, under the ‘EU-Turkey Statement’ agreed in 2016, the EU provides six billion euros in assistance to help with the protection of refugees in Turkey, in exchange for a range of Turkish policy measures, including stepped-up coast guard activities that help reduce onward migration of refugees to the EU.
Empirical approach
To examine the structure of public preferences for asylum and refugee policy, we draw on an original conjoint experiment. Conjoint experiments are particularly useful for studying public preferences towards multi-dimensional policy issues. Rather than asking respondents to assess and rate certain aspects of policies independently of one another, conjoint experiments require respondents to make a series of constrained choices between pairs of hypothetical policy options that differ across several dimensions. Since the features of the policy dimensions are fully randomized across each respondent, it is possible to identify the relative causal impact of the different policy features of these dimensions on public support for the overall asylum and refugee policy. Moreover, a conjoint design minimizes social desirability bias, which is of concern when surveying respondents about sensitive issues such as refugees and asylum-seekers. It does so by minimizing the ability of respondents to select a response they believe to be politically correct (Hainmueller et al., 2015), since the policy options vary across several dimensions.
We conducted a fully randomized choice-based conjoint experiment that was fielded online by the survey company Respondi in May 2019 in eight European countries: Austria, France, Germany, Hungary, Italy, Poland, Spain, and Sweden. 2 The countries have been selected on the basis that they represent a variety of experiences with refugees and asylum seekers, cover several geographic areas of the EU, and include a wide variety of labour market, welfare, and cultural institutions. These countries are also among the most populous countries in the EU, making our overall sample more representative of overall European public preferences. 3 In each country, we conducted a survey using a nationally representative sample of individuals (n = 1500) who are 18 years of age or older.
After a short introduction 4 that explained the exercise and briefly defined the terms asylum-seeker, refugee, and resettlement, individuals participating in the survey were given five conjoint tasks, consisting of two policies each. This means that, over the course of the experiment, each participant evaluated 10 randomly generated policies. In each of the five conjoint tasks, respondents were presented with two policies side by side which differed randomly in their policy features across six policy dimensions. The policy dimensions presented in the conjoint experiment, and the various policy features that are randomly assigned within each dimension, were informed by our conceptualization of multi-dimensional asylum and refugee policy discussed earlier in this article. Each of the six dimensions could take on two or three possible features, which are all listed in Table 1. 5 The order in which the dimensions were listed was randomized for each respondent, but the order remained the same across the five conjoint tasks. An example of a conjoint task is shown in English (translation of the German version) in Figure 1.
Experimental policy features, by six dimensions of asylum and refugee policy.
Sample Conjoint Task (English translation of a conjoint task shown in Germany).
In each conjoint task, respondents were asked to make their policy choice in two ways. First, respondents had to make a binary choice about which policy they preferred. In our analysis, a policy takes on the value of 1 if the person chose the policy in a conjoint task or 0 if the person did not choose the policy. The binary choice constrained respondents to make trade-offs and decide between the two policies they faced. Second, respondents were asked to rank their support for each policy on a scale from 1 (highly unsupportive) to 7 (highly supportive). 6
Immediately following the five conjoint tasks, the survey asked respondents a series of questions about their age, gender, education, political orientation, preferred scale of immigration, and political trust. The order of these questions was randomly assigned although they always followed the conjoint tasks. The precise wording of all these questions can be found in the Online appendix.
We analyze the results of the experiment by computing the average marginal component effects (AMCEs). The AMCE represents the average difference in the probability that a policy is supported when comparing two possible features within the same policy dimension, where the average is taken over all possible combinations of other policy dimensions (Hainmueller and Hopkins, 2015). The AMCE is estimated by regressing an indicator for whether the respondent chooses a given policy on the various dimensions of the policies that are listed above. It should be noted that AMCEs represent the average preferences but not necessarily the majority, since an average can be due to a majority preference or a strong preference of a minority instead (Abramson et al., 2019). We keep the number of profiles low as Ambramson et al. (2019) point out that this is a necessary condition for which AMCEs most closely approximate majority preferences.
The unit of analysis is the rated policy, meaning that we have 120,000 observations in our models. Each of the respondents (n = 12,000; 1500 per country) has made five choices, each of which included two different policies. When computing the AMCEs, we follow standard practice and apply cluster-robust standard errors at the respondent level to correct for possible within-respondent clustering. In all our analyses, we use entropy-balancing survey weights to correct for sampling error. In addition, we account for multiple testing to safeguard against type I error.
Results
The effects of policy features on support for asylum and refugee policy
We begin by examining what types of asylum and refugee policies are supported or opposed by our European respondents. Figure 2 shows the results of our conjoint experiment. It displays AMCEs to indicate the effects of policy features on the probability of accepting an asylum and refugee policy relative to the dimension’s reference category. 7 The dots report point estimates and the bars indicate the 95% confidence intervals around those points. The dots that do not have confidence intervals show the reference categories for each policy dimension. The reference category for each policy dimension is meant to capture the closest approximation of the current status quo in asylum and refugee policy.
Effects of policy features on the probability of accepting an asylum and refugee policy.
Our results suggest that Europeans are generally committed to providing protection to asylum-seekers and refugees, meaning that they do not systematically prefer the most restrictive policy feature in each dimension. For example, policies that return refused asylum seekers to situations where they could face harm moderately 8 reduce public support for the asylum and refugee policy when compared to policies which never return refused asylum-seekers to such situations (AMCE = –0.037 meaning that public support for the asylum and refugee policy is reduced by about four percentage points). Similarly, policies that do not provide any opportunities for family reunification for recognized refugees moderately reduce public support when compared to policies that always allow family reunification (AMCE = –0.047).
However, our results also suggest that Europeans’ commitment to providing asylum and refugee protection is contingent upon policy features which allow for a means of control, namely through the implementation of limits or conditions. As shown in Figure 2 below, along several dimensions, the public is most supportive of policy features which include limitations and conditions without fully eliminating opportunities for protection. For instance, when it comes to family reunification, policies that require refugees to cover the cost of living of their family members are significantly preferred to family reunification without this condition. In this case, the effect size, 7.4 percentage points, is large compared to the other effect sizes. Similarly, the public is slightly more supportive of policies that condition financial assistance for non-EU countries that host refugees on these countries’ efforts to help reduce the number of asylum-seekers coming to Europe than to policies that provide no financial assistance at all (AMCE = 0.026). Unconditional financial assistance has a moderate negative effect on policy support compared to the baseline of no support (AMCE = –0.050). While respondents have not penalized policies that include low levels of refugee resettlement, they have penalized policies that include high resettlement: these policies are 3.1 percentage points less likely to be chosen than policies that include the baseline of ’no resettlement’.
Underlining the focus on limits and conditions, policies that place annual limits on the number of asylum applications are moderately more supported than those that do not apply limits (AMCE = 0.051). Finally, the issue of whether asylum-seeking is governed at the national or the EU level has an important impact on public preferences. Policies that involve an EU central agency for processing and deciding asylum applications in Europe are significantly more penalized than policies where asylum assessments and decisions are made by national governments (AMCE = –0.054).
Country-level results
Remarkably, people prefer similar types of asylum and refugee policies in all the eight European countries we surveyed. Figure 3 reports the AMCEs for each country separately. 9 In almost all countries, there is a clear preference for decisions on asylum applications to be made by national government rather than a centralized European agency. Regarding the types of policies that are preferred, we find that the public systematically prefers policies that feature controls and conditions over policy alternatives that either abolish protection or provide it unconditionally, with few exceptions. As shown in Figure 3, respondents in all countries place a premium in terms of their support on policies that place annual limits on asylum-seekers over those that do not feature limits and, in almost all countries, penalize policies which feature high levels of refugee resettlement. Moreover, in almost all countries, the public places a premium on policies that feature conditional family reunification for recognized refugees. Most countries also have publics that prefer policies that provide conditional financial support to non-European countries near or in conflict regions.
Effects of policy features on the probability of accepting an asylum and refugee policy, by country (estimated AMCEs).
We also observe a strong commitment to upholding certain principles of protection for asylum seekers and refugees across the countries in our study. None of the countries in our study favours policies that include returning refused asylum seekers to places where they might face serious harm. This is true even in Hungary – a country with a government well-known for its anti-refugee policy positions in recent years: Hungarians are significantly more likely to penalize policies that feature returning refused asylum seekers if they might face serious harm. Moreover, in almost all countries, the public penalizes policies that feature the abolition of family reunification for recognized refugees.
Despite these commonalities across European countries, there are still some differences: For instance, in Hungary the public penalizes both high and low resettlement. In contrast, people in Spain place a premium on policies that include low or high levels of refugee resettlement. In a few countries, there is no premium for conditional financial assistance to non-EU countries hosting refugees compared to no financial solidarity (France, Italy, and Poland). In Spain, the public does not give a premium to policies that provide conditional family reunification nor does it penalize policies for providing unconditional financial assistance to low-income countries. Finally, in Italy there is no penalty for policies where a central European agency assesses and decides on asylum applications in Europe. This study is not intended to analyze the causes of these differences across countries. Our research design allows us only to speculate that these country differences could be attributed to a range of issues including, for example, the heterogeneous exposure to the ‘refugee crisis’ in 2015–2016, historical experiences with immigration, welfare state generosity, the state of the national economy, and various other socio-cultural differences.
Robustness
We have subjected our findings to a series of robustness checks which can be found in the Online appendix. First, our results are consistent when we include additional controls in our analysis of how different policy features across the six dimensions affect support for asylum and refugee policy. To account for the fact that this is a conjoint experiment, one cannot merely add control variables for the sub-groups to the model. Instead, it is necessary to add interactions between each of the sub-groups (e.g. age, gender, etc.) and each policy feature. As shown in the Online appendix, the AMCEs remain similar across sub-groups: they only sometimes decrease slightly in absolute value and remain statistically significant throughout. These results show that our experimental estimates are remarkably robust, across all six policy dimensions, to the inclusion of additional control variables at the individual level. Among other things, this confirms that the experimental setup has successfully randomized policy features across different individuals, thus underscoring the internal validity of the results.
To ensure that the covariates (e.g. EU trust or attitude towards migration) were unaffected by the treatments, we conduct tests of sample stability included in the Online appendix. These results underscore that the reported pre-existing political views and characteristics of the respondents were not affected by the different treatments.
By calculating false discovery rates (Anderson, 2008; Benjamini et al., 2006) as well as a Bonferroni correction, we account for the potential risks of type I error when analysing outcomes in six dimensions at the same time. The results indicate that all the main results in Figure 2 are fully robust to multiple hypothesis testing corrections (see the Online appendix).
Conjoint tasks can be cognitively taxing on respondents because they require a certain degree of concentration. Typically, we would expect fatigued respondents to choose a random policy, thus creating a bias towards non-effects. Still, there could be other heuristics at play such as stronger reactions to particular policy features. To reduce the risk of bias from survey fatigue we restricted the number of tasks to five per individual, which is well within the number of tasks that a respondent can complete before survey sufficing downgrades response quality (Bansak et al., 2018). To help participants focus on the conjoint tasks, we required them to participate only on a computer and not on a smart device such as a tablet or cell phone. 10 In addition, we analyze whether estimated preferences change as more tasks are conducted to ensure that any remaining form of fatigue does not affect our results strongly. The estimates, displayed in the supplementary material, show no statistically significant changes in estimates. Survey fatigue thus does not seem to pose a problem within our experiment. If anything, it biases our results towards insignificance.
While our results can be interpreted as internally valid, the results have to be interpreted subject to our choice of applying equal weights for each country (due to the similar sample size in each of the countries included in our analysis). To arrive at the estimates for the marginal means for an average citizen across the eight countries, the reweighted results using the size of the represented population in each country are shown in the Online appendix.
Discussion and conclusion
The protection of asylum seekers and refugees has become one of the most politicized and contested political issues in the EU, yet there has been no research on the structure of public preferences for asylum and refugee policies. We find that Europeans are supportive of policies that provide protection to asylum-seekers and refugees but this support tends to be contingent upon policy features which allow for a means of control, namely through the implementation of limits or conditions. Our research design allows us to examine simultaneously the separate impacts of multiple dimensions of policy on public preferences, revealing that Europeans generally prefer to uphold protections for individuals who face persecution but in a more controlled way, using limits and conditions, than is currently the case.
In contrast to the highly divisive political debates between European countries about how to reform asylum and refugee policies, we find remarkable consistency in public asylum and refugee policy preferences across the different countries we surveyed. In spite of some cross-country differences in public preferences, in each of the eight countries included in our study, the public tends to prefer asylum and refugee policies that feature limits and conditions over policy alternatives which either abolish protection or provide it unconditionally. Even in Hungary – which since the 2015 refugee crisis has become well-known for the government’s anti-refugee policy positions – the public supports policies that provide certain conditional and limited protections, rather than rescinding protection altogether.
Our findings urge the current scholarship to move beyond a simplistic continuum of restrictive vs. permissive preferences for asylum and refugee policy. Our study goes beyond existing approaches to demonstrate that public preferences tend to opt for an alternative path which neither endorses the unlimited and unconditional provision of rights and protections nor the other extreme of abandoning them entirely. In our view, this alternative path emerges from citizens’ attempts to reconcile an internal conflict between humanitarian concerns and perceived national interests when thinking about asylum and refugee issues. Our findings are also relevant for the larger body of literature on immigration policy preferences in general, where a great deal of scholarship has focused on what migrant characteristics the public prefers, such as skills, gender, labour market status, or ethnicity rather than on public preferences for the multidimensional policies and procedures that govern the admission and rights of migrants.
In terms of public policy debates, our research suggests that some aspects of the current model of the international refugee system are misaligned with the more control-based model that Europeans would prefer. For example, an annual limit on the number of asylum seekers would be incompatible with the 1951 Geneva Convention which most high-income countries (including all EU countries) have ratified. While our research only provides a snapshot of public opinion at a specific point in time and public policies do not always reflect public preferences, in light of the recent politicization of asylum and refugee policy across European countries, our results raise questions about the political sustainability of some aspects of the status quo of international asylum and refugee policies.
Therefore, our findings also demonstrate that policy-makers in Europe face some important challenges in taking account of the political demands of the public, while at the same time not violating international treaties. A fruitful avenue for future research would be to consider the theoretical and empirical basis of political legitimacy for asylum and refugee policy. Our research draws attention to a potential crisis of legitimacy around the governance of asylum and refugee protection in the EU. We need to understand better the causes of the disconnect between the policy preferences of European citizens and aspects of the current asylum system based on long-standing international norms and agreements. This line of research might illuminate potential policy solutions that could restore political legitimacy and help the public (re-)gain confidence in their national government and Europe’s ability to govern in this policy area.
Supplemental Material
sj-zip-1-eup-10.1177_14651165211006838 - Supplemental material for What asylum and refugee policies do Europeans want? Evidence from a cross-national conjoint experiment
Supplemental material, sj-zip-1-eup-10.1177_14651165211006838 for What asylum and refugee policies do Europeans want? Evidence from a cross-national conjoint experiment by Anne-Marie Jeannet, Tobias Heidland and Martin Ruhs in European Union Politics
Supplemental Material
sj-pdf-2-eup-10.1177_14651165211006838 - Supplemental material for What asylum and refugee policies do Europeans want? Evidence from a cross-national conjoint experiment
Supplemental material, sj-pdf-2-eup-10.1177_14651165211006838 for What asylum and refugee policies do Europeans want? Evidence from a cross-national conjoint experiment by Anne-Marie Jeannet, Tobias Heidland and Martin Ruhs in European Union Politics
Footnotes
Acknowledgements
The authors would like to thank Esther Ademmer for her contribution to various stages of the project. We are also grateful to Matthias Lücke, Saime Özçürümez, Elias Dinas, the editor, and three anonymous referees for helpful comments.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research for this paper is part of the Mercator Dialogue on Asylum and Migration (MEDAM), an international research and policy initiative funded by Stiftung Mercator.
Supplemental material
Supplemental material for this article is available online.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
