SISP Conference - Milan, September 15-17, 2016

Immigration crisis as EU crisis? Nicolò Conti* Danilo Di Mauro* Vincenzo Memoli^

*Unitelma Sapienza University of Rome ^University of Catania

Abstract In recent years, the number of immig rants reaching the EU has grown dramat ically. Problems ass ociated with immigrat ion have become more pressing, particularly as massive flows of asylu m seekers escaping war in the Midd le East have started to reach Europe. National reception systems show clear organizational prob lems especially in those member states most exposed to the immigrat ion pressure, while so me govern ments have revised their immig ration policies to pugnaciously limit the nu mber of arrivals. The refugee crisis and its polit ical repercussions have been felt with d ifferent intensity and timing across Europe, fro m the Southern and Eastern borders to France, Germany, the United Kingdom and Northern Europe. In this crit ical situation, coordination has proved problemat ic due to the nested interests of the member states. A lack of efficacy in finding solutions at the European level to the immig ration crisis has clearly been perceived by the citizens and has certainly influenced their attitudes toward the EU. This paper addresses the political feasibility of an integrated EU immig ration policy. What are the attitudes of citizens toward immigrants? What role do they think the EU should play to address the immigration crisis? Using data fro m the EUENGA GE project, we exp lore the main components that inform citizens’ attitudes toward immigrants in ten EU cou ntries and we test a set of theoretical arguments that aim at explaining their attitudes.

Introduction In 2015, more than 1 million of people arrived to Europe crossing the Mediterranean Sea, against 250 thousands in 2014 and 60 thousands in 2013 (UNCHR1 ). At the same time, asylum seekers also increased dramatically. The Migration Commissioner Dimitris Avramopoulos defined the one of the past two years “the worst refugees crisis since World War II” 2 . The EU is experiencing a refugee crisis that is also a humanitarian one and that, according to some observers, has transformed 1

See http://data.unhcr.org/mediterranean/regional.php (last access, August 2016) European Commission press release, 14th August 2015, http://europa.eu/rapid/press-release_SPEECH-155498_en.htm . 2

its Southern borders into a death trap (see Fargues, 2015). This critical scenario has made immigration a more salient issue in many EU countries: according to Eurobarometer data, in 2015 it was actually the main concern among Europeans. 3 The connection often made by media and by many successful radical parties between immigration, crime and terrorist attacks – a connection that is now also rooted in people’s minds (Fitzgerald et al., 2012) – have contributed to spread moral panic within society and to bring the immigration issue on top of the agenda. In this paper, we focus on citizens’ attitudes toward the EU immigration policy. Our aim is to understand the political feasibility of an integrated immigration policy, notably if, how and why citizens would support greater EU coordination in this specific policy field. We argue this is an important question for the current debate about the EU capacity to react to the recent immigration crisis and to other (economic, security) crises more in general. Social sciences have largely focused on the explanatory factors of attitudes toward immigrants and immigration. Scholars of different disciplines have tested the influence of a long list of explanatory factors on the formation of public opinion on both immigration nd immigrants. Some of them emphasize the role played by identity (Sides and Citrin, 2007; Bail, 2008; Levanon and LewinEpstein, 2010; Haubert and Fussell, 2006; Verkuyten and Martinovic, 2015). Others stress the influence of economic interests, sociotropic explanations (Hainmueller and Hopkins, 2015), the state of national economy (Sides and Citrin, 2007; Ruist, 2016; Rodolph, 2002), while in recent times some scholars have questioned the role of the economic and financial crisis on such attitudes (Creighton et al., 2015). Other contextual variables may also play a significant role in shaping citizens’ attitudes. The number of immigrants in a territory along with their socialization with natives are explanatory factors that have been investigated (Hopkins, 2010; Hjerm, 2009). People react differently to immigration also depending on the characteristics of immigrants (Hopkins, 2015), particularly on their belief system (Beierlein et al., 2016; Davidov and Meuleman, 2012).

3

See Standard Eurobarometer 83-First Results, Spring 2015, http://ec.europa.eu/public_opinion/archives/eb/eb83/eb83_en.htm (last access August 2016)

2

In the EU, the influence of interests is bound to differ across countries (Kunovich, 2004) and recent studies show that in the last decade (and before the migration crisis), attitudes toward immigration have become more positive in Northern Europe - such as Sweden or Denmark - but more negative in Southern Europe - especially in Spain and Greece (see Turner and Cross 2015). Despite a large literature on attitudes toward immigrants, studies on citizens’ attitudes toward EU immigration policy are quite rare. The limited number of works that are available (and that are not up-dated to the most recent developments) reveal that, along with individual economic interests, identity is relevant for both attitudes toward the EU and immigration (Luedtke, 2005). Some argue that the attitudes toward immigration and the support for EU integration are (negatively) related (De Vreese and Boomgarden, 2005), but this relationship has remained underexplored. Under what conditions would citizens agree to support stronger EU integration of immigration policy? Our analysis has benefited from the availability of original data from the first wave (2016) of the EUENGAGE survey of public opinion that was specifically designed to test people’s reactions to the most urgent challenges faced by the EU. The data allowed us to investigate the above research question in light of the most recent developments, hence to test the validity of past theory in the current critical scenario. The paper is structured in three sections. The first section introduces our theoretical framework and working hypotheses. The second section presents the data and the method used to test our hypotheses. The third section discusses our main results, while a short conclusion summarizes the main findings of the paper.

3

Attitudes toward immigrants and support for EU action in immigration: defining some hypotheses

Traditionally, member states have been rather reluctant to delegate sovereignty to the EU in the immigration field. Despite this negative attitude, in the last ten years, EU common policies on immigration and asylum have marked important developments (see Lavenex, 2006). However, the scope of the EU immigration policy is still limited, as decision on entrance of non-EU migrants and on asylum are taken by the member states. To address our research question, we focus on the concept of EU integration considered as EU coordinated action on immigration (including asylum policy). Along with a large set of explanatory factors, studies on public opinion and immigration have only rarely addressed the multidimensional nature of the phenomenon (see Hellwig and Kweon 2016). Citizens may distinguish among different components and generate distinctive attitudes toward immigration, immigrants and immigrations policies (for a short review see Ceobanu and Escandell, 2010). Building on these premises, we empirically test whether EU citizens have distinct attitudes toward immigration, immigrants and EU immigration policy. This is the first step of our analysis; the second step consists of the accurate examination of the relationships among these different components. The determinants of attitudes toward immigration have been investigated by scholars of different disciplines: socialization and socio-psychological factors, along with economic, contextual, institutional and political factors are at the core of the kaleidoscopic set of elements explaining how people perceive immigration and immigrants. At the same time, a bulk of studies have investigated citizens’ attitudes toward the EU (among others, see Gabel, 1998; Boomgaarden et al., 2011). Although these two subjects have largely been considered in isolation, some common explanatory factors have emerged in the literature. For example, some studies found national identity to negatively affect support for European integration at any level (see McLaren, 2002; Marks and Hooghe, 2003) as well as perceptions of immigrants/immigration (Luedtke, 2005). Moreover, other 4

studies show that general attitudes toward a broad phenomenon affect people’s views (and voting choice) on more specific aspects and institutional settlements regarding the same phenomenon (Burstein, 2003; Simon and Lynch, 1999; Page and Shapiro, 1983; Soroka and Wlezien, 2010). Notwithstanding our baseline expectation about the existence of different dimensions of attitudes, in the paper we test the hypothesis that attitudes toward immigration influence support for EU integration in the same policy field (H1). Although the nature of the EU remains ambiguous - it is not a true federal state nor an international organization – its institutions are a prominent part of the integrated multilevel governance system operating in Europe (see Fossum and Menendez, 2010). Different levels are responsible to address different policy problems, some of them are by definition more transnational and require coordination at the European level. EU citizens are part and (either positively or negatively) evaluate this unique process of regional integration (see Hobolt and Tilley, 2014). They recognize that the EU institutions are authorities that have been created to guarantee policy outcomes that address their needs (Hobolt and Tilley, 2014; Di Mauro and Memoli, 2016). Research shows that EU citizens are also able to distinguish between support for the EU system as a whole (diffuse support) and for its actions (specific support) (Boomgarden et al, 2011; Beaudonnet and Di Mauro, 2012). Because immigration has a transnational scope, we expect that people perceiving immigrants as a threat would be ready to support more EU action to limit their presence, as opposed to single country interventions that could prove ineffective given the transnational scope of the phenomenon. When immigration into the EU is perceived as a threat support for EU policy integration should increase (H2a). On the opposite, when immigration is perceived as – either individual or collective – benefit support for EU integration should decrease (H2b). These two conjectures have been put forward by past research (Luedtke, 2005) but have not been tested yet in EU’s current reality. It should also be noted that these conjectures are rather contested, for example De Vreese and Boomgaarden (2005) argue that anti- immigrant attitudes produce higher opposition to European integration as a whole. Since support for the EU is a rather complex stance and can be of several 5

types (Boomgaarden et al 2011; Beaudonnet and Di Mauro, 2012; Fuchs, 2009) to confirm our hypotheses would give evidence to the fact that attitudes toward immigration have an opposite effect on support for EU policy integration. This would be an interesting finding pointing to a relevant phenomenon: those subjects who feel more threatened by immigration are also potential supporters of EU integration in this policy field.

Data, variables and methodology

In this paper we use the data collected by the Horizon2020 EUENGAGE project. In June 2016, the project conducted the first wave of a public opinion survey in ten countries 4 addressing some urgent issues such as immigration, the economic crisis, Brexit and international security, along with more general attitudes toward national and EU institutions. The survey was conducted through a mix of CAWI/CATI methods. From the available data, we selected information on attitudes toward immigration in general, 5 together with more specific information on postures about EU policy integration.6

4

Czech Republic, France, Germany, Greece, Italy, the Netherlands, Poland, Portugal, Spain and the United Kingdom. Questions read as follows: Can you please indicate to what extent you agree or disagree with the following sta tements about the general impact of immigration on [COUNTRY] and [NATIONALITY]? 1) Immig rants increase the likelihood of a terrorist attack in (COUNTRY); 2) Immig rants are a significant cause of crime in the (country) 3) Immigrants take jobs that [COUNTRY] people do not want to take anymore; 4) Immig rants contribute more in taxes than they benefit fro m health and welfare services; 5) Immigration in general will imp rove our culture with new ideas and customs; 6) The religious practices of immigrants are a threat to the [nationality] way of life and its traditions. 6 Questions read as follows: In recent years, several EU countries have been facing extensive immigration from non -EU countries. Different policies have been suggested and we would like to know your view. For each of the following policy alternatives, please position yourself on a scale from 0 to 10, where 0 means that you fully support the policy at the top/left and 10 means that you fully support the policy at the bottom/right. If your views are s omewhere in between, you can choose any number that best describes your position . 1) [COUNTRY] should decide for itself how many immigrants to accept each year vs. The European Union should decide how many immig rants should be accepted by each Member State each year (0= [COUNTRY] should decide for itself how many immig rants to accept each year; 10= The European Un ion should decide how many immigrants should be accepted by each Member State each year); 2) The country immigrants arrive in should be responsible for hosting them vs. All the European Union's Member States should be responsible for the hosting of immigrants receives (0= The country immigrants arrive in should be responsible for hosting them; 10= All the European Union's Member States should be res ponsible for the hosting of immigrants); 3) The costs of providing asylum should be shared among all the Eu ropean Union's Member States vs. Each country should bear the costs depending on how many asylum seekers it receives (0= The costs of providing 5

6

In the analysis, we first proceeded running a factor analysis to find whether the ‘observed variables are linear combinations of underlying (unobservable) factors’ (Kim and Mueller, 1978:8). This method allows to detect measures of objective associations within different dimensions that are more parsimonious than the sum of information used to extract them (see Di Stefano et al. 2009). This first step in the analysis tests the validity of our baseline expectation that citizens’ attitudes are multi-dimensional and provides the dependent variables that are then used to test our working hypotheses. For the test of the hypotheses, we thus built as many multiple regression models as the number of dimensions given by the factor analysis. The independent variables included in those models are institutional trust (in general and in government), territorial attachment/identity (regional, national, European, cosmopolitan), ideology, social capital, and a set of sociodemographic variables (education, occupation, gender, age). These are the predictors used in the first two models, while model three also includes general attitudes toward immigration. The initial model has been built as follows: Y = α+βx1+βx2+βx3+βx4+βx5+βx6+βx7+βx8+βx9+βx10+βx11+ e Where: Y = dependent variables resulting from the factor analysis; 𝛽𝑥1=trust;7 𝛽𝑥2= confidence in government;8 𝛽𝑥3=membership;9

asylum should be shared among all the European Un ion's Member States; 10=Each country should bear the costs depending on how many asylum seekers it receives). 7 The question addressed to citizens is the fo llowing: Generally speaking, would you say that most peo ple can be trusted, or that you can't be too careful in dealing with others? The variable was recoding in the following way : 0= You can’t be too careful, 1= Most people can be trusted. 8 The question addressed to citizens is the following: How much of the time do you think you can trust [COUNTRY] government to do what is right? The variable was recoding in the following way: 0= Never -On ly some of the time, 1=Most of the time-Just about always. 9 The question addressed to citizens is the following: Do you actively participate in any of the following…? 1) An association of any kind; 2) A trade union; 3) A political party; 4) A citizens’ co mmittee; 5) A church or religious community. They are coded in the following way: 0=no, 1=yes. Aggregating the four variab les and recoding the index obtained, we have a new du mmy variable where 0=who do not belong to any association; 1= who belong to at least one association.

7

𝛽𝑥4=network size;10 𝛽𝑥5=regional identity;11 𝛽𝑥6=national identity;

12

𝛽𝑥7=European identity; 𝛽𝑥8=world identity;

13

14

𝛽𝑥9= redistribution of wealth; 𝛽𝑥10= left-right; 𝛽𝑥11= blaming;

15

16 17

10

The question addressed to citizens is the following: Do you actively participate in any of the following…? 1) An association of any kind; 2) A trade union; 3) A political party; 4) A citizens’ co mmittee; 5) A church or religious community. They are coded in the following way: 0=no, 1=yes. Aggregating the four informat ion in an additive index, we obtain a new measure which goes fro m 0, who do not belong to any association to, to 5, those who participate in all the associations. 11 The question addressed to citizens is the following: People feel different degrees of attachment to their region, to their country, to Europe, and world. What about you? Do you feel very attached, somewhat attached, not very attached or not at all attached to (your region)? The variable was recoding in the following way : 0=not at all attached/not very attached, 1=somewhat attached/very attached. 12 The question addressed to citizens is the following: People feel different degrees of attachment to their region, to their country, to Europe, and world. What about you? Do you feel very attached, somewhat attached, not very attached or not at all attached to (your country)? The variable was recoding in the following way: 0=not at all attached/not very attached, 1=somewhat attached/very attached. 13 The question addressed to citizens is the following: People feel different degrees of attachment to their region, to their country, to Europe, and world. What about you? Do you feel very attached, somewhat attached, not very attached or not at all attached to (Europe)? The variable was recoding in the following way: 0=not at all attached/not ve ry attached, 1=somewhat attached/very attached. 14 The question addressed to citizens is the following: People feel different degrees of attachment to their region, to their country, to Europe, and world. What about you? Do you feel very attached, somewhat attached, not very attached or not at all attached to (the world)? The variable was recoding in the following way : 0=not at all attached/not very attached, 1=somewhat attached/very attached. 15 The question addressed to citizens is the follo wing : Where would you place your views on this scale? 0 means you agree completely with the statement on the top/left; 10 means you agree completely with the following statement? 1) Redistribution of wealth fro m the rich to the poor. The variable was coded in the followin g way: 0= Fu lly in favor of redistribution of wealth fro m the rich to the poor, 10= Fu lly opposed to redistribution of wealth fro m the rich to the poor. 16 The question addressed to citizens is the following: In politics, people sometimes talk of left and right- Where would you place yourself on a scale from 0 to 10, where '0' means the extreme left and '10' means the extreme right? The variable is coded in the following way: 0=left, 10=right. 17 The question addressed to citizens is the following: How responsible do you think each of the following factors are for the present migration crisis that is affecting a number of EU countries? Use the scale from 0 to 10, where 0 means they are not responsible at all and 10 that they are fully responsible for the present economic crisis. If your views are somewhere in between, you can choose any number that best describes your position. 1) The lack of co mmon Eu ropean Union action against violence and poverty in the immigrants’ countries of origin, 2) The lack of co mmo n and unified European border controls. The variab les were coded in the following way: 0=not responsible at all, 10=fully responsible.

8

Our models allow to test different theoretical perspectives and to assess the direction of the relationship (if any) between attitudes toward immigration and EU policy integration. In the following section, we present the results of our analyses.

Analysis and discussion

In order to differentiate the many dimensions characterizing public attitudes toward immigration, we applied a factor analysis to nine variables selected from the EUENGAGE 2016 mass survey.18 The analysis unveils the presence of three dimensions explaining 64.3% of the total variance. As Table 1 shows, all indicators are positively interrelated to each respective dimension with values above 0.5. Citizens mainly appear to distinguish three dimensions. The first two relate to general attitudes toward immigration, while the third dimension concerns immigration policy. This result corroborates our expectation about the multi-dimensional nature of citizens’ attitudes. More specifically, the first factor explains 26.6% of the total variance. It relates to the benefits that immigrants bring to the hosting country in both economic and cultural terms. The second dimension refers to feelings of insecurity perceived by those citizens who associate immigration to threats such as terrorism and crime. This dimension explains 21% of the total variance. Finally, the third dimension – accounting for 16.7% of the total variance – refers to the role of the EU (vs. the national government) in immigration policy.

TAB. 1 HERE

18

To avoid problems of under or over representation in the sample we used weighted data. We rotated factors using the Varimax method that allows to minimize “the nu mber of variables that have high loadings on each factor and works to make s mall loadings even smaller” (Young and Pearce 2013: 84). As it is customary in this type of analysis, we adopted the Kaiser’s rule (1960) according to wh ich the main co mponents are selected when they show an eigenvalue greater than (or equal to) one.

9

We labeled these three dimensions Benefit, Insecurity and EU Integration, respectively. Then, we made use of their factor scores to build three corresponding indexes. Benefits measures the perceived gains stemming from immigration. Insecurity assesses sentiments of fear and concern about the impact of immigration on security. Integration measures the degree of public support for EU coordination of immigration policy.

TAB. 2 HERE

Moving now to the explanatory part of the analysis, Model one (table 2) shows that a general trust in others and confidence in government (beta=0.119) increase the probability of perceiving immigration as mainly beneficial. Social capital has a more complex relationship, as being a member of an association does not favor the perception of benefits of immigration (the opposite is true), but a wider networking does. Holding a strong European identity (beta=0.101) increases the perception of benefits, a result that confirms the influence of more unbounded identities on positive views on immigration (Sides and Citrin, 2007; Ceobanu and Escandell, 2010). On the contrary, regional and national identities are negative ly related to benefits. Although it has a limited impact, ideology also affects attitudes: those who self- locate toward the right of the political spectrum tend to see immigrants as a disadvantage (beta=-0.077). When the action of the EU is perceived as weak (blaming EU efficacy19 ), citizens adopt a reward-punishment logic and negatively evaluate the EU (beta=-0.039). Females are usually less likely to perceive immigration as beneficial (beta=-0.042), while we do not find different patterns across ages. The same is true for education. Although according to many education ‘has a marked effect’ on sentiments toward immigrants (Freeman et

19

This index was built applying a factor analysis (principal co mponent extraction) to our three variables on blame for immigrat ion policies. The result shows a single factor (exp lained variance =66.2%; reliability (Alpha of Cronback)=0.488). The index ranges from -2.966 (no blame for the EU) to 1.510 (blame).

10

al., 2015), from our data its impact does not appear statistically significant. 20 Preference for redistribution of wealth does not have a significant effect either, as well as different working conditions. The validity of the argument that those who perceive immigrants competing in the own labor market should also perceive them as a threat cannot be confirmed here (although a note of caution is necessary since our measure not only considers the economic benefits of immigration, but also the cultural ones). Looking more closely at the analyzed countries, we found that the perception of benefits of immigration outweighs that of costs. An interesting case in point is represented by the UK. Despite the supposed rampant anti- foreigner sentiment that was exacerbated by the EU Referendum campaign, immigrants in this country are mainly seen as beneficial (beta=0.286). There could be a rational explanation for that, as the immigrants that have arrived to the UK between 2001 and 2011 have paid 22 billion pounds more than they have actually received (see Dustmann and Frattini 2014). This is an interesting finding showing that, opposite to recurrent explanations about the vote in favor of Brexit, the British public is not so scared about the economic and cultural costs of immigration. The model on Insecurity (model 2) shows that trust, confidence in government, blame for lack of EU coordinated action in immigration, group membership, a more cosmopolitan identity, higher education and gender (female) create less sense of insecurity among citizens. On the contrary, membership of associations, employers and manual workers 21 and those leaning toward the right of the political spectrum feel more threatened by the sense of insecurity induced by immigrants. Overall, the association between insecurity and immigration does not appear so straightforward, nor it is so prominent in the analyzed countries.

20

It is worth mentioning that Hain mueller and Hiscox (2007) foun d a positive relationship between the educational level and the positive attitudes toward immigrat ion. In the model here presented, however, we focus on benefits coming from immigration and not specifically on racists orientations toward immigrants. 21 On the wages of native-born workers effect on perception of immigrants see also Hericourt and Spielvogel (2014).

11

After having examined the effects of the independent variables on general attitudes toward immigration, we can now focus on the origins of attitudes toward EU integration of immigration policy. Model three shows the results of our analysis. EU integration (vs. member state actions) is favored by trust and confidence in government. This result may appear contradictory at first glance: people who trust their own government tend to prefer the EU competence over the national competence in immigration. But theories of proxies (Anderson, 1998) show that people supporting their national institutions often support EU integration and our analysis corroborates this argument. National and regional identities create a more negative predisposition, while European identity increases the request for more EU integration in immigration. Right wing citizens are mainly against EU coordination in this field. The older generations are more supportive of EU policy integration, in the past they have also been described as more supportive of EU integration as a whole (see Gabel, 1998). Higher levels of education are positively related to Integration. A clear division appears among different types of works: self-employed citizens tend to support integration while the other categories tend to oppose it. However only results for manual workers are significant in the model. Since it is not possible to predict whether EU policies would be restrictive or more open to immigration, the role of labor market rationality may not be so straightforward in this relationship, although a conservative attitude of manual workers who tend to prefer the status quo (or the own member state action) is clearly visible. We can now test the impact of the general attitudes toward immigration on the support for EU coordination in this field. Both Benefit and Insecurity have a significant relationship (p