we grew here, you flew here. individual and country-level...

27
We grew here, you flew here. Individual and country-level determinants of natives’ attitudes towards immigrants in Europe Paper prepared for the IPSA ECPR Conference Whatever Happened to North-South? Sao Paulo, February 16-19, 2010 Work in progress – Please do not quote without authors’ permission – Comments are welcome This version February 2011 Abstract What determines the attitudes towards immigrants? Previous research has shown the strong and significant impact of individual level characteristics such as level of education, position on the labour market or position on a right-left scale of political attitudes. The role of country-level characteristics has only recently entered in the spotlight, as research has shown that level of wealth or existing policies can influence attitudes. This paper takes the economic explanation of attitudes towards immigrants and puts it in the context of immigrant-targeted policies. Using data from Eurobarometer 2006 combined with country level data collected by the authors, the analysis shows that the job market openness towards immigrants influence not only the attitudes towards immigrants, but also the relationship between these and the respondents’ position on the job market. Romana Careja Research Institute for Sociology University of Cologne [email protected] Hans-Jürgen Andreß Chair for Empirical Economic and Social Research University of Cologne [email protected]

Upload: vantuong

Post on 16-May-2018

215 views

Category:

Documents


1 download

TRANSCRIPT

We grew here, you flew here. Individual and country-level determinants of natives’ attitudes towards immigrants in Europe

Paper prepared for the IPSA ECPR Conference Whatever Happened to North-South?

Sao Paulo, February 16-19, 2010

Work in progress – Please do not quote without authors’ permission – Comments are welcome

This version February 2011

Abstract What determines the attitudes towards immigrants? Previous research has shown the strong and significant impact of individual level characteristics such as level of education, position on the labour market or position on a right-left scale of political attitudes. The role of country-level characteristics has only recently entered in the spotlight, as research has shown that level of wealth or existing policies can influence attitudes. This paper takes the economic explanation of attitudes towards immigrants and puts it in the context of immigrant-targeted policies. Using data from Eurobarometer 2006 combined with country level data collected by the authors, the analysis shows that the job market openness towards immigrants influence not only the attitudes towards immigrants, but also the relationship between these and the respondents’ position on the job market. Romana Careja Research Institute for Sociology University of Cologne [email protected] Hans-Jürgen Andreß Chair for Empirical Economic and Social Research University of Cologne [email protected]

1

1 Introduction It is time to change the approach to analyse attitudes towards immigrants. Since question of why natives feel a certain way towards immigrants has been first asked, models based on socio-economic status have dominated the analyses. Although this approach has brought major contributions to our understanding of how individuals relate to each other, it underestimates the significance of macro-level determinants, especially policies. To a large extent this „neglect“ is due to the fact that while socio-economic indicators are readily available, policy data generally are not. However, a wind of change is blowing and it is time to bring the new data into our models. Attitudes towards migrants have been explained by individual characteristics such as political orientations, social characteristics (education, sex), or position on the labour market. These models allowed to test individual-level theories, but did little to explained cross-country variation. The most prolific theory in the area of cross-country comparative analysis of attitudes towards immigrants is the group threat approach, whose basic tenet is that the perceived competition for limited resources, and the ensuing perceived threat to privileged position, leads members of the dominant group (nationals) to express negative attitudes towards the out-group. $$$ These models included two main macro determinants of attitudes size of immigrant group and poor economic situation together with a host of individual level variables (Quillian 1995; Scheepers, Gijberts, and Coenders 2002); (Kunovich 2004). The results so far are inconclusive. Therefore including policy as explanatory variable might shed some light about on the mechanisms behind opinions about immigrants. Policies concerning admission of immigrants and their rights determine largely the size of immigrant group. Generous policies determine an increase of incoming immigrant wave, restrictive policies lead to lower number of new arrivals (Emmenegger and Careja 2011; OECD 2009; SOPEMI 2008). In addition to influence on the size of immigrant group, policies have two other major effects, which influence the attitudes towards immigrants. First, policies influence the quality of newcomers (high or low skilled). Many Western European countries have enacted guest workers programmes or have encouraged immigration to supplement their depleted labour force in the wake of WWII. After a period of zero economic immigration following the 1973 oil crisis, they have replaced such programmes with a targeted recruitment of low and high skilled immigrants for specific economic needs (Emmenegger and Careja 2011). Second, policies are accompanied by an official discourse which sets the frameworks within which the government treats the issue of immigration. For example, the „zero immigration“ motto was launched in the wake of the 1973 oil crisis that prompted high inflation and high levels of unemployement in all Western countries. The governments responded, among other measures, with a complete stop of economic immigration programes presented as a means to protect the internal labour markets and the welfare standards of the Western countries. Re-enactements of the “zero immigration” policy have been launched also recently, for example in 1993 in France, with the famous Pasqua law (Guiraudon 2002) or in UK, most recently called for by Lord Tebbit (The Telegraph, 08 February 2010) and were interceded with policies which selected immigrants according to their skills. The official discourse surrounding these policy measures creates a certain image of immigrants and of their impact on the host society and economy. For example, natives can develop an overall feeling of need for protection, which in turn can increase the distance between themselves and newcomers. In addition, natives can embrace the view that immigrants can be useful, but only in certain respects, an attitude which combines a general reluctance with a limited acceptance.

2

Summing up, the policies which manage the immigration flows have the potential to affect natives’ attitudes towards immigrants. Exploration of the impact of policies has produced so far scarce, but extremely interesting evidence. For example, Hjerm (Heijm 2007) found that restrictive immigration policies are associated to higher levels of xenophobia, while Weldon (Weldon 2006) showed that natives of countries with more ethnic foundations of their citizenship regimes express more restrictive attitudes towards non-nationals. This paper sets to explore whether yet another group of policies, namely labor market policies, have an impact on the attitudes of natives vis-à-vis non-nationals. Although economic migration, as it was conceived in the form of guest workers programmes, does not exist anymore, all countries have changed their labour market legislation to respond to the arrival of foreigners on their territories. In this paper we show that the regulations which concern the opening or closing of the labour market to immigrants, shape the attitudes of natives towards the newcomers. The foreigners’ right to work in a Western European country can be granted in various forms, depending on the form of entry: it can be automatic, as in the case of (most) EU citizens, it can be subject to certain temporal restrictions, it can be granted automatically as soon as an asylum seeker has been officially recognised or can be granted within the limits fo special work programmes, such as seasonal work or IT Green Cards. Given that the right to work for foreigners comes with other entitlements, such as access to certain categories of benefits, or creates a new level of competition on the labour market between natives and newly arrived, not all countries open their labour markets to foreigners to the same extent. For example, UK has an advanced programme of selecting immigrants in function of their skills, while Germany is much more protective, relying on bilateral agreements concerning seasonal (low skilled) work coupled with more relaxed administrative requirements concerning the employment of highly skilled (Emmenegger and Careja 2011). We argue that the degree to which the labour market is open or closed to the arrival of immigrants influences natives’ attitudes towards immigrants. In order to explore the impact of labour market policies, we use a newly developped index which measures the degree of openess of labour markets of European countries to third country nationals (Labour Market Acess sub-index of Migrant Integration Policy Index, hereafter LMA). This paper contributes to the literature of determinants of anti-immigrant attitudes in several ways. First, it introduces a new independent variable at macro level, namely the labour market policies. Labour market policies are strongly connected through at least three important channels to the formation of attitudes towards immigrants: they influence the number of incomers, their quality and they promote a certain official discourse and image of immigrants. All these three dimensions are directly connected with public perceptions of, and attitudes towards immigrants. Second, it introduces a new measure, the Labour Market Access (LMA) subindex of Migrant Integration Policy Index (MIPEX). LMA evaluates the practice in European countries against European standards and EU policy recommendations concerning the access of foreigners on domestic labour markets. Of particular interest are the regulations concerning the access of the third-country nationals, which unlike (most) EU citizens, who have access to the labour markets on equal basis with the nationals, are subject of numerous restrictions. Although MIPEX exists since 2004 and it is widely used in country studies and country reports, it is remarkably absent from academic literature.

3

Third, it shows that by including policies, a more nuanced view than the one proposed by the classical version of group conflict theory can be formulated. While the socio-economic context preferred by the group conflict theory proposes a quasi-deterministic view of contexts’ influence on attitudes, the introduction of policies allows us to nuance this view, given the fact that policies are an intermediary element between contexts and individuals, and are often intended to shape contexts in a certain direction. The paper is organised as follows. The next section reviews the literature and formulates the hypotheses. The third section presents the data and the method, while the fourth presents the analysis. The fifth section discusses the results and concludes. 2 Theoretical background 2.1 Policies as determinants of attitudes After the WWII immigration has become a hot spot on the agenda of Western governments. Initially, these governments more or less explicitly encouraged immigration, as immigrants were needed to compensate for the depleted workforce, but after the 1973 oil price crisis they adopted policies designed to stop the economic migration and limit the entry of family members and asylum seekers. Currently, immigration to Western European countries of is managed through three main channels: entry regulations, welfare benefits and labour market policies (Emmenegger and Careja 2011). Emmenegger and Careja (2011) have showed that Western governments are combining these three instruments in such as way as to simultaneously fulfil the expectations of voters (job protection, limited redistribution) and the needs of the labour market (workers for certain sectors). In doing so, the governments create and entertain a discourse which differentiates between desired and undesired immigrants. The former are either highly skilled, working in top economic sectors, or low skilled, working in very specific sectors, with „dirty, dangerous and degrading“ jobs, which are refused by the nationals. The latter are a broad category which encompasses low skilled, asylum seekers and sometimes family members of legally residing immigrants. Labour market policies are particularly efficient at maintaining the distinction between desired and undesired immigrants, in particular through the regulations concerning the rights of foreigners to work. We argue that these regulations influence the attitudes of nationals towards immigrants. Before discussing the specifics of labour market policies and their influence on attitudes towards immigrants, a brief discussion of the relationship between policies in general, on the one hand, and the natives-immigrants interactions, on the other hand, is needed. This relationship is often stated, but the concrete mechanisms remain largely unexplored, and the evidence is scarce (Meuleman 2009). First, policies can increase or decrease the distance between „us“ (the national community) and „them“ (the immigrants). For example, increasing the difficulty of obtaining a job for non-nationals might contribute to keep them in a marginalised status, characterised not only by limited access to resources, but also by mistrustful attitudes towards the majority and the

4

state which represents it. In contrast, integration policies are meant to bridge this distance (Castles 1995; Castles and Davidson 2000). Second, simultaneously to producing policies, the state is producing a discourse which accompanies these policies. It is this “policy+discourse” package that is reflected in the attitudes of natives (Meuleman 2009). Policies can constitute an anchor for the public discourse crystallised around the issue of immigrants and their rights. The policy-makers’ vocabulary is taken over and used to define the categories in which the public constructs the image of immigrants. This link has been illustrated by Lahav (Lahav 2004), which has shown that public policy discourse contributes to the creation of categories such as “good” and “bad” immigrants. Third, this relationship is bi-directional. Policies not only influence the attitudes of those affected by them, but they are also the result of certain attitudes. Widespread ethnic prejudice can be the fuel which propels parties with anti-immigrant stance into Parliament, thereby increasing the likelihood that anti-immigrant measures are adopted (Lahav 2004). A telling example is the rise to power of the extreme right party in Switzerland which has successfully convinced Swiss to vote in referendum against the relaxation of conditions for the naturalisation of foreigners, thus ensuring the exclusion of immigrants from access to citizenship rights over generations (Riaño and Wastl-Walter 2006). Fourth, attitudes towards immigrants stem also from a more principial make-up of the country, more specifically the philosophical foundations of the nation and the resulting citizenship rules. Weldon (2006) combined a country’s cultural and legal traditions to obtain three types of citizenship regimes: collectivistic-ethnic, collectivistic-civic and individualistic-civic. He showed that people living in collectivistic-ethnic countries tend to be more intolerant towards foreigners, compared to people living in individualistic-civic. Keeping in mind these interactions between policies and attitudes, we see labour market policies (LMPs) as responses to citizens’ interests and to the needs of society and economy. In general, labour market policies regulate the conditions in which individuals take and terminate employment, as well as what everything that falls in between. Traditionally, labour market policies were formulated with the national labour market in mind, where the typical actors are domestic firms and national workers. However, in a new context in which the European Union, with its emphasis on circulation of labour, plays an increasing role, in which multinationals and domestic firms with external branches dominate the economic sphere, and in which there is a continuous incoming wave of foreigners from non-EU countries, Western governments change their labour market policies to accommodate to, or to protect from these pressures. For the argument of this paper, the regulations which open the labour market to foreigners and which manage the conditions for such employment are most relevant. We will refer to them in terms of openness/closeness: the more these regulations make it easy for foreigners to access the primary labour market, the more open the labour market. The entry of foreigners on the domestic labour market responds to the need of economy for labour force. On the one hand, highly skilled workers are needed to increase the competitiveness of the top economic sectors, or to increase the quality of education provided to the new generations. On the other hand, low skilled workers are needed in specific jobs in sectors in which nationals are reluctant to take up jobs (Piore 1979), or sectors which need a temporary surplus of labour (such as agriculture, construction, care or hospitality). Besides covering specific needs of the economy, immigrants are welcomed by the governments and business sector because they can help with alleviating the aging problem which lurks above

5

European economies and societies (Howse 2006; Muysken, Corvers, and Ziesemer 2008)1 and are considered important sources of innovation and entrepreneurship (Chellaraj, Maskus, and Mattoo 2006; Constant and Zimmermann 2006). In addition to satisfying the needs of the economy and the interests of business, the entry of foreigners on the domestic labour market is regulated in such a way to satisfy the interests of citizens. Although these are less clearly articulated, it is assumed that they are a reflection of the attitudes and opinions expressed in surveys. Citizens have an overwhelming tendency of stereotyping: immigrants abuse the welfare state, are dangerous, cost extra tax money and take away the job from natives (Paparella 2004; Crepaz and Damron 2009). From this image it is only natural to assume that the interests of citizens are geared towards less incoming immigration, more control over those who enter, less redistribution from nationals to foreigners. The labour market regulations answer these concerns and needs by a. selecting the immigrants who are needed through special programs and b. limiting the access to the labour market of those who come „uninvited“ by conditioning the right to work on the absence of qualified nationals or by not recognising the qualifications received outside Europe. These measures creatively combined create more or less open labour markets. As previously discussed, it is expected these policies produce certain responses within the population. As the impact of labour market openness has not been explored so far, the evidence as to the direction of this relationship lacks. We develop here few theoretical possibilities. $$$ An open labour market approach means that immigrants are welcome and that once they acquire a job they enjoy the benefits derived from this status. This can be perceived by the resident population in two ways. First, it can be perceived as threat: open labour market means more competition for jobs. Natives compete not only with natives, but also with non-natives who might be willing to work for less. Therefore, natives are likely to be dissatisfied with the openness and display negative attitudes towards immigrants. Second, it can be perceived in positive terms: open labour market means that immigrants have access to legal employment and therefore have no incentives to join the shadow economy. In addition, the opening of labour market is justified in terms of satisfying specific economic needs. Immigrants become thus tax payers and contributors to the general good, and are used for specific jobs, all of which means that they are likely to be seen in a better light by natives. A closed labour market has mostly negative effects on attitudes. On the one hand, natives can interpret the closeness as a means through which their governments protect them. However, this protection fits a pattern in which natives construct a self-image of a group threatened by the immigrants. If this is the case, then the closeness of labour markets contributes to re-enforcing the negative and rejective attitudes of natives. On the other hand, a closed labour market means more barriers to access the legal job market, which means that immigrants are likely to turn to the shadow economy. This situation enforces the natives’ view of immigrants as outlaws, and boosts their negatives opinions about them. Summing up, there are reasons to expect that LMPs trigger different responses: while a closed LMP reinforces the distance between natives and immigrants, feeding either into perceptions

1 For a different point of view see (Bermingham 2001)

6

of threat to the group or threat to the law and order of society, an open LMP can either increase the perception of threat to personal labour market position or create a positive perception of immigrants as contributors to the common good. How these responses can be explained? The negative attitudes can be explained by resorting to the logic proposed by the group threat theory (Blumer 1958; Bobo 1988; Kunovich 2004; Quillian 1995; Scheepers, Gijberts, and Coenders 2002). According to this theory the relationship between individual characteristics and expressed prejudice is heightened in contexts in which the dominant (native) group perceives threats from another group, be it an existing minority or an incoming immigrant group. The topic was thoroughly researched in American context, and only relatively recently has attracted the attention of scholars focusing on Europe. They have already produced an impressive amount of research which tests the effects of threats, usually conceptualised as size of minority group and bad economic situation, on natives’ attitudes towards immigrants, but the findings remain inconclusive, due mainly to model specification (Meulemann 2009). The breaking ground work has been done by Quillian (1995), who argued that in addition to individual level characteristics, the perceived threats for the group to which respondents belong, such as large size of non-native groups and poor economic conditions, determine attitudes towards the out-group individuals. His results showed that the two perceived threats intensify each other’s effects and in conjunction affect attitudes towards immigrants. In what concerns cross-level interactions, his results showed that the effects of some of the individual level variables included in the model (socio-demographic and status variables) vary with either one of the two causes of threat (high proportion of immigrants and poor economic conditions). Subsequent respecifications of his basic model (Ceobanu and Escandrell 2008; Kunovich 2004; Scheepers, Gijberts, and Coenders 2002) confirmed and nuanced these findings and expanded the line of research. Sheepers et al. (2002) found only percent of non-EU citizens to have an effect on the dependent variable, ethnic exclusionism. The cross-level interaction effects essentially replicated Quillian’s (1995). A main difference concerns the effect of manual workers: while Quillian did not find an effect, Scheepers and his colleagues found that in countries with higher proportion of non-EU citizens, the manual workers expressed higher levels of ethnic exclusionism. As opposed to Quillian (1995) and Scheepers et al. (2002), who’s basic contention was that the perceived threats intensify the relationships between individual characteristics and prejudice, Kunovich (2004) argued that the perceived threats can also dampen these relationships. He finds that indeed, poor economic conditions and relative group size interaction with individual characteristics affects prejudice in different directions: while the former wash away the positive effects of status among advantaged groups and reduce prejudice among disadvantaged groups, the latter strengthen the negative effect of student status and the positive effect of blue-collar and not-in-workforce status. These divergent effects of country characteristics prompt Kunovich to suggest that the group-threat theory might be more complex than initially thought. Ceobanu and Escandrell (2008) propose to test a more complex model, involving a comparison between East and West Europe across time. Contrary to some of the previous studies, economic conditions are not found to have an effect on anti-immigrant sentiment, while the size of immigrant group is found to have a surprising significant negative effect. Their conclusion is that civic West vs. ethnic East difference is not supported by evidence,

7

and that moreover, in both regions the impact of individual characteristics on anti-immigrants attitudes changed in time. The positive attitudes can be explained by two theories. Firstly, the contact theory maintains that contact between groups can lead the individuals belonging to those groups to accept diversity more readily. Contact theory provides mostly an explanation at individual level, expecting individuals belonging to the majority group who have contact with individuals belonging to the minority group to express more positive attitudes towards the latter. Hjerm (2007) finds indeed that individual contact with foreigners decreases the level of xenophobia, confirming earlier findings of Welch et al. (Welch et al. 2001) and McLaren (McLaren 2003). Transposed at macro-level, contact theory means that in countries where the integration of minority groups is stronger, the individuals belonging to the majority hold more positive attitudes towards individuals belonging to the minority group(s). Secondly, the utilitarian argument informs that individuals prefer options which increase their utility (well-being). In this vein, Dustmann and Preston propose a theoretical model that suggests that “economic-self interest points to an assessment of the benefits and costs from immigration that encompasses not only labour market competition, but also taxes and public burden, as well as general welfare effects determined by efficiency considerations” p29 (Dustmann and Preston 2006). Their results indicate that the perceptions about possible harmful effects of immigrants are associated mostly with the worries about immigrants being a fiscal burden on the economy and less with the labour market competition. The implication is that if immigrants are given the possibility to contribute to the general product of society and to the tax pool, which is subsequently redistributed, it is likely that the beneficiaries of this redistribution, among others the members of the dominant group, would prefer this option. If the preferred option (immigrants legally working) is producing the expected results (paid taxed, legal work), it is likely the natives would develop positive attitudes towards imigrants. Although policies have been hinted at as important determinants of attitudes, relatively little cross-country empirical evidence exists. One reason is undoubtedly the lack of data at country level concerning the policies in place. Nevertheless, few recent works found interesting results. Hjerm (2007) re-assessed the group threat theory following the idea that in order to properly test the implications of the theory one needs to take into account not only the relationships between the variables directly implied by the theory, but also the ones indirectly implied. More specifically, he argued that the group threat, as measured by relative size of minority group, is not a threat per se, but in certain conditions, such as the political context. In his words, “It is dominant people and instiutions amongst the majority who express, constitute and consolidate views about the minority.” (Hjerm 2007: 1256, our emphasis). His expectations concerning institutions were only partly supported: He finds that size of minority group per se does not influences the attitudes of natives vis-à-vis non-natives, and neither does the interaction between size of minority group and immigration policy, but the immigration policy per se does. More specifically, the more restrictive the immigration policy, the higher the xenophobia level expressed by the respondents.

Weldon (2006) brings evidence from a different perspective. His argument is that the philosophical and political foundations of citizenship rules influence the attitudes that citizens display vis-à-vis foreigners. He finds that indeed the legal and cultural dimensions influence the both political and social tolerance. In addition, legal dimension is found to interact with

8

individual characteristics such as left/right political orientation, democratic satisfaction in the case of political tolerance, and with left/right political orientation in the case of social tolerance. We argue that the character of labour market policies, captured by openness and the justificatory discourse) influences the attitudes of natives towards immigrants. Emmenegger and Careja (Emmenegger and Careja 2011) showed that governments implement policies in the area of immigration, welfare and labour market which diminish the attractiveness of developed countries for most categories of immigrants, while attracting immigrants considered “desirable” from an economic point of view. In addition to fulfilling their direct roles, such policies transmit signals and, through the accompanying justifications, form opinions. For example, a policy which closes the labour market to outsiders policies can be interpreted by the nationals as both a sign of governments’ care for them and also as a sign that outsiders are not to be welcome. Thus, policies not only respond to concerns in population, but also create / or perpetuate certain attitudes towards foreigners. An earlier study of Bauer and his colleagues brings evidence pointing to this direction. The worries expressed by nationals concerning the effects of immigrants differ in function of the type of immigrants entering the countries: in countries where asylum seekers predominate nationals are concerned about crime rates, while in countries where economic immigrants predominate nationals are concerned about losing jobs to immigrants. Thus, selection of immigrants according to the needs of the labour market is likely to be approved and is accompanied by more positive views about immigration (Bauer, Lofstrom, and Zimmermann 2000). 2.2 Other micro and macro determinants of attitudes As policies are relatively new explanatory factors for attitudes, the next few paragraphs survey briefly other micro- and macro-level determinants which have been combined in the literature, by now classical, on attitudes towards immigrants.

Three main approaches provide the theoretical support for individual-level explanations of attitudes towards immigrants and immigration. First, socio-economic factors have been repeatedly proven extremely stable both over time and cross-country. Education has been consistently showed to be positively associated with attitudes: the higher the education, the more positive the attitudes (Quillian 1995; Hainmueller and Hiscox 2002). Labour force status (employed/unemployed) and occupational classification (blue vs. while collar) have been found to be associated to attitudes toward immigrants. Blue collar workers’ and unemployed individuals’ negative attitudes towards immigrants have been mostly explained in terms of labour market competition for jobs and welfare support. Effect of income on attitudes is less clear (Kehrberg 2007), but it has been consistently found Western Europe as opposed to Eastern Europe (Kunovich 2004). Several studies found that older respondents, men and people living in rural areas hold more negative attitudes compared to younger individuals, woman or city dwellers, confirming the expectations derived from labour market and contact theories.

Second, identities and values have been used to explain attitudes towards migration, especially in studies focusing on nationalism and formation of attachments to and identification with national or supranational entities. Perception about size of immigrant group and consequences of immigration for one’s group standards, left-right political values,

9

identification with one’s group have been all found to be associated with attitudes towards immigrants. In particular, in time of economic stress, these values and perceptions have been found to be associated with an increase of intolerance and rejection of foreigners (Hjerm 2007; Citrin and Sides 2008; Ceobanu and Escandrell 2008).

Third, contact theory provides a more positive perspective arguing that where there is contact between individuals belonging to the native and immigrant group, the former are likely to display more positive attitudes, compared to natives who do not contact immigrants. There is some evidence supporting this theory (Pettigrew 1998a), although it is still unclear in which conditions and which type of contact produces such positive attitudes (Ceobanu and Escandrell 2010).

The theoretical approaches which incorporate macro-level explanations of attitudes build upon group threat theory, which posits the intergroup relations as a zero-sum game, which leads to formation of negative attitudes. The threats can be real or perceived, and can concern either competition for scarce resources (such as jobs or welfare benefits) or changes to the previous way of living and homogeneity of cultural and living standards and values. These approaches use a conceptualization of threat either in terms of size of immigrant group or worsening economic situation (low GDP per capita or high unemployment). Results so far are inconclusive: while some authors find that these factors make a difference (Quillian 1995; Scheepers et al. 2002), other find less convincing results (Hjerm 2007; Kunovich 2004).

2.3 Research hypotheses In this paper we use two variable which capture native’s attitudes towards immigrants. The first variable, ethnic, is an index of three stereotypes about foreigners (see details in Table 1 in Annex), while the second (immigrant) is an index of two evaluations of the role of immigrants in the economy. Besides their different nature (one stereotyping, the other rational evaluations), the two variables depict foreigners in different ways: the former refers to “people of different ethnicity”, thereby not emphasizing the idea of newcomers, and suggesting that the people to which it refers might be already settled in the country, while the second explicitly uses the word “immigrants”. Arguing that these differences between the two sets of questions prompt respondents to relate differently to the group of reference, we have opted for two dependent variables, instead of combining the five items in one index. We also expect differences in the impact of our explanatory factors, therefore we formulate different hypotheses for the two dependent variables. In the following paragraph, we formulate the theoretical expectations concerning the relationship between the main independent variable (LMA) and the two dependent variables. Detailed explanations concerning the effects of other variables included in the model will be given in the next section, when those variables are presented. Level Hypotheses IMMIGRANT Based on Kunovich’s insight (2004), we assume that the effect of country level characteristics can be either positive or negative. The classical group threat theory would expect a negative outcome: an open LMP means immigrants’ easier access to jobs, which increases the competition between natives and immigrants, leading to the formation of negative attitudes among the former. However, an explanation for a positive outcome can be formulated, if a rational approach is assumed. An open LMP implies on the one hand the immigrants’ easier access to legal jobs and on the other hand, a policy justification emphasizing the idea of

10

immigrants as useful for host society. Consequently, this implies that immigrants are perceived as contributing to the pool of resources, and as law-abiding members of the host community. H1a (threat theory): An open labour market policy is associated with a negative evaluation of immigrants’ economic role. H1b(alternative to threat theory): An open labour market policy is associated with a positive evaluation of immigrants’ economic role. Level hypotheses ETHNIC We remind the reader that the variable ethnic captures respondents’ tendency to uphold stereotypical assertions about immigrants impact on various domains of social, cultural and economic life of a community. As such, we assume that the propensity of individuals to express stereotypes is less influenced by rationalisations about the role of foreigners, but it is likely to be stimulated by perceptions about the presence of foreigners in the immediate environment. Therefore, an open LMP, with its likely effect of attracting foreigners, is likely to be associated with more stereotypical views of people of different ethnicity. However, as argued previously, a closed LMP, can reinforce the stereotypes by accommodating the argument that government protects the natives (i.e. it closes the labour market), because people of different ethnicity are a threat. H1c (direct threat theory) Open labour market policies are associated with support for stereotypes towards foreigners. H1d (indirect threat theory) Closed labour market policies are associated with support for stereotypes towards foreigners. Interaction effects This study is largely exploratory. However, of the large number of possible interaction effects, we have chosen to test only a few, which are strongly supported by the theory. Concerning the opinions about immigrants’ economic role (immigrant), we expect that LMA mediate the effect of status variables, in particular employment status (i.e., being unemployed or having a blue collar job). If group threat theory is right, and LMA is perceived as having negative effects, we expect a negative interaction effect. If LMA is perceived as having beneficial effects, as our analysis of which is our alternative explanation, we expect a positive interaction effect. In what concerns the stereotypical attitudes (ethnic), we expect LMA to mediate the categories most likely to express stereotypes. We test the effect of LMA on the relationship between employment status (being unemployed, having a blue-collar job), right ideology and education and ethnic. We expect LMA to be interpreted in negative terms as a instrument which leads to increasing the number of foretigners, and therefore to have a negative effect on the individual level relationships. If LMA is interpret positively, as being the means to complement labour force for predefined needs, we also expect a negative relationship because such a rational view does not impede a change of view of foreigners as elements alien to the ways of life of host society.

11

3 Data and Methods We are using Eurobarometer (EB) data from 2006 to test our hypotheses.2 We have retained in the analysis only countries which received in recent years relatively high numbers of immigrants according to the European Migration Network and OECD. All in all, our analysis uses data from 13757 respondents in 23 European countries. Although the sample covers countries from both Eastern and Western Europe, the group of countries is relatively homogeneous in that it comprises only medium-high to high income countries. Apart from the economic cycle specific to each country, no major economic crises affected the countries in the sample in the period previous to the survey. Therefore, no peculiar reasons to artificially strengthen the negative attitudes towards immigrants. Some descriptive information about the sample is summarized in Table 2 in the Annex. The sample contains only nationals of the countries included in the analysis and respondents for which both dependent variables were present. 3.1 Dependent variables The two dependent variables, immigrant and ethnic, were presented in the previous section. We just mention here that they are measured on a continuous scale from 0 to 1, with low values indicating negative attitudes. More specifically, a low value on the immigrant scale indicates that the respondent does not think that immigrants have a positive economic role, and a low value on the ethnic scale indicates a respondent’s tendency of thinking about people of different ethnicity in terms of stereotypes. 3.2 Independent variables Country level The main explanatory variable at the country level is the index of labour market openness, LMA. This subindex of the Migrant Integration Policy Index (MIPEX)3 observes four main dimensions of migrants’ integration in the labour market: eligibility, labour market integration measures, security of employment and the rights associated with the labour market status. On each dimension, the regulations in force are evaluated on a 0 to 100% scale against the EU standards. An additional policy variable was added, to capture the generosity of states concerning the right of foreigners on their territories to be reunited with their families. Family reunification has been long used as an entry opportunity, when economic migration was not an option. The 2 The EUROBAROMETER 66.3 covers the population of the respective nationalities of the European Union Member States, resident in each of the Member States and aged 15 years and over. The EUROBAROMETER 66.3 has also been conducted in the two acceding countries (Bulgaria and Romania) and in the two candidate countries (Croatia and Turkey) and in the Turkish Cypriot Community. In these countries, the survey covers the national population of citizens of the respective nationalities and the population of citizens of all the European Union Member States that are residents in those countries and have a sufficient command of one of the respective national language(s) to answer the questionnaire. The basic sample design applied in all states is a multi-stage, random (probability) one. In each country, a number of sampling points was drawn with probability proportional to population size (for a total coverage of the country) and to population density. 3 The data are provided by the MIPEX II network, which comprises 25 partner organisations, led by British Council and Migration Policy Group, with Research Partners at Sheffield University and the Universite Libre de Bruxelles.

12

rationale of this policy is quite different from the labour market policies, in that family reunification is much more influenced by international documents which settle human rights. Family reunion (family) is also a subindex of MIPEX, which evaluates the rules that regulate the right to family reunion: eligibility, acquisition conditions, security of status and the associated rights. On each dimension, the regulations in force are evaluated on a 0 to 100% scale against the EU standards. We control for GDP grow (GDPgrow) and trend in unemployment (unempltrd) to take into account the expectations of group threat theory which places the increased sense of threat in the worsening of economic conditions. We also control for the size of the immigrant population using the size of the foreign-born population (foreign born), as a proportion of the total population. In addition to the expected effects of the main explanatory country-level variable, LMA, which were detailed in the previous section, we expect the following effects of the other country level variables:

• We expect immigrant to be influenced by unemployment trend and GDP growth in the directions suggested by group threat theory: negative economic conditions are associated with negative opinions about immigrants’ economic role. The size of the foreign population does not have a clear theoretical influence on opinions about the economic role of immigrants, both positive and negative correlations being possible. On the one hand, group threat theory predicts a negative relationship, interpreting large number of foreigners as threat. On the other hand, larger foreign population may imply more contact between the two communities, hence less negative attitudes from natives. The other policy variable that we are exploring (family) has no theoretical implications supporting a direct effect.

• Since the expression of stereotypes is connected to perceptions about the presence of

the “others”, we expect ethnic to be influenced by the size of the foreign population in the direction predicted by group threat theory: a large group of non-natives is associated with stereotypes. The same is true for negative economic conditions as measured by GDP growth and unemployment trend. Our second policy variable family may be directly correlated to ethnic: permissive family reunification rules lead to an increase of the number of foreigners. The impact is unclear theoretically: on the one hand, it can encourage natives to express stereotypes, because the number of foreigners increases (group-threat theory). On the other hand, the justification of liberalisation of family reunification policy, with its emphasis on human rights (Soysal 1994), might make natives more sensitive to foreigners and therefore it might influence them to give up using stereotypes (rational).4

Individual level The choice of individual level variables is determined by the theoretical perspective adopted in this paper, which builds on group-threat theory and on a rational (utility) theory. However, the choice of variables is constrained by the availability of corresponding items in the EB. Table 1 in the Annex includes a description of the variables. The main explanatory individual-level variables focus on socio-economic status, especially labour market position (status). We 4 Contact theory also provides an explanation for the positive relationship: the arrival of new immigrants might also imply that natives become familiar with people of different ethnicity, and therefore they are likely to give up using stereotypes.

13

expect individuals who find themselves in a secure labor-market position (white collar, status 5, used as reference category) to be more positively oriented towards foreigners compared to those in less secure positions, such as blue collar workers or the unemployed. We have added variables which capture an self-assessment of personal economic trajectory (past) and level of optimism regarding the future personal situations (future). In accordance with group threat theory, we expect persons who see themselves on an ascending economic trajectory to display more positive attitudes towards migrants (as migrants are not perceived as a threat). However, a different explanation can also support a negative relationship. If the improving economic situation is perceived as the result of “good protection by the government against the “threat”, then the distance between nationals and newcomers is enforced, translated by the expression of negative assessments of immigrants. We control for age, gender, education (educ), place of residence (residence) and political orientation (right). In accordance with results of previous research from group threat theory, we expect more negative attitudes to be associated with older, male, or low educated individuals or with individuals residing in urban areas, i.e., groups of natives which are most likely to enter in direct competition with the immigrants for jobs and welfare resources. However, we expect nationals residing in urban areas (i.e., small towns or big cities) to be more positively oriented towards immigrants compared to rural residents (reference category), as these respondents a. come more often in contact with immigrants and b. have more flexibility on the job market. The last remark deserves an explanation: the arrival of immigrants, usually in urban areas, means that they will take up jobs which natives are unwilling to take. Thus natives either reorient themselves towards better paying jobs, or continue to live on welfare support. In both situations, they are better off, hence positive attitudes. We also control for migration background (migrant). As a rule, the survey was not applied if the respondent was not citizen of the country. However, being citizen does not preclude a migration background. and, according to group contact theory, we expect these respondents to be more positive about other migrants. Additional interaction effects Besides the interaction effects of LMA, we explore other possible interactions. Since opinions about immigrants’ economic usefulness are likely to be influenced by economic conditions, we expect that an increasing unemployment trend negatively affects the relationship between status variables (in particular being unemployed or a blue collar worker) and immigrant. We do not test the impact of GDP growth because we expect that the unemployment trend is more strongly felt by individuals, as a closer threat, than an abstract growth of GDP which is not immediately felt. Since stereotypes are determined primarily by perceptions about foreigners and can be “stimulated” by worsening economic conditions, we test the following interactions: increasing unemployment is expected to negatively affect the relationship between status (especially being unemployed and blue-collar) and ethnic, and between right political orientation and ethnic. The size of foreign population is also expected to enhance the relationship between right political orientation and ethnic. Since family reunification has been the gate to developed countries for many foreigners, and has been the focus of numerous political and policy debates, it is expected to influence the relationship of right ideology and education to ethnic.

14

3.3 Methods In the following multivariate analysis, we use hierarchical linear regression models to test our hypotheses. Data are weighted to control for different population sizes of the 23 countries and for the specific sampling design in each country. 4 Analysis /Results 4.1 Descriptive Analysis The following two graphs show mean attitudes across the 23 countries in our sample (West and East Germany (gew and gee) were treated as two “different” countries). ***** Figures 1 and 2 about here***** The two figures indicate that the opinions about immigrants’ economic benefits (Figure 1) and stereotypes about people of different ethnic origin (Figure 2) vary widely across the countries in our sample. Besides the usual suspects at the individual level, it is worth exploring whether this variation is linked to different policies that these countries implement to target immigrants. The policies we look at are labour market access (LMA) and family reunion regulations. As the following correlation coefficients indicate, the labour market access index (LMA) correlates positively with immigrant (r=0.59) and negatively with ethnic (r=-0.38).5 However, the ethnic-LMA correlation is not very strong. The direction of the statistical association is the same with the other policy indicator family, positive in the case of immigrant and negative in the case of ethnic, but the coefficients are mostly smaller. All in all, attitudes towards immigrants are more positive in countries with migration-friendly policies, while attitudes towards ethnic minorities are more positive in countries with sub-(EU)-standard migration policies. The only policy variable which shows stronger results is the labour market access, confirming our decision to focus our analysis on it. Other country-level conditions, for which we control, are the size of foreign born population (foreign_born), the growth of GDP per capita (GDPgrow) and the trend of unemployment rate (unempltrd). As the Table indicates, the correlations of these variables with the dependent variables are weak. *****Table 1 about here***** 4.2 Models In search for a theoretically meaningful, but parsimonious model for the data, several models of increasing complexity were tested. According to the intra-class correlation coefficient in the empty model, about 7% of the ethnic variance and about 8% of the immigrant variance is due to differences between countries. This leaves not much room for macro explanations, but is not hopeless. 5 In order to control for the problem that small countries (in terms of their population size) dominate the results, if we treat all countries alike, all calculations referring to cross-country comparisons are made using population weights provided with the EB (see section on methods).

15

Tables 2 summarizes the results for the first dependent variable, immigrant, while Table 3 summarizes the results for ethnic. Variables were introduced stepwise. Due to space restrictions, we present here only the empty model (Models 0 and 4), the model including only individual-level effects (Model 1 and 5), the model with country-level variables (Model 2 and 6) and the final model, which includes also the cross-level interactions (Model 3 and 7). ****Tables 2 and 3 about here.**** At the individual level the effects are quite stable across different models and dependent variables. Most of them are in line with our theoretical assumptions. Hence, they are robust independent of what kind of macro explanation we use. More specifically, in the case of opinions about the economic role of immigrants (Table 2, Model 1), respondents who are unemployed, blue collar workers and hold rightist political values do not consider that immigrants can have a positive economic role, i.e. that they are needed in specific economic sectors and that they can help solving the ageing problem. These findings are in line with theoretical expectations derived from theories of competition for jobs and political values. All other individual level values behave as expected, in the sense that respondents who are more educated, live in urban areas, have a good economic situation and are optimistic about the future are inclined to agree that immigrants have a positive economic role. No surprise, European citizens who have migrant background display also positive opinions. The only country level variable which is found to have an effect on immigrant is LMA (Model 2). The positive sign of this direct effect indicates that the more open labour market policies are the more likely the respondents are to support the idea that immigrants can fulfill a positive economic role. The result is robust to various model specifications. It seems that on average, the rational discourse which surrounds the opening of labour markets (“We need immigrants for specific jobs and to remedy the problems we have.”) finds an echo within the population. The interaction effects in Model 3 reveal a couple of interesting findings. We remind the readers that we did not explore the whole range of possible interaction effects, but we have chosen the ones more theoretically plausible. We have interacted labour market access with the two employment status variables, which capture individuals in labour market positions most likely to compete with immigrants (unemployed and blue-collar) and also with future, which captures optimism about future. The only significant result is observed for the interaction between LMA and blue-collar workers: the negative relation between blue-collar and immigrant is made less negative by LMA. This result seems to indicate that the rational reasons provided by the government to justify the opening of the labour market are accepted even by some categories which are directly “in danger”. Due to the small size of the coefficient, we interpret this finding more as an indication of a tendency. The positive interaction effect between unemployment trend and being unemployed completes the picture. It is a counterintuitive finding, as group threat theory predicts a negative interaction. It seems to suggest that, under the constraints of an unfavourable economic context, unemployed respondents are more inclined to recognise that immigrants can play a positive economic role. Even more so, if these immigrants take up jobs that the natives (although unemployed) prefer to reject. This finding supports our idea that when it comes to evaluating strictly the economic usefulness of immigrants, the natives adopt a rational-utilitarian reasoning. Table 3 presents the results for the variable ethnic. The individual level variables in Model 5 behave largely as expected theoretically, and they are stable across the three models. Retirees,

16

students, unemployed, blue-collar workers, and respondents holding rightist political orientations are all categories which think that people of different ethnicity increase unemployment, increase insecurity and do not add to the cultural makeup of the country of residence. As expected, urban dwellers, more educated people, those who perceived themselves on an upward general trend and are optimistic about the future are less likely to display stereotypical opinions. The country level variables in Model 6 depict a more complex picture. The policies, whose application would lead to increase the number of immigrants, are present with opposite effects on respondents’ propensity to uphold stereotypical views of foreigners. This finding would be disturbing if we were to explain it in terms of classical theories such as contact or group threat theory, which emphasize the effect of increasing number of foreigners. However, they all can be coherently explained, if we focus explanation not on the consequences of policies, but on the discursive context surrounding the adoption of policy measures. Opening the labour market to foreigners is often justified in rational terms, emphasizing their potential positive economic role, while the liberalisation of family reunification rules is justified in terms of human rights. These different rationalisations are reflected in respondents’ stereotyping: On the one hand, the rational discourse concerning labour market policy does not directly contradicts the stereotype “foreigners do not fit”, and the negative sign can be interpreted as indicating the “natural” tendency of natives to uphold stereotypes about foreigners. On the other hand, the positive sign that accompanies the family variable seems to suggest that the human rights argument, which emphasize the human qualities of foreigners and their universal rights, convinces natives to give up the use of stereotypes concerning foreigners. The unemployment trend influences attitudes in the opposite direction than expected by group threat theory: increasing unemployment is associated with a low propensity to support stereotypical views. This finding suggests that on average, when economic conditions worsen, natives re-assign blame for bad economic policies (Rustenbach 2010), maybe to the government, and are likely to admit that foreigners are not responsible for increased crime rates or unemployment. Adding the interaction terms in Model 7, does not change the direction and strength of the direct relationships involving country-level variables, but shows that their effect on selected individual characteristics is only partly as expected. First note that the policy variables do not have an effect on the individual characteristics most susceptible to be associated with expression of stereotypes. Second, the structural country characteristics – unemployment trend and size of foreign population do. The positive sign of the interaction between unemployment trend and being unemployed can be interpreted in the same vein as the positive direct effect of unemployment, as an indication of blame re-assingment (Rustenbach 2010). The negative sign of interactions involving rightist political orientation indicates that we capture a more complex situation: respondents holding rightist values are sensitive to a certain political discourse that interprets the increase in unemployment and in the size of the foreign population as problems directly connected to foreigners and their characteristics. Such an interpretation is supportive for expressing stereotypical views about foreigners. 5 Discussion and conclusion Not so long time ago, Ceobanu and Escandrell (2010) made a convincing argument in favour of increasing the diversity of explanations of attitudes towards and opinions about foreigners. This paper follows this advice and explores the impact of policies on two types of opinions:

17

rational opinions about the economic usefulness of immigrants and stereotypical opinions about foreigners in general. Using a multilevel model and data from the Eurobarometer and collected by the authors, this analysis has showed that the two types of opinions have different sets of determinants. How our results speak to the existing literature? As the literature review has showed, most explanations which include a macro-level explanation rest on one of two theoretical foundations: either group threat theory (most common), or a variant of contact theory. Our results show that these theories explain only part of the story, and that when analysing policies none of these explanations does a good job to incorporate effects which run in seemingly different directions. We show that a coherent explanation can be formulated if one looks not at the consequences of policies (as group threat and contact theory would invite us to do, by focusing our attention on increase/decrease in the number of foreigners as result of policies), but on the policies themselves and the discourse which accompanies their adoption. At the individual level, indeed, respondents whose job market status would put them more in job competition with foreigners were likely to disagree with a positive economic role of immigrants and to express stereotypical opinions. These findings confirm inferences from group threat theory, and are complemented by the positive opinions expressed by urban dwellers, educated respondents and persons optimistic about their future. Negative opinions were voiced by respondents holding rightist values, a finding which confirms the findings of a rich literature focusing on the impact of political values on attitudes about foreigners. The country-level characteristics and their interactions with individual characteristics reveal more details into this picture, which seem to support our hunch. We have argued that it is not only the policies themselves that influence opinions, but also the discourse that surrounds the adoption of those policies. Even more so, as the results of policies is sometimes not immediately apparent, but the public debates are quickly accessible. Our results show that the direct effects of policy variables on opinions reflect this impact: the rational discourse surrounding the opening of the labour market to foreigners leads to agreement that immigrants are needed for specific economic reasons, but does nothing to change the stereotypical view of foreigners. In other words, “immigrants can come and work here because we need them, but we still think that they are a disturbance for our way of life.” The more human-rights oriented discourse surrounding the family reunification policies does not have an effect on opinions about the economic role of immigrants. This is not surprising given the fact that the arguments are of different kind. However, these arguments seem to diminish natives’ propensity of thinking about foreigners in negative stereotypes. The interaction effects indicate that among our policy variables, only labour market openness has an effect: it influences only the relationship between blue-collar status and opinions about the economic role of immigrants. The effect is small, but the direction is consistent with the direct effect of LMA, pointing at the power of arguments to mitigate negative opinions. In other words, these findings show that in order to meaningfully interpret the policy-related findings, we have to quit the framework set by threat theory, which invites us to look at the outcomes of policies (more openness, more immigrants), and adopt a theoretical framework in which the rational individuals react to the public debates surrounding those policies. (to be continued)

18

Tables and figures Table1 Correlations between macro-level variables and dependent variables immigrant and ethnic. ethnic immigrant LMA family foreign born GDPgrow unempltrd ethnic 1.0000 immigrant 0.1517 1.0000 LMA -0.3785 0.5886 1.0000 family -0.0720 0.2921 0.6005 1.0000 foreign born 0.0798 0.0387 -0.0157 -0.3643 1.0000 GDPgrow -0.1125 0.0573 -0.1933 -0.2333 -0.3536 1.0000 unempltrd 0.4129 -0.2055 -0.2479 -0.3709 0.5751 -0.4245 1.0000

19

Table 2. Results of models for the dependent variable immigrant Model 0 Model 1 Model 2 Model 3 Coeff

(s.e.) Coeff

(s.e.) Coeff

(s.e.) Coeff

(s.e.)

individual constant 0.494*** 0.473***

(0.034) 0.48***

(0.052) 0.48***

(0.033)

retiree -0.007 (0.011)

-0.007 (0.011)

-0.006 (0.010)

student 0.011 (0.021)

0.011 (0.021)

0.008 (0.013)

unemployed -0.036 (0.020)

-0.036 (0.020)

-0.073*** (0.020)

selfemployed 0.012 (0.013)

0.012 (0.013)

0.011 (0.012)

white collar ref ref Ref blue collar -0.061***

(0.013) -0.061***

(0.013) -0.053***

(0.015)

educ 0.014*** (0.002)

0.015*** (0.002)

0.015*** (0.001)

past 0.029*** (0.004)

0.029*** (0.004)

0.028*** (0.004)

future 0.019** (0.009)

0.019** (0.009)

0.03*** (0.007)

village ref ref Ref Ref small town 0.03

(0.016) 0.029

(0.016) 0.028***

(0.007)

big city 0.069*** (0.014)

0.069*** (0.014)

0.068*** (0.008)

migrant 0.058** (0.027)

0.058** (0.027)

0.058*** (0.009)

right -0.013** (0.004)

-0.013** (0.004)

-0.013*** (0.001)

male 0.004 (0.017)

0.004 (0.017)

0.004 (0.006)

age 0.002*** (0.000)

0.001*** (0.000)

0.002*** (0.000)

country LMA 0.003**

(0.001) 0.003**

(0.000)

family GDPgrow 0.015

(0.016) 0.016

(0.015)

unempltrd -0.01 (0.056)

-0.008 (0.048)

foreign_born 0.003 (0.006)

0.003 (0.005)

interactions LMA*unepmployed 0.0005

(0.000)

LMA*blue collar 0.001** (0.000)

LMA*future -0.0008 (0.000)

unempltrd*unemployed 0.065** (0.026)

unempltrd*blue collar -0.022 (0.019)

unempltrd*future -0.018 (0.009)

*** p< .001, ** p< .01, *p< .05 (robust standard errors in parentheses)

20

Table 3. Results of models for the dependent variable ethnic Model 4 Model 5 Model 6 Model 7 Coeff

(s.e.) Coeff

(s.e.) Coeff

(s.e.) Coeff

(s.e.)

individual constant 0.540*** 0.557***

(0.032) 0.510***

(0.031) 0.510***

(0.024)

retiree -0.04** (0.014)

-0.037** (0.015)

-0.038*** (0.008)

student -0.058*** (0.014)

-0.065*** (0.013)

-0.062*** (0.011)

unemployed -0.071*** (0.016)

-0.064*** (0.016)

-0.102*** (0.017)

selfemployed -0.002 (0.013)

-0.011 (0.013)

-0.004 (0.010)

white collar ref Ref Ref Ref blue collar -0.045***

(0.010) -0.039***

(0.010) -0.053***

(0.012)

educ 0.022*** (0.003)

0.025*** (0.003)

0.024*** (0.001)

past 0.020*** (0.005)

0.020*** (0.005)

0.019*** (0.003)

future 0.029*** (0.008)

0.036*** (0.008)

0.028*** (0.004)

village ref Ref Ref Ref small town 0.005

(0.009) 0.0008

(0.009) 0.004

(0.006)

big city 0.028** (0.014)

0.016** (0.014)

0.027*** (0.007)

migrant 0.066** (0.023)

0.065*** (0.023)

0.064*** (0.008)

right -0.028*** (0.004)

-0.029*** (0.004)

-0.027*** (0.002)

male -0.011 (0.011)

-0.012** (0.011)

-0.011** (0.005)

age 0.00002 (0.000)

0.0002 (0.000)

0.00003 (0.000)

country LMA -0.002**

(0.000) -0.002**

(0.000)

family 0.003** (0.001)

0.003** (0.001)

GDPgrow -0.002 (0.012)

0.005 (0.012)

unempltrd 0.082** (0.039)

0.082** (0.035)

foreign_born -0.0008 (0.003)

-0.0002 (0.004)

interactions LMA*unepmployed -0.0007

(0.000)

LMA*blue collar 0.0008 (0.000)

LMA*educ -0.0002*** (0.000)

LMA*right -0.00009 (0.000)

unempltrd*unemployed 0.059** (0.022)

unempltrd*blue collar 0.01 (0.016)

21

unempltrd*right -0.007** (0.003)

family*educ -0.0001 (0.000)

family*right -0.0002 (0.000)

foreignborn*right -0.002*** (0.000)

*** p< .001, ** p< .01, *p< .05 (robust standard errors in parentheses)

22

Figure 1. Mean opinions about the economic usefulness of immigrants (immigrant); by country

0.7250.694

0.6540.645

0.6160.561

0.5390.537

0.5290.524

0.4840.474

0.4660.4650.464

0.4260.4190.416

0.3890.325

0.2730.239

0.227

negative indifferent positiveAttitude towards immigrants (mean)

esseludkfiieptfrnlit

cypl

beat

gewsi

gbczgr

geemthusk

Figure 2. Mean attitudes towards ethnic minorities (ethnic); by country

0.7130.679

0.6710.6400.635

0.6170.594

0.5760.570

0.5450.535

0.5280.5060.502

0.4910.488

0.4590.455

0.4410.409

0.3110.260

0.220

negative indifferent positiveAttitude towards ethnic minorities (mean)

fiseplfr

dknl

gewlusi

geebeieskhuesptgb

itatczgrcymt

23

Annex Table1 List of variables name question details Dependent variable

immigrant Average of two questions 1.Immigrants are needed for economy 2.Immigrants are needed to solve aging problem

Tend to agree-tend to disagree Recoded, higher values indicate positive evaluation. Source: EB

ethnic Average of three questions 1.People of different ethnicity enrich cultural life 2.People of different ethnicity cause insecurity 3.People of different ethnicity increase unemployment

Tend to agree-tend to disagree Recoded, higher values indicate positive evaluation (no sterotypes). Source: EB

Independent variables micro status (Labor market status) Current occupation

1.Pensioneer and housewifes 2.Student 3.Unemployed 4.Self-employed 5.White collar (reference) 6.Blue collar

Source: EB

past If you compare your present situation with five years ago, would you say it has improved, stayed about the same, or got worse?

Recoded, higher value indicates improvement Source: EB 66.3

future In the course of the next five years, do you expect your personal situation to improve, to stay about the same or to get worse?

Recoded, higher value indicates improvement Source: EB

left Left-right scale, 1 left 10 right Source: EB age years Source: EB sex 1 male 0 female Source: EB education age when left school Source: EB residence 1.rural 2. small town 3. big town Source: EB migrant 1 if born outside country of residence

or if parents immigrants 0 otherwise

Source: EB

Independent variables macro Labour market access LMA 0 – 100, high numbers indicate more

open labour markets Source: MIPEX

Family reunification family 0 – 100, high numbers indicate more liberal reunification rules

Source: MIPEX

Unemployment trend uempltrd 1 increasing unemployment 0 decreasing unemployment

Source: Eurostat

GDP growth GDPgrow Difference between 2005-2002 of GDP rates

Source: World Bank

Foreign born population foreign born

Proportion of foreign born population, of total population

Source: Migration Information Website (cites OECD and World Bank data)

24

Table 2 Descriptive statistics of variables used in the analysis Variable N mean sd Min max

micro level: individuals ethnic 13757 0.53 0.35 0.00 1.00 immigrant 13757 0.48 0.39 0.00 1.00 retiree 13757 0.36 0.48 0.00 1.00 student 13757 0.07 0.26 0.00 1.00 unemployed 13757 0.05 0.23 0.00 1.00 selfemployed 13757 0.08 0.27 0.00 1.00 Blue collar 13757 0.13 0.34 0.00 1.00 Educ 13757 18.30 3.15 5.00 22.00 Past 13757 2.16 0.79 1.00 3.00 Future 13757 2.24 0.70 1.00 3.00 Small town 13757 0.38 0.48 0.00 1.00 Big city 13757 0.27 0.44 0.00 1.00 Migrant 13757 0.11 0.32 0.00 1.00 Right 13757 5.29 2.12 1.00 10.00 male 13757 0.48 0.50 0.00 1.00 age 13757 48.29 17.44 15.00 98.00 macro level: countries LMA 23 56.96 20.16 25.00 100.00 Family 23 58.22 16.31 32.00 92.00 GDPgrow 23 2.62 1.54 0.44 5.50 unempltrd 23 0.61 0.50 0.00 1.00 Foreign born 23 9.40 7.35 1.80 37.40

25

References Bauer, Thomas K., Magnus Lofstrom, and Klaus F. Zimmermann. 2000. Immigration Policy,

Assimilation of Immigrants and Natives' Sentiments towards Immigrants: Evidence from 12 OECD-Countries. Edited by I. D. N. 187, F. z. Z. d. Arbeit and I. D. N. Institute for the Study of Labor.

Bermingham, John R. 2001. Immigration: Not a solution to problems of population decline and aging. Population and Environment: A Journal of Interdisciplinary Studies 22 (4):355-363.

Blumer, Herbert. 1958. Race prejudice as a sense of group position. Pacific Sociological Review 1:3-7.

Bobo, Lawrence. 1988. Group conflict, prejudice, and the paradox of contemporary racial attitudes. In Eliminating Racism: Profiles in controversy edited by P. K. a. D. Taylor. New York: Plenum Press.

Castles, S. . 1995. How nation-states respond to immigration and ethnic diversity. New Community 21 (3):293-308.

Castles, S., and A. Davidson. 2000. Citizenship and migration. Globalization and the politics of belonging. New York: Routledge.

Ceobanu, Alin M., and Xavier Escandrell. 2008. East is West? National feelings and anti-immigrant sentiment in Europe. Social Science Research 37:1147-1170.

———. 2010. Comparative analyses of public attitudes toward immigrants and immigration using multinational survey data: A review pf theories and research. Annual Review of Sociology 36:309-328.

Chellaraj, Gnanaraj, Keith E. Maskus, and Aaditya Mattoo. 2006. Skilled immigrants, higher education and U.S. innovation. In International migration, remittances and the brain drain, edited by M. S. Ça�lar Özden. Washington DC: World Bank and Palgrave Macmillan.

Constant, Amelie, and Klaus F. Zimmermann. 2006. The making of entrepreneurs in Germany: Are native men and immigrants alike? Small Business Economics 26 (3):279-300.

Crepaz, Markus M. L. , and Regan Damron. 2009. Constructing tolerance: How the welfare state shapes attitudes about immigrants. Comparative Political Studies 42 (3):437-463.

Dustmann, Christian, and Ian Preston. 2006. Is immigration good or bad for the economy? Analysis of attitudinal responses. Research in Labor Economics 24:3-34.

Emmenegger, Patrick, and Romana Careja. 2011. From dilemma to dualisation: Social and migration policies in the ‘reluctant countries off immigration’ In The Age of Dualization:Structures, Policies, Politics, and Divided Outcomes, edited by S. H. Patrick Emmenegger, Bruno Palier, and Martin Seeleib-Kaiser. Oxford, UK: Oxford University Press.

Hainmueller, Jens, and M.J. Hiscox. 2002. Educated preferences: Explaining attitudes toward immigration in Europe. International Organizations 61:399-442.

Heijm, Mikael. 2007. Do number really count? Group threat theory revisited. Journal of Ethnic and Migration Studies 33 (8):1253-1275.

Howse, Kenneth. 2006. The benefits of migration for an ageing Europe. Ageing Horizons (4):12-18.

Kehrberg, J.E. 2007. Public opinion on immigration in Western Europe: Economics, tolerance, and exposure. Comparative European Politics 5:264-281.

Kunovich, Robert M. 2004. Social structural position and prejudice: an exploration of cross-national differences in regression slopes. Social Science Research 33:20-44.

26

Lahav, Gallya 2004. Immigration and politics in the new Europe: reinventing borders: Cambridge University Press.

McLaren, L.M. . 2003. Anti-immigrant prejucide in Europe: Contact, threat perception, and preferences for exclusion of immigrants Social Forces 81 (3):909-936.

Meuleman, Bart. 2009. The influence of macro-sociological factors on attitudes toward immigration in Europe. A cross-cultural and contextual approach, Faculty of Social Sciences, Catholic University Leuven, Leuven.

Muysken, Joan, Frank Corvers, and Thomas Ziesemer. 2008. Immigration can alleviate the ageing problem, METEOR, Maastricht Research School of Economics of Technology and Organisation, Research Memoranda No 004.

OECD. 2009. OECD Factbook 2009. Paris: OECD. Paparella, Domenico. 2004. Social Security Coverage for Migrants: Critical Aspects. Edited

by A.-. Paper read at European regional meeting "Migrants and Social Protection' of the International Social Security Association, Oslo.

Pettigrew, T.F. 1998a. Intergroup contact theory. Annual Review of Psychology 49:65-85. Piore, Michael J. 1979. Birds of Passage. Migrant labor and industrial societies. Cambridge:

Cambridge University Press. Quillian, Lincoln. 1995. Prejudice as a response to perceived group threat: population

composition and anti-immigrant and racial prejudice in Europe. American Sociological Review 60 (1):586-611.

Riaño, Yvonne , and Doris Wastl-Walter. 2006. Immigration policies, state discourses on foreigners and the politics of identity in Switzerland. Environment and Planning A 38 (9):1693-1713.

Rustenbach, Elisa. 2010. Sources of negative attitudes toward immigrants in Europe: A multi-level analysis. International Migration Review 44 (1):53-77.

Scheepers, Peer, Merove Gijberts, and Marcel Coenders. 2002. Ethnic exclusionism in European countries. Public opposition to civil rights for legal migrants asa response to perceived ethnic threat. European Sociological Review 18 (1):17-34.

SOPEMI. 2008. Trends in International Migration/International Migration Outlook. Paris: OECD.

Soysal, Yasemin Nuhoglu. 1994. Limits of Citizenship: Migrants and Postnational Membership in Europe. Chicago IL/London: University of Chicago Press.

Welch, S., L. Sigelman, T. Bedsoe, and M. Combs. 2001. Race and place. Race relations in an American city. New York: Cambridge University Press.

Weldon, Steven A. . 2006. The Institutional Context of Tolerance for Ethnic Minorities: A Comparative, Multilevel Analysis of Western Europe. American Journal of Political Science 50 (2):331-349.