Labour Migration in the European Union:
The EU Eastern enlargement’s impact on opinions and
confidence in the EU as a whole
Master thesis
in the study programme Public Administration MSc
(Specialisation: Economics and Governance, 17/18)
by Elisabeth Teske (s2087162)
Supervisor: Dr Max van Lent
Abstract
With the Eastern enlargement of the European Union (EU) in the first decade of the 21st
cen-tury, labour migration has been a highly discussed topic in the European public sphere. The facilitation of migration due to open borders promoted fears of mass immigration especially in the old member states. Moreover, as many policy-making competencies have shifted from the national to the European level and people blame the EU for unpleasant developments, a nega-tive opinion towards labour migration might correlate with lower confidence in the EU. An analysis with micro data from the European Value survey from 27 EU members over a period of 18 years suggests that the opinion on labour migration in general became more favourable over time but was thwarted by the Eastern enlargement. Furthermore, the opinion depends greatly on the respondent’s nationality and skill level as well as other socio-demographic fac-tors. Unlike economic theory suggests, a higher skill level has always a positive impact on the opinion on labour migration. A correlation between the opinion on labour migration and con-fidence in the European Union exists, as with a more positive attitude towards labour migration, the confidence also rises. The findings point to short-comings in the existing economic theory and provide important insights for policy-makers.
Contents
1. Introduction ... 1
2. The EU - Enlargement and Migration ... 3
2.1 Labour Migration within and to the EU ... 5
3. Literature Review ... 7
3.1 Labour Migration in the EU ... 7
3.2 Support for Labour Immigration... 8
3.3 EU Support... 9
3.4 Own Contribution ... 10
4. Theoretical Background ... 11
4.1 The Economic Effects and Mechanisms of Labour Migration ... 11
4.2 Budget and Welfare State ... 13
5. Hypotheses ... 14
5.1 Labour Migration ... 14
5.2 Mediating Effect of Skill Levels ... 15
5.3 EU Support... 16
6. Research Design ... 17
6.1 Data Collection and Case Selection ... 18
6.2 Other Data ... 20
6.3 Methodology and Operationalisation ... 20
6.3.1 Dependent Variables ... 21
6.3.2 Independent Variables ... 22
7. Description of the Data ... 24
7.1 Labour Migration ... 24 7.2 Confidence in the EU ... 27 8. Results ... 29 8.1 Hypothesis 1 ... 30 8.2 Hypothesis 2 ... 34 8.3 Hypotheses 3 and 4 ... 41 9. Robustness Checks ... 46 9.1 Change of Variables ... 47
9.2 Common Trend Assumption (New Member States and Non-Member States) ... 49
9.2.1 Descriptive Analysis ... 50
9.2.2 Regression Analysis ... 52
9.3 Reversed Causality ... 55
10. Conclusion ... 56
12. Appendix ... 62
Table of Figures
Figure 1 - Descriptive Statistics - Opinion on Labour migration ... 24Figure 2 - Descriptive Statistics – Confidence in the EU ... 27
Figure 3 - Regression Analysis - Opinion on Labour Migration (General Sample) – Macro Level ... 31
Figure 4 - Regression Analysis - Opinion on Labour Migration (New Member States) – Macro Level ... 32
Figure 5 - Regression Analysis - Opinion on Labour Migration (Old Member States) – Macro Level 33 Figure 6 - Regression Analysis - Opinion on Labour Migration (General Sample) – Micro Level ... 35
Figure 7 - Regression Analysis – Opinion on Labour Migration (New Member States) – Micro Level ... 36
Figure 8 - Regression Analysis - Opinion on Labour Migration (General Sample and Control Variables) – ... 38
Figure 9 - Regression Analysis - Opinion on Labour Migration (New Member States and Control Variables) – Micro Level ... 39
Figure 10 - Regression Analysis - Opinion on Labour Migration (Old Member States and Control Variables) – Micro Level ... 40
Figure 11 - Regression Analysis - Confidence in the EU (General Sample) – Micro Level ... 41
Figure 12 - Regression Analysis - Confidence in the EU (New Member States) – Micro Level ... 42
Figure 13 - Regression Analysis - Confidence in the EU (Old Member States) – Micro Level ... 43
Figure 14 - Regression Analysis - Confidence in the EU (General Sample and Control Variables) – Micro Level ... 44
Figure 15 - Descriptive Statistics - Opinion on Labour Migration (New and Non-EU Member States) ... 50
Figure 16 - Descriptive Statistics – Confidence in the EU (New and Non-EU Member States) ... 51
Figure 17 - Regression Analysis on Opinion - Labour Migration (New and Non-EU Member States) – ... 52
Figure 18 - Regression Analysis - Opinion on Labour Migration (New and Non-EU Member States) – ... 53
1
1. Introduction
Migration in general has always been a highly polarising topic in Europe. The International Organization for Migration (IOM) defines migration as “the movement of a person or a group
of persons, either across an international border, or within a State.” (IOM, 2018).1 In order to
secure continuous economic growth, Western European countries relied on migration in the aftermaths of World War II, to satisfy the growing need for labour. While recruitment first focused on poorer European states in the South, over time former colonies and other parts of the world also became important source countries (McLaren, 2015). But when economies shifted to downward trends governments tried to reduce the labour migration they had initiated, and policy-making focused on economic needs of the Western European Societies. The economic impact of immigration differs among social classes or groups. Besides economic consequences, immigration can also impact a society’s identity as it brings along different values and traditions. These various migration effects shape the individual opinion on immigration. Some people perceive immigration as threatening while others see migrants as an enrichment to the economy and society. Hence, when deciding on migration policies in democratic polities, policy-makers must, besides economic outcomes, also take into account these voters’ attitudes towards immigration opinions (Facchini and Mayda, 2008). Yet, opinions are not set in stone and can change over time as internal, e.g. the economic outlook, or external circumstances such as migration inflows evolve forcing policy-makers to adjust. Therefore, in order to understand policies and the process of their creation, studying public and individual attitudes towards a certain topic is crucial. Comprehending and explaining the process of policy-making is a substantial part of the academic discipline of public administration, ”the study of government decision-making, the analysis of the policies themselves, the various inputs that have produced them, and the inputs necessary to produce alternative policies.” (Morrow, 1975 as cited in McKinney and Howard, 1998, 62)
The present work will focus on labour migration in Europe. The research interest is thereby twofold. First, the evolvement of the opinion on labour migration and its influencing factors will be investigated. Second, the research is concerned with the connection between the opinion on labour migration and the support for the European Union (EU). With the introduction of the European Single Market in 1993 in the then-called European Community (EC) which is now the EU, labour migration between the in the beginning mainly Western European member states
1 Although migration can happen for many reasons such as flight, displacement or family reunion, this work focuses on labour migration, i.e. “movement of persons […] for the purpose of employment.” (ibid.)
2
was facilitated to a great extent. The goal of free movement and a single European Economic Area was already set in the Treaty of Rome in 1957, when the group of member states was largely homogenous in economic terms. With the accession of economically less prosperous states in Southern Europe, fears of mass immigration from these countries and a consequential burdening of the labour markets and welfare systems rose but the anxieties proved to be ill-founded (Kvist, 2004). Yet, with the EU’s largest enlargement, the accession of in total 12 countries including 10 Central and Eastern European countries (CEECs) in 2004 and 2007, concerns emerged again given the cleavages in terms of economic performance and political systems between old and more rich member states and new mostly less wealthy acceding states. These fears most likely shape opinions on labour migration. Given the fears in the older mostly Western EU member states, a change in opinion towards labour migration can be expected over time and particularly with the accession of the new mostly Eastern EU member states. On the other hand, during the time of the Cold war, most CEECs were part of the Warsaw Pact and immigration as well as emigration was restricted. With acceding the EU, both were suddenly more likely, a phenomenon with no prior experience of reference. In these countries, too, a change in opinion can be expected. This change may differ from that in the old member states. Moreover, different factors on the macro and micro level can shape the opinion. One part of the present Master thesis will therefore investigate the opinion on labour migration in various European Countries over a timeframe of 18 years lasting from 1990, so well before the Eastern enlargement, until 2008, a few years after the accession rounds.
Secondly, the opinion on labour migration is closely connected to support for the EU. With
the deeper positive and negative (economic) integration of the EU,2 many competencies were
transferred from the national level to the EU. Thus, individuals often blame the EU for immigration. When doing so, they point at the freedom of movement as well as at its extensive possibilities to influence policy-making (Barbulescu and Beaudonnet, 2014). A quite recent example are the arguments used by the pro-Brexit camp in the UK referendum on whether Great Britain should continue to be part of the EU, claiming that the EU is responsible for (perceived) high levels of immigration and that immigrants, mainly from the new member states, lower wages (Lichfield, 2016). Since the campaign was successful, the Brexit is an exemplary political event, at which perceptions mattered more than actual economic facts. A second part of this thesis will therefore focus on the confidence in the EU among the member
2 Negative integration constituted the dismantling of barriers whereas positive integration is the creation of common policies (Nello, 2012).
3
states and how the opinion on labour migration impacts support for the EU. The research question guiding this work then presents itself as follows:
How did the opinion on labour migration in Europe develop over time, especially with the EU enlargement, which factors influenced it and to what extent does it relate to
the confidence in the EU?
The thesis will be structured as follows: First, some background information on the EU enlargement and migration as well as an overview over the existing literature researching the opinion on labour migration and EU support will be given. It is followed by a theoretical background explaining the economic effects of migration for host and source countries using standard economic theory. Deriving from that, the hypotheses of this thesis will be explained, followed by the research design and methodological framework. The used data will be examined first in descriptive statistics in order to identify some preliminary trends. These will then be followed by regressions both on the macro and micro level and some robustness checks. A concluding discussion will present the results.
2. The EU - Enlargement and Migration
On 1 may, 2004, ten new countries joined the European Union, namely Cyprus, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia and Slovenia, increasing the number of member states abruptly from 15 to 25. It constituted the largest accession wave, in terms of numbers of countries as well as population size, in the history of the EU which emerged from the first-called European Coal and Steel Community (ECSC) and the European Economic Community (EEC). With the accession of Bulgaria and Romania on 1 January 2007, which were not seen fit in 2004 to join the EU, the so-called 10+2 Round was completed. Given the large number of Central and Eastern European Countries joining, these enlargement rounds are referred to as a transformation of “what had been a process of Western European Integration into a near Europe-wide process of integration.” (Nugent, 2010, 35, emphasis in original) The integration of the mostly former communist states into the liberal Western European order was, besides tighter economic connections and liberalisation, the main motivation for the accession negotiations, that were rapidly held right after the collapse of the Soviet Union (ibid).
On the one hand, although highly desired by both the joining countries and old member states, the accession was also feared by the Western European countries. They acknowledged the large differences on economic, cultural and political terms. Despite already existing trade
4
and co-operation agreements and structural aid programmes such as PHARE (Poland and Hungary Assistance for the Restructuring of the Economy) and EBRD (European Bank for Reconstruction and Development), which aimed to prepare the accession candidates for a membership in the EU (Nello, 2012), the economic performance of the acceding countries was considerably lower than that of the EU countries. The gap had never been this great in any previous enlargement round. The differences stemmed from the huge transformations from planned economies to free market systems and from authoritarian communist regimes to democracies the CEECs’ political systems and economies had just undergone. However, accession of the CEECs was highly promoted by the EU and, in order to appease hesitating member states, the so-called Copenhagen conditions were introduced containing political as well as economic criteria to “provide some kind of objective basis for selecting countries ready to join the EU, as well as indicating to the applicant countries the tasks they are expected to perform.” (ibid., 422) Nevertheless, in 2002 the GDP of the acceding countries amounted for 4.8% of the GDP of the EU-15 and, although variation existed among the new member states, all had lower GDP than the old member states and, on average, the per capita income amounted of one fourth of the old member states (Kvist, 2004). Moreover, the EU-15 had established more comprehensive social protection than the CEECs (Gaston and Rajaguru, 2013). These constituted push factors for emigration from the new states (i.e. domestic factors that drive individual out of their country of origin) and pull factors for immigration into the old member states (i.e. domestic factors that attract individuals to a foreign country) and made policy makers in the Western countries fear for mass immigration due to the freedom of movement after the accessions (Kahanec, Zaiceva and Zimmermann, 2010).
With becoming members of the European Union, the CEECs gained access to the common market with the free movement of persons, goods, services and capital, the so-called four freedoms. Besides benefits for tourists, the freedom of movement also applies to workers under Article 45 of the treaty of the Functioning of the European Union (TFEU), which gives EU nationals the freedom to seek employment EU wide and protects them against discrimination as well as grants them the right to remain in the state chosen after ending their employment. Article 49 of the TFEU grants the right of establishment or setting up businesses in other member states. In theory, new member states enjoy these freedoms with the moment of accession. In practice, given the fears in the old member states, transitional arrangements were put into place by almost all old member states. In some countries, access to the EU-15 labour markets for workers from CEECs was restricted for a time period of up to 7 years after the
5
accession (Kahanec, Zaiceva and Zimmermann, 2010). Other countries did not restrict access but adjusted their social policies to ‘protect’ them from migrants from the from CEECs (Kvist, 2004). On the side of new member states, arrangements were equally made to restrict access to their national labour markets for workers from the old member states (Nello, 2012).
2.1 Labour Migration within and to the EU
The accession of new members to the EU naturally impact labour migration flows, especially after the introduction of the four freedoms. Fears of mass immigration were not supported by experiences from previous accession rounds as they had not shown massive increase of immigrants from acceding countries to older EU member states, particularly after economic conditions in new and old member states converged. During the cold war, no migration from the CEECs to Western Europe was allowed, thus after 1989, waves of immigrants were feared as the migration potential from these countries may not have been exhausted as it was the case in prior acceding countries (Kvist, 2004). In a study for the European Commission, Boeri and Brücker (2001) projected a net migration of 335, 000 people immediately after the Eastern enlargement, due to the high differences in per-capita income without transitional arrangements. The inflow, although declining, would consist over a longer period of time, hence contradicting fears of mass immigration. Although the Eastern enlargement would not affect employment and wages on an aggregate level, they argue that specific sectors and regions would be more exposed than others.
The stock of foreign residents from the eight CEECs acceding in 2004 in the old member states increased to 1,910,000 by the end of 2007 compared to 893,000 in 2003, and 1,860,000 residents from Bulgaria and Romania in 2007, amounting for 0.5% of the population, respectively (Kahanec, Zaiceva and Zimmermann, 2010). Since most immigration from the CEECs prior to the enlargement concentrated on Austria and Germany, these countries were expected to receive the largest share of immigrants given their proximity to the CEECs and network effects. Ireland and Portugal which had the smallest population share of CEEC citizens were expected to receive the smallest number of immigrants. Yet, given the implementation transitional arrangements, the labour market access to these countries was restricted and instead the UK and Ireland, which had not imposed restrictions, constituted the main destination for emigration after the 2004 enlargement (Nello, 2012). Regarding Romania and Bulgaria, Spain and Italy attracted the largest share of immigrants from these countries. Nevertheless, it should be kept in mind that with the expiration of the transitional agreements, migration might increase
6
to member states such as Austria and Germany as with transition periods migration will only be postponed but not lowered in levels overall (Kvist, 2004).
Most labour migration into the EU, especially into the Western European countries, however, comes from outside Europe and the stocks of non-EU foreign residents are significantly higher (4.5% in 2007) (Kahanec, Zaiceva and Zimmermann, 2010). According to the International Organization for Migration (IOM), the main EU countries of destination in 2008 were Germany, the UK, France, Italy and Spain. Immigration policy is considered a
common interest and was incorporated in the third pillar of the Maastricht treaty,3 yet the
cooperation focuses mainly of the combat of illegal immigration, asylum policy and the external border protection (Nugent, 2010). An achievement regarding labour migration, the focus of this work, is the introduction of the so-called blue-card system that facilitates EU work permits for skilled workers and is inspired by the US green cards (Nello, 2012).
In the last decades, immigration into the EU increased due to, inter alia, the end of the cold war and facilitated moving conditions. Popular countries receiving large total numbers of immigrants among Western Europe are Germany and the UK: For example, in 1989/90 Germany received by far the largest total inflow of immigrants (1.5 Million or 1.9% of the
population)4 followed by the UK (250,000 or 0.4% of the population). This trend continues in
later years, yet the inflows into Germany diminish (802,500 in 1998 and 680,800 in 2008) whereas the numbers steadily increase for the UK (332,400 in 1998 and 590,200 in 2008). Nevertheless, these inflows still constitute a small fraction (less than 1%) compared to the population. Countries with high inflows of immigrants compared to their size of population are Luxembourg (2.7% in 1998 and 3.5% in 2007) and Iceland (1.1% in 1989, 1.6% in 1998 and 3.3% in 2008). In the last years, Spain (958,300 in 2007 compared to 81,200 in 1998), Italy (534,700 in 2008 compared to 156,900 in 1998) and France (294,000 in 2007 compared to 57,800 in 1999) became more attractive. Regarding the new member states, fewer numbers exist, but in general, the inflows are lower. Countries that receive the highest total amounts of immigrants are Slovenia (6,000 in 1992, 4,600 in 1998 and 29,200 in 2008) and the Czech Republic (9,900 in 1999 and 104,400 in 2007). In general, inflows rarely exceed a fraction of 1% compared to the native population.
3 The Treaty of Maastricht, entering into force in 1993, constituted the creation of the European Union basing it on three so-called pillars: 1. The European Communities, 2. The common Foreign and Security Policy and, 3. The Cooperation in Justice and Home Affairs (Nugent, 2010).
7
3. Literature Review
After migration in the context of EU enlargement was portrayed, this section reviews the existing literature on labour migration with a particular focus on the public opinion on immigration and the public support for the European Union. The selected strands of literature provide relevant background knowledge. They also explain how the present thesis fits into the academic literature on labour migration.7
3.1 Labour Migration in the EU
Whereas the current public debate in Europe focuses on refugees and their impact on Europe, a lot of the earlier academic literature discussed labour migration, examining push and pull factors, such as policies or economic circumstances, as well as the effects. Regarding the factors influencing migration decisions, there are three main positions. Firstly, many scholars argue that welfare state generosity is important (cf. De Giorgi and Pellizzari, 2006). The generosity of the social system can act as a magnet influencing the decision of emigrating and the country of destination. Yet, labour market conditions such as lower unemployment rates and higher real wages matter too, which is a second, competing, position. Warin and Svaton (2008) for example conclude that besides network, geospatial and linguistic determinants, labour market conditions and social expenditure levels affect migration decisions from CEECs, Eastern European Countries and developing countries into the EU-15. A third group stresses the impact of non-economic influences (cf. Pedersen et al., 2004). They argue that factors such as cultural distance or network effects also impact the decision. Moreover, also policies matter as Kraus and Schwager (2000) discover. For instance, the prospect of accession to the EU likely leads to a postponement or cancellation of immediate emigration.
Besides the determinants for migration, the economic effects of migration, primarily in the policy fields of taxation and welfare state in the host country, gain much academic attention. The effects often depend on the composition of immigrant flows and their skill levels which will be further explained in the theoretical background. In general, much research focuses on the United States (US) with important contributions for example by Borjas (1995), who argues that if the immigrants’ skills differ sufficiently from those of the native stock, immigration is beneficial for the host country. The competing literature branches disputes whether migration leads to higher taxes and more welfare generosity (cf. Gaston and Rajaguru, 2013), i.e. a compensation against the risk of job insecurity due to a higher influx of immigrants, or lower taxes and less welfare state generosity, such as Razin et al. (2002), who argue that tax rates are lowered in the case of low-skilled immigration as the native population is less willing to pay
8
taxes to support low-skilled immigrants who will receive an increasing share of revenues. Moreover, strategic interaction regarding tax and benefit levels can occur between countries in order to attract more high-skilled workers, leading to a race to the bottom, i.e. a competition about the lowest tax rates and benefit levels (Hindriks, 1998). In light of the EU enlargement 2004, interaction between old member states took place and decisions on restriction of labour markets were made because other EU-15 member states did so too (Kvist, 2004).
3.2 Support for Labour Immigration
Because labour migration always has an effect, be it economic or social, on the host but even more so on the source country, it leads to public debates. As debates emerge, individuals form and express opinions on labour migration. Put simply, they either support or reject it. This, in turn, shapes policy-making so that much research is conducted on the individual support for labour migration in the host country as it is done in this thesis. In the beginning, scholars mainly looked at migration flows into the US, but recently attention also shifted towards the European Union. Support is mostly conditioned on macro and micro level factors. On the macro level, the number of incoming migrants and the perception whether migration affects social benefits and salaries matters, e.g. if labour market competition is enhanced by incoming migrants, more restrictive policies as well as higher benefit levels are preferred (Murard, 2017). Regarding the micro level variables, the opinion is shaped by the individual labour market position, i.e. the possibility of enhanced competition for employment due to immigration, and the impact of altered welfare policies on the individual. Scholars such as Scheve and Slaughter (1999) and O’Rourke and Sinnott (2006) discover a link from skills to wages and immigration-policy preferences. Lower skilled workers associate a greater influx of immigrants with lower wages, hence preferring a restrictive immigration policy. This is further conditioned on the host country’s economic performance. In general, natives will be most hostile towards immigrants with the same skill level as this weakens their own position in the labour market (Hainmueller and Hiscox, 2007). Nevertheless, cultural causes can alter attitudes, such as xenophobia, cultural background and religion as well as socio-demographic factors and with higher education people are in general more prone to immigration regardless of the immigrants’ skill level and their country of origin (ibid.).
The link between skill levels or income and preferences for immigration policies can also be explained by concerns for the fiscal burden of public services. In general, high-skilled immigration is preferred as it usually burdens the social system less (Hanson, Scheve and Slaughter, 2007) and immigration from richer countries than from poorer countries since this
9
lowers the risk of welfare migration (Malchow-Møller et al., 2009). Moreover, rich natives are more hostile towards low-skilled immigration as they usually share a higher tax burden (Faccini and Mayda, 2009; Hainmueller and Hiscox, 2010). Besides shaping preferences about the skill mix of incoming migrants, migration can also lead to preferences of higher welfare state generosity as a tool of compensation. Especially in occupations exposed to the effects of immigration, support for government redistribution increases (Burgoon, Koster and Van Egmond, 2012).
Lastly, attention is paid to how individual attitudes are transformed into policy measures, i.e. important contributions to the academic field of public administration. Facchini and Mayda (2008), for example, argue that although migration is economically beneficial and was facilitated compared to the past, numbers are still limited to a great amount due to restrictive migration policies. In a democratic society these are largely determined by individual preferences and only a minority supports open migration policies.
3.3 EU Support
As with the support for labour migration, much academic attention is paid to the drivers of support for the EU and European integration. Various determinants such as economic, cultural and domestic factors have been identified. Many scholars argue that on a micro level differences in skill levels matter for the support of the EU, an example being Anderson and Reichert (1995) who state that membership naturally creates winners and losers, and approval depends on the status of being a winner (more supportive) or loser (less supportive). High-skilled individuals who benefit indirectly from the EU are in general more supportive. Moreover, low-skilled support is related the position of their relative wage in EU comparison, e.g. high-wage low-skilled workers will be more hostile, whereas high-skilled support is influenced by their relative human capital value (Gabel, 1998). Support for the EU becomes stronger if their occupational opportunities rise due to the liberalised EU market meaning that with a relative high native wage support declines, and with relative high value of human capital support increases. Besides the individual skill level, the skill endowment on a national level is important as individuals from lower-skilled countries are more supportive than individuals from higher-skilled countries (Bringear and Jolly, 2005).
Yet, others emphasize socio-demographic and national differences in explaining variation such as the influence of identity-based factors on the support for European Integration (cf. Hooghe and Marks, 2004). National identity is constructed through socialisation and political
10
conflicts. Following this, it can restrain as well as enhance support for European integration depending on whether national identity is considered inclusive or exclusive (ibid.).
Others argue that fears of immigration are a key to understanding public support for EU integration, thus combining migration and EU support which is also the approach of the present thesis. McLaren (2002), for instance, argues that hostility towards the EU is not only due to calculations of costs and benefits but because of the fear of other cultures. The nation-state is seen as the main point of reference for identity and EU integration softens the integrity of the nation-state. People identify with their fellow natives as an in-group and seek to protect it. The more this is the case, the more hostile individuals are towards the EU. The categorisation of in- and out-groups is also picked up by other researchers such as De Vreese and Boomgaarden (2005). If other Europeans are seen as outsiders, people are less favourable towards further European Integration. Other factors such as a positive attitude towards the own national government, feelings of economic security and political sophistication are support-enhancing. The direct effect of immigration on attitudes towards the EU is investigated by Barbulescu and Beaudonnet (2014). They argue that support has declined because the EU is perceived as having a positive and liberal opinion on migration and being responsible for increased inflows of immigrants whereas individuals see immigration as a threat.
3.4 Own Contribution
The previous literature review reveals a quite extensive already existing literature on the support for labour migration as well as the EU which constitute the foci of the present study, but gaps still exist that this work will try to fill. The contribution is threefold. First, most work is static and often presents only a ‘snapshot’ at a certain point of time. However, attitudes are fluid because the circumstances that shape opinions change over time. Opinions are not static. This is acknowledged by Murard’s work (2017) exclusively. She studies the impact of migration inflows in Europe on the individual attitudes being “one of the first to look into the
evolution of attitudes towards immigration.” (ibid., 3) The present paper studies the same
phenomenon, yet examining the impact of a significant event, the EU Eastern enlargement. So far, no study has been conducted on the effect of the EU enlargement on the opinion towards labour migration making it the first contribution.
Moreover, the study connects different branches of the existing literature. Like Barbulescu and Beaudonnet (2014), a connection is made from the opinion on labour migration to the support for the EU. Yet, their study is limited to only one country, Italy. The second contribution, therefore, is the extension of the sample to various European countries over time, again examining the time before and after the Eastern enlargement in particular. Additionally,
11
when looking at the enlargement impact, academic focus mainly lies on the old member states, so the analysis looks at both old and new member states and the differences, the third contribution. In sum, the study will provide a quite extensive picture on the attitude towards labour migration and the consequential support for the EU in different countries over a longer period of time analysing both the macro and micro level, thus re-examining the insights from separate works and contributing especially to the research on the EU Eastern enlargement.
4. Theoretical Background
The following section explains some simple economic insights why migration occurs and how it affects the economies of both the host and source country. The first part regards the gains of migration for production neglecting the welfare states effects which will be discussed in the second part.
4.1 The Economic Effects and Mechanisms of Labour Migration
The determinants of migration are manifold and interact in a very complex way. Influencing factors can be of economic, cultural, geographical and political nature but the focus here will be set on economic factors. Migration occurs when, first, differences in income (equality) or labour demand between the host and the source country exists and, secondly, the expected benefits offset costs for the individual migrating. The welfare effects of international migration depend on the level of analysis (Gaston and Rajaguru, 2013). If one takes a global focus, international migration produces overall gains (Kemp, 1993). Applying the gains-from-trade theorem shows that benefits from free international migration are even higher than those from free international trade. But these gains have to be re-distributed either inter-nationally or intra-nationally. Global redistribution, however, is, in practice, hardly possible.
Looking at the national level, the host country and its population benefit from immigration. In a host country whose technology consists of two production inputs, labour and capital, immigration leads to an increase in labour supply and a loss in market wage. National income for natives increases, thus creating the immigration surplus, since more can be produced to a lower price. This benefits the capital owners and offsets the losses for workers. Immigration in this situation causes a wealth redistribution from labour to capital. Yet, a surplus only emerges if the native wages are responsive to immigration, i.e. an inelastic labour demand curve, and do not contribute the same amount of capital as the native population since otherwise only migrants would gain and not the natives (Borjas, 1995). Additionally, if wages are not fully flexible, for example due to minimum wages, an increase in labour supply can lead to unemployment (Kahanec and Zimmerman, 2011). Immigration can also lead to increasing
12
returns to scale, e.g. through knowledge transfers and produce an increased marginal product of labour and capital thus more output and a higher immigration surplus. Like the capital owners, immigrants benefit from migration since they receive higher wages than in their country of origin. In the source country, the picture is reversed. Labour gains given a decrease in supply leading to higher wages and capital owners lose. Moreover, the country gains with the likely received remittances. Hence, migration decreases the inequality between host and source country (ibid.).
So far, the theory only focused on capital and labour, thereby assuming no differences in skills between native and host country as well as within a country. The increase in national income depends on the skill composition of the incoming workers, more specifically to what extend the skill composition differs (Borjas 1995; Kahanec and Zimmermann, 2011). If skill compositions between native population and migrants differ, the immigration surplus remains positive. To simplify the production model, only two forms, skilled and unskilled workers, are distinguished. With the same proportion of skilled and unskilled workers as the native population, wages are not affected and no immigration surplus for natives emerges, hence migrants labour force has to be complementary to the native labour force (Borjas, 1995). The surplus maximises if the immigrants’ skills complement the native skills, so in an extreme case if native population consists only of skilled workers, an inflow of purely unskilled workers could maximise the increase in national income and the surplus (in the absence of capital which
will be neglected in this section).5 But as before, the surplus depends on elasticities.
Since it is almost impossible that the host country’s labour market consists of workers with only one skill level, migrant inflows have different impact on skilled and unskilled labour market equilibria. In general, immigrants depress wages or cause unemployment for natives with the same skill level and, since unskilled and skilled labour are seen as complements, increase wages for people with the complement skill level. Therefore, skilled immigrants benefit unskilled and harm skilled natives, leading to a compression of wages, whereas unskilled immigrants benefit skilled and harm unskilled natives, leading to more inequality. With the aim of reducing inequality, high-skilled immigration is preferred.
As said in the beginning, migration occurs if differences in wages and labour demand exist. Building on the Heckscher-Ohlin model, countries differ in their endowments of skilled and
5 Borjas (1995) shows that with taking capital into account the immigration surplus may be higher with an influx of skilled labour as the net fiscal costs are lower with a large share of high-skilled immigrants.
13
unskilled labour. In the country that is skilled labour abundant, wages for skilled labour will be relatively lower and wages for unskilled labour relatively higher and vice versa in a country that is unskilled labour abundant. Hence, skilled migration will occur from the skilled labour abundant country to the country where skilled work is scarce, thereby hurting skilled work in the unskilled abundant country. Unskilled migration works the other way around, thus hurting unskilled work in the skilled abundant country (O’Rourke and Sinnott, 2006).
4.2 Budget and Welfare State
Besides labour market effects, migration also has budgetary effects. As shown in studies mentioned in the literature review, migrants’ decision to move can be influenced inter alia by the size of the welfare states (De Giorgi and Pellizzari, 2006; Warin and Svaton, 2008). Building on the income-maximisation hypothesis, the so-called welfare magnet hypothesis states that the immigrants torn to more generous welfare states are usually unskilled workers (Borjas, 1999). Unskilled workers are usually net welfare-recipients whereas skilled workers are net contributors given that most welfare systems are publicly funded by taxes. Skilled workers therefore prefer to move to countries with smaller welfare states to pay less taxes for a system from which they proportionately gain less than they receive. Moreover, integration of migrants into the labour market affects the welfare state. With a smaller welfare state, the risk of moral hazard decreases and a better integration into the labour market will be accomplished (Nannestad, 2007).
Migration is not only influenced by the size of the welfare state but can also impact its size in return, e.g. through policy changes pull workers with favourable skills or to catalyse integration. In order to attract skilled migration with positive budgetary effects, countries might downsize their welfare state which could lead to a race to the bottom between countries (Kvist, 2004). On the other hand, it can also be used to soften negative effects of immigration for certain groups in form of redistribution. Especially if unskilled natives suffer losses due to immigration, they can be compensated by increased social benefits. In accordance with theories of international trade (cf. Rodrik, 1998), social spending increases in more exposed economies, so that workers who feel economically more insecure will be better protected constituting the so-called exposure or insurance effect (Gaston and Rajaguru, 2013). Increased social spending, will be accomplished by higher taxation. This, in turn, is opposed by skilled workers as they usually share a higher burden and profit less from higher social benefit levels. Moreover, it can have negative consequences for the preferred skill composition of immigrants given the above explained self-selection. Consequently, the aim to attract more high-skilled workers and to
14
appear less preferable to unskilled workers can pressure governments to lower their level of taxation and benefits, the so-called redistribution or tax effect (ibid.).
5. Hypotheses
5.1 Labour Migration
In the public opinion, the focus lies on distributional effects rather than efficiency gains (Borjas, 1995). So, although a certain measure can be welfare enhancing, the perception of disadvantages or the actual experience of losses of certain groups can dominate the public opinion. Since immigration increases the supply of labour in a country which either leads to lower wages or unemployment, this may feed concerns about job security in general or identification rather with fellow natives who could lose their jobs than with migrants with the same skill level and consequently to a rejective opinion towards labour migration. Hence, on a macro level, it can be assumed that higher inflows of migrants decrease the support for labour migration.
This work focuses on the differences in opinion between old and new member states, so despite a general more hostile attitude with a larger fraction of migrants, the initial level of support might differ between states. Migration towards (Western) Europe is characterised by lower level of education and skills (Kahanec and Zimmermann, 2011). These countries can be defined as skill abundant and building on the theory, unskilled immigration benefits skilled natives and harms unskilled natives. Moreover, in general skilled immigration burdens the welfare system less than unskilled immigration, providing the possibility of a greater immigration surplus. Therefore, in the old member states which are mainly in Western Europe, immigration is seen less favourable. On the other hand, the Central and East European countries which are in the majority of the acceding countries in the EU enlargement, are considered to be unskilled abundant as the majority of emigrants to Western Europe have lower skills (ibid.). According to economic theory, they can expect skilled immigration which fiscally speaking is more beneficial than unskilled migration. These countries then are likely to be more in favour of migration, also because they can gain from remittances.
With the enlargement and open borders between old and new member states, migration is facilitated and, theoretically, unskilled migration flows from the CEECs and skilled migration from the Western countries intensify, thus reinforcing the prevailing opinion over time. Important here is that although the EU enlargement might not impact the actual overall inflows of migration drastically, it might be perceived as if, thus creating a more hostile attitude. Again,
15
political effects outweigh the economic effects and perception matters more than actual facts. In that sense, the first hypothesis is the following:
Hypothesis 1: (a) With a higher ratio of immigrants compared to the native population, the average opinion on labour migration becomes more hostile. (b) Old EU member states are in general more hostile towards labour immigration than new EU member states. (c) With the EU enlargement, the old member states become even more hostile whereas the new member states become more favourable towards labour migration.
Yet, as mentioned before, during the time of the cold war, migration in general was more complicated in the CEECs, be it immigration or emigration. Hence, these countries have less experience regarding this topic than Western European countries, thus increasing the possibility of xenophobia. As a consequence, the countries might be more hostile than Western European countries given their worse economic performance. The native employment could suffer with large inflows of immigrants, stimulating fears of unemployment and the loss of the most productive workers further weakens the economy. With the facilitation of entering the countries as a labour migrant and again the perception of increased inflows, the countries might even get more hostile leading to the following hypothesis:
Alternative Hypothesis: (b) New EU member states are more hostile towards labour immigration than old EU member states given their historical background and economic performance. (c) With the EU enlargement, both groups become more hostile.
5.2 Mediating Effect of Skill Levels
On the macro level, immigration inflows influence the public opinion. On the micro level, individual determinants are more important in shaping the personal opinion, especially one’s skill level. Different parts of the population have heterogenous attitudes towards labour migration since they are not all affected to the same extent, be it labour market competition or welfare effects such as taxation or benefit levels. With the expected unskilled immigration into the West European countries, unskilled natives there would lose due to lower wages and pressure on the welfare system, hence making them more hostile than skilled workers. They, on the other hand, would benefit from higher wages, and, in general, are often more educated and less xenophobic. In the unskilled abundant Central and Eastern European Countries unskilled workers are expected to gain from immigration since the supply of labour decreases
16
and whereas skilled labour relatively loses because of possible inflows of skilled migration. It can therefore be expected that unskilled labour perceives migration more positively than skilled workers, reversing the above predicted effect for old member states.
With the possibility to migrate due to the enlargement, the groups benefitting from migration, skilled labour in the old member states and unskilled labour in the new member states, are expected to be more positively disposed towards migration whereas the migration losers, unskilled workers in the new member states and skilled workers in the new member states, will be more negatively inclined.
Hypothesis 2: (a) The skill level works as a mediator for attitudes. (b) In old member states, unskilled workers are more hostile towards labour migration than skilled workers. In new member states, the picture is reversed. (c) With the EU enlargement, the different attitudes between labour groups and countries are reinforced.
5.3 EU Support
The history of the EU has been a success story. It brought peace, stability and prosperity to the (Western) European continent. Therefore, it is expected to enjoy broad support in the old member states. Regarding the individual determinants, skilled labour is expected to be more positive than unskilled labour as they are often more educated about the benefits of the EU. The CEECs sought a stronger connection to the West after the fall of the Iron Curtain by joining NATO and the EU. Given that they applied for EU membership very soon after 1991 and the substantial economic and political gains from the EU, individual support for the EU is also expected to be high, again assuming that the approval rates among higher skilled individuals exceed the ones among unskilled labour.
Building on the results of the articles discussed in the literature review, support is also determined by identity-based factors. Before the EU enlargement, the member states constituted a rather homogenous group with similar economic performance and political systems as well as a shared history and identity as part of the West. With the accession of the Central and Eastern European Countries, more distinct cultures and systems became part of the Union, which can undermine the support. Hence, approval for the EU could have decreased. The CEECs on the other hand enjoy the actual benefits of membership, thus the projection is that the support remains on similar high levels or even grows.
17
Hypothesis 3: (a) Both old and new member states’ populations show high support for the EU. (b) In the post-enlargement time, approval declined in old member states, whereas peoples from the new member states still remain high in their support or even show increasing levels.
Support for the EU can also be correlated to the support for labour migration (De Vreese and Boomgarden, 2005; Barbulescu and Beaudonnet, 2014). Individuals can perceive the EU integration as responsible for immigration as EU directives shape immigration policies such as the EU Blue Card. Hence, with lower support for immigration, lower support for the EU can be assumed. Applying that to the first hypothesis, old member states would be less supportive of the EU since they are more hostile towards labour migration than new member states. As the hostility is projected to increase in the old member states due to the enlargement, EU support would also decrease. In the new member states, approval of the EU increases as the labour migration is projected to be received more positively after the accession.
Hypothesis 4: (a) With greater rejection of labour migration, less support for the EU is expressed. (b) Old member states show less support for the EU than new member states given their hostility towards labour migration. (c) This gap deepens with the EU enlargement.
6. Research Design
The retrospectively conducted research tries to explain the public and individual attitudes in the EU towards labour migration and, subsequently, towards the EU with a special attention to the effect of the EU Eastern enlargement. The topic is concerned with (1) a development over time and (2) the comparison of different countries. Put differently, observations have (1) a longitudinal character as well as (2) a cross-sectional. A panel-design is, therefore, the most suited research design (Toshkov, 2016).
Given that the study investigates the opinions in the EU, it is only logic that the desired and appropriate cases are countries that are part of the EU or members of the European
Economic Area (EEA)6at the end of the observed period as well as their citizens. At the end of
2008, when the investigation period ends, the EU was constituted by 27 member states: Austria, Belgium, Bulgaria, Republic of Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands,
6 The EEA constitutes a trading agreement between the EU and Iceland, Norway and Liechtenstein extending all four freedoms to these countries (Nugent, 2010).
18
Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden and the UK. The EEA included, besides all EU member states, Iceland, Liechtenstein and Norway.
6.1 Data Collection and Case Selection
Measuring opinions on a certain topic is not easy in an objective scientific work as they are always subjective. The best research type therefore is a survey, in this specific case a survey that was conducted over a long period of time, in various countries with a significant high number of respondents and standardised questionnaires to make generalisations and comparisons possible. A few longitudinal surveys exist such as the World Value Survey (WVS), the European Values Survey (EVS) and the International Social Survey Programme (ISSP) providing data from the 1980s on various topics important to social science. The data are usually collected in intervals of 5-10 years (waves) on varying countries. Given the desired cases, the data base chosen for the present work is the EVS. The European Values Survey is a cross-national longitudinal research survey focusing on European thoughts and beliefs regarding family, work, religion, politics, society and life in general. The representative samples are chosen randomly of the countries’ populations conducting face-to-face interviews
with adults older than 18 years with a standardized questionnaire.7 So far, four waves of surveys
have been conducted starting in 1981 in a cycle of nine years. In every wave, more countries are included; in the last wave in 2008 47 European Countries participated (EVS, 2018). In this work, data from wave 2 (1990), wave 3 (1999) and wave 4 (2008) are used. The first wave (1981) is not included because it does not contain the targeted questions. The data are freely available for research purposes and provided by Leibniz-Institut für Sozialwissenschaften (GESIS) (GESIS, 2018).
With the EVS that provides micro data over a period of time, a change in attitudes can be observed. It provides one of the longest running surveys and similar questions are included in all waves which makes them comparable. Other surveys such as the WVS or the ISSP also provide useful data on the researched topic and are conducted in shorter intervals than the EVS. Yet, they do not include enough European Countries over a consistent time period (WVS) or conduct surveys on different topics in different years which makes it hard to compare the data (ISSP). Hence, the EVS is the better choice for the focus of this research.
7 Reports on the Methods and Variables can be accessed for all conducted waves separately on http://www.europeanvaluesstudy.eu/page/longitudinal-file-1981-2008.html. An overview of the selection method, mode of data collection, data collectors and the dates of collection is available on https://dbk.gesis.org/dbksearch/sdesc2.asp?no=4804&db=e&doi=10.4232/1.12253.
19
However, given that the survey catches the personal opinion, the data can suffer from a lack of comparability as individuals might understand the questions asked differently than intended or cultural differences lead to other understandings of concepts. Nonetheless, “the data from EVS are generally accepted and used by researches around the globe in comparative studies in different areas, though with some delayed interest by the economists.” (Mojsoska-Blazevski & Petreski 2011, 201). Given the high standards in random and representative selection, the conducted data gives a good sample to generalise the results of the case population for the relevant population, the citizens of the EU.
Regarding the case selection, most countries mentioned above are included in every of the observed waves, except for Luxembourg and Greece which did not participate in Wave 2 (1990). As this study is mainly interested in the change between the third and fourth wave, these countries are nevertheless included to obtain more observations. The group of old member states therefore consists of Austria, Belgium, Denmark, Finland, France, Germany, UK, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain and Sweden, i.e. the countries that were part of the EU prior to the EU Eastern enlargement.
Initially, the group of new member states should only focus on the countries exceeding in the Eastern enlargement in 2004. Yet, with Bulgaria and Romania, two more Eastern European Countries joined the EU in 2007. Since they fit the profile that characterises the Eastern European Countries and the fourth wave was conducted after their accession, they will be included in the analysis for the same reason as Luxembourg and Greece, for the sake of having more observations. Moreover, in academia the two accessions are often put together under one big enlargement round (cf. Nugent, 2010). Hence, the group of new member states consists of Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Romania, Slovak Republic, Slovenia, i.e. countries that became EU member with the Eastern enlargement. Unfortunately, Cyprus which also joined the European Union had to be excluded as its citizens only participated in the fourth wave which makes it impossible to analyse a change in time.
Regarding the cases of Iceland, Liechtenstein and Norway which are not part of the European Union but of the European Economic Area that enjoys the same rights of free movement, only Iceland participated in all three waves whereas Liechtenstein never participated, and Norway only did so in wave 2 and 4. Consequently, Iceland and Norway will also be included in the analysis. Switzerland which is part of the single market only participated
20
in the last wave. Croatia, the newest member of the EU will not be included in the analysis as it only joined the EU in 2013 and the last wave was conducted in 2008.
In sum, the type of research is mixed. The macro sample is small-N made up of 28 countries with maximum three times of data collection providing a total number of macro observations of 81. The micro sample is a large-N design with a total number of observation of 109,854.
6.2 Other Data
The most important is derived from the EVS, but other data is also used for the part of the analysis on the macro level. The first hypothesis regards the assumption that the opinion is conditioned on the size of the migrant inflows. In order to analyse that, data on the immigrant inflows into a country at a certain point of time as well as the total population of this country at that point in time are needed. The most comprehensive data bases on immigration inflows are provided by the IOM and the Organisation for Economic and Co-operation and Development (OECD). Both provide data that cover the researched period. Unfortunately, neither of them covers all countries of the sample at all required time points but it was not possible to find a data base that can. Comparing both, the IOM data base can provide more numbers than the OECD data base which also excludes categorically the sample countries Bulgaria, Lithuania, Malta and Romania. The data on immigrations flows are therefore taken from the Migration Data Portal of the IOM that was launched in December 2017.
As the inflow of immigration has to be examined in relation to the country’s population, also data on the sample countries’ population is required to compute percentages. This data is taken from the United Nations Department of Economic and Social Affairs (UN DESA), precisely its World Population Prospects data base, since it is also the data of reference for the IOM’S Migration Data Portal.
6.3 Methodology and Operationalisation
Researching both the public and individual opinion means that the level of observation and analysis will be the macro as well as the micro level, thus a multi-level analysis will be conducted, although the data was only collected on the micro level, i.e. the individual responses in the different countries and waves. On the macro level, the average opinion in a country will be examined as well as the inflow of migrants into this country. Here, the unit of observation and analysis are the states. On the micro level, individual opinions and characteristics are observed, so the unit of observation and analysis, here, is the individual. First, a rather general analysis seeks to examine the effect of immigration inflows, i.e. the independent variable, on the average opinion towards labour migration in EU countries, i.e. the dependent variable,
21
hereby paying attention especially on the waves immediately before and after the enlargements in 2004 and 2007 and the differences between old and new member states. Unfortunately, the data on immigration inflows does not depict the migrants’ composition of skills, yet as said before, immigration towards Europe is usually characterised by lower levels of skills and education. Following that, a more detailed regression on the micro level investigates the effect of different individual characteristics on the personal opinion on labour migration with special attention on skill levels, the main explanatory variable. In a third regression, again on the micro level, the support for the EU will be tested and how it depends on an individual’s opinion on labour migration, the main explanatory variable.
The type of regression will be an Ordinary Least Square (OLS) regression which tries to produce a regression with the smallest deviations. Yet, as both dependent variables are ordinarily coded, an ordered probit regression will also be conducted and the results be
compared. 8 Moreover, since the sample size for hypothesis 1 is rather small (81 observations)
and heteroskedasticity cannot be ruled out, robust standard errors will be estimated.9
6.3.1 Dependent Variables
Hypotheses 1 and 2 regard the opinion on labour migration on the macro and micro level. This will be tested with the question “When jobs are scarce, employers should give priority to
[NATIONALITY] people over immigrants”. Respondents could choose in their answers
between “agree” (Value 1), “neither” (Value 2) and “disagree” (Value 3). For the macro level analysis, the median of a countries’ respondents was used. Although in wave 4 more specific questions are asked regarding migration, this question item was chosen as it was asked during wave 2-4, which makes the responses comparable and allows the identification of a potential time trend. However, it does not distinguish between EU immigrants and immigrants from the rest of the world, thus the immigration inflows were also not restricted only to EU immigrants. But since no question more specific towards EU migration was asked, this question is the best possibility. For the micro analysis on the hypothesis on labour migration, the same question item will be used but then not with the country’s median but the individual response. Following the hypothesis 1(b), the assumption is that the values differ between the old and new member states with the median as well as individual opinion in the old member states taking the value
8 With a typical OLS regression, it is assumed that the dependent variable is continuously coded. In an ordered probit regression, the dependent variable can have more than two (ordered) outcomes, for instance, in the present case, “agree”, “neither” or “disagree” (Greene, 2003).
9 Usually, when working with standard errors, homoskedasticity is assumed, i.e. the variance of all residuals is unrelated to values of the independent variable. Robust standard errors allow for a softening of this assumption, so the existence of heteroskedasticity (Angrist and Pischke, 2015).
22
1 or 2 and, according to hypothesis 1(c), decreasing over time whereas for the new member states and its citizens it is expected to be higher and increasing over time.
Hypotheses 3 and 4 will be tested with the question “Please look at this card and tell me,
for each item listed, how much confidence you have in them, is it a great deal, quite a lot, not very much or none at all? - The European Union”. Respondents could choose on an ordered
scale: “1 - A great deal”, “2 - Quite a lot”, “3 - Not very much”, “4 - None at all”. According to the hypothesis 3(a), the value is expected to be low, between 1 and 2, thus showing great confidence in the EU and, following hypothesis 3(b), to increase for individuals from the old member states, i.e. depicting less support, and to remain low or decrease for individuals from the new member states.
6.3.2 Independent Variables
The main explanatory variable in the macro level analysis is the inflow of immigrants into one country. This will be measured as the percentage of incoming migrants of the host countries population in the year prior to the survey wave. The expectation is a negative impact of a higher inflow of migrants on the opinion of labour migration as natives fear for jobs and changes in welfare policies (hypothesis 1(a)). The regression controls for time- and country-specific trends, since these show the evolvement over time and the differences between countries which concerns the second part of the first hypothesis. According to it, the expectation is a positive impact on labour migration’s attitude if the country is a new member state and a reinforcing of the effect over time.
The main explanatory variable for the micro level analysis on labour migration is the individual’s skill level. A usual determinant for skill levels is the education level (cf. Facchini and Mayda, 2009). The EVS data set, unfortunately is inconsistent in their question items over different waves. The only related question item in all three examined waves regards the age of finishing education, an ordered variable reaching from value 0 (“no formal education”) to 10
(“21 and more years”). Therefore, the main regression will include this question item as the
main explanatory variable. Another measurement often used is the type of profession (cf. O’Rourke and Sinnott, 2006). The survey also asks about professions but again is not consistent over time. A solution here is to focus only on wave 3 and 4 which have similar questions items and are the more important ones. The skill level will then be tested with a recoded variable on a lower, medium or higher skill level.
23
The EVS also includes various other socio-demographic information which can be used as to rule out confounding variables. These control variables on age (ordered variable, coded in intervals), gender (dummy) and employment status (dummy) will be included in the analysis. According to Mayda (2006), men and younger people are less hostile towards labour migration, so a positive correlation will be assumed. People who have a job should be less hostile towards labour migration than unemployed individuals who have to compete with an increased number of applicants. Again, country- and time specific trends will be controlled for.
The regression on EU support includes the opinion on labour migration as a main explanatory variable, using the same question item that was the dependent variable in the micro level analysis on labour migration. Following hypothesis 4(a), opinion on labour migration and the EU are assumedly positively correlated, so with a more positive opinion on labour migration the confidence in the EU is higher. The same control variables on age, gender and employment status are used expecting the same correlations as in the analysis on labour migration. The question on the age when finishing one’s formal education is also included. It is expected that with a lower level of education, individuals have less confidence. Confidence in the EU usually depends on the individual position in the political spectrum. Problematic is the fact that both far-right and far-left are critical towards the EU. As the question on political affiliation focuses on the direction and not on the extent (value 1 being “left” and value 10 being “right”), incorporating this variable seems little conclusive. A more suitable variable is the confidence in the own parliament since individuals with more extreme views usually have less confidence in the own political system and in the EU. The variable is coded in the same way as the question on the confidence in the EU. A question on the confidence in the government may have been more suitable but was only asked in the last wave. Lastly, a variable for national identity is included as a control variable asking for the geographical groups the individuals feels it belongs to first (value 1 being “Locality or town where you live”, value 2 being “Region or country where you live”, value 3 “Country as a whole”, value 4 “Europe” and value 5 “The world as a whole”). The more narrow or local this group is defined, the less supportive of the EU, the individual is expected to be. As the differences between countries and over time in the support for the EU should be measured, the country and time effects are included again. According to the hypothesis 4(c), when being a new member state the support grows over time whereas in old member states the support declines.