Public support for free movement of European Union (EU)
citizens in the EU
An empirical study of the effect of the Brexit referendum
Lisa van Winden, s1513338
Thesis
Master of Science in Public Administration: International and European Governance
Dr. D.D. Toshkov
Table of contents
Abstract 2
1. Introduction 3
2. Literature review 5
2.1 Freedom of movement in the EU 5
2.2 EU citizens’ opinion on freedom of movement 6
2.2.1 Determinants: utilitarian perspective 8
2.2.2 Determinants: affective perspective 9
2.2.3 Levels: individual and country 10
2.3 Possible influence of the Brexit referendum on the opinion of EU citizens 11 2.4 Combining the opinion on freedom of movement and the Brexit referendum 12
3. Theory and hypotheses 12
3.1 Utilitarian theory and benchmarking theory 13
3.2 Affective theory and benchmarking theory 14
4. Research design 15
4.1 Operationalization and measurement 15
4.2 Case selection 17
4.3 Research method 17
4.3.1 Data collection 17
4.3.2 Data analysis 18
4.4 Validity and reliability 19
5. Analysis 20
5.1 Descriptive statistics 20
5.2 Inferential statistics 22
5.2.1 Support for free movement (after the Brexit referendum) 22
5.2.2 Changes in support for free movement of specific groups after the Brexit referendum
24
6. Conclusion 35
References 39
Abstract
Support for free movement among European Union (EU) citizens is consistently high. This right however receives increasingly negative attention and has been a controversial policy issue, especially during the Brexit debate. This study examines the changes in support of EU citizens for freedom of movement in the EU after the Brexit referendum. I examine these possible changes in general and for specific groups of individuals and countries, based on utilitarian and affective perspectives combined with benchmarking theory.
By using Eurobarometer data from May 2016, November 2016 and November 2017, I perform linear regressions (with interaction effects) to identify changes in support between the different time periods and groups of individuals and countries. The results suggest a small increase in general support for free movement among EU citizens after the Brexit referendum. This effect however seems to have diminished over time and has been heterogeneous. Effects that have hold over time and have indicated a unification among EU citizens concern GDP per capita and attachment to the nation state. The effect of GDP per capita on the opinion on free movement has diminished after the Brexit referendum and citizens who feel strongly attached to the nation state have experienced a greater positive influence compared to citizens who also feel European after the Brexit referendum. EU citizens living in a country of origin of EU immigrants have experienced a less negative change compared to the change of citizens who do not live in a country of origin or destination after the Brexit referendum. While EU citizens living in a country of destination of EU immigrants also have experienced a less negative (or perhaps more positive) change compared to the change of their reference group, this only has happened after a longer period of time. In contrast, EU citizens with lower human capital have become even more negative towards free movement compared to citizens with higher human capital after the Brexit referendum than before, and this seems to have hold over time. Therefore, on the one hand the Brexit referendum might have led to more unification, but on the other hand it might have also increased polarization of opinions between individuals with different levels of human capital.
1. Introduction
Freedom of movement of persons is one of the pillars of the European Union (EU) and is part of the four economic freedoms of goods, services, labor and capital. It was established with the Treaty of Maastricht in 1992 and entails the right for all EU citizens to move and reside (live, work and study) freely throughout the EU (Article 45, TFEU).
This right is a cornerstone of the EU and is associated with consistently high public support, but it has been receiving increasingly negative attention from member states, especially since the last decade. Some of the member states frame intra-EU migration as problematic. For instance, in the built-up to the Brexit referendum, freedom of movement became a controversial policy issue and important part of the debate (Becker & Fetzer, 2016; Goodwin & Milazzo, 2017; Roos, 2019). Another example is Italy, which is experiencing a “brain drain”, while high-educated citizens leave the country to work in more promising EU countries (Kaplan, 2019). Also in the Netherlands, politics are recently arguing against freedom of movement, because it is perceived to create uncertainty (Brandsma, 2020).
Since in recent times the right to free movement receives increasingly negative attention, this study examines public opinion on freedom of movement of EU citizens by focusing on recent changes and attempts to explain these changes. The focus is on changes after the Brexit referendum. This event is chosen because freedom of movement has played an important role in the corresponding debate (Becker & Fetzer, 2016; Goodwin & Milazzo, 2017; Roos, 2019), and a member state actually leaving the EU is a first instance. Hence, it could either lead to a unifying or disuniting effect for the EU member states (De Vries, 2017; 2018). Therefore, the research question of this study is to what extent the Brexit referendum has affected the opinion of EU citizens on freedom of movement within the EU.
Freedom of movement is of interest to the academic literature on the EU, also to the strand studying public opinion (Vasilopoulou & Talving, 2019; Blinder & Markaki 2019; Lutz, 2020). This study adds to the literature, as the attitudes of EU citizens towards freedom of movement have not been studied in combination with an event, such as the Brexit referendum, before. There have been studies focused on factors shaping public attitudes towards free movement, often based on utilitarian and affective perspectives (e.g. Vasilopoulou & Talving, 2019), but not in combination with the possible influence of an event. The literature does find, based on benchmarking theory, that support for EU membership among EU citizens was higher immediately after the Brexit vote (De Vries, 2017). Attitudes towards free movement however differ from the more general EU opinion (Vasilopoulou & Talving, 2019). It is therefore interesting to focus on possible changes in the opinion on freedom of movement by the Brexit referendum. In this study, these strands in the literature and theory are therefore combined by giving insight into the possible effect of the Brexit referendum on the level of support of EU citizens for freedom of movement.
This is also of interest for politicians and policy makers, while it gives further insights about the possible effects of the Brexit referendum on the opinion of EU citizens on EU integration. In this study, this is specifically focused on one of the cornerstones of the EU, which has played an important role in
the Brexit vote and is in current times under negative attention; freedom of movement. Insights from this study can therefore enhance the understanding and knowledge of politicians and policy makers regarding EU citizens and their opinion on freedom of movement in reaction to an event as the Brexit referendum and can help them shape policy in the future. This is particularly interesting, while different EU member states are expressing negative stances towards this right to freedom of movement.
In order to study this research question, I use Eurobarometer data representative for all EU citizens from the 28 member states.1 The surveys from May 2016 (85.2), November 2016 (86.2) and November 2017 (88.3) are included in the analysis to capture the possible effect of the Brexit vote, which took place on 23 June 2016 (Gov.uk, 2016). I supplement these data with data on GDP per capita extracted from the World Bank (2020).
By performing a multivariate linear regression including interaction effects between the time periods and the independent variables of interest, changes in support for free movement between time periods and groups of individuals and countries can shine through. By using the conditioning strategy, I control for confounders. In addition, I examine possible mediating variables. I also conduct and report robustness checks that consist of (a) linear regressions including unclear (‘don’t know’) answers and (b) logistic regressions. These are included in the appendix.
The findings suggest that after the Brexit referendum, there has been an increase in support for free movement among EU citizens in general. This increase in support however seems to have diminished over time and is no longer observed in November 2017 compared to May 2016. In addition, this effect has been heterogeneous. After the Brexit referendum, the relationship between GDP per capita and the opinion on free movement has become less negative (i.e. has diminished) and this seems to have hold over time. A greater positive influence on the effect on the opinion on free movement of EU citizens who strongly attach to the nation state compared to citizens who also feel European after the Brexit referendum is also found to have hold over time. EU citizens who live in a country of origin2 have experienced a less negative change in the opinion on free movement than citizens who do not live in a country of origin or destination. EU citizens living in a country of destination3 also have experienced a less negative (or perhaps more positive) change in their effect on the opinion on free movement compared to the reference group and this change has been larger than the change in the effect of citizens living in a country of origin, but this is only found over a longer time period in November 2017. So, this cannot be attributed to the Brexit referendum with certainty. In contrast, EU citizens with lower human capital have experienced an even more negative change in their opinion on free movement compared to the negative change of citizens with higher human capital after the Brexit referendum, and there is some evidence that this has hold over time.
1 This includes the United Kingdom (UK), as in 2016-2017 the UK was still part of the EU. 2 A country of origin is a country which has large outflows of intra-EU migrants.
The Brexit referendum therefore has had two contrasting effects. On the one hand, it has led to a smaller difference in and therefore unification of the opinions on free movement of EU citizens concerning GDP per capita, and between EU citizens feeling strongly attached to the nation state and citizens also feeling European. On the other hand, it has led to an increased cleavage in and therefore polarization of support for free movement between citizens with different levels of human capital.
The found effects and changes should however be interpreted with caution. Due to the nature of the used data, it is not certain that the found effects are caused by the Brexit referendum. The data are pooled time series and are not conducted right before and after the event of the Brexit vote, so differences between samples or other events might have contributed to found changes in the opinion on free movement.
In section 2, I review and reflect upon the existing literature. This follows by section 3 on theory, from which I derive hypotheses. Hereafter in section 4, I explain and justify the research design. I present and analyze the results in section 5. In section 6, I discuss and conclude upon the main insights and theoretical and practical implications of this study.
2. Literature review
In this section, I review the literatures on both (the determinants of) the opinion on freedom of movement of EU citizens and the potential influence of the Brexit referendum by describing and reflecting upon the main thoughts and findings.
2.1 Freedom of movement in the EU
Bauböck (2018) describes freedom of movement as the core of EU citizenship. This is in line with the establishment of this right in the Treaty of Maastricht, with which also EU citizenship was established (Vasilopoulou & Talving, 2019). The right to free movement entails that EU citizens can move and reside (live, work and study) freely throughout the EU (Article 45, TFEU). A legal basis also ensures that EU citizens residing in another EU member state are treated equally to nationals in access to employment and housing, working conditions, and social and tax benefits (Kraatz, 2020).
The use of freedom of movement in the EU started out with low levels, but intra-EU migration increased over time (European Commission (EC), 2016a; 2018a; Vasilopoulou & Talving, 2019; Schmidt, Blauberger & Sindbjerg Martinsen 2018), especially after the accession of Central and Eastern Europe to the EU (Ruhs, 2017). EU member states that did not place restrictions on migration from these new member states (the UK, Ireland and Sweden) were most popular among these new EU migrants (Van Mol & De Valk, 2016). The economic and financial crisis increased intra-EU mobility, but also enhanced return migration (Van Mol & De Valk, 2016).
Freedom of movement should foster economic development, by matching supply and demand of workers (Fevre, 1998). Intra-EU migration however mainly concerns migration from poorer to richer
member states (Vasilopoulou & Talving, 2019; EC, 2018a), and therefore seems to promote economic self-interest and utility gains. An example is the labor migration from Central and Eastern European member states towards Western European member states (Favell, 2008; Ruhs, 2017). Also, the immigration from high-skilled Italians to more promising EU countries is an example, because unemployment rates in the own country among young individuals are high (Tintori & Romei, 2017).
Next to the expected economic gains from the right to free movement, there is also a more value-driven perspective to this right (Hooghe & Marks, 2018). This is derived from the common EU citizenship and feeling of identity. The value of free movement is found in the emancipation of the individual from the member state (De Witte, 2018). Vasilopoulou and Talving (2019) argue that the shared experiences through freedom of movement have contributed to the development of a common European identity and values. It is therefore important to make the difference between intra-EU and non-EU migration.
2.2 EU citizens’ opinion on freedom of movement
The opinion of EU citizens has been perceived as irrelevant for a long time, while the opinion of elites was seen as sufficient for further European integration (Haas, 1958; Hobolt & De Vries, 2016). Public opinion however has become increasingly important over time, introducing an interest in the opinion of EU citizens in the academic literature. Hooghe and Marks (2009) describe this trend towards public opinion becoming more included and decisive in the EU as a shift from “permissive consensus” to “constraining dissensus”.
The greatest part of the literature on EU public opinion focuses on public attitudes towards EU integration (e.g. McLaren, 2006; Hobolt, 2014; Hobolt & De Vries, 2016; De Vries 2018). Freedom of movement is, as core of EU citizenship, one of the pillars of EU integration. Some studies on the opinion on EU integration therefore make the combination with freedom of movement. For example, Toshkov and Kortenska (2015) describe how freedom of movement affects the public opinion on further EU integration by examining a link between immigration from Central and Eastern Europe after the enlargement to other EU member states and the support for EU integration. A negative relation is found on the regional level. In this way, they use freedom of movement as possible effect for changing public opinion towards EU integration. Furthermore, Toshkov (2019 – work in progress) finds a positive relation between the share of migrants from Central and Eastern Europe in localities and the vote for Eurosceptic parties in these localities in the Netherlands, Sweden and Finland. He therefore also uses the right to free movement to describe formulation of public attitudes and political views towards EU integration.
There is however also a strand of literature that focuses specifically on the public opinion on freedom of movement as an outcome. These studies mainly focus on its determinants. They find that in general EU citizens are quite positive towards free movement, but that this differs for different types of citizens and in different countries (Vasilopoulou & Talving, 2019; Blinder & Markaki, 2019; Lutz,
2020). This corresponds to the idea that public opinion is not consistent and unidimensional, but varies and is multidimensional (Hobolt & De Vries, 2016).
Studies focusing solely on attitudes towards freedom of movement, draw upon a large number of studies that focus on attitudes towards EU integration (e.g. McLaren, 2006; Hobolt, 2014; Hobolt & De Vries, 2016; Toshkov & Kortenska, 2015) and migration in general (e.g. Scheve & Slaughter, 2001; Hainmueller & Hiscox, 2007; Hatton, 2016). Since freedom of movement as a right is part of EU integration and also concerns migration, namely intra-EU migration, a combination of these two strands of literature form the base of the literature focusing on the opinion on freedom of movement as outcome.
In the literature focusing on the opinion on freedom of movement of EU citizens, cleavages are often studied. Bauböck (2018) identifies a cleavage between mobile and immobile Europeans and links this to freedom of movement. He does this by ‘classic’ divisions of old versus young, low- versus high-educated etcetera, but does not provide further empirical evidence.
Vasilopoulou and Talving (2019) do empirically study the determinants of and cleavages in the opinion on free movement of EU citizens. They find in a recent study based on Eurobarometer data from November 2015 to May 2017 that employed and higher-educated individuals are more likely to support freedom of movement, and that individuals with a strong national identity are less supportive of this right. They however find small explanatory power for these individual-level factors and argue that country affluence is of larger importance for shaping public opinion on freedom of movement. They find that citizens in richer member states are more negative towards freedom of movement than citizens in poorer member states (Vasilopoulou & Talving, 2019). In addition, they also find through interaction terms that citizens with low human capital in poorer member states are more supportive than these citizens in richer member states. They therefore argue that individual-level factors have more explanatory power in more affluent countries (Vasilopoulou & Talving, 2019).
The literature on the opinion on free movement also studies ambivalences. Lutz (2020) studies the ambivalence between the opinion on freedom of movement as a right for the person itself and the opinion as the right to it for others and thereby adds interesting insights to the literature. Based on Eurobarometer data from May 2016, he finds a trade-off between the own mobility rights and that of others; while people value their own mobility, they also fear immigration. He finds that determinants as country-level inflow of migrants and individual-level feeling of identity shape these in- and outward perspectives. Consistent with Vasilopoulou and Talving (2019) he finds that the negative effect of a strong national identity on the opinion on free movement is more salient in Western European member states. He however does not link this to country affluence, but to the large inflow of migrants.
Whereas the above-mentioned studies solely focus on intra-EU migration, there are also studies combining intra-EU migration with non-EU migration. Meltzer et al. (2018) find that there is a relation between the opinion on migration in general and freedom of movement. EU citizens who negatively view immigration are also more negative about free movement and vice versa. Blinder and Markaki (2019) go further and examine differences and ambivalences between the opinion on intra-EU migration
and immigration from outside the EU based on the European Social Survey. They find a small 10 percent of respondents who prefer intra-EU migration above immigration from outside the EU. This is mainly explained by a common (supranational) feeling of identity and not by economic arguments or cognitive mobilization (Blinder & Markaki, 2019). Cognitive mobilization corresponds to political engagement and resembles the idea that citizens who are cognitively engaged in politics take cues from elites in their political views and opinions (Zaller, 1992; Blinder & Markaki, 2019). Blinder and Markaki (2019) also find that the supranational EU identity leads to more support for free movement, as found by Vasilopoulou and Talving (2019) and Lutz (2020). McLaren (2001) also studies a possible difference between attitudes of EU citizens towards non-EU and EU migrants. She however finds that the vast majority of EU citizens views these two identically.
The literature on attitudes towards freedom of movement often mentions a few determinants as having explanatory power. These are human capital, country affluence, share/inflow of migrants and feeling of identity (Vasilopoulou & Talving, 2019; Blinder & Markaki, 2019; Lutz, 2020). So, there seem to be two main perspectives underlying the determinants of the opinion on freedom of movement: economic utility (self-interest) and identity. The literature indeed distinguishes between different perspectives contributing to attitudes towards EU integration and also freedom of movement in particular. The most common, which are also distinguished here, are the utilitarian perspective and the affective perspective, corresponding to respectively utility and identity arguments.
2.2.1 Determinants: utilitarian perspective
The utilitarian perspective focuses on economic utility and self-interest by looking at factors such as skills, education and affluence. It is a rational model, derived from rational choice, and uses cost-benefit analysis (McLaren, 2006). It stems from the idea that individuals with high human capital and income are expected to be more supportive of EU integration, because this favors them (Anderson & Reichert, 1996). So, if the benefits of membership or a specific policy are higher than the costs, this leads to support and vice versa.
The socioeconomic status of citizens can be combined with a different experience in costs and benefits of EU integration (Hooghe & Marks, 2018). Similarly, immigrants can be perceived as an economic threat by some more than others (Blinder & Markaki, 2019). The labor market competition hypothesis in the immigration literature entails the consideration of individuals about the impact of immigrants on wages for nationals with similar skills (Scheve & Slaughter, 2001; Mayda, 2006; Vasilopoulou & Talving, 2019). Both Scheve and Slaughter (2001) and Mayda (2006) find that skill, measured in years of education, significantly contributes to the opinion towards migrants. So, low-skilled individuals are less positive towards immigrants than high-low-skilled individuals. They find this for respectively the United States and cross-country. Lancee and Sarrasin (2015) however question this consistently found effect of education. Based on an analysis of the Swiss Household Panel, they argue that differences between educational levels are due to selection effects.
Concerning the opinion on free movement, Vasilopoulou and Talving (2019) and Blinder and Markaki (2019) follow the labor market competition hypothesis and hypothesize that low-educated or low-skilled workers, individuals with low human capital, have a more negative attitude towards free movement. This because intra-EU migration mainly concerns low-skilled jobs and could therefore negatively affect their wages (Blinder & Markaki, 2019). Vasilopoulou and Talving (2019) do indeed find a significant effect for individuals with low human capital being less positive to freedom of movement. As mentioned before, they find this effect to be more salient in more affluent member states. Vasilopoulou and Talving (2019) therefore argue that the utilitarian perspective has more explanatory power in affluent member states than in less affluent. Meltzer et al. (2018) also find a positive relationship between education and the opinion on free movement based on an analysis in seven EU member states. Blinder and Markaki (2019) do not find evidence for differences in skills in the preference for EU migrants over non-EU migrants. They therefore argue that skill-levels are not an important contributor to a difference in opinions on intra-EU versus non-EU migration.
Next to human capital, affluence is also an important factor for the utilitarian perspective. Vasilopoulou and Talving (2019) find that EU citizens living in richer (i.e. more affluent) member states are more negative towards free movement. This is against the expectation and findings of the literature on EU integration, in which citizens living in a richer country are more positive towards further integration (Anderson & Reichert, 1996; Hobolt, 2014). This difference occurs because affluent countries attract more migrants, and thereby increase the salience of migrants and the probability of a more negative attitude (Vasilopoulou & Talving, 2019; Lutz, 2020; Toshkov & Kortenska, 2015). Boeri (2010) argues that this negative attitude against migrants is higher in richer countries, because of perceived fiscal burdens for the welfare state (Hatton, 2016). Schmidt et al. (2018) indeed also find that the institutional differences among EU member states imply that the right to freedom of movement has fine-grained implications which are not always foreseen.
The utilitarian perspective is an important contributor to the opinion on free movement, according to most of the studies. Human capital, country affluence and flows of migrants seem to shape the opinion on free movement of EU citizens (Vasilopoulou & Talving, 2019; Lutz, 2020).
2.2.2 Determinants: affective perspective
The feeling of identity forms the affective perspective. Individuals are hesitant against immigrants, because they worry that resources, as jobs, are given away to them, but also that they threaten their culture (McLaren, 2006; Hainmueller & Hopkins, 2014). This corresponds to sociotropic concerns as in- and out-group perspectives, group conflict and perceived threats (Hainmueller & Hopkins, 2014; McLaren, 2006; Quillian, 1995). A stronger feeling of national identity is often associated with lower levels of support for the EU and its integration (McLaren, 2006; Carey, 2002; Hobolt, 2014), and also a less positive view towards immigrants (Quillian, 1995).
This association of feeling of identity is confirmed for intra-EU migration in the literature. Vasilopoulou and Talving (2019) find that the feeling of attachment to the nation state is an important contributor to a more negative view towards free movement. In addition, they find that this feeling of attachment is more important for citizens living in affluent member states than in less affluent member states. This because intra-EU migration is more salient in these countries (Vasilopoulou & Talving, 2019). Lutz (2020) describes the importance of identity towards in- and outward perspectives and perceived threats, consistently with the ideas of McLaren (2006) and Quillian (1995). He describes that this in-group perspective is more noticeable in countries with high shares of EU migrants and negatively influences the opinion on free movement (Lutz, 2020). He however does not link this to country affluence as Vasilopoulou and Talving (2019) do.
Blinder and Markaki (2019) also find that identification with the supranational EU identity leads to a more positive opinion on free movement. Interestingly, Blinder and Markaki (2019) find that when comparing opinions on intra-EU migration to migration from outside the EU, the supranational EU identity also contributes to a preference of intra-EU migration above immigration from outside the EU. This is nonetheless found for a very small number of EU citizens. McLaren (2001) does not find such a difference in an earlier study.
It seems that the affective approach of identity is important, because most studies find a significant negative effect of a strong national identity on the opinion on freedom of movement (Vasilopoulou & Talving, 2019; Blinder & Markaki, 2019; Lutz, 2020). The feeling of identity is associated with in- and outgroup effects and group conflict, in line with the literature concerning the opinion on migrants (Hainmueller & Hopkins, 2014; Lutz, 2020).
2.2.3 Levels: individual and country
The literature on the opinion on freedom of movement also demonstrates that the determinants or examined factors are both on individual- and on country-level. Individual-level factors are among others human capital and feeling of identity, whereas country-level factors are among others affluence and inflow of migrants.
Vasilopoulou and Talving (2019) indicate that the individual-level factors offer little explanatory power and that most of the variance can be explained by country-level factors – mainly country affluence. They find that utilitarian and affective perspectives both have more explanatory power in affluent member states and that citizens in less affluent member states are more positive towards freedom of movement, regardless individual characteristics (Vasilopoulou & Talving, 2019).
Lutz (2020) discusses both individual- and country-level factors as possible explanation of the found discrepancy between the opinion on own mobility rights and that of others, and freedom of movement in general. He finds evidence for both, but the effect of the individual-level feeling of identity is more salient in countries which receive higher shares of intra-EU migrants. Consistently, Blinder and Markaki (2019) find some evidence for ambivalence between the opinion on freedom of movement for
EU citizens and immigration from outside the EU and the opinion on free movement in general on the individual-level explained by the feeling of supranational identity.
Country-level factors seem important, because these shape the environment or perspective in which an individual formulates his or her opinion towards free movement. This might be related to country affluence (Vasilopoulou and Talving, 2019), but can also include other factors as the share and in- and outflow of migrants (Lutz, 2020). Individual-level factors are however also important, while these have explanatory power and can explain differences within countries (Vasilopoulou & Talving, 2019; Lutz, 2020; Blinder & Markaki, 2019).
2.3 Possible influence of the Brexit referendum on the opinion of EU citizens
Crises and events often form critical junctures for change (Hooghe & Marks, 2018). It is also found that these have an influence on the opinion of EU citizens towards EU integration, and can thereby lead to different opinions within the EU. For example, after the Eurozone crisis, support for monetary integration remained high within the euro area, but dropped in the other EU member states and therefore led to divisions within the EU (Hobolt & Wratil, 2015).
As the Brexit referendum is a unique event, because it is a first instance that a member state has decided to leave the EU, a lot of studies have focused on it. Most studies however examine reasons why the Brexit vote occurred (e.g. Hobolt, 2016; Vasilopoulou, 2016; Clarke, Goodwin & Whiteley, 2017). Studies about consequences of the Brexit referendum mainly contain economic arguments and do not explore possible influences on the public opinion on the EU. There are however some exceptions, such as Taggart and Szczerbiak (2018) who use the Brexit vote, among other events, to describe the advance of Euroscepticism, and De Vries (2017) who examines public support for the EU after the Brexit vote. There are different suggestions on what the influence of the Brexit vote on the future of the EU will be. Some argue this can be a unifying moment for the EU, others that it will drive the EU further apart (De Vries, 2017). De Vries (2017) studies these speculations and examines the possible influence of the Brexit vote on the opinion of EU citizens on EU membership by conducting two surveys; one before the Brexit vote and one right after. She finds that support for the EU from EU citizens was higher directly after the Brexit vote, especially for those who think consequences of the Brexit for the UK will be bad.
De Vries (2017; 2018) uses benchmarking theory of public opinion towards the EU in studying the effect of the Brexit vote on public support for EU membership. This theory holds that support for the EU is formed by a comparison. Namely, a comparison between the benefits of the status quo (e.g. EU membership) versus those of an alternative state (e.g. leaving the EU). The Brexit referendum gives insight into the alternative state of leaving the EU (De Vries, 2017). The finding of De Vries (2017) that support for EU membership is higher after the Brexit referendum confirms the idea that political and economic uncertainty and costs make this alternative state less favorable. Benchmarking theory is derived from management studies and is built upon comparisons of performance (Watson, 1993). It is
also used in theory regarding public opinion towards the EU and its integration by the use of national proxies, as economy or institutions (Hobolt & De Vries, 2014).
Since the Brexit vote, EU citizens have acquired insights into the alternative state of withdrawing from EU membership. The UK hereby has become the example of leaving the EU and the corresponding consequences. It therefore has provided EU citizens with insightful information on costs and benefits of this alternative state. When consequences are perceived to be bad, costs are high and this shapes opinion. Hobolt and Wratil (2015) find that utilitarian theory has been most important for shaping public opinion on the EU after the Eurozone crisis. One might therefore ask whether in times of crises or disruption, theories regarding costs and benefits gain influence.
2.4 Combining the opinion on freedom of movement and the Brexit referendum
Immigration is perceived as an important driver for the Brexit vote. Goodwin and Milazzo (2017) confirm that concerns over migration and perceived consequences explain a large part of the Brexit vote, based on individual- and aggregate-level data from the British Election Study. Increasing migration also led to switches from Remain to Leave (Goodwin & Milazzo, 2017). They however do not make a distinction between intra-EU and non-EU migration
Roos (2019) describes that through framing and opportunity, freedom of movement has become a salient issue in the UK as opposed to Germany, in which this policy issue has been depoliticized. In this way, free movement has become an important part of the Brexit debate. Heinkelmann-Wild, Kriegmair, Rittberger and Zangl (2020) indeed argue that getting rid of freedom of movement has been used as “red line” by some Conservatives in politics leading up to the Brexit withdrawal agreement.
Studies that empirically combine the opinion on freedom of movement within the EU and the Brexit vote, often do this by examining the link between the (negative) perception and/or framing of this right and intra-EU migration in Britain and the vote to leave the EU. Becker and Fetzer (2017) for example show that migration from the Central and Eastern European member states has contributed to anti-European sentiment and the vote for independence parties, because it has lowered wages from low-skilled workers and has increased pressure on public services as schooling, housing and health care.
This shows that freedom of movement is in the literature actively associated with the Brexit vote as a perceived driver. The Brexit referendum however also might have influenced the opinion of EU citizens on freedom of movement. This because the alternative state of leaving the EU has become visible by the Brexit vote and therefore has given insightful information about corresponding consequences (uncertainties, costs and benefits) (De Vries, 2017; 2018). This might have led to changes in the opinion on freedom of movement of EU citizens.
3. Theory and hypotheses
In the previous section, I already mentioned some theories from the literature on both public opinion and the possible influence of the Brexit referendum. These are utilitarian theory, affective theory and
benchmarking theory. In this section, I derive the hypotheses which are examined in this study from these theories.
3.1 Utilitarian theory and benchmarking theory
Utilitarian theory concerns rationality and economic self-interest (McLaren, 2006). Perceived costs and benefits shape opinions. Rationality and cost-benefit analysis therefore lead to either support for integration or a certain policy for “winners”, or opposition to it for “losers” (Anderson & Reichert, 1996; Vasilopoulou & Talving, 2019; Hobolt, 2014).
Benchmarking theory also involves costs and benefits. It entails a comparison between the costs and benefits of the status quo and the alternative state (De Vries, 2017; 2018). It therefore looks to the favorability of the alternative state relative to the status quo. Since the Brexit referendum has given form to the alternative state of leaving the EU, this has informed EU citizens about corresponding costs and benefits of no longer being able to move and reside freely. Because the Brexit vote has showed a lot of uncertainty, negotiations, obstacles and costs (also think about businesses leaving the UK), it is argued that this alternative state does not look favorable. Therefore, I propose the following hypothesis:
H1: The Brexit referendum has led to a more positive view of EU citizens on freedom of movement. While the attitudes towards free movement differ for groups of citizens and countries (Vasilopoulou & Talving, 2019; Lutz, 2020; Blinder & Markaki, 2019; Meltzer et al., 2018), it is argued that the Brexit referendum has led to unification of the opinions on free movement among EU citizens. This means that the Brexit referendum is expected to have had a greater positive influence on the opinion of citizens who are more negative towards free movement compared to the citizens who are more positive.
Citizens living in a country of destination probably have a more negative view on freedom of movement than citizens in countries of origin before the Brexit referendum, because these countries have larger inflows of intra-EU migrants (Lutz, 2020; Vasilopoulou & Talving, 2019). The negative consequences of the Brexit referendum however might have given insights in the economic benefits of free movement and/or the costs of leaving this right (De Vries, 2017). This might have contributed to a more positive increase in the opinion on freedom of movement for EU citizens who live in a country of destination compared to citizens who live in a country of origin. This because of both the perceived costs of leaving this right and the larger potential for an increase in support, while EU citizens living in a country of origin already show higher levels of support. This leads to the following hypothesis:
H2: The Brexit referendum has had a greater positive influence on the opinion on freedom of movement for EU citizens living in a country of destination than for EU citizens living in a country of
Utilitarian theory is based on economic (self-)interest and therefore often includes affluence. Vasilopoulou and Talving (2019) find that country affluence contributes to a more negative stance towards freedom of movement. It is argued that the Brexit vote has given insights into the alternative state of leaving the EU and thereby the right to free movement, and the corresponding uncertainties and costs (De Vries, 2017). This might have contributed to a larger positive increase on the opinion on freedom of movement for EU citizens living in a member state with a higher GDP per capita, while these have most potential for an increase. The following hypothesis reflects this thought:
H3: The Brexit referendum has had a greater positive influence on the opinion on freedom of movement for EU citizens living in a country with a higher GDP per capita.
The labor market competition hypothesis also relates to utilitarian theory. Citizens with low human capital are expected to be more negative towards freedom of movement, because intra-EU labor migration mainly consists of low-skilled workers and this might lead to downwards pressure on their wages (Vasilopoulou & Talving, 2019; Blinder & Markaki, 2019). The Brexit referendum however might have shown the benefits of free movement, also to EU citizens with low human capital. These insights in the alternative state and the larger potential for an increase in support compared to their counter group might have led to a larger increase in support for free movement of citizens with lower human capital compared to citizens with higher human capital. I therefore propose the following hypothesis:
H4: The Brexit referendum has had a greater positive influence on the opinion on freedom of movement for EU citizens with lower human capital than for EU citizens with higher human capital.
3.2 Affective theory and benchmarking theory
Affective theory relates to the feeling of identity. EU citizens who feel a stronger national bond, tend to see the EU as a threat and are more negative towards free movement (Vasilopoulou & Talving, 2019; Lutz, 2020; Blinder & Markaki, 2019). This can also be combined with benchmarking theory. The Brexit referendum might have shown the benefits of free movement, and while individuals with a stronger feeling of national identity are more negative towards free movement and therefore have more potential for an increase in support compared to their counter group, they might have become more positive towards this right compared to citizens who also feel European. The following hypothesis indicates this line of thinking:
H5: The Brexit referendum has had a greater positive influence on the opinion on freedom of movement for EU citizens who strongly attach to the nation state than for EU citizens who also feel
It is interesting to examine the addition of affective theory to benchmarking theory, because benchmarking theory focuses on costs and benefits and the affective perspective concerns identity.
4. Research design
In this section, I explain and discuss the research design of this study. First, I operationalize the used variables and explain their measurement. Hereafter, I justify the case selection. In the research method, I explain the methods of data collection and data analysis. Finally, I reflect upon the validity and reliability of this study.
4.1 Operationalization and measurement
I operationalize and measure the concepts by using the Eurobarometer4 and World Bank as data sources. The Eurobarometer is a public opinion survey, carried out in all member states of the EU on behalf of the EC (GESIS, 2019a). I supplement these data with data on GDP per capita of the World Bank (2020). I further explain the collection of these data in section 4.3.1.
The dependent variable is the opinion of EU citizens on freedom of movement. Freedom of movement is the right for all EU citizens to move and reside (live, work and study) freely throughout the EU (Article 45, TFEU). The Eurobarometer measures the opinion of EU citizens towards free movement with the following question: “What is your opinion on the free movement of EU citizens who can live, work, study and do business anywhere in the EU?” (European Communities, 2016a; 2016b; 2017). Possible answers are: for, against or don’t know. In this study, for takes a value of 1 and against a value of 0. This variable is thus a binary variable. I omit the unclear (‘don’t know’) observations from the analysis.5
I use the Brexit referendum as event, which could have affected a change in the dependent variable: the opinion of EU citizens on freedom of movement. The Brexit referendum took place on 23 June 2016. On this date, the UK citizens voted in favor of leaving the EU (Gov.uk, 2016). To capture this event, I use the Eurobarometer of May 2016 (85.2, data collected from 21 to 31 May 2016; EC, 2016b, 2016c) and November 2016 (86.2, data collected from 3 to 16 November 2016; EC, 2016d, 2016e), as the Brexit referendum occurs in between these two surveys. I also take into account the Eurobarometer of November 2017 (88.3, data collected from 5 to 19 November 2017; EC, 2017a, 2017b) to be able to say more about the possible effect over time.
Countries of origin and destination are countries which have respectively large outflows and inflows of intra-EU migrants. I define countries of origin and destination according to the Intra-EU Labour Mobility Reports of the EC (2016a; 2018a) and data on in- and outflow of EU migrants from
4 All exact questions and answer options from the Eurobarometer surveys for the variables that I use in this study are included in table A1 in the appendix.
5 Results including unclear (‘don’t know’) answers as ‘against’ are included in the appendix, because it can be argued that respondents answering ‘don’t know’ are hesitant to admit that they are against freedom of movement.
Eurostat (2020a; 2020b). Both are included as dummy variables. The countries of origin are Bulgaria, Hungary, Latvia, Lithuania, Poland, Romania and Slovenia. These countries have a value of 1, all others of 0. The countries of destination are Austria, Belgium, France, Germany, Ireland, Italy, the Netherlands, Spain, Sweden and the UK. These countries have a value of 1, all others of 0.
GDP is a measure for economic activity and a country’s macroeconomic performance. A higher level of GDP resembles a wealthier (i.e. more affluent) nation state (Vasilopoulou & Talving, 2019). In this study, I use GDP per capita with purchasing power parity constant in 2011 international dollars. In this way, GDP can be compared between countries. These data are gathered from the World Bank (2020). I merge data of 2016 with the May and November 2016 waves of the Eurobarometer and data of 2017 with the November 2017 wave. I divide the GDP per capita by 1,000 to enhance the interpretation of the results.
Human capital concerns cognitive ability and skills (Vasilopoulou & Talving, 2019; Blinder & Markaki, 2019). Therefore, I measure human capital by both educational level and current occupation. I recode educational levels in the same way as Vasilopoulou and Talving (2019). Answers to the Eurobarometer question how old the respondent was when stopping fulltime education are recoded as the following: up to 15 years into low education,6 16 to 20 years into medium education and 21 years and above into high education. I place respondents who are still studying into the category corresponding to their current age. For example, I code a respondent who is still studying and is 18 years old as medium educated. Refusals and unclear answers are omitted from the analysis. The Eurobarometer recodes occupation in the following categories: self-employed, managers, other white collars, manual workers, house persons, unemployed, retired and students (European Communities, 2016a; 2016b; 2017). I regard self-employed, managers and students as high human capital.
The feeling of identity concerns the attachment to the nation state (Vasilopoulou & Talving, 2019; Lutz, 2020). The Eurobarometer measures this in four different categories: identification with the own nationality only, the own nationality and European, European and the own nationality or European only (European Communities, 2016a; 2016b; 2017). For the purpose of this study, I recode these categories into a dummy variable. This variable captures exclusive attachment to the own nationality (a value of 1) or also to the EU (a value of 0). Refusals, unclear and attachment to ‘none’ answers are omitted from the analysis.7
The regressions also include control and possible mediating variables. Demographic controls control for gender (man or woman), age and the area the respondent lives in (rural, small/middle sized town or large town), in addition I control for political views (left to right, on a scale from 1 to 10).8 A separate model also includes the opinion on immigrants from both in- and outside the EU (very positive
6 This category includes respondents who never had full-time education.
7 Results including unclear (‘don’t know’) and attachment to ‘none’ answers in the category ‘nationality only’ are included in the appendix.
8 Refusals and unclear (‘don’t know) answers are omitted from the analysis. Results including unclear (‘don’t know’) answers in the far-right category are included in the appendix.
to very negative, on a scale from 1 to 4),9 and the image of the EU (very positive to very negative, on a scale from 1 to 5),10 because these can be mediating variables.
4.2 Case selection
The analysis consists of EU citizens living in the 28 member states consistent with the years 2016 and 2017, who are questioned by the Eurobarometer. Since the UK was still a member state during this time, I also include citizens from the UK in the analysis. Observations of candidate countries are omitted. The Eurobarometer consists of a random probability sample (EC, 2016b), which contributes to the generalization of the results to the whole population of interest: EU citizens.
I omit EU citizens living in a different EU country than their official nationality from the analysis.11 This because these citizens use the right to freedom of movement, so they are probably very positive about it (as found by Kuhn, 2015) and can therefore skew the results. I omit these observations by omitting respondents with a different nationality than the country in which the survey is conducted.
I use the Eurobarometer waves of May 2016 and November 2016, because the Brexit referendum occurs in between these two surveys. I also take into account the survey in November 2017 to look at a possible effect of the Brexit referendum over a longer period of time. In this way, I examine the possible effect of the Brexit referendum on the opinion on free movement.
4.3 Research method 4.3.1 Data collection
I extract Eurobarometer data from the GESIS data archive (EC 2016c, 2016e, 2017b; GESIS, 2019b). TNS opinion and social collects the Eurobarometer data. The Eurobarometer data I use in this study are collected from 21 to 31 May 2016 (EC, 2016b), 3 to 16 November 2016 (EC, 2016d) and 5 to 19 November 2017 (EC, 2017a). The data are collected in all 28 member states, and also five candidate countries (EC, 2016b). For this study, I only use the data from the 28 member states. The drawn sample for the Eurobarometer is a random probability sample with a total coverage of all countries (EC, 2016b). In all countries around 1,000 interviews are conducted, with the exception of Cyprus, Luxembourg and Malta in which 500 interviews are conducted (EC, 2016b). A new independent sample is drawn from the population for every wave of the Eurobarometer (GESIS, 2019a).
I extract data on GDP per capita from the World Bank (2020). The World Bank and the International Comparison Program derive these GDP per capita (constant in 2011 international dollars) by adding “the sum of gross value added by all resident producers in the economy plus any product
9 Results including unclear (‘don’t know’) answers in the category ‘very negative’ are included in the appendix. 10 Results including unclear (‘don’t know’) answers in the category ‘very negative’ are included in the appendix. 11 EU citizens with double nationality are not omitted if one of the nationalities is the same as the country in which the survey is conducted.
taxes and minus any subsidies not included in the value of the products” (World Bank, 2020). There are no deductions made for depreciation of manufactured assets or depletion of natural resources.
4.3.2 Data analysis
This large-N observational research uses a conditioning strategy to try to make inferences. Data are pooled time series, as I use multiple Eurobarometer waves and for each a new and independent random sample is drawn from the population (GESIS, 2019a). The data on GDP per capita that I add to the dataset are panel data. While an experimental design (with panel data) including the counterfactual of no Brexit referendum would be ideal, this is not possible. Therefore, large-N observational research can best be used to make inferences about the population by using the conditioning strategy. By controlling for confounding variables via the conditioning strategy, their influence is taken away by adjusting for their effects and the relation between the independent variables of interest and the opinion on free movement can shine through (Toshkov, 2016).
In addition, I add possible mediating variables to examine whether these are mechanisms which mediate the support for free movement. These are the opinions on (non-)EU migrants and the EU in general. Mediating variables mediate the effect between the independent variable(s) of interest and the dependent variable (Toshkov, 2016). The Brexit referendum could have changed the opinions on (non-)EU migrants and the EU in general, resulting in a change in the opinion on free movement. In addition, these variables can be dependent on the independent variables of interest. Including these in a model and comparing these to the other models can therefore shed light on possible mechanisms through which the effect on the opinion on free movement takes place.
I perform the analysis with multivariate linear (ordinary least squares; OLS) regressions. I estimate and report one model to identify a change in support for free movement over time in general and several models including interactions to identify changes between groups and over time. These interactions concern the change of the independent variables of interest in both November 2016 and November 2017 compared to May 2016. In this way, I can identify possible changes in support for the right to free movement between the separate hypothesized groups.
The use of linear regressions is however not ideal, because the dependent variable – the opinion on freedom of movement – is a binary variable taking the value for or against. While in this case a logit or logistic regression would be more appropriate, I also make use of interaction effects between points in time and the independent variables of interest. These become clearer in linear regressions. Therefore, after consideration I have decided to use linear regressions including interaction effects.
As a robustness check, I include the corresponding logistic regressions in the appendix. The appendix also contains linear regressions including unclear answers in the most negative answer options of the opinion on free movement and some of the independent variables of interest, confounders and possible mediating variables. This because it is interesting to see whether this gives (large) differences compared to the analysis excluding the unclear answers.
4.4 Validity and reliability
I control for alternative explanations by including confounding variables as age, gender, the area the respondent lives in and political views in the analysis. This benefits the internal validity, because it controls to some extent for other possible interferences and processes. There are however some alternative explanations which cannot be controlled for. These for example consist of other events that could have led towards a change in the opinion on freedom of movement, e.g. terrorist attacks, and that have occurred in the same time period as between the used Eurobarometer surveys. So, because the Eurobarometer surveys are not conducted right before and after the Brexit referendum, this diminishes the internal validity of this study.
An additional downside of the Eurobarometer is that for every survey a new, independent sample is drawn. This means that possible observed changes cannot be attributed to the Brexit referendum with complete certainty but could also have emerged due to a different sample. So, because the opinion on free movement entails pooled time series and not panel data, other factors than the Brexit referendum also might have played a role in observed changes in opinion. This is a limitation, which cannot be controlled for with the research design and diminishes the internal validity of the study. The results should therefore be interpreted with caution.
Next to this, omitting the unclear answers from the analysis might influence the precision and credibility of the results. To check for this, the appendix consists of results including these answers in the most negative answer category. This is however still suboptimal, because it is never clear what the true meaning of these unclear answers is.
Less precision of the results can also stem from the fact that I use linear regressions, while the dependent variable is binary. Therefore, the use of a logit or logistic regression would be more appropriate. The included interaction effects in the analysis however have led to the decision to use linear regressions. As a robustness check, I include logistic regression models in the appendix.
Also, while on the one hand the inclusion of the possible mediating variables of the opinion on (non-)EU immigrants and the image of the EU in general in the analysis can give a more extensive picture of the mechanisms through which the opinion on free movement changes, it can on the other hand lead to less precision. This because their mediating effect is uncertain and they are not independent from the influence of the independent variables of interest and the Brexit referendum, like the included confounders. Therefore, only one model includes these possible mediating variables.
Attention should also go to possible reversed causality. The opinion on free movement can influence the independent variable of interest of attachment to the nation state. This creates a problem of reversed causality. It could be that a negative opinion on freedom of movement shapes the feeling of identity of a respondent. Next to this, the inclusion of UK citizens also leads to a problem of reversed causality. The opinion on freedom of movement can lead to a vote for Brexit for UK citizens, as these are the only EU citizens with the right to vote in the Brexit referendum.
The use of a great number of observations rules out that empirical associations are due to chance and contributes to the generalization of the results to all EU citizens (Toshkov, 2016). The sample of the Eurobarometer is a random probability sample (GESIS, 2019a). Although this creates a limitation, as mentioned above because this sample is drawn independently every survey, this does create a good reflection of the whole population of EU citizens and benefits generalizability and thereby external validity.
I maintain internal reliability of this study by using quantitative data, which are less subjective to interpretation. Because Eurobarometer and World Bank data are publicly accessible, this enhances the replicability and thereby external reliability of the study. I furthermore enhance replicability and external reliability by describing the research design and methods, and reporting choices in operationalization, measurement and case selection.
5. Analysis
In this section, I present and analyze the empirical results. First, I give more insights about the trend in the opinion on free movement through descriptive statistics. Hereafter, the results of the regressions follow, which I analyze and compare to the expectations from the hypotheses.
5.1 Descriptive statistics
Figure 1 gives insights about the average support for free movement in each country in November 2017. This figure shows the countries from least to most supportive towards this right.
Figure 1. Mean opinion on freedom of movement per country (November 2017)
It is interesting to note that Austria is the country in which the opinion on free movement is most negative, followed by the UK. Figure 1 however shows that the overall opinion on free movement is quite high in all countries. This consistent high public support in all countries also holds over time (EC, 2018b). This is shown in figure 2, which is based on data of the Eurobarometer surveys from November 2015 until May 2018 and is extracted from the report of the Eurobarometer in May 2018 (EC, 2018b).
Figure 2. Opinion on freedom of movement in the EU over time (EC, 2018b)
Figure 2 shows a consistent high public opinion on free movement over time. In addition, it indicates an interesting increase in support in November 2016, the Eurobarometer after the Brexit referendum, of two percentage points. This increase in support remains in 2017 and becomes even one percentage point higher in May 2018.12
Figure 3 zooms in on the public opinion on freedom of movement from the used Eurobarometer surveys in this study (May 2016, November 2016 and November 2017) and splits this out in countries of origin, countries of destination and countries which are neither. Each line takes into account the weighted average of each country in the group.
Figure 3. Weighted average of opinion on freedom of movement shown in groups of countries
12 It is important and interesting to notice that the one percentage point increase in May 2018 (as visible in figure 2) is associated with a one percentage point decrease in the answer category ‘don’t know’, while the two percentage points increase in support in 2016 is associated with a two percentage points decrease in the answer category ‘against’.
Figure 3 shows that EU citizens in countries of origin have become more positive towards free movement in November 2016 and then slightly less positive in November 2017. Citizens in countries which are neither a country of origin nor destination have changed relatively little in their opinion. They have become a bit more positive in November 2016 and then in November 2017 a bit more negative again. Citizens in countries of destination show the largest positive increase in the opinion on free movement. The support in these countries has increased in both November 2016 and November 2017. This suggests that citizens in all countries have become more positive towards free movement after the Brexit referendum. This effect however has diminished for citizens in countries of origin and countries which are neither a country of origin nor destination. Citizens in countries of destination do have a continuing positive trend in November 2017. These trends are however only descriptive. To be able to make empirical statements, I present the results of the regression analyses in the next section.
5.2 Inferential statistics
The results of the regressions now follow. I use these to analyze the data and reject or confirm the hypotheses in table 1, which I formulated in section 3 based on the theory and existing literature:
Table 1. Hypotheses
H1 The Brexit referendum has led to a more positive view of EU citizens on freedom of movement. H2 The Brexit referendum has had a greater positive influence on the opinion on freedom of
movement for EU citizens living in a country of destination than for EU citizens living in a country of origin.
H3 The Brexit referendum has had a greater positive influence on the opinion on freedom of
movement for EU citizens living in a country with a higher GDP per capita.
H4 The Brexit referendum has had a greater positive influence on the opinion on freedom of
movement for EU citizens with lower human capital than for EU citizens with higher human capital.
H5 The Brexit referendum has had a greater positive influence on the opinion on freedom of
movement for EU citizens who strongly attach to the nation state than for EU citizens who also feel European.
5.2.1 Support for free movement (after the Brexit referendum)
Table 2 shows the results of a linear regression without interaction effects. These results concern data of all three surveys. This gives a general image of the contributions of the independent variables of interest and the control variables to the opinion on free movement and tests the first hypothesis. The R-squared is 4.3 percent. This means that this model only explains a small part of the opinion on free movement.
Table 2. Linear regression without interaction effects
Model (1)
Including control variables
Survey
May 2016 (reference)
November 2016 0.009*** (0.003)
November 2017 0.003 (0.003)
Country type
Neither origin nor destination (reference)
Country of origin 0.042*** (0.004) Country of destination -0.041*** (0.003) GDP per capita -0.000** (0.000) Education Low education -0.036*** (0.005) Medium education -0.027*** (0.003)
High education (reference) Occupation
Self-employed (reference)
Managers 0.005 (0.006)
Other white collars -0.020*** (0.006)
Manual workers -0.019*** (0.006)
House persons -0.010 (0.008)
Unemployed -0.026*** (0.007)
Retired -0.014** (0.006)
Students 0.033*** (0.008) Strongly attached to nation state -0.115*** (0.003)
Control variables
Gender 0.008*** (0.003)
Age 0.000** (0.000)
Area 0.000 (0.002)
Political placement (left-right) -0.006*** (0.001)
Constant 0.956*** (0.011)
R-squared 0.043
Observations 62,186
Standard errors in parentheses, * denotes significant at 10% level, ** at 5% level and *** at 1% level; Reference categories are May 2016, countries who are neither origin nor destination, high education and self-employed.
The general opinion on free movement is slightly (roughly one percentage point) more positive in the Eurobarometer of November 2016 than in May 2016. This increase in the support for free movement in November 2016 compared to May 2016 might therefore suggest that the Brexit referendum has functioned as a benchmark. Following benchmarking theory, the Brexit referendum and thereby the alternative state of leaving the EU has looked less favorable than the status quo, because it might have shown the obstacles and costs of leaving the EU and the right to free movement (De Vries, 2017). This result therefore seems to confirm the first hypothesis, which states that the Brexit referendum has led to a more positive view of EU citizens on freedom of movement. This found effect however has been heterogeneous, because it has differed for different groups of citizens (as also becomes clear in table 3). The change in opinion on free movement in November 2017 compared to May 2016 is smaller, but still positive. This change is however not significant. Therefore, it seems that the significant increase in the
support on free movement only has been temporary. This suggests that the effect of benchmarking theory only holds for some time, but diminishes after the event (here: the Brexit referendum) has passed. Table 2 shows significant effects of all independent variables of interest. EU citizens living in a country of origin are more positive towards free movement than citizens who do not live in a country of origin or destination. In contrast, EU citizens living in a country of destination are more negative towards free movement than citizens who do not live in a country of origin or destination. This corresponds to Lutz (2020), who finds less support for free movement among countries with higher shares of EU migrants. GDP per capita has a small negative effect. This means that EU citizens living in a country with a higher GDP per capita are less supportive of free movement. This is consistent with the finding of Vasilopoulou and Talving (2019). Lower educated and medium educated EU citizens are more negative towards free movement than citizens who are higher educated. Lower educated are however more negative than medium educated citizens. In addition, compared to self-employed, other white collars, manual workers, unemployed and retired are more negative towards free movement. Students have a more positive view compared to self-employed, but also have equal or higher human capital. EU citizens with lower human capital therefore have a more negative opinion on freedom of movement. This is in line with the findings of Vasilopoulou and Talving (2019) and Meltzer et al. (2018). EU citizens who strongly attach to the nation state are less supportive of free movement than citizens who also feel European. This corresponds to the findings in the literature (Vasilopoulou and Talving, 2019; Blinder & Markaki, 2019; Lutz 2020). These significant effects of all independent variables of interest are in line with the existing literature and indicate that both utilitarian and affective perspectives as well as individual- and country-level factors explain the (differences in) opinion on free movement.
All control variables also show significant contributions, except for the area the respondent lives in. Women are slightly more positive towards free movement than men. The effect of age is positively significant, but very small. EU citizens who place themselves on ‘the right’ of the political spectrum are slightly more negative towards free movement.
5.2.2 Changes in support for free movement of specific groups after the Brexit referendum
Table 3 presents the results of the linear models with interaction effects between the time periods and the independent variables of interest. This gives interaction effects for November 2016 and November 2017 compared to May 2016 for all independent variables of interest. The first model concerns only the independent variables of interest, the second model includes the control variables and the third model also adds the possible mediating variables. In the appendix, results including unclear answers (table A2) and results from logistic regressions (table A3) can be found.
