Politics, Not Economic Interests: Determinants of Migration Policies in the European Union 1

Politics, Not Economic Interests: Blackwell Oxford, IMRE International 0197-9183 Spring 41 1Original ¾ 2007 2007 by UK Article ⅞ Publishing, the ⅞ Migration Center for Ltd. Review Migration Studies of New York. All rights reserved. Determinants of Migration Policies in the European Union 1 Simon Hix London School of Economics and Political Science Abdul Noury Université Libre de Bruxelles In this article we examine the determinants of European Union (EU) migration policies. We look at the passage of six pieces of migration and immigrant integration legislation in the fifth European Parliament (1999 2004). Based on the sixty-one roll-call votes on these bills we create a migration score for each Member of the European Parliament. We then use regression analysis to investigate the determinants of these scores. We find that the strongest determinants of policy outcomes on migration issues in this arena are the left-right preferences of EU legislators. These are stronger predictors than the economic preferences of national parties constituents or the economic interests or political preferences of the member states. INTRODUCTION On the issue of immigration in Europe, economic interests and political opinions often seem to be in conflict. Europe is an ageing continent and public finances are straining with the ever-growing burden of public pensions. Also, the fast growing service sector is generating the sort of jobs that most resident workers are reluctant to take up. Hence, in economic terms, most European countries need more rather than less migrants. Yet, the rise of anti-immigrant parties and movements in several countries, and the increase in violence targeted at ethnic minorities, suggests that political preferences rather than economic interests are the main determinants of the views of many European citizens on immigration and migration-related issues, such as the rights of ethnic minorities. Migration and immigrant integration policies in Europe are increasingly determined at the European Union (EU) level as well as at the national level 1 Research for this paper was funded by Migration Policy Group, <http://www. migpolgroup.com>. We would like to thank Adrian Favell and the referees of International Migration Review for comments on an earlier version of this paper. 2007 by the Center for Migration Studies of New York. All rights reserved. DOI: 10.1111/j.1747-7379.2007.00061.x 182 IMR Volume 41 Number 1 (Spring 2007):182 205

DETERMINANTS OF EU MIGRATION POLICIES 183 (cf. Guiraudon, 2003). The Maastricht Treaty (1993) established an EU Justice and Home Affairs policy, which was then reformed by the Amsterdam Treaty (1999) as part of the EU s Area of Freedom Security and Justice. Since Amsterdam, the EU has adopted a raft of migration and immigrant integration legislation. EU rules now cover, inter alia, the right of third-country nationals to family reunification, the working rights of third-country nationals, the admission of third-country nationals as students or volunteers, financial and technical assistance to third countries facing migration and asylum crises, the temporary protection of persons displaced by economic, political, or environmental disasters, which member state is responsible for examining an asylum application, standards for the treatment of asylum seekers, and non-discrimination on the grounds of racial or ethnic origin. Hence, EU rules now cover the full gamut of migration policies, from entry, residence, and economic rights of immigrants to societal integration of immigrants and their descendents. Furthermore, the European Parliament now has significant amendment and veto powers in the adoption of these policies. Most EU migration and related policies are now adopted under the so-called co-decision procedure. 2 This procedure is a classic bicameral procedure, where the European Parliament, which is elected every five years in EU-wide elections, has coequal legislative authority with the EU Council, which is composed of government ministers of the EU member states. Under the codecision procedure, legislation is initiated by the EU Commission, amended by either a qualified-majority vote (QMV) in the Council or a simple majority in the European Parliament, and finally adopted by QMV in the Council and a simple majority in the European Parliament. So, when making migration and immigrant integration policies, do EU politicians promote the labor market interests or the political and cultural preferences of their constituents? Do they follow their own personal experiences or do they simply vote along ideologically driven party lines? These questions have direct implications for understanding the evolution of migration policies in Europe. However, these questions are also important for the broader issue of whether economic interests or political preferences ultimately determine migration policies in liberal democracies. To answer these questions we look at the passage of six pieces of migrationrelated legislation in the fifth directly elected European Parliament (1999 2004). 2 Formally, the codecision procedure was first used in the area of EU asylum policies after the Amsterdam Treaty. The Nice Treaty (2003) then introduced a passaralle clause, which allowed the EU governments to decide to extend the codecision procedure to cover most other areas of migration and related policies, which they duly invoked in November 2004.

184 INTERNATIONAL MIGRATION REVIEW From the sixty-one recorded ( roll-call ) votes in the European Parliament on these bills we create an index for each Member of the European Parliament (MEP). The index allows us to study the MEPs behavior on migration issues. We then use regression analysis to investigate the determinants of the score each MEP received on this index. To check the robustness of our findings we also analyze the behavior of the MEPs on each separate piece of legislation. We find that the strongest determinants of behavior in the European Parliament on migration and related issues are the general left-right ideological positions of the European parliamentarians and the transnational European parties. These are stronger predictors of political behavior in the European Parliament on these issues than the economic preferences of the European parliamentarians constituents, or the economic interests or political preferences of the EU member states. In other words, left-wing politicians support liberal migration policies, despite the economic interests of many of their voters, who often compete with immigrants for unskilled jobs. Meanwhile, right-wing politicians support restrictive migration policies, despite the economic interests of many of their supporters, who benefit from increasing returns on capital investment which results from greater immigration. The rest of the article is organized as follows. The next section discusses some of the existing explanations of actors attitudes and behavior on migration issues. Then we introduce the six pieces of EU migration-related legislation on which we focus and how we calculate a migration index from MEP voting behavior on these bills. We also describe the statistical models and independent variables we use to investigate the determinants of MEP voting behavior. In the final sections, we present the statistical results and a short conclusion. ECONOMIC AND POLITICAL DETERMINANTS OF MIGRATION POLICY PREFERENCES The literature on migration policy and politics highlights economic as well as political determinants of citizens and politicians views on migration policy. Starting with the economic factors, at the individual level, unskilled workers and unemployed workers are more likely to be opposed to immigration than more highly skilled workers or owners of capital (esp. Scheve and Slaughter, 2001). On average, immigrants are willing to accept lower-skilled and lower-paid jobs than the indigenous workforce (e.g., Borjas, Freeman, and Katz, 1997). Hence, unskilled workers and those already unemployed are more likely to be in competition for jobs with immigrants than higher-skilled workers. There is evidence, for example, that the annual volume of immigration

DETERMINANTS OF EU MIGRATION POLICIES 185 is inversely related to the average wages of unskilled workers (Butcher and Card, 1991; LaLonde and Topel, 1991). Although these results are based on evidence from the labor market in the United States, it is reasonable to assume that a similar effect holds in the EU. Most EU member states have a more rigid and inflexible job market than the United States. For example, hiring and firing costs as well as minimum wages are higher in Europe (Nickell, 1997). Hence, the likely impact of greater immigration is higher unemployment amongst lower-skilled workers rather than lower wages. Nevertheless, against the economic interests of lower-skilled workers or the unemployed are the interests of high-skilled workers, the owners of capital and pensioners all of whom stand to benefit from higher levels of immigration, either because more immigration will push down factor production costs or because more immigrants will mean a larger workforce and more payments into national pensions systems. At an aggregate level, on the one hand, European countries with lower growth rates stand to benefit most from higher levels of immigration, as more immigration would reduce wage costs, and so boost growth. On the other hand, European countries with higher growth rates may demand more migration, to alleviate bottlenecks in the labor market and increase the labor supply. Nevertheless, a clearer aggregate economic interest related to migration is the size of the public pensions burden. Throughout Europe, the state is the main provider of pensions. As a result, countries with a higher proportion of the population that is of pensionable age will tend to have higher pensions expenditure as a proportion of total public expenditure. These states are consequently more likely to be in favor of promoting immigration, to increase the pool of workers paying in to the tax system, than states that have smaller populations of pensionable age. However, politics, rather than economics, also determines individuals and countries attitudes towards migration. At the individual level, an important factor is whether an individual feels that more immigration will change the nature of their society and whether they think such changes are desirable. For example, people who are racist are likely to be opposed to migration, while ethnic minorities are likely to be more liberal towards further migration. Furthermore, rights of immigrants and minorities are part of the new social left-right dimension in European politics, where voters and parties positions are shaped more by attitudes towards social and political rights and duties rather than individual economic rights and obligations (cf. Kitschelt, 1995). Specifically, left-libertarian voters and politicians, who favor equal opportunities and equal treatment, are likely to have more liberal attitudes towards immigration

186 INTERNATIONAL MIGRATION REVIEW and the rights of migrant communities. In contrast, right-authoritarian voters and politicians, who favor traditional social and cultural values, are likely to have more conservative attitudes towards immigration and the rights of migrant communities. These ideologically based preferences can conflict with economic interests. For example, unskilled workers and the unemployed, the economic groups who are more likely to be negatively affected by immigration, are core voters of many socialist and social democratic parties in Europe. But, as these parties become more left-libertarian in their orientations, they are likely to promote liberal policies towards immigration and the rights of ethnic and religious minorities. On the other side, pensioners and the owners of capital are more prone to support right-wing parties, but as these parties become more right-authoritarian, they are likely to favor more restrictive immigration and social integration policies. At an aggregate level, a variety of national historical, cultural, and political factors shape countries attitudes towards immigration (e.g., Brubaker, 1992; Hollifield, 1992; Money, 1997; Favell, 1998; Joppke, 1999; Neumayer, 2005). Numerous factors could be included here. However, two are perhaps worth highlighting for in the context of EU policies on migration and societal integration. First, countries with colonial pasts, such as the United Kingdom, France, the Netherlands, Belgium, Spain, and Portugal generally have larger minority populations and have historically had more liberal policies towards immigration and asylum. Second, the general liberalness of a society, such as public attitudes towards minority rights and solidarity with the Third World, have generally resulted in more open immigration and asylum policies (Hollifield, 1992; Joppke, 1998). One issue in any polity, however, is how these individual or aggregate economic and political preferences translate into actions by elected representatives, via the design of electoral and political institutions (cf. Powell, 2000; Strøm, Müller, and Bergman, 2003). For example, in a first-past-the-post electoral system and a separation of powers between the legislature and the executive (as in the United States), elected politicians are likely to respond issue by issue to the preferences of their constituents, rather than follow their personal ideological preferences or the voting instructions of their party leaders. In contrast, in parliamentary systems, with strong party organizations and where elections are fought on broad ideological issues, elected politicians are more likely to follow their general left-right preferences or the voting instructions of their party leaders. In the EU context, there is a separation of powers between the EU executive (the European Commission) and the European Parliament, which suggests

DETERMINANTS OF EU MIGRATION POLICIES 187 that MEPs are relatively independent from the Commission and national governments. Nevertheless, in almost all member states, national party leaderships control the selection of candidates in European Parliament elections, and as the powers of the European Parliament have grown national parties have used these powers to ensure that MEPs have similar preferences to their national party leaders on all the key policy issues on the EU agenda. National parties have also increased their contacts with their MEPs once elected, for example by issuing voting instructions to MEPs on highly salient issues. Also, European elections tend to be fought on the general popularity of national parties rather than on specific policy issues on the EU policy agenda and the behavior of the MEPs (e.g., van der Eijk and Franklin, 1996). As a result, national parties have a strong influence on the behavior of their MEPs. However, the main political organizations inside the European Parliament are the transnational political groups, who bring together the national parties with similar policy preferences (Raunio, 1997; Kreppel, 2002). The political groups have significant organizational resources, hierarchical leadership structures. The political group leaderships influence the behavior of their backbenchers by issuing voting instructions, employing whips to enforce their instructions, and by using coordinators in the committees of the Parliament to monitor and coordinate the behavior of their members in the dayto-day legislative scrutiny work of the Parliament. The political groups in the Parliament are mainly organized around the traditional European party families. The two largest groups are the European People s Party European Democrats (EPP-ED), which brings together all the main conservative and Christian democratic parties on the center-right in Europe, and the Party of European Socialists (PES), which brings together the various socialist, social democratic, labor, and other center-left parties in Europe. Together these two political groups contain almost 70 percent of the MEPs. The third largest group in 1999 2004 parliament was the European Liberal Democrat and Reform Party (ELDR), which voted in equal proportion with the EPP and PES. 3 To the left of the PES are two smaller groups: a group of Green and left-wing regionalist MEPs, in the Greens/European Free Alliance (G/EFA); and a group of radical-left and ex-communist MEPs, in the European United Left/Nordic Green Left group (EUL/NGL). To the right of the EPP is a group of national-conservative parties in the Union for a Europe of the Nations (UEN) group. In addition to this general left-right pattern of the 3 In the current, 2004 2009, European Parliament, the ELDR group changed its name to the Alliance of Liberals and Democrats for Europe.

188 INTERNATIONAL MIGRATION REVIEW groups, several anti-european MEPs from the extreme left and extreme right sit in the Europe of Democracies and Diversities (EDD) group. 4 Finally, several MEPs are not members of any of the political groups, and sit as non-attached members. Research on general voting patterns in the European Parliament has shown that as a result of the organizational structure and membership of the political groups, the cohesion of the political groups in the European Parliament has grown over time and competition between the groups is mainly structured along left-right rather than pro/anti-europe lines (Kreppel and Tsebelis, 1999; Hix, 2001; Noury, 2002; Hix, Noury, and Roland, 2005, 2006). This consequently suggests that when voting on migration and related issues, MEPs are more likely to act on their personal ideological opinions or the voting instructions of their national or European party leaders than the specific migration policy preferences of the citizens in their member state or the supporters of their national party. This proposition points in the same direction as some existing research on EU migration policies. In particular, Gallya Lahav (2004a, 2004b) has shown that there is a gap between the preferences of the public and the attitudes and behavior of the EU elites in this policy area. Despite growing public opposition to immigration, there is a consensus amongst European elites in favor of delegating migration policies to the EU and promoting relatively liberal migration and immigrant integration policies through the EU. If this holds in the European Parliament, then we should expect the median voter in the parliament to be in favor of liberal EU migration policies, and MEPs general attitudes towards European integration to influence their voting behavior on EU migration legislation. MEP VOTING BEHAVIOR ON EU MIGRATION AND IMMIGRANT INTEGRATION LEGISLATION Six Bills in the Fifth European Parliament Table 1 shows the six main pieces of EU migration and immigrant integration legislation that passed through the fifth directly elected European Parliament (1999 2004). The first bill, on the equal treatment between persons without racial and ethnic discrimination, aimed to establish an EU-wide definition of discrimination on the grounds of racial and ethnic origin, to shift the burden of proof to the defendant in certain circumstances, and to provide a minimum 4 In the current, 2004 2009, European Parliament, the EDD group changed its name to the Independence/Democracy group.

DETERMINANTS OF EU MIGRATION POLICIES 189 TABLE 1 MIGRATION LEGISLATION IN THE FIFTH EUROPEAN PARLIAMENT No. of Legislation Date Initiated Date Passed RCVs in EP Date of EP Votes (1) Equal treatment between Nov 25, 1999 Jun 29, 2000 8 May 18, 2000 persons without racial and ethnic discrimination (CNS/1999/0253) (2) Third-country nationals Dec 1, 1999 (dropped) 9 Sep 6, 2000 right to family reunification (CNS/1999/0258) (3) Equal treatment between Nov 25, 1999 Nov 27, 2000 2 Oct 5, 2000 persons-general nondiscrimination (CNS/1999/0225) (4) Third-country nationals long-term resident status Mar 13, 2001 Nov 25, 2003 18 Feb 5, 2002 (CNS/2001/0074) (5) Entry and residence of third-country nationals for the purpose of employment (CNS/2001/0154) (6) Third-country nationals right to family reunification (CNS/1999/0258) Jul 11, 2001 (awaiting Council approval) 15 Feb 12, 2003 May 2, 2002 Sep 22, 2003 9 Apr 9, 2003 Total no. of roll-call votes in the EP = 61 level of redress for people who have suffered discrimination. As such, this legislation represented the most advanced level of protection against racial and ethnic discrimination anywhere in the world. The European Parliament debated and voted on a series of amendments to the legislation in May 2000, the EU Commission accepted some of the amendments proposed by the parliament, and the legislation was finally adopted by the Council in June 2000. The second bill, on family reunification of third-country nationals, proposed that third-country nationals residing lawfully in the EU, refugees and other persons enjoying subsidiary protection, and EU citizens whose family are third-country nationals, should have full rights to family reunification. The European Parliament voted on a series of amendments to the legislation in September 2000, and the Commission accepted most of the parliament s amendments, but the Council failed to reach an agreement on the bill. The third bill aimed to establish a general framework to combat discrimination and ensure equal treatment in employment, and covered nondiscrimination on the grounds of religion or belief (as well as disability, age, or sexual orientation). The legislation covered direct and indirect discrimination as well

190 INTERNATIONAL MIGRATION REVIEW as harassment. The European Parliament voted on a series of amendments in October 2000, many of which were then accepted by the Commission, and the Council finally adopted the legislation in November 2000. The fourth bill, on the status of third-country nationals who are longterm residents, aimed to facilitate the freedom of movement for resident third-country nationals, to better regulate the status of family members of citizens of the EU, and to define the possibilities for refusing or withdrawing the right of residence. The European Parliament voted on several amendments to the legislation in February 2002 and the Council adopted the legislation in November 2003, accepting a number of parliament s amendments. The fifth bill aimed to harmonize EU rules on the entry and residence of third-country nationals for the purpose of paid employment and self-employment. The proposal set out common definitions, criteria, and procedures regarding the conditions of entry and residence of third-country nationals for the purpose of employment and a single national application procedure encompassing both residence and work permit within one administrative act (a one-stop shop ). The European Parliament passed several amendments in February 2002, most of which were accepted by the Commission. However, as of July 2005, the Council had not yet passed the legislation into law. Finally, the sixth bill was a new proposal of the second bill we look at, on third-country nationals rights to family reunification. There were two main changes to the original proposal: a standstill clause, which would restrict national discretion in the application of the terms of the legislation; and a deadline clause, which set a deadline of two years after the transposition of the legislation for the initiation of a more fully harmonized law on family reunification. The European Parliament voted on several amendments in April 2003, and the Council adopted the law in September 2003. A Migration Index from European Parliament Roll-Call Votes Not all votes in the European Parliament are by roll-call, where how each MEP votes (Yes, No, or Abstain) is recorded in the minutes of the Parliament s plenary session. Most votes are taken either by a show-of-hands or by an electronic vote. But, either 32 MEPs or a political group can request that a roll-call vote is taken on any issue. Hence, roll-call votes are usually used for high-profile or controversial issues. As Table 1 shows, there were 61 roll-call votes on these six pieces of EU legislation. To create an index from MEP voting behavior in these 61 votes we first looked at the exact subject of each vote, to determine the direction, or policy

DETERMINANTS OF EU MIGRATION POLICIES 191 implication, of the outcome of each vote. Some proposals aimed to extend the rights and opportunities for migrants in the EU (a pro-migration issue), while others aimed to reduce the rights or opportunities of migrants in the EU (an anti-migration issue). Pro-migration MEPs should vote Yes on a pro-migration issue and No on an anti-migration issue. We consequently granted 1 point if an MEP voted in a pro-migration way (voting Yes if the issue was pro-migration or No if the issue was anti-migration), and 0 if the MEP did not vote in a pro-migration way (No on a pro-migration proposal or Yes on an anti-migration proposal). The final score for an MEP on the index was then calculated as the sum of the points each MEP achieved, divided by the number of migration roll-call votes (61), and multiplied by 100. So, if an MEP voted in a pro-migration way in all 61 votes he or she scored 100, and if an MEP voted in an anti-migration way in all 61 votes he or she scored 0. We only calculated scores for MEPs who participated in at least 25 percent of the votes (16 votes). 5 Figure I shows the mean score for each member states delegation of MEPs and the MEPs in each European political group. Swedish, Danish, Greek, and Belgian MEPs are on average more pro-migration than Italian, British, German, or Irish MEPs. Also, MEPs on the left, in the green/regionalist group (G/EFA), the radical left group (EUL/NGL), and the socialist group (PES), are on average more pro-migration than MEPs on the right, in the nationalconservative group (UEN) and the Christian democrat/conservative group (EPP-ED). Moreover, the difference between the political groups is considerably greater than the difference between the national delegations, which suggests that party-political preferences are more important than national interests on these issues. However, this conclusion cannot necessarily be inferred without more detailed statistical analysis. MODELS AND VARIABLES: FACTORS INFLUENCING MEP VOTING ON MIGRATION ISSUES To determine the factors influencing MEP voting on migration and immigrant integration issues we estimate a series of multivariate regression models. These models take the following general form: 5 This is very similar to the one computed by Kalt and Zupan (1984), and more generally to the method used by many interest groups in the United States to rank Congressmen on the issues they care about. For example, see the liberal-conservative index produced by the Americans for Democratic Action, <http://www.adaction.org/votingrecords.htm>.

192 INTERNATIONAL MIGRATION REVIEW Figure I. MEP Migration Scores by Member State and Political Group Y m = α + β 1 PERSONAL m + β 2 IDEOLOGY m + β 3 CONSTITUENCY m + β 4 MS m + β 5 EPG m + ε m m = 1,, 618 where Y is the dependent variable, α is a constant, β 1 to β 5 are regression coefficients, ε is an error term, and each MEP is indexed by m. We were able to calculate pro-migration scores for 618 MEPs in the fifth European Parliament.

DETERMINANTS OF EU MIGRATION POLICIES 193 We estimate the same basic model for two types of dependent variable: (1) the score of each MEP on the migration index; and (2) the behavior of each MEP in the set of roll-call votes on each of the six migration bills separately. To create the second dependent variable, we recoded the MEPs vote decisions so that 1 represented a pro-migration vote decision and 0 represented an anti-migration vote decision. We use OLS linear regression for the models of the first dependent variable and logistic regression for the models of the second set of dependent variables. The term PERSONAL combines three variables which capture the effect of the personal characteristics of MEPs on their voting behavior on migration issues. The first variable, Female (MEP), is coded 1 if the MEP is female and 0 if the MEP is male. In our study, 198 of the 618 MEPs were women (32 percent). The second variable, Minority (MEP), is coded 1 if the MEP is a minority or 0 otherwise. We counted as a minority those MEPs who were from an ethnic or national minority, such as a British Muslim MEP or a French MEP of African or North African origin, or a German MEP of Turkish or Polish origin. We identified 21 of the 618 MEPs in our study who were ethnic minorities (3.4 percent). The third variable, Age (MEP), is the age in years of an MEP. If women, minorities, or older members are more in favor liberal migration and immigrant integration policies than men, nonminorities or younger members, then these variables should be positively related to MEPs scores on the migration index and the propensity of MEPs to vote for pro-migration legislation. The term IDEOLOGY combines two variables which look at the relationship between MEPs general ideological preferences and their voting behavior on migration and immigrant integration issues. These two variables are produced by applying the NOMINATE scaling method (Poole and Rosenthal, 1997) to all the 5,190 roll-call votes in the 1999 2004 European Parliament. Research has shown that for the fifth European Parliament, the first and second dimensions revealed by this method are strongly related to the left-right and pro/anti EU preferences of MEPs, respectively (Hix et al., 2006). As a result, the first variable here, Left-Right Position (MEP), is the location of the MEPs on the first dimension produced by NOMINATE in the fifth parliament, and second variable, Pro/Anti EU Position (MEP), is the location of the MEPs on the second dimension produced by this scaling method. On the left-right variable, 1 is the furthest left position and +1 is the furthest right position, and on the pro/anti EU position variable, 1 is the most anti-eu position and +1 is the most pro-eu position. Hence, if more left-wing MEPs are in favor of liberal migration and immigrant integration policies, the left-right variable should be negatively related to MEPs scores on the migration

194 INTERNATIONAL MIGRATION REVIEW index and the propensity of MEPs to vote for pro-migration legislation. Similarly, if more pro-eu MEPs are in favor of liberal migration and immigrant integration policies, the pro/anti EU variable should be positively related to MEPs scores on the migration index and the propensity of MEPs to vote for pro-migration legislation. The term CONSTITUENCY combines three variables which look at how variations in the political preferences and economic interests of MEPs constituents influence MEPs voting on migration and immigrant integration issues. Because European elections are fought by national parties in national contests, and because national parties control the selection of candidates for seats in the European Parliament, as we discussed above, the main constituents for each MEP are the supporters of his or her national party. To calculate the political preferences and economic interests of national party supporters we use data from the Eurobarometer 47 survey, where face-to-face interviews were conducted with more than 16,000 citizens in the fifteen EU member states in March April 1997. We chose this particular survey because it included a set of questions on EU citizens attitudes towards race, ethnicity, and migration as well as the standard questions on which national party each citizen supports and citizens personal economic situation. For the first variable here, Racism (Constituency), we took the mean position of each national party s supporters on a ten-point racism scale, where 1 represented not at all racist and 10 represented very racist. For the second and third variables, Unemployed (Constituency) and Unskilled (Constituency), we calculated the percent of each national party s supporters who were unemployed and unskilled workers, respectively. Whereas the personal characteristics and ideology measures vary between each MEP these three constituency measures vary at the level of each national party, as these measures assume that all the MEPs from the same national party have the same constituents. If MEPs respond to their constituents political preferences, we expect MEPs from national parties whose supporters are more racist to have lower scores on our migration index and be less likely to support pro-migration legislation. If MEPs respond to their constituents economic interests, we expect MEPs from national parties who have a higher proportion of unemployed or unskilled workers amongst their supporters to have lower scores on our migration index and be less likely to support pro-migration legislation. The term MS combines six variables which look at how variations in member states political and economic interests influences behavior in the European Parliament on migration and immigrant integration issues. Here, we use three variables which capture the politics of migration in each EU member state: Minorities (MS) is the proportion of each EU member state s population

DETERMINANTS OF EU MIGRATION POLICIES 195 that is an ethnic or racial minority, as measured by the Minority Rights Group (1997); Foreign Born (MS) is the proportion of each EU member state s population that is foreign born, as measured by the Organization for Economic Cooperation in Europe (OECD, 2003); and Asylum Seekers (MS) is the number of asylum applications each member state received in 2002 as a percent of its population (OECD, 2003). These variables consequently vary at the level of each member state. The member states with colonial pasts tend to have larger minority and foreign born populations, and are likely to be more supportive of liberal EU migration and immigrant integration policies. Similarly, although one might expect countries with more asylum seekers to be opposed to more liberal migration and immigrant integration policies, member states with larger numbers of asylum seekers are likely to be in favor of common EU migration policies, as these policies are likely to force other member states to share the burden of asylum seekers (cf. Thielemann, 2003, 2004). The fourth member-state level variable, Growth Rate (MS), is the average annual growth rate of each EU member states between 1995 and 2004, as reported by Eurostat. 6 This variable captures the broad economic interests of each member state vis-à-vis migration, and tests whether member states with lower levels of growth or higher rates of growth are in favor of liberal migration policies. The fifth member-state level variable, Foreign Aid (MS), is the total foreign aid of each EU member state as a percent of the member state s Gross National Income in 2003, as reported by the OECD. 7 This variable is a proxy for the general political preferences of a member state towards migrants, in that higher levels of foreign aid are highly correlated with liberal policies on migration (e.g., Neumayer, 2005). The sixth member-state level variable, Pensions Burden (MS), is the population of a member state aged 65 and over as a percent of the total population aged 15 to 64 in 2004, as reported by Eurostat. 8 MEPs from member states with higher pensions burdens are likely to be more supportive of pro-migration policies. In addition to these member state-level continuous variables, in some models we include member state-specific dummy variables. These variables control for other member state-level economic and political interests that might influence MEP voting on migration and immigrant integration issues. 6 See <http://epp.eurostat.cec.eu.int>. 7 See <http://www.oecd.org>. 8 See Eurostat news release 48/2005, April 8, 2005.

196 INTERNATIONAL MIGRATION REVIEW Finally, the term EPG represents a set of dummy variables for the political groups in the fifth European Parliament. Based on their average behavior in all votes in the fifth European Parliament, the general left to right ordering of these groups is as follows: EUL/NGL, G/EFA, PES, ELDR, EDD, EPP-ED, and UEN (Hix et al., 2006). The baseline group for these variables are the MEPs who sit as non-attached members (na). Including these variables in a model allows us to look at which political groups are more in favor of liberal migration and immigrant integration policies and also allows us to investigate variations in MEP behavior within the political groups. Descriptive statistics for all the variables are listed in the Appendix. STATISTICAL RESULTS Table 2 shows the results of a series of models of the determinants of MEP scores on our migration index. The main finding is that MEPs ideological preferences are dominant. The two measures of ideological preferences left-right position and pro/anti-eu position are always highly significant with the expected signs. Also, when the ideological preferences of MEPs are included (in models 4 and 5), the explanatory power of the models increases enormously. The results also show that women and minority MEPs are more promigration than men or nonminority MEPs, and significantly so. For example, in model 1 women score on average 8 points higher on our migration index than men, and minorities score on average 11 points higher than nonminorities. However, once left-right and pro/anti-eu preferences are included, these effects disappear (in models 4 and 5). What this means is that women and minority MEP are predominantly on the left and pro-european, and left-wing and pro-european women and minority MEPs are indistinguishable in their voting behavior on migration issues from left-wing and pro-european men and nonminority MEPs. In contrast, the age of an MEP has no significant effect in any specification. Regarding the preferences of MEPs constituents, politics again trumps economic interests. Regardless of the model specification, MEPs from national parties who are supported by unskilled workers and the unemployed are more rather than less pro-migration. This is because these parties are on the left, and the MEPs from these parties follow the general ideological positions of their parties rather than the specific economic interests of their supporters on migration issues. In contrast, party supporters attitudes towards racism are significant predictors of MEP voting in all model specifications. Specifically, within the left and the right, and within each European political group, MEPs from

DETERMINANTS OF EU MIGRATION POLICIES 197 TABLE 2 DETERMINANTS OF MEP VOTING ON MIGRATION ISSUES (1) (2) (3) (4) (5) (6) Constant 47.284*** 70.996*** 55.186** 32.993*** 26.838*** 41.354*** (6.865) (7.709) (23.703) (9.777) (8.850) (5.431) Female (MEP) 7.705*** 8.187*** 6.519*** 0.219 0.202 0.342 (2.497) (2.258) (2.196) (0.849) (0.673) (0.645) Minority (MEP) 11.470* 9.658 7.641 1.440 2.692 2.407 (6.014) (6.643) (6.151) (2.504) (2.418) (2.098) Age (MEP) 0.004 0.079 0.113 0.045 0.037 0.023 (0.129) (0.122) (0.116) (0.042) (0.034) (0.033) Left-Right Position 47.701*** 32.520*** 32.727*** (MEP) (0.785) (4.553) (5.295) Pro/Anti-EU 17.974*** 9.004*** 14.248*** Position (MEP) (1.016) (2.219) (2.440) Racism 13.209*** 16.205*** 1.214** 1.204* 0.570 (Constituency) (1.084) (1.355) (0.585) (0.682) (0.824) Unemployed 100.328*** 116.253*** 34.622*** 43.640*** 45.786*** (Constituency) (12.936) (12.613) (7.400) (7.475) (6.991) Unskilled 103.667*** 111.709*** 26.880*** 22.989*** 12.614 (Constituency) (20.648) (22.837) (7.727) (7.408) (8.870) Minorities (MS) 1.654*** 0.052 0.419* (0.613) (0.261) (0.221) Foreign Born (MS) 0.277 0.478** 0.001 (0.446) (0.197) (0.111) Asylum Applications 2.030 9.618** 5.236 (MS) (13.133) (4.774) (3.566) Growth Rate (MS) 0.656 0.855 0.833* (1.523) (0.569) (0.442) Foreign Aid (MS) 30.673*** 20.910*** 8.367*** (7.026) (3.123) (2.518) Pensions Burden 0.214 0.475 0.415* (MS) (0.715) (0.297) (0.229) ELDR 25.548*** 25.913*** (3.159) (3.491) PES 19.474*** 17.481*** (4.007) (4.822) EUL/NGL 18.852*** 21.469*** (6.255) (7.048) G/EFA 16.818*** 18.583*** (6.069) (6.974) EPP-ED 0.810 0.697 (3.195) (3.287) EDD 1.751 0.198 (2.511) (2.766) UEN 2.772 1.446 (2.784) (2.925) Adjusted R-squared 0.02 0.21 0.26 0.90 0.94 0.94 Observations 618 592 592 592 592 592 Notes: Dependent variable: MEP score on migration index. Method: OLS linear regression. Robust standard errors in parentheses. In model 6, dummy variables for member states are included but estimates of those variables are not reported. *Significant at 10%. **Significant at 5%. ***Significant at 1%.

198 INTERNATIONAL MIGRATION REVIEW Figure II. Relationship Between Left-Right Location and Migration Score national parties whose supporters are more racist have lower scores on our migration index. Political factors are also more influential than economic factors when considering the effect of the member state-level variables. The only variable here which is significant in all models is the level of foreign aid. This suggests that MEPs from member states with general more liberal political cultures are more pro-migration, regardless of their left-right location or political party affiliation. On contrast, the main measures of the economic interests of the member states, the average annual growth rate and pension burden, are generally not significant. Also, the number of minorities, foreign born, or asylum applications in a member state are not consistently significant. Finally, the magnitudes and signs of the European political group variables are as expected, with the liberals (ELDR), socialists (PES), radical left (EUL/ NGL), and greens/regionalists (G/EFA) being more pro-migration. The strength of the relationship between left-right preferences and MEP voting behavior on migration and immigrant integration issues is illustrated in Figure II. The further left an MEP is, the higher he or she scores on our migration index. Also, the effects of political group membership can be seen by the clustering of the MEPs. The cluster on the furthest left are the MEPs in the green/regionalist (G/EFA) and radical left (EUL/NGL) groups, next to them are the socialists (PES), then in the middle are the liberals (ELDR), and on the

DETERMINANTS OF EU MIGRATION POLICIES 199 right are the anti-europeans (EDD), national-conservatives (UEN), and Christian democrats/conservatives (EPP-ED). Interestingly, there is a dominant pro-migration coalition in the European Parliament, of greens, radical left, socialists, and liberals, most of whose MEPs score well above 50 percent on our index. Nevertheless, there are some variations within the European political groups which are not explained by left-right positions. Our results in Table 2 suggest that the main factors explaining these differences within the political groups are whether an MEP is personally pro- or anti-european, how racist his or her voters are, and how liberal his or her member state is. The estimates reported in Table 2 are based on an aggregate score that combines various pieces of legislation. If MEPs perceive or interpret those pieces of legislation as substantially different, then aggregation would not lead to a meaningful score. To check the robustness of these findings, we replicated models 3 and 5 from Table 2 for all the votes on each piece of legislation separately. Table 3 presents the results of this analysis. The findings, above, generally hold. In particularly, left-right preferences of MEPs are the dominant determinants of MEP voting behavior on all six pieces of legislation. The magnitude of this effect is large, the effect is highly significant, and the total explanatory power of the models (as measured by the pseudo R-squareds) more than doubles with the addition of this variable. Nevertheless, there are some interesting additional findings in this legislation-by-legislation analysis. Economic interests of constituents had a significant, and expected, effect on the one piece of legislation which was most directly related to the economic rights of migrants: the directive on third-country nationals entry and residence for the purpose of employment (legislation 5). This was the only piece of legislation where once left-right and European political group positions are controlled for (in model 10), MEPs from national parties who are supported by unskilled workers and the unemployed were significantly more likely to vote against a more liberal EU migration policy in this area. For example, on the left, MEPs from the socialist and radical left parties, whose main support base is unskilled workers and the unemployed, voted against more liberal employment rights for migrants, while MEPs from the social democratic parties and green parties, whose main support base is public sector workers, were free to vote with their ideological convictions, and so supported a more liberal migration policy in this area. In addition, the pension burden of a member state had a significant positive effect in two bills (legislations 5 and 6) that related more to the rights of new immigrants than the integration of existing migrants. In other cases, it was either insignificant or did not have the expected sign.

TABLE 3 DETERMINANTS OF MEP VOTING ON INDIVIDUAL MIGRATION BILLS (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Legislation 1 Legislation 2 Legislation 3 Legislation 4 Legislation 5 Legislation 6 Constant 1.301 1.921 1.168 2.021 9.237 66.850*** 6.568*** 3.008*** 2.001** 0.958 0.686 13.032*** (1.692) (2.902) (0.978) (1.258) (6.858) (14.614) (0.707) (1.157) (0.804) (1.494) (1.033) (2.842) Female 0.237*** 0.226 0.475*** 0.064 0.656* 2.406 0.511*** 0.061 0.360*** 0.023 0.501*** 0.330 (MEP) (0.135) (0.192) (0.076) (0.099) (0.391) (1.770) (0.058) (0.082) (0.061) (0.101) (0.079) (0.226) Minority 0.592 0.636 0.758*** 0.166 0.280 11.949 0.288* 0.113 0.613*** 0.093 0.649*** 0.724 (MEP) (0.369) (0.481) (0.223) (0.312) (0.722) (0.000) (0.173) (0.368) (0.161) (0.360) (0.225) (0.806) Age (MEP) 0.005 0.044*** 0.009** 0.005 0.010 0.024 0.010*** 0.002 0.006* 0.002 0.010** 0.000 (0.008) (0.012) (0.004) (0.005) (0.021) (0.082) (0.003) (0.005) (0.003) (0.005) (0.004) (0.015) Left Right 6.798*** 4.284*** 44.003* 5.142*** 6.104*** 11.126*** Position (MEP) (1.183) (0.593) (25.602) (0.545) (0.685) (1.239) Pro/Anti EU 1.948** 0.323 51.400*** 4.595*** 2.575*** 3.919*** Position (MEP) (0.861) (0.381) (10.966) (0.326) (0.457) (1.240) Racism 1.071*** 0.231 1.413*** 0.095 1.251*** 4.907*** 1.546*** 0.118 2.001*** 0.770*** 1.864*** 0.209 (Constituency) (0.132) (0.213) (0.100) (0.096) (0.140) (1.105) (0.086) (0.102) (0.085) (0.142) (0.127) (0.205) Unemployed 18.286*** 1.767 11.402*** 1.153 3.526 67.831** 18.075*** 2.043 12.944*** 3.572** 21.903*** 2.925 (Constituency) (2.228) (3.503) (1.106) (1.605) (5.683) (33.229) (1.014) (1.596) (0.929) (1.689) (1.540) (3.334) Unskilled 2.756 5.332* 3.615*** 1.315 7.151 3.120 9.134*** 7.438*** 0.042 10.016*** 9.893*** 7.026** (Constituency) (1.870) (3.185) (1.028) (1.404) (5.061) (16.011) (0.778) (1.570) (0.824) (1.777) (1.221) (2.771) Minorities 0.013 0.109* 0.145*** 0.030 0.210 3.903*** 0.100*** 0.116*** 0.279*** 0.055 0.246*** 0.028 (MS) (0.043) (0.059) (0.023) (0.030) (0.128) (0.957) (0.019) (0.028) (0.020) (0.036) (0.028) (0.078) Foreign Born 0.031 0.016 0.078*** 0.054*** 0.420*** 0.378*** 0.003 0.043** 0.137*** 0.065*** 0.099*** 0.012 (MS) (0.025) (0.036) (0.015) (0.017) (0.132) (0.141) (0.014) (0.019) (0.011) (0.021) (0.015) (0.033) 200 INTERNATIONAL MIGRATION REVIEW

Asylum Applications (MS) Growth Rate (MS) Foreign Aid (MS) Pensions TABLE 3 (CONTINUED) DETERMINANTS OF MEP VOTING ON INDIVIDUAL MIGRATION BILLS (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Legislation 1 Legislation 2 Legislation 3 Legislation 4 Legislation 5 Legislation 6 0.860 1.483 2.344*** 1.200** 6.325*** 49.173*** 1.066*** 0.290 2.866*** 2.662*** 2.753*** 1.039 (1.071) (1.008) (0.564) (0.558) (2.040) (12.318) (0.393) (0.531) (0.539) (0.618) (0.597) (1.074) 0.374*** 0.329** 0.111** 0.072 0.194 9.866*** 0.298*** 0.049 0.112** 0.115* 0.285*** 0.533*** (0.104) (0.134) (0.053) (0.068) (0.343) (3.101) (0.039) (0.056) (0.046) (0.068) (0.056) (0.163) 3.242*** 1.999*** 3.303*** 2.277*** 2.250 37.581*** 0.704*** 0.626* 3.705*** 0.603 3.504*** 4.577*** (0.583) (0.737) (0.335) (0.430) (1.530) (7.909) (0.201) (0.341) (0.289) (0.438) (0.328) (0.928) 0.088 0.004 0.020 0.044 0.095 1.719*** 0.097*** 0.023 0.029 0.066* 0.063** 0.268*** (0.056) (0.068) (0.031) (0.037) (0.216) (0.611) (0.019) (0.028) (0.024) (0.037) (0.031) (0.091) Burden (MS) Party Dummies No Yes No Yes No Yes No Yes No Yes No Yes Pseudo R-squared 0.36 0.68 0.19 0.48 0.25 0.88 0.31 0.70 0.26 0.75 0.28 0.89 Observations 1,739 1,739 4,367 4,367 832 347 8,785 8,785 7,570 7,570 4,351 4,042 Note: Dependent variable: MEP voted in a pro-migration way in a vote on a particular piece of legislation. Method: logistic regression. Dummy variables for each vote on each piece of legislation are included but not reported. Robust standard errors in parentheses. *Significant at 10%. **Significant at 5%. ***Significant at 1%. DETERMINANTS OF EU MIGRATION POLICIES 201

202 INTERNATIONAL MIGRATION REVIEW CONCLUSION Migration and immigrant integration policies in Europe are now decided at the European level as well as at the national level, and the European Parliament has an increasing role in amending and passing EU legislation in this policy area. The European Parliament is also an interesting environment to look at the general issue of what factors shape the migration policy behavior of elected politicians. Hence, how Members of the European Parliament vote on migration issues is important, first, for understanding what factors influence EU migration policies today and will influence these policies in the future and, second, for understanding whether elected politicians are primarily driven by economic interests or political preferences when making migration policies. Looking at the passage of six key pieces of migration legislation through the 1999 2004 European Parliament we find that politics rather than economic interests ultimately shapes migration policy outcomes in this chamber. General ideological preferences of MEPs, and in particular their location on the left-right dimension, as well as the positions of the transnational European parties, are stronger predictors of MEPs voting behavior on migration issues than the economic preferences of the supporters of the MEPs national parties or the economic interests of the MEPs member states. The dominance of political factors over economic interests of MEPs constituents meant that the fifth European Parliament supported relatively liberal migration policies, as the median voter in that parliament had relatively liberal preferences on migration policy. Moreover, because of the way the European Parliament is elected by some form of proportional representation in each member state a pro-migration coalition in the European Parliament, comprising socialists, liberals, greens, and radical left MEPs, is likely to dominate the European Parliament for some time to come. Nevertheless, if European citizens turn against their political elites, and vote in greater numbers for more openly anti-immigrant parties, then it is conceivable that a more anti-migration majority might emerge in the European Parliament at some point in the future. More generally, our findings suggest that when elected representatives adopt migration policies, their ideological and political preferences may be more significant than their own or their constituents economic interests. But, how far can one generalize from behavior in the European Parliament? MEPs are quite disconnected from their voters, since the electoral connection between MEPs and voting behavior in European Parliament elections is weak. If European parliamentarians are more easily punished or rewarded by voters, then perhaps economic interests would play a stronger role. Nevertheless,