Ethnicizing citizenship, questioning membership. Explaining the decreasing family migration rights of citizens in Europe

Abstract Citizenship does not equal belonging. In this paper, we investigate how the disjunction between the ‘imagined community’ and the formal citizenry impacts on citizens’ rights. In particular, we analyse decision-making on the family migration rights of citizens in France, Germany and the Netherlands. Our analysis shows that in these three countries, notwithstanding their different migration and citizenship regimes, the reduction of citizens’ family migration rights is based on the same discursive mechanism: the ‘membership’ of citizens of migrant origin who marry a partner from abroad is called into question. As they are excluded from membership of the imagined community, their entitlement to family migration rights is decreased. Ethnic conceptions of national community, intersecting with gender and class, play a crucial role in shaping the rights attached to citizenship in Europe today.


Introduction
On 13 January 2015, David Pujadas, lead newsreader of France's major public television news bulletin, messed up. In the 20 o' clock news, Pujadas described a French citizen as 'a Muslim, married to a French woman' . This slip might have gone unnoticed if a week later, a young NGO-leader called Najoua Arduini-ElAtfani had not confronted him on live television, stating that 'this is dramatic, Mr. Pujadas. This is not a Muslim married to a French woman; this is a man who is French, and a Muslim, married to a woman who is French, and perhaps not a Muslim, perhaps a Catholic…' . The video of his slip-up and her protest was widely shared on social media. 1 Citizenship is about belonging and about the boundaries of belonging: it is about demarcating who 'we' are, who is part of 'us' , and who is not (Yuval-Davis 2007). In the most obvious sense, citizenship policies are part of the politics of belonging because they regulate access to citizenship, i.e. who gets to be citizen and who does not. However, Pujadas' 'unhappy' formulation, as he himself called it, is emblematic for another crucial but often overlooked aspect of the politics of citizenship and belonging. The 'imagined community' and the citizenry do not overlap perfectly. As a result, possessing citizenship of a state does not always equal being considered a member of the political or national community of that state: 'People who are constructed to be members of other ethnic, racial and national collectivities, are not considered "to belong" to the nation state community, even if formally they are entitled to' (Yuval-Davis 2007, 563). In Pujadas' eyes, as in the eyes of many others, Muslims cannot be 'really' French -regardless of whether they are French citizens or not. In this paper, we investigate when and how the problematization of the membership of certain citizens impacts on the status and entitlements attached to citizenship. In particular, we focus on the family migration rights granted to citizens as instances of 'politics of belonging' , by which those in political power (re)produce the limits of the 'community of belonging' (Yuval-Davis 2006, 205).
Since the 1970s, citizens have enjoyed strong family migration rights in most European states, based on the principle that citizens should not be forced to choose between living in their country of citizenship or living with their (foreign) spouses, children or parents. In recent years however, a trend of a deterioration of citizens' family migration rights is discernible. In Belgium, the United Kingdom, Denmark, Norway and the Netherlands, citizens have to fulfil the same conditions as settled foreigners to bring over foreign spouses or children. In 2007, both France and Germany introduced restrictive requirements for citizens' unification with foreign spouses. This apparent convergence raises the question whether the reduction of citizens' family migration rights is driven by similar mechanisms in different countries. Studies of family migration policies in the United Kingdom, Denmark and Norway have shown that recent restrictive reforms target non-white citizens, or citizens of migrant origin, in particular (Muller Myrdahl 2010;Schmidt 2011, 261;Wray 2011, 230, 231). 2 In this paper, we inquire to which extent the reduction of family migration rights of citizens is the result of the problematization of the membership of particular categories of citizens.
In order to evaluate whether similar mechanisms drive the reduction of citizens' family migration rights in different contexts, we will undertake a comparative analysis of decision-making processes in France, Germany and the Netherlands, three countries which have historically adopted very different migration and citizenship regimes. We will analyse decision-making processes about the family migration rights of citizens both at the administrative and the political level, based on records of parliamentary debates, archives of the ministries responsible for migration and citizenship policies, and interviews with politicians and civil servants involved in the decision-making process. We investigate whether, when and how civil servants and politicians problematize the membership of citizens of migrant origin, and whether, when and how this results in reducing citizens' family migration rights. Especially in politicians' and civil servants' discourses, conceptions of membership based on ethnicity or race often remain implicit. Such conceptions can barely be spoken, because they are at the same time self-evident (almost unconscious) and inadmissible (illegitimate). Our analysis must therefore be interpretive, paying close attention to context and omission so as to 'explore connotation and inference, the unspoken level at which texts so often construct meaning' (Anderson 2012, 327). The use of non-public sources such as ministerial archives and interviews is particularly relevant here, because ethnicized conceptions of belonging are more likely to surface in situations where politicians and civil servants are not directly exposed to public scrutiny.
Assuming like Yuval-Davis that intersecting vectors of inclusion and exclusion are mutually constitutive, in the sense that 'there is no meaning to the notion of "black" (…) which is not gendered and classed ' (2007, 565), we will inquire throughout our analysis how constructions of (non)membership based on ethnicity and race intersect with gender and class. We take both ethnicity and race to be (self)ascribed, socially construed identity categories, where ethnicity is defined along cultural, religious, historical and linguistic lines and where race is defined according to presumed biological or physical characteristics.

Citizenship and migration: identity and rights
The question whether and how the status, rights and identities tied to citizenship have changed in recent decades as a result of global migration and the increased salience of migration control has been the subject of intense academic scrutiny since the 1990s. At first, scholarly attention focused on 'levelling up' , i.e. on decreasing inequality between citizens and non-citizens through increasing the rights granted to resident non-citizens (Hammar 1990). Scholars such as Soysal (1994) and Jacobson (1996) argued that as international human rights treaties had replaced the nation-state as a source of rights, national citizenship had become irrelevant as a category of membership.
In recent years, scholars have observed a different way in which national citizenship is declining as a category of membership: not through 'levelling up' but through 'levelling down' . The difference between citizens and non-citizens is seen to diminish not through an increase of foreigners' rights, but through a decrease of citizens' rights. Spiro (2008, 6), writing about the evolution of US citizenship, argues that globalization has deprived the state of the capacity to define the nation as a bounded community with a distinct identity. As US citizenship has become more inclusive and easier to obtain, US citizens have lost their privilege over foreign residents: 'If citizenship does not meaningfully coincide with actual community, it is unlikely to be determinative of rights and duties ' (2008, 107). The central claim of Joppke's recent book Citizenship and Immigration (2010) is in many ways similar to Spiro's. Joppke observes that as most countries in Europe have come to tolerate dual nationality and to complement ethnic criteria for access to citizenship with ius soli, citizenries have become ethnically more diverse. Whereas 'classically' , state membership implied membership of a 'cultural community ' (2010, 111), Joppke now observes a 'weakening of the particularistic, identity-lending dimension of citizenship ' (2010, 142). In parallel to this 'decoupling of citizenship and nationhood ' (2010, 143) the rights tied to citizenship have decreased. Joppke considers these three developments -liberalization of access to citizenship, decrease of rights tied to citizenship and weakening of identity tied to citizenship -'logically complementary' , i.e. matching and mutually reinforcing (2010,152). Like Spiro, Joppke argues that 'as citizenship has become more accessible, it inevitably must mean less in terms of rights and identity' (2010, 33).
Joppke's optimistic view of a European trend towards liberalization of access to citizenship is contested, as comparative analyses of the regulation of naturalization and birth right citizenship show a much more mixed picture of restrictive and liberal tendencies (Goodman 2010;Vink and De Groot 2010). We leave aside the discussions on the liberalization of citizenship to focus on a different part of the argument brought forward by Spiro and Joppke. Almost implicit but crucial to their argument is the claim that ethnic diversity among the citizenry leads to a reduction of citizens' rights. It is this claim which we critically assess in this paper.
In particular, we take issue with the assumption shared by Spiro and Joppke that ethnic diversity among the citizenry 'inevitably' coincides with a weaker collective identity and therefore with a reduction of rights. The belief that a citizenry must be ethnically homogeneous to be meaningful reflects a communitarian conception of citizenship, where citizenship is defined as membership of a community with its own particular culture and tradition. In other conceptions of citizenship, where citizenship is defined through individual rights (liberalism), participation in the political community (republicanism) or participation in the welfare state (Scandinavian welfare state ideology), there is no 'inevitable' link between ethnic diversity, identity and rights. Furthermore, both Joppke and Spiro appear to take ethnic diversity as a given. In contrast, we assume that ethnicity and diversity are social and political constructs: which 'differences' are considered to matter and be problematic depends on time and place. If there is a connection between 'ethnic diversity' of the citizenry and a reduction of citizens' rights then it is not 'inevitable' but politically construed. Therefore, we set out to examine whether and how questioning the belonging of certain citizens has affected the rights tied to citizenship.
In this, we follow the work of postcolonial and feminist authors who have shown that there always have been citizens perceived as 'strangers in our midst' . Nation-states have never been able to make their citizenries match perfectly with their 'imagined communities' (Glenn 2002;Smith 1997;Vogel 1994). From its very inception, liberal citizenship was shaped not only by universalistic and egalitarian aspirations, but also by highly exclusionary representations of 'the citizen' , who has been imagined white and masculine, often also wealthy and Catholic or Protestant. Those who failed to fit this image were often excluded from citizenship or from the full enjoyment of citizenship rights. Today as in the past, formal citizens may be symbolically excluded from the nation, in contrast to 'autochtons' or 'real citizens' (Ceuppens and Geschiere 2005), because of their race or ethnic background (Reiter 2008;Tsuda 2014).
Researchers analysing citizenship, identity and rights in migration societies often focus on political and social rights such as the right to vote or to welfare, rather than on family migration rights. However, citizens' family migration rights are of crucial importance both analytically and politically, because they derive directly from the primary entitlement that sets citizens apart from foreigners: the right of citizens to remain in their country of citizenship. Considering people's 'deep and vital interest in being able to live with their immediate family members' , what is at stake in disputes over citizens' family migration rights is the extent to which they may be 'forced by the state to choose between home and family' (Carens 2003, 97). In other words, citizens' right to remain in their home country is eroded if they are not allowed to live there with their family. This is why citizens have been granted privileged family migration rights. Because citizens' family migration rights are directly grounded in their 'right to remain' , i.e. in citizens' 'membership' and 'rootedness' (Block 2012, 3-5), family migration rights offer particularly valuable testing ground to evaluate the policy impact of the problematization of citizens' membership in different European societies.

Citizenship, 'diversity' and family migration: three different starting points
In 2010, the percentage of the population that was foreign-born was 11.5% in France and the Netherlands, and 13% in Germany (OECD 2015). In France and the Netherlands, (de) colonization led to large numbers of residents from (ex)colonies, who were considered ethnically 'different' , being granted citizenship. In Germany, many migrants from Central and Eastern Europe ('ethnic resettlers') were also granted automatic or easy access to citizenship after the Second World War until the 2000s, but this was because they were considered ethnically German. These different migration histories have led to more perceived ethnic diversity among the citizenry in the Netherlands and France than in Germany. This was further reinforced by differences in citizenship policies and naturalization rates. Be it due to differences in the migrant population or to differences in policies, naturalization rates have been consistently higher in the Netherlands (4-9%) and France (2-5%) than in Germany (less than 2%) since the 1980s (Reichel 2012;Waldrauch 2006). Moreover, France has offered grandchildren and children of migrants French citizenship since 1851 and 1889, respectively. The Netherlands applied ius soli until the end of the nineteenth century and re-opened access to citizenship for grandchildren of migrants in 1953 and to their children in 1985. In contrast, Germany only opened up its citizenship to children of migrants in 1991 and obliged them to choose between the nationality inherited via their parents and German nationality until 2014. As a result, the number of citizens of migrant background perceived as 'ethnically different' from the majority population has been higher over past decades in France and the Netherlands than in Germany.
These three countries have historically also dealt differently with the allocation of family migration rights to citizens. Until the 1970s, citizenship laws throughout Europe were based on the principles of patriarchy and unity of the family. In France, Germany and the Netherlands, the foreign wife of a male citizen automatically obtained his citizenship, while a female citizen who married a foreigner automatically lost her citizenship. Only from the 1970s onwards were citizenship policies adapted to the emerging norm of gender equality, so that foreign spouses of citizens -male and female -were granted facilitated access to naturalization (Knop 2001). Since naturalization was now neither automatic nor immediate, policies had to be devised to regulate the entry and stay of foreign partners.
In France, citizens were attributed strong family migration rights. This reflects French legal tradition which bases French nationality neither on ethnicity nor on a voluntary act, but on 'socialization' , i.e. 'the acquisition of social codes' and 'an effective sociological bond' (Weil 2008, 53). Marriage to a French citizen has long been considered an obvious example of such an 'effective sociological bond' with France. Until recently, the only restrictions which applied were controls regarding the authenticity of the marriage, including minimum residence and marriage periods before a permanent residence permit was granted. Still today, the large majority of conditions imposed on migrants who wish to bring their families to France, including housing and income conditions, do not apply to citizens.
Similarly, in Germany, the conditions governing family migration have been much more favourable for citizens than for foreigners, the government following the logic that 'German citizens' interest in establishing family unity in the federal territory is especially protected' . 3 Just as in France, spousal migration sponsored by German citizens was subjected only to controls on sham marriages including a 'probationary period' , but not to income or housing conditions.
In contrast, in the Netherlands, citizens have not had privileged access to family migration over the past decades. This privilege was reduced already in the 1980s -not by levelling down, but by levelling up. In the 1970s and early 1980s, Dutch politics were dominated by progressive tendencies. Equal treatment and emancipation were key values for right-wing as well as left-wing governments. In consequence, an explicit effort was made to eliminate differential treatment between Dutch citizens and settled migrants from Dutch legislation wherever possible. As part of this effort, it was decided in 1978 that, like Dutch citizens, settled migrants would be allowed to bring over non-married partners and be exempted from the income requirement (Bonjour 2009, 142-148). This indeed was the type of levelling up of migrants' rights to equal citizens' rights that induced Hammar (1990) to speak of 'denizens' , and Soysal (1994) of 'post-national membership' .
Thus, while the citizenry in France comprised many more migrants perceived as ethnically different than the German citizenry, both countries granted citizens privileged family migration rights. In the Netherlands, in the same period that citizenship was opened up to migrants' children, the family migration rights of migrants were improved to match citizens' rights. Both the French and the Dutch case show that opening up citizenship to migrants does not inevitably coincide with a reduction of citizens' rights. We seek more context-specific explanations and wonder to what extent the different histories of migration and citizenship in France, Germany and the Netherlands are reflected in different degrees or ways of problematization of the membership and family migration rights of citizens of migrant origin.

France
For a long time, French citizens have enjoyed strong family migration rights, with restrictions limited to controls on fraud. Most recently however, citizens' marriages have been subjected to restrictive measures even if they were considered genuine. This restrictive move must be understood as part of the increased politicization of migration in France in the early 2000s, as a result of electoral pressure from the Far Right Front National and of events such as the 9-11 terrorist attacks and the riots in French suburbs in the autumn of 2005. French political debates about migration and integration in the 2000s were marked by the fear that French society and identity were endangered as a result of past and present immigration flows and 'failing' immigrant integration. In this context, Nicolas Sarkozy, leader of the right-wing UMP, advocated more selective migration policies, so that migration would be 'choisie' rather than 'subie' , i.e. 'chosen' rather than 'undergone' . To Sarkozy, the most prominent example of unwanted immigratie subie was family migration. In 2003, family migration made up 73% of total immigration inflow in France. Over 60% of these family migrants came to join a French citizen (Kofman, Rogoz, and Lévy 2010, 9, 10). Any attempt at reducing the number of family migrants was much more likely to be successful if it included the spouses of citizens. This quantitative goal was entwined with a qualitative problematization of family migration, which became entangled with the problematization of the membership of French citizens of migrant background.
In 2003, Sarkozy, then minister of the Interior, pointed out that the number of French citizens marrying a foreigner had quadrupled over the last five years and that most of these marriages were celebrated abroad, especially in Algeria, Tunisia and Morocco. He took this to be a clear indication that a substantial portion of these marriages was either fraudulent or forced. 4 Sarkozy spoke of 'young girls of migrant origin' who were forcibly married off during summer holidays in their parents' country of origin as a 'gaping breach' in control policies which 'weighs heavily on the fate of girls of migrant origin' . 5 Implicitly but unmistakeably, Sarkozy targeted Muslim citizens by referring to forced marriages and to Northern Africa. Ethnicity and gender intersect in Sarkozy's representation of female citizens of migrant background as passive victims of an oppressive culture. This 'Othering' of citizens paved the way towards restricting their family migration rights.
In 2007, Sarkozy, now president, initiated the introduction of a pre-departure integration condition for foreign spouses not only of resident migrants, but also of French citizens. Henceforth, foreign spouses of French citizens would have to participate in an exam on French language and society before migrating to France. If their knowledge proved insufficient, they were obliged to participate in a course before being granted an entry visa. For the first time, the right of French citizens to bring a foreign spouse to France was made conditional.
The UMP-government proposal to apply pre-departure integration requirements to spouses of French citizens had been highly controversial in Parliament. In the Legislative Committee of the Senate, a majority -including members of the governing UMP partyvoted to exempt the spouses of French citizens. These senators argued that 'marriage with a French citizen is a strong sign of integration and the spouse of a French citizen should therefore be assumed to be integrated' . 6 To salvage his reform proposal, Interior Minister Brice Hortefeux pointed out that while Senators often talked about well-integrated Canadian and Australian spouses, their number was negligible. Instead, Hortefeux emphasized, out of 60 thousand foreigners marrying a French citizen in 2004, '43 thousand were of African origin, of whom more than 12 thousand from Sub-Saharan Africa' . 'Family migration has become a major source of immigration to France' , minister Hortefeux argued, 'and therefore we believe a pre-departure integration test to be necessary' . However, in 'particular cases the test might represent a constraint rather than an advantage' , notably for 'French expatriates who marry abroad and wish to resettle in France to pursue their career' . 7 Thus, Hortefeux presented two different categories of citizens marrying foreigners and suggested treating them differently. The first category of citizens was marrying Africans. Implicitly, the image conjured up is that of citizens of migrant origin marrying spouses from their own or their parents' country of origin -thus signalling by their partner choice that these citizens 'really' belong to Africa rather than to France. The second category concerned citizens moving abroad for career purposes. Implicitly, the image is that of a white, male, well-earning citizen. The first category of citizens are formally insiders, but their origin and their partner choice combine to make them symbolic outsiders. Marriage to these citizens is no automatic guarantee for integration: a test and a course are necessary. The upper-class citizen -probably white, probably male -is very much an insider: his spouse's integration does not need to be tested. Ethnicity, gender and class combine to produce insiders and outsiders of the imagined community.
This categorization was taken up by UMP Senator Del Picchia in an amendment proposing to reintroduce the pre-departure integration requirement for spouses of French citizens, but exempting the spouses of 'citizens residing abroad who wish to resettle in France for professional reasons' . Clearly targeting expat citizens imagined to be male, Del Picchia argued that 'a young French executive sent abroad, who marries a local woman' should 'be entirely exempted from the formalities of the test and the course' if he returned to France with his foreign wife. 8 UMP Senator Longuet agreed that 'certain mixed marriages are marriages of openness, while others are marriages of closure': The international mixed marriage presents an aspect of openness: I am delighted to see all our young people who go abroad to serve France, be it in development aid or as executives, and who marry there, and allow others to share in our culture. However, we are surprized to observe the systematic occurrence of mixed marriages of closure, of a community that withdraws into itself. (…) What if these marriages systematically served the purpose of permitting certain French citizens to have wives who have been subjected to the mutilation of circumcision? 9 The 'marriage of closure' is clearly imagined as practiced by citizens of migrant origin. This 'Other' citizen is represented not as a female victim here, but as a male oppressor who adheres to marriage and sexuality norms repulsive to 'real' French citizens.
As amended by Del Picchia, the pre-departure integration measure for citizens' spouses was adopted by both Houses of Parliament. The exemption for expats aimed at treating different categories of citizens differently, but it does not prevent 'native' citizens who have not worked abroad but met their foreign partner in France, on holidays, or online from being confronted with the integration requirement. Its effective impact was therefore most likely very limited. More 'refined' measures were proposed in 2007 by the Interministerial Committee on Immigration Control, which advocated evaluating 'the intensity of the ties of the couple with France compared to the intensity of the ties with the country of origin of the foreign spouse' . 10 Clearly, this proposal targeted citizens of foreign origin with partners from their own or their parents' country of origin. However, these proposals have hitherto not been implemented, perhaps because the government chose to avoid the risk of being accused of discrimination on ethnic and racial grounds. 11 In a liberal democracy, the possibilities for legally treating citizens unequally on ethnic grounds are very limited. New restrictions of family migration rights must target (almost) all citizens, as the integration requirement does, or none at all.
Thus, our analysis of the French case shows that the push for reduced family migration rights resulted from the questioning of the membership of citizens of migrant background in a context of strong politicization of the migration and integration issue.

Germany
Family migration constitutes a significant share of total migration into Germany: in 2007, 29% of all residence permits granted to third-country nationals were based on family-related claims, three-fourths of which was spousal migration (BAMF 2008, 32).
Until recently, German citizens had comparatively extensive family migration rights. However, in 2007 restrictive reforms were introduced that applied to third-country-national and German sponsors alike. Important changes introduced were a minimum age of 18 years and a pre-entry language requirement. Furthermore, a proof of income could be demanded of German sponsors in 'exceptional' cases, defined as situations where the couple could be 'reasonably expected' to settle together abroad, for instance when the German sponsor had dual citizenship or had substantive links to the foreign spouse's country of origin. 12 This provision was designed to reduce spousal migration of citizens of migrant background. Ethnic resettlers (Aussiedler) were explicitly exempted from this income requirement, 13 illustrating that lawmakers consider them to possess a 'stronger' membership than other ethnic minorities.
Especially parliamentarians of the governing Christian Democratic Union (CDU) strongly problematized family migration as an unwanted flow that needed control. 14 Family migration was also increasingly connected to integration (deficits) and forced or arranged marriages. Interior minister Wolfgang Schäuble stated the reason for introducing a pre-entry language requirement in parliament: 'Up to 50 per cent of the third generation of certain migrants have spouses who did not grow up in Germany. This indicates that these are often arranged marriages (…) This is an abuse that inhibits integration, which we have to fight precisely in the spirit of Article 6 [which protects marriage and family] of our constitution. ' 15 At the beginning of this speech, Schäuble stated that 'up to 50 per cent of the second and third generation of Turkish origin marr[y] partners from abroad. ' It is thus obvious that with 'certain migrants' Schäuble addresses residents of Turkish origin -roughly half of whom actually possess German citizenship (Auswärtiges Amt 2015). Just as in France, the immigration of spouses of Muslims was problematized. The strong rhetorical concentration on forced or arranged marriages enabled a highly gendered discourse portraying incoming spouses as female victims in need of policy intervention (Block 2014, 248-250).
At the same time, politicians and civil servants realized that any effort to reduce the number of spousal migrants needed to also include citizen sponsors. Since 2000, more citizens than foreigners had been sponsoring spousal migration, and government reports explained this at least partly with higher naturalization rates and spouses joining ethnic German resettlers (BAMF 2007, 37). While in 1996, 65% of spouses joined foreigners and 35% joined citizens, this ratio was inverted 10 years later (Block 2012, 102).
In 2006, the Ministry of the Interior first suggested in an evaluation report to extend the income requirement to citizens since 'especially for recently naturalized Germans (…) an income requirement will provide an incentive to integrate' , also since 'many foreigners migrating to Germany with a long-term perspective immigrate directly into the welfare system' . 16 To sustain this argument, the ministry referred to former Berlin integration commissioner Barbara John's (CDU) statement at an expert meeting: German sponsors who reunite with foreign spouses, regardless of whether it is a German marrying a Filipina or a naturalized Turk who brings in his wife from the outskirts of Izmir, do not have to provide a proof of income. I consider this wrong (…) for integration reasons. How will we create an incentive, and an incentive is also always pressure, how will we sustain an incentive to strive for qualification, to strive for employment, to be resourceful, to get trained, if [they] can become German at the age of 16 and have no perspective? 17 This quote displays a particular paternalism: the citizen of migrant background (addressed not as 'German' but as 'naturalized Turk') needs 'incentive' and 'pressure' to seek employment. It also underlines that the income requirement for citizens was devised specifically for Germans of migrant background. Here, ethnicity and class intersect: citizens of migrant background are assumed to be uneducated, unemployed and prone to abuse welfare.
In interviews with the authors of this paper, MPs and civil servants related the restriction of family migration rights of German citizens to a perceived 'devaluation of German citizenship' resulting from the increasing ethnic diversification of the German citizenry since the liberalization of citizenship law in 2000. Put differently, as more and more individuals hold citizenship whose membership to the community is questioned due to their ethnicity, politicians and civil servants consider citizenship to be devalued and thus justify decreasing the rights attached to it: German citizenship has lost the significance it used to have. We recently had a case concerning the expulsion of a pregnant Nigerian prostitute. The child's father was also Nigerian, had a residence permit and been here longer than eight years. This child, born in Germany, will be German. Of two Nigerians! That is the new citizenship law. This is ius soli, of course, but the significance of citizenship has really been a bit relativized now. 18 [The decreasing rights of German sponsors] are certainly also motivated by the increasing naturalization rates, in the sense that 'reunification with Germans' often means that these Germans are former foreigners who have been naturalized, and then still bring in their spouse from their original country of origin. This was surely an argument to have no reason to treat [German citizens] differently [from foreigners] in this context. 19 A CDU MP justified extending the language requirement to citizens by clearly positioning citizens with Turkish heritage outside the national German community: Many constellations exist, and will only increase in the future through the 'option model' , where the male family members who bring in the Turkish women, formally are German citizens, but possibly live just as much in a parallel world as a Turkish citizen. 20 Especially when discussing the 'exceptional' income requirement, CDU MPs and civil servants describe ethnic minority citizens' belonging as much weaker than that of 'natives': The constitutional principle states: same things must be treated in the same way, and different things must be treated differently. And there is certainly a difference between dealing with an only-German, who has lived his entire life only in Germany, and dealing with somebody who has lived in Turkey for 30 years and now for eight years in Germany and has managed to be naturalized. 21 In 2012, the 'exceptional' income requirement for citizens was invalidated by the Federal Administrative Court, which ruled that German citizens cannot be required to lead their marriage abroad and that dual citizenship cannot be a reason to impose an income requirement. 22 The political attempt to specifically target ethnic minority citizens clashed with the norm of equal treatment that is crucial to liberal values and rule of law. If German policy-makers wish to reduce the family migration rights of ethnic minority citizens, they can only do so by reducing the rights of all citizens.
Our analysis has shown that even though access to citizenship was substantially liberalized in 2000, ethnic conceptions of who can be a 'real' German still prevail among many politicians and civil servants. German policy developments appear to neatly confirm Spiro's and Joppke's line of thought: as German citizenship became easier to obtain for persons of migrant background, the (family migration) rights attached to citizenship decreased. However, the similarity of the discursive mechanisms observed in France and Germany shows that it is not liberal reform of citizenship policies in itself that leads to a reduction of rights, but the questioning of the membership of citizens of migrant background, in a context of strong politicisation of the migration and integration issue.

The Netherlands
In the late 1970s, Dutch politicians improved the family migration rights of settled migrants so as to match the rights of citizens. Paradoxically, this levelling up paved the way to an early and sharp levelling down. In the early 1990s, the progressive ideology that had dominated Dutch politics in the 1980s was replaced by a more neoliberal approach, which emphasized that rights ought to be coupled to duties. Rising immigration numbers and poor migrant performance in education and employment caused increasing concern among Dutch politicians. In 1993, the Lubbers III government, composed of Christian Democrats and Social Democrats, introduced the first restrictive reform of Dutch family migration policies in over a decade: the income requirement was reintroduced at 70% of the welfare level. Importantly, this income requirement was to apply to settled migrants and Dutch citizens alike.
This decision was very controversially debated behind the closed doors of cabinet negotiations. Civil servants suggested that the income requirement would lead to a very limited decrease in immigration if it was applied only to migrant sponsors but not to Dutch citizens. 23 This quantitative aspect was coupled to a qualitative argument, when civil servants pointed out that many citizens applying for family migration were of Surinamese origin. 24 Eventually, the decision was made at the highest political level. In a handwritten note to prime minister Ruud Lubbers, deputy minister of Justice Aad Kosto pleaded in favour of applying the income requirement to Dutch citizens. He emphasized that '34 per cent of incoming flow of family migrants were Dutch citizens in 1991' and that 'many of these Dutch citizens are naturalized Surinamese and others' . 25 Apparently, these decision-makers considered the migration background of citizens to be relevant when determining what their family migration rights must be. If the citizens making use of family migration rights were mostly naturalized citizens, rather than 'real' Dutch citizens, it was deemed legitimate to curtail these rights.
However, when the discussion moved from secret government negotiations to public parliamentary debate a couple of weeks later, neither the quantitative nor the qualitative argument for applying the income requirement to citizens with foreign spouses was mentioned. When prime minister Lubbers presented the reform proposal in Parliament, he emphasized repeatedly that 'equal treatment of Dutch and non-Dutch' was a 'very important starting point' for his government. 26 Thus, through a paradoxical path dependency, the principle of equal treatment introduced in the 1970s to improve the rights of migrants, now served to justify decreasing the rights of citizens.
The legacy of the progressive 1970s and 1980s contributed to the decrease of citizens' rights in yet another manner. The migrant integration policy conducted by the Dutch government in the 1980s was called 'ethnic minorities policy' . It was as ethnic minorities that migrants were expected to find their place in Dutch multicultural society, regardless of whether they chose to adopt Dutch citizenship. The assumption that origin and ethnicity are crucial to a person's integration into Dutch society, regardless of their citizenship, shapes Dutch policies and debates until today. While the term 'ethnic minority' was dropped in the early 1990s, it was replaced by the term 'allochtoon' or 'person from elsewhere' . Sixty per cent of migrants living in the Netherlands and 80% of children of migrants born in the Netherlands possess Dutch citizenship. 27 However, they are hardly ever addressed as Dutch citizens. In political discourse as well as in official statistics, they are routinely categorized as 'allochtons' .
In the early 2000s, the politicization of migration peaked in the Netherlands as it did in France and Germany, with marriage migration as a focal point of political attention. Debates focused exclusively on marriage migrants from Morocco and Turkey, who were assumed to come to join young persons of Turkish and Moroccan origin. This image of family migration is persistent even though it is false: between 1995 and 2011, only 20% of all family migrants were born in Morocco or Turkey, 28 and between 2003 and 2006 only 14% of sponsors were 'second-generation' migrants (WODC 2009, 30). Ethnicity and class intersected in the association of 'import brides and grooms' from Morocco and Turkey with poor education and unemployment, as well as with 'traditional' , patriarchal gender and family norms contrary to 'progressive' 'Dutch' values (Bonjour and De Hart 2013). The children or grandchildren of migrants living in the Netherlands are almost all Dutch citizens. However, parliamentarians almost invariably referred to them as 'Turks and Moroccans' , 'Turkish youngsters' or 'young Moroccans' -never as Dutch. 29 Even the government wrote about 'Turks and Moroccans of the second generation who marry an often poorly educated person from the country of origin' . 30 Born and raised in the Netherlands, their 'country of origin' was thought to be elsewhere. Their citizenship was never once mentioned: it was simply irrelevant in these political debates. The notion that citizenship might entail privileged family migration rights never came up, and the application of the restrictive reforms of 2004 and 2005 -ranging from a minimum age of 21 and an income requirement of 120% of the minimum wage to pre-departure integration requirements -to citizens and migrants alike was not questioned once.
In fact, 35% of the persons bringing over a foreign partner to the Netherlands are Dutch citizens without a migration background (WODC 2009, 30). The restrictive policy reforms would apply to these 'native' Dutch as well. The only politician to mention this fact was Green MP Azough, who objected to obstacles being imposed on the foreign partners of 'problemless Dutch' , that is 'autochtonous Dutch' (assumed to be 'problemless') and 'well integrated, highly educated allochtonous Dutch' . 31 Other politicians never even considered whether or how restrictions might affect 'native' Dutch and their foreign partners. It seems that class intersected with ethnicity in politicians' imagination of these couples, so that it was assumed that 'native' citizens with foreign spouses would possess the financial and educational resources to meet the income and integration requirements.
In 2010, parliamentarians raised questions about the obstacles put in the way of a particular group of Dutch citizens with foreign partners: expats. All examples brought forward in the debate were of Dutch men working abroad. The Christian Union mentioned 'a man working for an international company in Thailand' while the Socialist Party referred to a Dutch navigating officer who had met and married a 'very fine woman' while working abroad: 'Surely we are not going to make things hard on these people!' 32 The government promised that the Immigration Office would improve its service to expats so as to take their particular circumstances into account. 33 In the Netherlands as in France, the expat, imagined white, male and upper class, is the ultimate insider, and his marriage to a woman met abroad who follows him upon his return is the most positively valued of all international marriages.
Dutch politicians have let go of the idea that citizenship should entail privileged access to family migration a long time ago. In 2013, the government even went a step further by proposing that Dutch citizens be granted fewer family migration rights than resident foreigners. The EU Directive on Family Reunification of 2003 limits the Dutch government's room for manoeuvre to implement further restrictions. However, the Directive applies only to third country national sponsors, not to citizens. 34 Therefore, the government proposed to raise the minimum age for family migration to 24 (instead of 21, as the Directive allows) only for Dutch citizens. This proposal met with fierce criticism from the influential Advisory Committee on Migration Affairs however and has thus far not been implemented (ACVZ 2013).
The fact that early levelling up of migrants' rights as a result of egalitarian ideology of the 1970s and 1980s paved the way to early levelling down of citizens' rights in the 1990s is one of the paradoxical effects of the progressive political past of the Netherlands. Another such paradoxical effect is the notion, stemming from the ethnic minorities policy of the 1980s, that the ethnic background of a person is more relevant for policy purposes than his or her citizenship. Citizens of migrant background remain 'allochtons' , ethnicized Others, whose family migration rights are likely to be questioned along with their membership if they engage in marriage migration practices considered problematic.

Conclusion
This paper started from the observation of a common trend in North Western Europe towards reducing the family migration rights of citizens. Our analysis of the Dutch, French and German cases confirms that there is convergence: policies are moving in the same direction. However, this convergence is constrained substantially by path dependencies. In France and Germany, where citizens have had strong family migration rights since the 1980s, there are still substantial privileges for citizens over migrants, such as the exemption from the income requirement. In the Netherlands, where equal treatment has been the norm since the 1980s, citizens have no privileged rights to family migration at all, and are subjected to the same increasingly restrictive requirements as migrants.
However, our analysis also shows that in these three countries, notwithstanding their different migration and citizenship regimes, the reduction of citizens' family migration rights is based on the same discursive mechanism. The 'membership' of citizens of migrant origin who marry a partner from abroad is called into question, which justifies reducing their citizenship rights. Thus, ethnic conceptions of the national community play a crucial role in shaping the rights attached to citizenship in Europe today. This ethno-nationalism is racialized, in the sense that it is intertwined with implicit notions about skin colour and descent. It is also gendered and classed. The citizen of migrant (Muslim) background who marries someone from abroad is regularly presented either as a female victim of an oppressive culture, or as a male oppressor. White citizens are overwhelmingly assumed to be male and middle or upper class. Their transnational marriages are not problematized. Sometimes, they are explicitly exempted from restrictive measures, as expats are in France. However, attempts at making distinctions between different categories of citizens often fail due to the tension with the fundamental liberal norm of equal treatment, as illustrated by the judicial condemnation of the German income requirement. This liberal egalitarianism explains why most often generic policies applying to all citizens are formulated even if a very particular category of citizens is targeted. Another explanation for the generic formulation of most policies is the intersection of ethnicity and class in the construction of belonging: politicians and civil servants implicitly assume that 'real' (white) citizens will have sufficient resources to fulfil new restrictive requirements.
In all three countries, the family migration rights of citizens were reduced in a context of strong politicization of migration and integration. In such a context, transnational marriages of citizens of migrant background are doubly problematized, as evidence of 'failing integration' of these citizens and as generating new inflows. It is no wonder then, that the policy impact of the questioning of the 'membership' of citizens is especially visible when it comes to citizens' family migration rights. However, in the current context of almost continuous politicization of migration and integration, the questioning of the 'membership' of certain categories of citizens is likely to affect other policy fields as well. For instance, in March 2015, the Dutch Parliament adopted a law subjecting access to basic social security to language requirements, a restriction of welfare rights which is much more likely to affect citizens of migrant background. There is an urgent need for further critical scrutiny of the ways in which the disjunction between the formal citizenry and the imagined community is affecting the rights of citizens. Notes 32. TK Plenary February 10, 2010. 33. TK 32175 (14). 34. In 2014, the Dutch Council of State ruled that the income requirements for foreign spouses of Dutch citizens should also be conform to the Family Reunification Directive (Raad van State 201400027/1/V3, judgement of December 17, 2014). Scholars interpret this ruling as implying that henceforth, the Directive generally applies also to Dutch citizens (Boeles 2014).

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the Netherlands Organisation for Scientific Research (NWO)[grant number 275-69-002].