Enter at your own risk: free movement of EU citizens in practice

Abstract The citizenship jurisprudence of the European Court of Justice has raised hopes for a more social Europe and triggered fierce debates about ‘social tourism’. The article analyses how this case law is applied by EU member state administrations and argues that they are actively containing the Court’s influence. As a result, rather than reconciling the logics of ‘opening’ and ‘closure’, they are heading towards an uneasy coexistence between free movement and exclusive welfare states. The argument here is illustrated with empirical evidence from Austria and Germany. Although both countries have taken different approaches to EU migrants’ residency and social rights, they produce similar effects in practice: increasingly, EU migrants are being tolerated as residents with precarious status without access to minimum subsistence benefits. Ironically, attempts to restrict residency rights have resulted in a temporary extension of EU migrants’ access to welfare in some instances.

description, the 'logics of closure' and 'opening' are coming into conflict (Ferrera 2005: 2; see also Martinsen and Vollard 2014). On the one hand, social solidarity in Europe has typically evolved within the boundaries of national welfare states. On the other hand, European integration transcends and undermines national boundaries in many respects, not least by enabling the free movement of EU citizens. Against this backdrop, Ferrera (2005: 2) asked whether EU member states would find ways to reconcile these contradictory logics and to reconstruct solidarity within Europe.
Our article takes up this question and focuses on EU member state administrations: (how) are free movement and welfare rights reconciled in practice? Rather than reconciling opening and closure, we argue, EU member state administrations are heading toward a situation in which there is an uneasy coexistence between these two logics. EU citizens are de facto allowed to move and reside ever more freely in Europe. However, access to social protection, which had begun to slowly open up due to the case law of the European Court of Justice (ECJ), is being restricted again, mostly along traditional national boundaries, and is only granted to workers from other EU countries. Economically inactive EU migrants and jobseekers 'enter at their own risk' .
Our analysis adds to existing studies in at least two important respects. Firstly, we develop a simple framework to systematise and analyse recent developments at the intersection between EU free movement and national welfare states. Our conceptualisation of four combinations of opening and closure allows for the capturing of both the Court's recent shift regarding EU citizenship and the divergent practices of member state administrations. Secondly, by tracing the application of ECJ citizenship jurisprudence by national administrations, we add to the emerging literature on Europeanisation through case law, which has mainly focused on domestic legislative responses so far (Blauberger 2014;Treib 2014: 13). The Court's case law has raised both hopes (for 'social citizenship') and fears (of 'social tourism'), but member state administrations have been able to limit its domestic influence. Rather than simply ignoring case law, however, they actively 'contained compliance' (Conant 2002) and, in doing so, struggled with significant unintended consequences.
The next section presents our analytical framework. We briefly revisit Ferrera's abstract concepts of closure and opening and spell out in greater detail four conceptions of closure and opening, as we find them in the evolving ECJ citizenship jurisprudence and its divergent application by member state administrations. We then turn to the literature on Europeanisation through case law to develop our own theoretical argument about (re)active containment. The second main section presents empirical evidence from the Austrian and German case studies. As we will show, the two countries have taken very different approaches de jure to administering the free movement of EU citizens, but these different approaches have led to similar outcomes in practice. Essentially, EU migrants enter Austria and Germany at their own risk -in other words, their residence is not questioned, even if they do not work and do not have sufficient independent resources, but, at the same time, they cannot count on social assistance from their host state. Ironically, attempts to overcome this tension between opening and closure by restricting residency rights have resulted in a temporary extension of access to welfare in some instances. The final section addresses the generalisability of our findings and returns to the initial conflict between German social courts.

Analytical framework: closure versus opening
To describe the difficult relationship between national welfare states and European integration, Maurizio Ferrera (2005: 2) coined the concepts of opening and closure: Social sharing builds on 'closure' . … Since the nineteenth century (in some countries since much earlier) the nation state has provided the closure conditions for the development of sharing sentiments and practices within its own territory. European integration, on the contrary, rests on 'opening': on weakening or tearing apart those spatial demarcations and closure practices that nation states have built to protect themselves. EU law includes both elements but tells us little about their mutual relation. Most important in terms of opening, the free movement of workers and equal treatment are enshrined in Article 45 of the Treaty on the Functioning of the European Union (TFEU). At the same time, EU treaty law also expresses respect for 'closed' national welfare states. According to Article 153(4) TFEU, the EU's competence in social policy 'shall not affect the right of Member States to define the fundamental principles of their social security systems and … shall not prevent any Member State from maintaining or introducing more stringent protective measures compatible with the Treaties' . In some treaty provisions, opening and closure intersect: whereas EU migrant workers enjoy full equal treatment, including access to welfare systems in the country of employment, economically inactive EU citizens' right to move and reside freely and their protection against discrimination is 'subject to the limitations and conditions laid down in the Treaties and by the measures adopted to give them effect' (Article 21 TFEU).
Both logics are reflected in EU secondary legislation as well, but, again, they are treated separately for the most part. On the one hand, Regulation (EC) 883/2004 does not aim to harmonise or even replace national welfare systems but only coordinates EU member states' social security systems. On the other hand, Directive 2004/38/EC regulates the right of EU citizens and their family members to move and reside freely within the EU. All EU citizens with a valid passport can reside in another EU member state for three months and for longer -even if they are economically inactive -as long as they have 'sufficient resources … not to become a burden on the social assistance system of the host Member State' (Article 7(1b) Directive 2004/38/EC). The reference to being a 'burden on the social assistance system' links residence rights and social welfare. Yet even though Regulation (EC) 883/2004 and Directive 2004/38/EC were adopted on the same day, the relationship between social security and residence was not systematically addressed in the legislative process (Couchier et al. 2008: 3). 'The problem is, however' , Ferrera (2005: 2) wrote, 'that the "separate tracks" solution -that is, the insulation of national social protection systems from the dynamic of economic integration and from supranational interference -long ago ceased to be viable' .

Opening and closure in practice: four conceptions
How, then, are conflicts between free movement in the EU and exclusive welfare systems, or between opening and closure, resolved in practice? Often, these politically sensitive conflicts are left to supposedly non-political actors: judges and bureaucrats. Whether they like it or not, courts and administrations have to reach decisions on these conflicts by interpreting EU free movement and welfare rights. Basically, their choice is between four -admittedly broadly conceived -options (see Table 1). As we will show below, the two main phases of ECJ citizenship jurisprudence largely fit into two of these categories ('solidarity reconstructed' , 'return to the roots'), but administrations apply case law differently, and their practice matches two other conceptions ('entry at own risk' , 'generosity by accident').

Solidarity reconstructed
An expansive interpretation not only of EU free movement provisions, but to a certain extent also an opening up of welfare systems to EU migrants, is probably what Ferrera (2005: 8) had in mind when he referred to the 'restructuring of social sharing … at the EU level' . The ECJ has long been perceived as pioneering in this expansive direction by gradually developing its doctrine of EU 'social citizenship' (Wollenschläger 2011). For example, in Grzelczyk (C-184/99, No. 44), the Court held that EU member states had to show 'a certain degree of financial solidarity' between their nationals. Building on this expansive interpretation, the Commission has long been critical of national welfare restrictions such as the British 'right to reside' test. According to the Commission, EU citizens habitually residing abroad should have been able to draw on social benefits in

Return to the roots
By contrast, the Court's more recent case law appears to return to the roots of European integration. Originally, and for the first decades of the integration process, free movement was interpreted mainly as the right of workers and the self-employed. Even though Directive 2004/38/EC regulates the freedom of movement for all EU citizens, it still allows for restrictions: economically inactive EU citizens without sufficient resources may not only be excluded from welfare benefits abroad, but they may also be denied the right to reside abroad for more than three months in the first place (if they pose an 'unreasonable burden' on the social assistance system). The Court drew on these provisions to develop a more restrictive interpretation of both residence and social rights in a series of recent judgments (C-333/13 Dano; C-67/14 Alimanovic; C-299/14 Garcia-Nieto; and C-308/14 Commission v. United Kingdom). Legal scholars have been critical of the Court, warning against 'throw[ing] us back to the era before the introduction of European citizenship' (Verschueren 2015: 388) and identifying a 'reactionary phase' of ECJ jurisprudence (Spaventa Forthcoming).
In the context of the failed 're-negotiation deal' with the UK, which provided for discriminatory access to in-work benefits, some even warned against 'regressing … to pre-Rome' (O'Brien 2016: 938).

Entry at own risk
Apart from the aforementioned coherently expansive and restrictive interpretations of both residence and welfare rights by the ECJ, we find a more uncomfortable coexistence of opening and closure in domestic administrative practice. On the one hand, free movement is interpreted expansively, i.e. EU migrants' right of residence is not questioned or even revoked, even if migrants do not work and do not have sufficient independent resources to establish legal residence. On the other hand, welfare rights are interpreted restrictively, and economically inactive EU migrants are excluded from social benefits. Our case studies below will show this pattern, which we label entry at own risk, in greater detail.

Generosity by accident
Finally, a counter-intuitive situation may emerge if restrictive residence rules are supposed to block access simultaneously to national welfare systems, but these residence rules are not enforced properly by immigration authorities. Unintentionally, such an ineffective closure of residence rights may result in a relative opening up of welfare systems to newly arrived migrants. As long as restrictive rules for residency are not enforced effectively and no specific access barriers targeting migrants are introduced in national welfare provisions, the result is one of generosity by accident. Ironically, some of the most generous instances in terms of the access EU migrants gain to social benefits, which we discuss in detail in our case studies, have resulted precisely from EU member states' failed attempts to enforce restrictive residence policies. Hence, at least temporarily, welfare rights have been accidentally expanded to economically inactive migrants or jobseekers.
To sum up, the ECJ has produced two series of judgments, which stand in stark contrast to one another but are quite consistent when taken alone. Previously, expansive citizenship jurisprudence promoted not only free movement but also a certain opening up of welfare states (solidarity reconstructed). Recently, restrictive case law has reconfirmed the closure of national welfare systems and limited the free movement of poor, economically inactive EU citizens (return to the roots). By contrast, mixed combinations of closure and opening have actually prevailed in administrative practice through entry at own risk and generosity by accident. Before we trace these mixed responses in our case studies, we discuss possible explanations for the limited impact of ECJ case law on national administrations.

Explaining differential Europeanisation through case law
Research on 'integration through law' often assumes that the Court triggers policy changes without actually studying them (Conant 2002: 15). This neglect of real Court influence is also common in many accounts of ECJ citizenship jurisprudence. Despite little empirical evidence, great hopes and fears have been associated with this particular strand of case law. On the one hand, Caporaso and Tarrow (2009: 600) welcomed the Court's earlier citizenship jurisprudence as a contribution to the 're-embedding of social regulation at the supranational level' as a complement to European market liberalisation. On the other hand, Höpner and Schäfer (2012: 448) warned that 'the possibility that this line of ECJ case law will trigger welfare state retrenchment is at least as plausible as a perspective that interprets it as the nucleus of an emerging European welfare state' .
In contrast to these far-reaching expectations, we begin on a more cautious note that is broadly shared in the emerging literature on 'Europeanisation through law' (Treib 2014: 13). We argue that, faced with expansive ECJ jurisprudence, EU member states will -often successfully -try to limit the Court's influence at the implementation stage. Most famously, Lisa Conant (2002) asserted that EU member states contain compliance by applying ECJ rulings only to the individual conflict but otherwise remaining passive. As a consequence of contained compliance, ECJ case law often does not exert much influence beyond an individual case. In addition, Conant identified another political response with the same goal of limiting the ECJ's domestic influence. 'Pre-emption' refers to situations in which member states 'carefully construct European or domestic law to avoid future judicial interference in particular areas' (Conant 2002: 32).
We share the expectation that member states try to limit the impact of the ECJ but argue that the underlying process is more complex in at least two respects. Firstly, containing compliance by simply remaining passive is easier to sustain for some domestic actors than for others. For example, when confronted with an unwelcome ruling of the ECJ, national legislatures may opt for a 'wait and see' approach (Slagter 2009), and yet national administrations can be confronted with an increasing number of individual claims based on directly effective ECJ jurisprudence. Under these circumstances, active containment may become necessary for member state administrations. Moreover, the costs of actively containing compliance, e.g. in terms of workload or expenditure, may be distributed unequally even within the context of a single member state administration.
In our field of interest, the ECJ's citizenship jurisprudence has created differential incentives for national immigration and welfare authorities. On the one hand, national immigration authorities have little incentive to actively challenge the residence rights of EU citizens, even where ECJ case law explicitly allows them to do so. Controlling legal residence is costly in terms of the administrative workload and, apart from being politically sensitive, expelling EU migrants without legal residence would place a significant financial burden on immigration authorities. EU migrants can delay expulsions through litigation and return immediately after an expulsion anyway, since EU law only permits re-entry bans under exceptional circumstances. Thus, rather than actively challenging the residence rights of EU migrants, national immigration authorities often prefer to remain passive. On the other hand, national welfare authorities have a strong incentive to actively contain compliance with expansive ECJ rulings and to reject the benefit claims of economically inactive EU migrants in order to reduce welfare expenditure. Moreover, to avoid these kinds of benefit claims in the first place, and thus to reduce their administrative burden, welfare authorities prefer to establish general barriers to welfare access, and recent, more restrictive ECJ jurisprudence partly enables them to do so. As a result, welfare authorities are actively seeking ways to contain the Court's influence on welfare access, and this challenge has even been exacerbated by the passivity of immigration authorities.
Secondly, containing compliance with ECJ case law may require complex domestic adjustments for another reason. Given the great legal ambiguity inherent in much ECJ case law, we doubt that EU member states will always be successful in carefully constructing legislation in order to 'anticipate' (Blauberger 2014) and 'pre-empt … future judicial intervention' (Conant 2002: 35). Legal ambiguity not only creates loopholes for creatively containing compliance but also opportunities for potential litigants (Schmidt 2008). And where domestic legislation is explicitly designed to minimise the influence of ECJ case law, it may produce unintended consequences at the implementation stage itself. Hence, domestic efforts to contain compliance with ECJ case law may, at least temporarily, prove unsuccessful or even counter-productive and take on a reactive rather than proactive character. In our case studies, we will support this argument by the recurrent observance of generosity by accident, that is to say, member states' ineffective application of restrictive residence rules resulting in a temporary opening up of welfare access. Even though member states were ultimately able to overcome the unintended opening up of welfare, the examples shown here demonstrate that containing compliance involves active and complex adjustments for national administrations.
In sum, we add to a more differentiated understanding of contained compliance and of Europeanisation through law more generally. As the example of ECJ citizenship jurisprudence will show, the Court has had a considerable impact in terms of administrative adjustments but only limited effect on outcomes in terms of access to welfare.

Empirical analysis
We compare evidence from Austria and Germany indicating how EU migrants' rights of residence and their access to social benefits are administered in practice. These countries were chosen because, on paper, their approaches in relation to EU migrants' residence rights represent two extremes of a range of options. Directive 2004/38/EC defines the framework conditions under which EU migrants are allowed to reside abroad but leaves it to the host member state to examine this right of residence, i.e. whether it wants to introduce a registration certificate, what kind of evidence citizens have to provide and whether there is a penalty for non-registration: 'For periods of residence of longer than three months, Member States should have the possibility to require Union citizens to register with the competent authorities in the place of residence, attested by a registration certificate issued to that effect' (recital 12 of Directive 2004/38/EC).
EU member states have implemented this provision quite differently (see Table 2). While some countries (including Germany) do not request EU migrants to register at all, others (including Austria) make use of the option to control free movement rather strictly, for example by requiring comprehensive evidence (rather than just self-declarations) of sufficient resources and penalising failure to supply this.
In order to examine how free movement and welfare rights are applied in practice in the two member states, we analysed the documents and websites of the competent authorities as well as parliamentary requests. Moreover, we conducted 52 semi-structured, background interviews, mainly with case workers and civil servants from national welfare and immigration authorities but also at the ministerial level, including rights advocacy groups in the field (see Table 3).

Austria
In Austria, Directive 2004/38/EC was transposed restrictively through the Residence and Settlement Act (Niederlassungs-und Aufenthaltsgesetz, NAG) and the Aliens Police Act (Fremdenpolizeigesetz, FPG), which both came into force in 2006. Means-tested minimum benefits (Bedarfsorientierte Mindestsicherung, BMS) are regulated at the regional level but partly harmonised by a framework agreement between the federal and regional levels (BMS-Vereinbarung). Despite its hybrid character, the Mindestsicherung is considered to be social assistance and not a measure to facilitate labour market access (Windisch-Graetz 2014). As we will show, Austrian administrative practice largely falls into our category of entry at own risk: restrictive residence rules are scarcely enforced in practice, but they serve to exclude all EU migrants other than workers or the self-employed from access to social benefits (with some exceptions of generosity by accident as an unintended consequence of Austrian residence regulation).
All people, regardless of their nationality, have to report their presence to the local authority when taking up residence in Austria. In addition, Austria has taken advantage of the option laid down in Article 8 of Directive 2004/38/EC and introduced a registration certificate (Anmeldebescheinigung). According to §53 of the NAG, EU migrants who want to stay for longer than three months in Austria have to get their right of residence certified within the first four months of their stay.  They have to apply for the registration certificate with the competent local immigration authority. In order to receive this certificate, EU citizens have to fulfil the conditions laid down in §51 of the NAG (which transposes Article 7 of Directive 2004/38/EC into national law). They need to prove that they are either workers or self-employed, or that they have sufficient resources and comprehensive health insurance for themselves and their family members so that they do not have to draw on social assistance benefits or the supplementary pension during their period of residence. De jure, Austria interprets free movement in a restrictive way. What initially appear to be clear rules in theory involve many uncertainties in practice: When should someone be considered a worker and when self-employed? And what resources are sufficient? While the EU's legal concepts of workers and the self-employed are already vague and not always easy to assess (Common Market Law Review editorial 2014: 733), the assessment of 'sufficient resources' is even fiddlier: Member States may not lay down a fixed amount which they regard as 'sufficient resources' , but they must take into account the personal situation of the person concerned. In all cases this amount shall not be higher than the threshold below which nationals of the host Member State become eligible for social assistance, or, where this criterion is not applicable, higher than the minimum social security pension paid by the host Member State. (Article 8 Directive 2004/38/EC) How are sufficient resources interpreted by the Austrian authorities? In line with EU law, Austrian legislation does not define sufficient resources in terms of a fixed amount. In practice, authorities usually work with thresholds equivalent to the basic levels of minimum benefits or supplementary pension (basic levels in 2016: minimum benefits €837.76; supplementary pension €882.78). One practical difficulty is the recognition of savings deposits. According to several interview partners, short-term 'travelling deposits' (Wanderkonten) were used in the past to meet the sufficient resources criterion in a potentially fraudulent way. Today, most authorities do not include savings deposits in their assessment or they argue that, in principle, savings would have to be high enough to meet living expenses from capital income. Finally, depending on the local authority's standards, sufficient resources are often equated with being employed. As a consequence, newly arrived jobseekers do not receive the registration certificate until they find employment. One interview partner from a local immigration authority even argued that EU law was unnecessarily complex in allowing for the legal residence of workers or citizens with sufficient resources, since both criteria were treated as being largely synonymous in practice.
But what consequences do EU migrants face when they are not issued with the registration certificate, i.e. migrants who are not given the right of residence for more than three months? De facto, there are hardly any consequences. EU migrants are told to consult the immigration authorities again as soon as they meet the criteria for lawful residence. Sometimes they are also told to leave the country voluntarily, but they usually do not receive expulsion orders. Interview partners from Austrian immigration authorities often justified their practice by pointing out that the mere presence of these people would not 'bother' anyone since they created no costs. As Table 4 shows, on average, only 125 EU migrants per year were expelled from Austria between 2008 and 2013. Moreover, while Austrian statistics do not list the different reasons for expulsion, several interview partners confirmed that a lack of sufficient resources was rarely the reason. On the basis of §66 FPG, EU citizens can be expelled on grounds of public order or security, because they no longer fulfil the conditions necessary for right of residence or because they do not provide sufficient documentation. The rationale of public order or security is also used to justify the much larger number of prohibitions on residence ( §67 FPG). Rather than being expelled, typically EU migrants without sufficient resources 'only' have to live with the threat of being expelled (Bundesministerium für Arbeit 2015: 13-14).
In sum, Austrian immigration authorities favour opening over closure with regard to residence. Welfare systems, however, remain closed, as EU migrants without registration certificates are denied access to minimum subsistence benefits in situations of hardship.
Austrian social legislation enables welfare authorities to largely exclude EU citizens from minimum benefits. EU citizens who are not workers and reside legally (but not yet permanently) in Austria are only eligible 'as long as they would not lose their right of residence by claiming those benefits' (Article 4 (3) BMS-Vereinbarung). Even if the latter provision is not enforced in practice, EU migrants bear the burden of uncertainty (Blauberger and Schmidt 2014). They are warned that by applying for the Mindestsicherung, they may risk their right of residence. 4 In practice, EU citizens have to prove their right of residence by presenting their registration certificate in most Austrian regions as a precondition for claiming minimum benefits. 5 Although the registration certificate has merely declaratory value in EU legal terms, it is often decisive in answering the question as to whether a person can claim benefits or not. According to Austrian authorities, cases such as Dano (C-333/13) in Germany could not occur in Austria, since the certificate is demanded as a precondition for receiving the Mindestsicherung.
In sum, we observed a mixture of opening and closure: de facto, the presence of EU migrants without 'sufficient resources' is tolerated by Austrian immigration authorities, but they do not receive the registration certificate and consequently are denied access to minimum benefits. EU citizens enter Austria at their own risk. From the point of its introduction, the registration certificate was designed as a means to keep EU migrants excluded from access to welfare benefits. The former Minister of Labour and Social Affairs, Rudolf Hundstorfer, commented that 'Europe could have learned from Austria. … Austria already exploited the opportunities offered by EU law to clearly restrict access to the welfare system some time ago' . 6 Nevertheless, in some instances administrative practice is more generous than originally intended. Austrian welfare authorities often neglect the declaratory value of the registration certificate and -partly consciously, partly unconsciously -consider it to be constitutive, i.e. they treat the registration certificate as being de facto 'valid' even if the conditions for its issuing are no longer fulfilled. On this basis, benefits are granted. For example, EU migrants could possibly work for just three months to meet the conditions necessary for receiving the registration certificate, and have recourse to benefits when losing their job afterwards. The respective welfare authorities sometimes inform immigration authorities about these changes, but there are no consequences. Immigration authorities point to the fact that they often could not expel people due to Article 8 of the European Convention on Human Rights, and expulsions of EU citizens without a re-entry ban are ineffective anyway. In practice, the registration certificate is neither reviewed nor revoked. As a result, EU citizens in possession of a registration certificate may be able to claim benefits even though they no longer fulfil the conditions for the registration certificate. Austria is generous by accident.
With regard to the Austrian supplementary pension (Ausgleichszulage), a constitutive interpretation of the registration certificate was even advanced by the Supreme Court of Justice (OGH). When reacting to Brey (C-140/12), the OGH decided that EU citizens otherwise meeting the outlined criteria were entitled to receive the supplementary pension, as long as immigration authorities did not withdraw the registration certificate. 7 In contrast to immigration authorities, which remained inactive, pension authorities (Pensionsversicherungsanstalten) were not allowed to assess the right of residence of the claimant themselves.
After Dano (C-333/13), the OGH changed its opinion -the registration certificate only documents right of residence, but it does not automatically entitle the individual to social benefits. As soon as EU migrants apply for the supplementary pension, their right of residence may be reviewed. Economically inactive EU migrants exercising their right of free movement solely in order to receive social assistance cannot claim benefits, such as the supplementary pension, on the basis of EU law. 8 Despite the shortcomings of the registration certificate, there are currently no plans to reform the Austrian Residence and Settlement Act. The registration certificate is still generally perceived positively, serving a certain controlling function and sending a signal of deterrence to potential 'welfare migrants' . Rather than reforming the registration certificate and its questionable enforcement record, therefore, the executive branch is planning to build more specific barriers to access into its social regulations. The Federal Ministry of Labour, Social Affairs and Consumer Protection and the regions propose to amend the BMS-Vereinbarung in order to explicitly limit the eligibility of EU migrants to workers or self-employed. This change would reinforce the trend towards restrictive welfare states without real restrictions of free movement, i.e. EU citizens' 'entry at own risk' . And it would make Austrian social regulation quite similar to that in Germany, to which we now turn.

Germany
In Germany, EU citizens' right of residence is regulated liberally (see Table 2) by the Law on Freedom of Movement of Union citizens (Freizügigkeitsgesetz/ EU), adopted in 2004. Directive 2004/38/EC was already known at that time, but parts of it were only implemented by later reforms, for instance in 2007 (Brinkmann 2014). German legislation (Sozialgesetzbuch II, SGB II) on minimum subsistence benefits (Grundsicherung für Arbeitsuchende, also Arbeitslosengeld II or simply Hartz IV) is restrictive vis-à-vis EU citizens and generally excludes economically inactive EU migrants and jobseekers. Hence, a mismatch between largely unrestricted free movement and restricted access to welfare benefits is already present at the legislative level. In terms of administrative practice, we will show that the results are very similar to the Austrian case: EU citizens enter Germany at their own risk, with some instances of politically unintended generosity, mainly due to the jurisprudence of German social courts.
Originally, the German legislation on EU citizens' right of residence also introduced a registration certificate (Freizügigkeitsbescheinigung) that was issued by immigration authorities. In 2013, however, Germany abolished this certificate, arguing that this would reduce the administrative burden 9 and that the previous registration procedure had not involved any substantive controls anyway (Thym 2014). As a consequence, the right of residence no longer needs to be 'declared' but simply exists if EU migrants meet the criteria of §2 (in conjunction with §4) of the Freizügigkeitsgesetz/EU. Workers and self-employed migrants enjoy the right of residence in any case; the right of residence of economically inactive migrants for more than three months is conditional upon having sufficient resources and health insurance. In contrast to Austria, however, EU migrants -as is the case for everyone else -only have to report their presence with the municipal registration office in order to meet the general reporting obligations in Germany. They do not have to prove that they fulfil the preconditions mentioned above. Moreover, since there is no registration requirement, there is also no potential penalty. In practice, national immigration authorities stay passive and effectively favour opening over closure, since they are not required to control EU migrants in the first place. With regard to jobseekers, the legislator somewhat tightened the rules in 2014. The reform sought 'to implement EU law as interpreted by the ECJ' 10 and limited jobseekers' right of residence to six months -or more, if they can provide evidence that they are continuing to seek employment and that they have a genuine chance of being engaged ( §2 (2) 1a Freizügigkeitsgesetz/EU).
If EU citizens do not or no longer fulfil the requirements for legal residence, according to §5 of the Freizügigkeitsgesetz/EU, local immigration authorities may state that they have lost the right of residence (Verlustfeststellung). This rarely happens in practice. The competent institutions do not withdraw EU migrants' rights of residence; they neither review jobseekers' 'chances of being engaged' nor impose consequences if people no longer meet the remaining criteria mentioned in §2 of the Freizügigkeitsgesetz/EU. First and foremost, immigration authorities are overloaded with work; even if they request that EU citizens should leave the country, the latter can contest the decision, which has the effect of suspending the order. Moreover, courts are overwhelmed as well. Mostly, Union citizens are expelled for reasons of public order, security or health.
As Table 5 shows, German authorities declared the loss of right of residence for an average of 485 EU migrants per year between 2009 and 2013 for those not or no longer meeting the criteria for right of residence ( §5(4) Freizügigkeitsgesetz/EU). Again, despite their increase, these numbers are small in both absolute terms and relative to other grounds on which individuals lose the right of residence in Germany, including fraud ( §2(7) Freizügigkeitsgesetz/ EU) and public order, security or health ( §6(1) Freizügigkeitsgesetz/EU). To sum up, the logic of opening clearly prevails in German residence legislation and administrative practice. Free movement is de facto unrestricted.
Contrary to its expansive interpretation of free movement, Germany handles EU migrants' access to welfare rights restrictively. Welfare authorities systematically deny access to social benefits: both jobseekers and economically inactive Union citizens are not eligible for minimum subsistence benefits. According to §7 SGB II, foreigners whose right of residence solely arises from the search for employment are generally excluded. In addition, §23 of the Social Code XII (SGB XII) excludes foreigners who enter the national territory solely in order to obtain social assistance. In practice, it is the local welfare authority that judges whether Union citizens meet the criteria necessary to claim the relevant benefit. It assesses their right of residence and generally does not grant benefits to people considered to be jobseekers or economically inactive. Since its introduction, the compatibility of these exclusions, especially those that exclude jobseekers, with EU law had been disputed. One important question was whether Hartz IV had to be defined as social assistance or a measure facilitating labour market access. This distinction is of particular importance since member states cannot exclude jobseekers from benefits that are intended to facilitate access to employment, whereas they may well do so in the case of benefits classified as social assistance (C-138/02 Collins; and C-22/08 and C-23/08 Vatsouras and Koupatantze). Hartz IV has both elements: on the one hand, its aim is to enable its beneficiaries to lead a life in human dignity ( §1 (1) SGB II); on the other hand, it seeks to provide support for access to the labour market ( §1 (2) SGB II), since beneficiaries must be available for work ( §2 SGB II). Whereas the ECJ first left it to the member states to decide upon the definition of the benefits at issue (C-22/08 and C-23/08 Vatsouras and Koupatantze), it finally confirmed the German point of view in its recent judgments in Dano (C-333/13), Alimanovic (C-67/14) and Garcia-Nieto (C-299/14). The Court classified Hartz IV as social assistance (within the meaning of Article 24(2) of Directive 2004/38/EC) since 'the predominant function of the benefits at issue in the main proceedings is in fact to cover the minimum subsistence costs necessary to lead a life in keeping with human dignity' (C-67/14 Alimanovic, para. 45).
In sum, the ECJ considers German practices and its restrictive interpretation with regard to welfare rights compatible with EU law. Here again, the welfare state remains exclusive. De facto, EU migrants' presence within the national territory is at no time under review, which means that they are tolerated but do not have access to social benefits. As mentioned in our introduction, Germany's Federal Social Court addressed this precarious situation with regard to jobseekers in its judgment in December 2015, thereby forcing politicians to deal with the conflict between free movement and exclusive welfare. The court highlighted the 'enforcement deficit of aliens law' and ruled that the state has the responsibility to ensure that anyone with a 'solidified' residence in Germany can lead a life in human dignity. After residing in Germany for six months, the Court held, Union citizens should become eligible for social assistance according to Social Code XII, Hilfe zum Lebensunterhalt.
Again, as in Austria, Germany's failed attempt to impose restrictive residence policies resulted in unintended welfare access -in this case regarding jobseekers. In 2014, the German legislative reform limited jobseekers' right of residence to six months. Unintentionally, these six months have become the threshold for a better residence status -a 'solidified' stay -which is linked to social assistance according to Social Code XII. The Federal Social Court's judgment thus reinterpreted and transformed the restriction of residence rights into a partial opening up of the welfare system. Before the judgment, jobseekers had an incentive not to become visible to German authorities at all in order to preserve their right of residence as jobseekers for as long as possible. After the judgment, EU jobseekers may conversely want to prove that they have already stayed in Germany for more than six months in order to become eligible for social benefits. By accident, welfare access for EU jobseekers has become more generous in Germany than originally intended.
As already mentioned in the introduction, the judgment of the Federal Social Court did not arrive without controversy. Diverse regional social courts and local social courts opposed it, which is extremely unusual given the hierarchy of courts in Germany. 11 The Federal Social Court triggered not only opposing decisions by lower courts but also a storm of protest from municipalities and districts, 12 first and foremost since the benefits accorded by the Federal Social Court are financed by them, in contrast to benefits accorded by SGB II. Moreover, the judgment created a new situation and new function for the local social offices dealing with benefits of SGB XII. In contrast to benefits granted by the SGB II, which are administered by the Jobcenter, the benefits of the SGB XII are intended for people who are unfit to work. Consequently, local social offices that had never been in charge of employment services before were given new responsibilities. Whereas some municipalities and districts proactively tried to develop strategies to deal with this new situation, others announced that they would reinforce expulsions, 13 and others simply did not implement the judgment of the Federal Social Court and waited for a reaction from the legislator.

Conclusion: a stable closure of welfare systems?
The relationship between free movement in the EU and nationally circumscribed welfare systems, or between opening and closure, has always been difficult and politically sensitive. Despite their political salience, however, EU leaders have left it to judges and bureaucrats to resolve many conflicts in this area. The ECJ has produced two opposing series of case law; the first of these promoted an expansive interpretation of free movement and a partial opening up of national welfare systems for EU citizens, and, more recently, the second series reconfirmed restricted access to national welfare states and questioned the free movement of economically inactive EU citizens without sufficient resources. National administrations, for their part, face many difficulties in applying the Court's citizenship jurisprudence consistently and have produced a rather mixed outcome: EU migrants enter other member states at their own risk. In other words, insufficient resources are not de facto an obstacle to free movement but only to welfare access. As we have shown with empirical evidence from Austria and Germany, administrative practices produce similar results even though, on paper, member states may have quite different legislative approaches.
Our analysis confirms a major argument from the literature on Europeanisation through case law -namely, that EU member states try to contain the influence of ECJ jurisprudence -while adding to a more differentiated understanding of contained compliance. Instead of staying passive, domestic welfare administrations have had to make significant adjustments in order to contain actively the influence of ECJ citizenship case law on welfare access. Moreover, domestic efforts to contain compliance have had unintended consequences themselves and have sometimes appeared to be a process of trial and error rather than careful anticipation and pre-emption. Ultimately, the Court's influence on actual outcomes in terms of welfare access was limited and temporary, but its impact on administrative practices was significant. While we expect these findings to hold across other EU member states as well, there are limits to generalising from our Austrian and German cases. For example, containing compliance might be less demanding for countries in which national courts are more reluctant to send references up to the ECJ and hence exert less pressure on national administrations (Wind et al. 2009: 63).
In practical terms, EU member states seem to be moving towards the stable closure of national welfare systems, facilitated by recent ECJ case law and reinforced by the crisis of the EU (Ferrera 2014). It is important to emphasise, however, that this development differs from a simple return to the roots of free movement of workers and the coordination of social security systems. The practice of entry at own risk is producing a new underclass of mobile EU citizens who do not enjoy the status of workers -they are jobseekers and economically inactive EU migrants, who are denied equal treatment with respect to social benefits but whose residence is de facto tolerated, even if it means they face potentially precarious situations. Essentially, this is the challenge raised by the German Federal Social Court's judgment presented in our introduction: Can EU member states shut their eyes to this development? One year after the judgment, the German legislator has answered 'yes and no' by revising social legislation. According to the new rules, 14 EU citizens who are not workers or permanent residents are generally excluded from both SGB II and SGB XII benefits, but a (very limited) social responsibility of the German state to those living on its territory is acknowledged: EU migrants without sufficient resources, who agree to return voluntarily to their country of origin, can benefit from short-term support for up to one month and a loan for their return ticket.

Notes
Michael Blauberger is Associate Professor in the Department of Political Science/ Salzburg Centre of European Union Studies (SCEUS) at the University of Salzburg. He has published widely on European integration and Europeanisation through case law. His current projects focus on alleged 'welfare migration' in the EU and on democratic backsliding in EU member states. [michael.blauberger@sbg.ac.at]