ECJ Judges read the morning papers. Explaining the turnaround of European citizenship jurisprudence

ABSTRACT

Recent jurisprudence of the European Court of Justice (ECJ) marks a striking shift towards a more restrictive interpretation of EU citizens’ rights. The Court's turnaround is not only highly relevant for practical debates about ‘Social Europe’ or ‘welfare migration’, but also enlightening from a more general, theoretical viewpoint. Several recent studies on the ECJ have argued that the Court is largely constrained by member state governments’ threats of legislative override and non-compliance. We show that an additional mechanism is necessary to explain the Court's turnaround on citizenship. While the ECJ extended EU citizens’ rights even against strong opposition by member state governments, its recent shift reflects changes in the broader political context, i.e., the politicization of free movement in the European Union (EU). The article theorises Court responsiveness to politicization and demonstrates empirically, how the Court's jurisprudence corresponds with changing public debates about EU citizenship.

KEYWORDS:

• EU citizenship
• European Court of Justice
• free movement
• judicial independence
• judicial responsiveness
• politicization

Introduction

The judgment of the ECJ on the Dano case (C-333/13) had been long awaited, but the outcome took many by surprise. The ruling addresses the question, whether an unemployed Romanian citizen residing in Germany can claim German social benefits and ‘the CJEU just says no’ (Peers 2014). Taken in isolation, an unexpected ruling does not pose much of a puzzle: if rulings were predictable, we would need no courts in the first place. Read together with a series of follow-up cases, however, the Dano judgment marks a striking break in the ECJ's – previously expansive – interpretation of EU citizens’ rights to receive social benefits abroad.

From a political science perspective, this conclusion of the Court-driven extension of cross-border social rights is highly interesting for both theoretical and practical reasons. Theoretically, most analysts of European ‘integration through law’ have come to agree that the ECJ possesses some sort of bounded independence (Larsson and Naurin 2016: 2, 27) from EU member state influence – but the precise boundaries and the potential mechanisms of political influence remain contested (Stone Sweet and Brunell 2012). The Court's changing interpretation of EU citizenship offers a unique opportunity to compare the explanatory power of different – political as well as legal – accounts. We offer a complementary conception of the ECJ's continuing role in shaping the European integration process. Practically, the Court's jurisprudence on EU citizens’ free movement and access to social benefits is highly salient and has been described as ‘the most ambitious and tantalizing line of case law in recent memory’ (Thym 2014). On the one hand, the Court was initially perceived as pioneering a more ‘social Europe’ by extending the social rights of EU citizens beyond national boundaries (Caporaso and Tarrow 2009). On the other hand, this cross-border extension of social rights has triggered fierce debates at the national level about alleged ‘welfare migration’ or ‘benefit tourism’, which culminated in the run-up to the Brexit referendum.

By way of a plausibility probe, we argue that the Court reoriented its citizenship case law in response to these changes in the broader political context. As our analysis shows, over the last decade, the public debate about EU citizenship has dramatically gained in salience and has become mostly associated with negative frames, e.g., regarding potential ‘welfare abuses’. In response, European judges deviated from their earlier case law towards a more restrictive approach, which is puzzling in terms of pure legal doctrine. The two most prominent political explanations of ECJ jurisprudence focus on EU governments’ threats against unwarranted judicial activism – either the threat of non-compliance (Carrubba and Gabel 2015) or of legislative override (Larsson and Naurin 2016). We find plenty of evidence for both threats, but they do not correlate with the evolution of the Court's case law on EU citizenship. EU governments signalled opposition to the Court's interpretation of social citizenship from the very beginning and the adoption of the Directive 2004/38 (hereafter referred to as Citizenship Directive) in 2004 can be regarded, at least, as an attempt to ‘modify’ (Martinsen 2015: 36) the Court's jurisprudence in a restrictive way. And yet, political threats alone did not prevent the Court from further expanding its interpretation of social citizenship until recently, when public opinion strongly sided with the position of adverse governments.

The next section briefly sets out our puzzle: the Court's turnaround from an expansive to a restrictive interpretation of EU ‘social citizenship’. In the theoretical section, we briefly review the literature on the bounded independence of the ECJ and discuss different mechanisms of political influence. In the empirical section, we demonstrate the added value of our explanation, which focuses on the broader public debate and its reflection in the Court's reasoning.

The Court's turnaround on EU ‘social citizenship’

A puzzling shift has taken place in the Court's case law on EU social citizenship. Around the turn of the century, the possibility to claim equal access to social benefits in other member states was extended from ‘economically active’ to ‘economically inactive’ EU nationals on the basis of the newly established status of Union citizenship (Wollenschläger 2011). Recent years witnessed a break from the Court's expansive interpretation of the social rights of Union citizens. The Court has not only stopped extending cross-border welfare access to further categories of EU citizens (Davies 2018), but it has partly reversed its legal doctrine towards a more restrictive approach.

The early phase of the case law, initiated with the ground breaking Martínez Sala judgement (C-85/96), developed an expansive vision for Union citizenship. Economically inactive EU nationals had only enjoyed a right of residence for more than three months since the adoption of the so-called residence Directives in the early 1990s (90/364, 90/365 and 93/96). The Court had always acknowledged the possibility for member states to enforce the conditions connected to their lawful residence, including the possession of sufficient resources to avoid becoming a burden on the social assistance system of the host member state and insurance against healthcare costs (eg Grzelczyk, C-184/99, para 42). Strengthened by the establishment of ‘Union citizenship’, however, the Court engaged in expanding the equal right to social benefits to students, jobseekers and other economically inactive EU citizens by striking down member states’ strict enforcement of these residence conditions on the basis of the principle of proportionality (Dougan 2013). In particular, it ruled that measures to enforce residence conditions should not become the automatic consequence of a Union citizen's recourse to social assistance (Grzelczyk, paras 42–43). Instead, the Court derived ‘a certain degree of financial solidarity’ between foreign EU citizens and their host member state from the preamble's statement that they should not become an ‘unreasonable’ burden on the host social assistance system (Grzelczyk, para 44). Declaring that Union citizenship was ‘destined to become the fundamental status of nationals of the member states’ (Grzelczyk, para 31), the Court extended the principle of equal treatment to lawfully residing Union citizens, not on the basis of their economic category but ‘purely’ as citizens of the Union (Baumbast, para 84, see also Trojani, C-456/02, para 31). Indeed, the right to freedom of movement was, as a fundamental principle of EU law, considered the primary aim and the general rule, whereas the conditions for lawful residence had to be construed narrowly and in compliance with the principle of proportionality (Brey, C-140/12, para 70). By requiring member state authorities to embark upon an assessment of the ‘personal circumstances’ of the individual applicant, the Court relied extensively on the individual's primary right to free movement, often overriding the ‘legitimate interests’ of the member states’ in securing collective dimension of social solidarity (De Witte 2012).

In contrast, the Dano judgement (2014) marked the end of Court's citizenship driven extension of social rights according to most EU legal scholars. Whereas the Court in its early case law strongly emphasized the Union citizen's right to equal treatment ‘simply as a citizen of the Union’ (Trojani, para 31), the Court stated in Dano that

so far as concerns access to social benefits, a Union citizen can claim equal treatment only if his or her residence in the territory of the host member state complies with the conditions for lawful residence of the Citizenship Directive. (C-333/13, para 69)

In other words, the Court poured the content of the Treaty based primary right to equal treatment into a statement in secondary law, turning the Grzelczyk approach to residential requirements on its ‘constitutional head – the latter no longer temper equal treatment rights; they constitute the rights’ (Shuibhne 2015). The restrictive approach of Dano was confirmed in Alimanovic (C-67/14) and García-Nieto (C-299/14). According to the Court in these cases, member states are not required to take into account the individual circumstances of the Union citizens as Directive 2004/38 itself already provided a ‘gradual system’ for the right of residence and access to social assistance (Alimanovic, para 60). With the UK infringement case (Commission v. UK, C-308/14), the restrictive judicial turn was further confirmed beyond the ‘specific situation of a non-integrated person’ such as Dano into a more general judicial reasoning with potential implications for a ‘wide array of different cases’ (Costamagna 2016).

Legal scholars are divided about the exact moment when the Court shifted its approach towards the social rights of Union citizens, but the shift itself is hardly disputed. Amongst others, it has been suggested that the early and ‘radical’ phase of citizenship case law already came to an end with the Förster and Vatsouras judgments in 2008, in which the Court started to explore the more restrictive provisions of Directive 2004/38 (Dougan 2013: 140). By yielding in to the Union legislator that students did not qualify for the grant of maintenance assistance during their first five years of residence and migrant jobseekers not for their job-hunting period, the question was raised whether the Court had retreated from assessing the proportionality of both Union legislation and member states’ enforcement measures. As a result, the post-2008 case law has been described as a period of ‘judicial hesitation’ about the proper interaction between Union citizenship rights and the conditions and limitations governing the right to residence and equal treatment (Dougan 2013: 140–145). The tipping point in balancing these competing interests arguably forms the Brey case (para 72), where the Court attached prominence to the member states’ need to protect public finances but also required member states to extensively assess individual circumstances before denying a recently arrived economically inactive Union citizen lawful residence. In sum, therefore, while hesitation was already visible, we submit that the Dano judgement marks the end of a justice-driven extension of social rights. Some scholars go further, by observing a clear restrictive judicial approach (Shuibhne 2015; Thym 2015), a ‘reactionary phase’ (Spaventa 2017) or even a ‘pre-Rome’ stage (O’Brien 2016: 938).

Court independence and mechanisms of political influence

How can we account for ECJ jurisprudence? And more specifically, what explains significant shifts in the Court's case law? Most explanations of ECJ behaviour fit into one of two opposed camps, which make strong assumptions about either judges’ independence from or their responsiveness to the political preferences of EU member state governments. These two opposing theoretical discussions go well along the traditional lines of supranationalist and intergovernmentalist integration theories. Our empirical analysis will show that neither of these two classic perspectives is sufficient to account for the Court's recent turnaround on EU citizens’ cross-border access to welfare benefits. We propose a complementing account, which gives greater attention to changes in the broader EU political context. ECJ judges may be independent enough to resist direct political threats from EU member state governments, but when EU legal issues become increasingly politicized, public opinion and political concerns are reflected in the Court's case law.

ECJ independence

The argument of a European Court that takes effective decisions independent of member states’ political influence has been accepted in the scholarly debate for long (Kelemen 2012; Stein 1981; Weiler 1991). Most importantly, heterogeneous member-state preferences and high decision-making thresholds are said to largely shield the Court's agency against collective political counter-measures (Stone Sweet and Brunell 2012: 205). Therefore, supranationalists have argued that this self-empowered Court successfully enhances expansionist European law (Burley and Mattli 1993). At the same time, the Court can rely on the EU's unique enforcement system to ensure compliance with its rulings. By establishing supremacy and direct effect, the Court has effectively enlisted private litigants and national courts as decentralized enforcers of European law (Kelemen 2011). In sum, political counter-measures to the Court's case law ‘are in theory improbable and in practice nearly impossible’ (Scharpf 2010: 217).

Apart from demonstrating the ECJ's great independence from political influence, these accounts also provide strong arguments why significant reversals of established case law should be unlikely. First, from an inner-legal perspective, judges have good reasons to honour established judicial precedent. This is not to say that precedent determines follow-up jurisprudence, but regular deviations or major breaks with previous case law come at the price of undermining legal coherence and certainty (Schmidt 2012: 10). Secondly, accounts emphasizing the great degree of ECJ independence typically depict the Court as an ‘engine of integration’. The Court is not only regarded as largely ‘unconstrained’, but also seen as ‘directed’ towards more integration (Scharpf 2012). If the Court tends to use its autonomy to promote integration, however, it should be particularly unlikely to reverse integration-enhancing jurisprudence and to return to lower levels of integration. Thus, significant changes in the Court's case law should only be likely after a reform of the European Treaties and, even then, supranationalists would expect the Court to seek new opportunities to promote integration rather than pulling back from established case law. Moreover, if individual attitudes were taken into account, EU Eastern enlargement should have made judges’ support for restrictions to free movement and equal treatment even less likely.

In sum, if we start from the assumption of great ECJ independence and try to understand the evolution of its case law in isolation from external political developments, we should expect the Court to generally avoid significant jurisprudential shifts for reasons of legal coherence and to deviate from established legal precedent mostly to promote further integration.

Political threats: legislative override and non-compliance

Other accounts have questioned the ‘unconstrained’ nature of the Court and theorized political limits of ECJ independence. From the perspectives of intergovernmentalism and principal-agent-theory, the preferences of EU member state governments are seen as the Court's major constraint. Pioneering in this theoretical tradition, Garrett et al. (Garrett et al. 1998) conceptualized the ECJ as a strategic actor which tries to anticipate potential political countervailing measures at the European and domestic levels. Several recent studies have theorized and tested the influence on the Court of two distinct political countervailing measures: legislative override (Larsson and Naurin 2016) and non-compliance (Carrubba and Gabel 2015; Carrubba et al. 2008).

While the two mechanisms are distinct – legislative override requires collective action at the EU level, non-compliance results from unilateral action at the domestic level – the underlying logic is very similar. In order to prevent the ECJ from issuing unwelcome rulings, member state governments threaten to take countervailing measures at the European or national levels. The Court is said not to be immune against these kinds of threats. Despite the EU's enforcement system, member state governments may threaten to infringe European rules if the domestic costs of compliance are getting too high (Carrubba and Gabel 2015: 11, 23). And even though actual legislative override requires a qualified majority or unanimous agreement among EU member states, thus making it a highly unlikely case in a fairly heterogeneous Union (Davies 2014), judges still operate under conditions of uncertainty and may already be influenced by the perceived risk of legislative override (Larsson and Naurin 2016: 382).

In addition to their similar logic, the same kind of empirical evidence is typically used to assess member states’ threats of legislative override and non-compliance. Written observations submitted during ECJ proceedings are interpreted as signals of EU governments’ preferences and as potential threats against unwelcome jurisprudence (Carrubba and Gabel 2015: 47; Larsson and Naurin 2016: 389). The two mechanisms are expected to work differently under different EU voting rules – the threat of legislative override being more credible under qualified majority voting, and non-compliance arguably being more likely in politically sensitive areas protected by unanimity rule (Larsson and Naurin 2016: 391; for an alternative way of distinguishing the two mechanisms empirically, see Carrubba and Gabel 2015: 146f.). More importantly for our puzzle, we can derive the same expectation regarding significant shifts of ECJ jurisprudence from both arguments.

Accordingly, if member states signal strong opposition to the Court's jurisprudence through their written observations, we should expect the Court to be responsive and, if necessary, to reverse unwelcome case law.

Political context: public debate and diffuse legitimacy

We believe that both perspectives discussed so far, emphasizing the independence of the ECJ and its responsiveness to EU governments’ preferences, capture important parts of the Court's behaviour. And yet, they overlook an essential piece for solving the puzzle of why the Court reverses its own jurisprudence. Even where the ECJ is independent enough to withstand strong opposition from EU governments, it does not rule in a ‘political vacuum’, but may be influenced by changes in the broader political context (Martinsen 2015: 12). Particularly since the 1990s, we have seen European political issues becoming increasingly contested and politicized – especially in the media as a reflection of broader public sentiments (cf. de Wilde et al. 2016). The previously accepted ‘permissive consensus’ where citizens content themselves with elite decision-making can no longer be regarded as the basis for European integration (Hooghe and Marks 2009).

Our argument that the ECJ is responsive to public debate and politicization is far from unique in the literature on courts. Various studies on the US Supreme Court have shown a strong link between public opinion and court jurisprudence (Epstein and Martin 2011). These studies find that the US Supreme Court is highly responsive to public opinion. Justices are public persons, reading newspapers, watching the news, give talks and engage with the public. Their work brings them in contact with ‘the temper of the times’ (Flemming and Wood 1997: 493). Especially in perilous political situations, justices will restrain themselves for fear of acting outside of the broad contours of public support. They can be expected to behave in anticipation of a lack of public support, especially when a lack of specific public support – i.e., for a specific decision or policy – risks affecting their ‘diffuse’ or aggregate support (Clark 2009: 973f.). Similar findings on judicial responsiveness were reported for the jurisprudence of the German Constitutional Court (Sternberg et al. 2015). However, the literature also notes that judicial responsiveness tends to be incremental. Courts are not weathervanes, changing direction with the wind (Flemming and Wood 1997: 494). Constitutions, legislation, precedents as well as judicial attitudes prevent this from happening. Within those constraints justices are left with some scope for interpretation which allows for changes in judicial reasoning if substantial swings in public mood occur.

As to the ECJ, one can easily find numerous case notes, in which claims about judges’ underlying motives are made, which fit this perspective: ‘Judges in Luxembourg are not autistic and listen to the general political context’ (Thym 2014) and ‘judges read the morning papers’ (Peers 2014). Often, however, these claims are not investigated empirically and the precise mechanisms through which political debates translate into ECJ jurisprudence remain unaddressed (Mantu and Minderhoud 2015: 11). We can think of at least two such mechanisms. On the one hand, judges themselves are part of the public and may simply be influenced by the same debates and developments as public opinion in general. On the other hand, judges may be concerned about their ‘diffuse legitimacy’ (Clark 2009: 973; Larsson and Naurin 2016: 386), i.e., about the acceptance and long-term support of their jurisprudence by the general public. While both mechanisms are highly plausible theoretically, they are impossible to distinguish empirically (Epstein and Martin 2011: 280–281; Sternberg et al. 2015: 589–590).

In sum, we expect the ECJ to be responsive not only to political threats by national governments, but also to substantial shifts in the broader public mood. Thus, even if the Court is able and willing to pursue a particular line of jurisprudence against strong member state opposition, it may shift or reverse previous case law when public opinion – as mediated through the media – sides with member states.

Explaining the Court's turnaround

The remainder of this article analyses empirically the ECJ's turnaround on EU social citizenship in light of the different theoretical accounts of Court behaviour set out above. We begin the empirical analysis by justifying our research design and methods, before discussing the limitations of existing explanations in accounting for the jurisprudential shift on citizenship. The two subsequent parts show the parallels between the increasing politicization and public debate on ‘welfare migration’ over the last decade and their reflection in the Court's case law on EU citizens’ access to welfare benefits.

Research design, method and limitations

Our empirical analysis focuses on the link between public opinion and the evolution of the ECJ's case law on EU ‘social citizenship’, i.e., EU citizens’ cross-border access to welfare benefits. In terms of salience, this case law was hardly matched by any other line of ECJ jurisprudence over the last decade, neither among EU legal scholars nor in the public debate. The question of EU citizens’ cross-border access to welfare benefits touches upon several core issues of integration, such as EU enlargement, citizens’ mobility, the viability of national welfare systems, Eurosceptic parties addressing the topic in a welfare-chauvinist way, as well as citizens’ eroding consent with the European integration project. At the same time, the scope of our analysis is clearly more limited than that of general accounts of the Court's jurisprudence (Carrubba and Gabel 2015; Larsson and Naurin 2016; Stone Sweet and Brunell 2012) and also more narrow than studies on the Court's entire citizenship case law (Šadl and Madsen 2016). Theoretically, our aim is not to generally test and refute any of the other explanations of Court behaviour discussed above, but to demonstrate the plausibility and the added-value of an account emphasizing the influence of the broader political context on the Court (for a similar view regarding the complementary character of different mechanisms of political influence, see Larsson and Naurin 2016: 385f.). Adopting a narrow empirical focus allows comparing closely the potential parallels between a clearly circumscribed public debate on ‘welfare tourism’ and its reflection in a limited number of Court cases on EU citizens’ welfare rights.

For that purpose, our empirical analysis brings together three different kinds of sources: member state written observations, newspaper articles and the Court's own jurisprudence.

To begin with, we replicated the analysis of Larsson and Naurin with regard to the most important cases on EU social citizenship (see Annex I for the Court cases and coding). For cases until 2012, we drew on the Court's ‘Reports for the Hearing’, in which the reporting judge used to summarize the key statements of all parties involved in a particular case, including all member states that filed an observation, the Commission and sometimes the Council. Since no such reports exist for cases after 2012, we had to rely directly on written observations and on the summaries of the different parties’ positions in the opinions of the advocate general.¹ On this basis, we first identified the main issue of the questions referred to the Court. Afterwards, we determined the respective opinions of the actors involved in these issues as well as the final position of the Court – and coded finally whether these positions were either in favour of ‘more Europe’, i.e., of an expansive interpretation of EU law, or in favour of preserving national sovereignty, i.e., a restrictive interpretation, or ambivalent (Larsson and Naurin 2016: 392).

Secondly, our media analysis covers both quality newspapers and tabloids from five EU member states: Austria, Denmark, Germany, the Netherlands and the UK. We selected these five member states because they are all net receivers of EU migration according to Eurostat comparative data. Moreover, political concerns have been expressed by the governments in all five member states concerning EU cross border welfare. In April 2013, the ministers of interior from Germany, Austria, the Netherlands and the UK sent a joint letter to the Council of the European Union stating their concerns and the view that free movement of persons and access to welfare should not be unconditional. This position has been supported by the Danish government. On the basis of a common coding manual, we searched and coded all newspaper articles between January 2003 and June 2016 which treated the theme of free movement and EU citizens’ access to welfare in another member state in three daily national newspapers; a centre-right paper, a centre-left paper and a tabloid (see Annex II for our search terms, selection of newspapers and coding strategy).

Finally, we analyzed the Court's case law qualitatively whether it reflects the public debate in terms of language and timing. Since ECJ deliberations are secret, we lack information about an important step in the process through which public debate supposedly influences the Court. Ultimately, our analysis can only demonstrate a strong correlation between public debate and Court reasoning, combined with a plausible theoretical mechanism. We therefore agree with Karen Alter's cautious note on explaining Court jurisprudence: ‘The real issue is that we can never really know why[…]judges make the decisions they do. The best we can do is to rely on theories of causation tested via proxy evidence’ (Alter 2014: 338).

The Court not bending to member state opposition

The most prominent existing accounts fall short of explaining the shift of the ECJ citizenship jurisprudence. A legal positivist approach cannot explain the Court's turnaround as described in section 2: the underlying Treaty law remained unchanged during the entire period of investigation and the Court continued its expansive jurisprudence even after the adoption of the Citizenship Directive. Moreover, political explanations focusing on member state ‘threats’, i.e., threats of non-compliance (Carrubba and Gabel 2015) or legislative override (Larsson and Naurin 2016), are insufficient to account for the Court's turnaround. When applying these explanations to the ECJ's citizenship case law, we would expect being able to trace the Court's shift back to a rise in member states’ opposition. Yet, our analysis shows that member states constantly opposed the Court's expansive citizenship jurisprudence from the beginning and actively contained compliance (Heindlmaier and Blauberger 2017).

For that purpose, we coded the positions of EU member states involved in EU social citizenship cases. From Martínez Sala to Vatsouras & Koupatantze, we identified 15 issues in total that were under discussion. Member states advocated a restrictive interpretation of these issues 61 times, while backing ‘more Europe’ only 7 times (see Annex I). It was only in Martínez Sala that the member state of origin argued in favour of its own national. Concerning Grzelczyk, only Portugal and the Commission opted for ‘more Europe’ when considering him as a worker which is finally linked to social benefits. Belgium, Denmark, France and the UK held a restrictive view on both issues under discussion, namely the exclusion of economically inactive EU migrants from benefits and the general exclusion of students even after their residence right was recognized. They argued for instance that Grzelczyk, as a student, did not fall within the scope of the Treaty and could therefore not rely on its prohibition of discrimination. Member states even submitted a coordinated response by the Council, opposing ‘more Europe’ when stressing that regulation 1612/68 on freedom of movement for workers did not apply to Grzelczyk.

With regard to Trojani, all member states involved positioned themselves in favour of a restrictive interpretation of both the first issue, whether there was a right of residence based on economic activity, and the second issue, regarding the ‘limitations and conditions’ of a right of residence based on Union citizenship. They declared, among others, that Trojani did not have a right of residence since he had neither sufficient resources nor a health insurance. And even though he had a residence permit valid for five months, he was only temporarily lawful staying there and could not derive any – not even temporary – right of residence from it.

As the written observations illustrate, member state governments already ‘threatened’ the Court in the earlier phase of citizenship jurisprudence. In addition, Directive 2004/38 constituted an attempt to at least modify the expansive jurisprudence of the Court (Martinsen 2015). In terms of non-compliance, the Commission found in 2008 that the transposition of the Directive was ‘rather disappointing’; and it already received numerous complaints and opened five infringement cases suggesting incorrect practical application as well (European Commission 2008). But still, these political threats or even disobedience proved not successful to prevent the Court from ruling expansively until 2014. The accounts are therefore insufficient to explain the Court's turnaround. We have to take a complementing explanation into account: changes in the broader political context.

The politicization of ‘welfare migration’

Figure 1 below shows the results of our media analysis. As Panel 1 of Figure 1 demonstrates, the public debate about cross-border welfare has dramatically gained in salience over the last decade. Although the issues already received some attention in 2004, the great majority of all relevant articles (about 75 per cent) were published during the last years (2013–16) with a clear peak in 2014. The issue of cross-border welfare was increasingly discussed under the heading of negative connoted terms like ‘welfare migration’ or ‘benefit tourism’. This framing came up and dominated the discussions especially in the years 2013 and 2014 when public attention peaked. Panel 2 documents that the issue of access to benefits for EU citizens was constantly and predominantly judged negatively. The number of negative statements regarding cross-border welfare clearly outweighed positive or neutral statements (about twice as much) – a proportion which is fairly constant over time and for all countries investigated.

Figure 1. Politicization of cross-border welfare.

Note: 2016 values cover the period 1 January to 30 June.

There are, of course, differences between the five member states under investigation, which are significant and in need of explanation (which in length is beyond this paper). In Denmark, for example, the debate was not only the most intense (Denmark had the highest number of all articles), but it was also the most concentrated (more than 50 per cent of all Danish articles were published in 2014). The British debate was less intense (in terms of the number of published articles), yet it took place over a longer time period. And in the Netherlands, the issue was discussed more steadily on a rather low intensity level (the least number of articles). Noteworthy is also that negative statements regarding cross-border welfare dominated the discussions most in the Netherlands and in the UK (where more than 75per cent of all statements were deprecatory; in Denmark ‘only’ 62 per cent). In Austria, a debate about ‘social tourism’ started already in 2004, mainly driven by the populist-right Freedom Party (FPÖ) and the right-leaning tabloid Kronen Zeitung. However, whereas this debate attracted only modest media attention from 2004 until 2012, we can identify a clear peak in 2014. In sum, despite all these interesting divergences, the general pattern is highly similar across all five countries: over the last decade, the issue of cross-border welfare was not only assessed and framed increasingly negatively but it also drew more and more the public's attention. Denmark might prove to have the most intense politicization of cross-border European welfare but it remains a trend across all five member states. No matter the level, we identify a clear negative turn in the public debate in all cases.

This overall finding reveals that the turn in the ECJ's case law on access to benefits for migrant EU citizens came at a time when the negative politicization of this issue peaked. This suggests strongly that the Court has been sensitive to public debate. Moreover, our evidence indicates a very close, issue-specific sensitivity of the Court to the public debate on ‘welfare migration’ rather than to other changes in the political context. Legal accounts of the Court's citizenship jurisprudence have discussed – yet, not systematically analyzed – various other potential influences on the Court such as the failure of the EU Constitutional Treaty in 2005 (Dougan 2013: 150), EU Eastern enlargement in 2004 and 2007 (Mantu and Minderhoud 2015: 4), or the financial and economic crisis since 2008 (Šadl and Madsen 2016). In contrast to these more general political developments, our evidence suggests that it was primarily the soaring public debate about the relation between European free movement and national welfare that brought the ECJ to change course. Moreover, if we can show that the Court responded to public contestation, this influence cannot be reduced to political threats, since threats of legislative override and actual non-compliance had not kept the ECJ from ruling expansively in the past.

The Court's responsiveness to the public debate

This section discusses the Court's responsiveness to the public debate and political concerns over welfare migration on the basis of the Court's own reasoning and the exposure of the Court to public debate. We present proxy evidence in four dimensions. First, we observe a shift in the structure of the Court's legal reasoning from an emphasis on primary law, the status of Union citizenship, to the residence conditions formulated in secondary legislation. Second, there is a shift in the Court's balancing act between the objectives of promoting freedom of movement and the protection of public finances in favour of the latter. Thirdly, the discourse the Court adopts in its judgments reflects the underlying concerns over welfare migration. And finally, we argue that the Court's exposure to the public debate is reflected in the timing of its judgements.

It should be recalled, first, that in its earlier case law, the Court assigned primacy to citizenship as a ‘fundamental status of nationals of the member states’ and emphasized the existence of ‘a certain degree of financial solidarity’ (C-184/99, Grzelczyk, para 44). This was done in marked contrast to the position of most member states, who argued that Union citizenship has no ‘autonomous content’ and should not grant new and more extensive rights to Union citizens (para 21). In the recent line of restrictive judgements, no reference is made any more to the principles of financial solidarity and citizenship. Whereas in Dano, the Court still refers to the ‘fundamental status’ of Union citizenship and explicitly affirms Article 24(2) as an exception to the principle of non-discrimination, Alimanovic skips any reference to Union citizenship and directly moves towards assessing residence conditions by stating that Union citizens can claim equal treatment only when they comply with the conditions of Directive 2004/38 (para 49).

This shift in reasoning structure is flanked by a shift in balancing the objectives guiding free movement law (Kramer 2016). Financial concerns have featured in the arguments of member states from the start, especially in their attempts to limit the ‘temporal effect’ of expansive judgments (see e,g. Grzelczyk para 49) and have generally been acknowledged by the Court as a legitimate interest to be pursued by the member states (see, e.g., Bidar, para 55–56). In the early citizenship case law, however, the pursuit of such objectives had to be construed narrowly and was constrained by the broader purpose of facilitating and strengthening the right of free movement and residence (see especially Baumbast, para 91). In recent years, the Court has gradually replaced its rhetorical emphasis on the objective of promoting free movement with an emphasis on the objective of creating conditions for the exercise of free movement and preventing Union citizens from becoming an unreasonable burden on the social assistance system of the host member state. The Court already embarked on this turn in its Ziolkowski and Szeja judgment, where it considered that

it is true that Directive 2004/38 aims to facilitate and strengthen the exercise of the primary and individual right to move and reside freely [ … ], the fact remains that the subject of the directive concerns [ … ] the conditions governing the exercise of that right and the right of permanent residence. (C-425/10, para 36)

With the tipping point arguably being the Brey case (para 72), recent cases attach prominence to the need to protect member states’ public finances. This is illustrated by the shift in the Court's interpretation of recital 10 of the Citizenship Directive. Whereas this recital was used in Brey to stress the requirement of proportionality of enforcing residence conditions and the existence of ‘a certain degree of solidarity’, in Dano and Alimanovic recital 10 was used to stress the objective of the Directive to prevent economically inactive Union citizens from becoming an unreasonable burden on the social assistance system (para 74 and 50 respectively). In its García-Nieto judgment, the Court went so far as stating that the Directive's provisions were consistent with ‘the objective of maintaining the financial equilibrium of the social assistance systems of the member states’ (para 45).

Thirdly, the argument can be made that it is not only the nature of its reasoning, but also the Court's use of language that reflects public and political sentiments on ‘welfare migration’. In Dano, for example, the Court interpreted article 7(1)(b) as a provision seeking to prevent economically inactive Union citizens from using the host member state's welfare system to fund their means of subsistence (para 76). More fundamentally, the Court accepted the possibility for a member state to refuse social assistance benefits to Union citizens who use their freedom of movement ‘solely in order to obtain another member state's social assistance’ (emphasis added, para 78). Although it should be stressed that this statement dealt with a provision in German legislation, it is clear that this formulation comes close to a definition of ‘welfare migration’ by acknowledging, for the first time, that the motives of economically inactive Union citizens may play a role in assessing their right to reside and equal access to social benefits (Verschueren 2015: 374). Although the ‘facts’ of recent litigants were arguably less ‘meritorious’ under EU law than their predecessors (Davies 2018), justifications provided by the Court for its decisions do matter. In terms of framing and discourse they effectively convey a different message to the member states: EU law does offer possibilities to restrict Union citizens’ access to national welfare benefits and counter concerns over ‘welfare migration’, whether actually present or not.

Finally, evidence suggests that these restrictive judgements not only followed public consternation and preceded major political events, but the judges were pointed to the political relevance of these judgements. It follows from our media analysis that the Court's judgment in the Dano case followed a period of increased media attention to ideas around ‘welfare migration’ (see Figure 2). Before deciding its next case, Alimanovic, the Court was informed by the Advocate General in his Opinion about ‘the unusual stir that that Court judgment [Dano] has caused in the European media and all the political interpretations that have accompanied it confirm the importance and sensitivity of the subject’ (Opinion, C-67/14, para 4). Not only the attention of the media was unusual, however, but also the Advocate General's remark about the political sensitivity of the Court's judgements on this topic can be considered unique in EU legal practice and culture. A final indicator is the timing of the publication of the Commission v. UK judgement, nine days before the UK's EU membership referendum. Whereas the timing as such cannot suffice as evidence, it is rather the decision not to postpone a politically sensitive decision on a heated subject of campaigning over continued membership of a key member state that demonstrates the Court's willingness not to avoid the impression of political intervention. This impression is shared by a significant part of legal scholars, with one of them noting that ‘the ECJ has played politics and lost. [ … ] the Court is a political actor. It is impossible to know the intersection of legal and political in the minds of judges, but those minds do not operate in a vacuum’ (O’Brien 2017: 209).

Figure 2. The Court's responsiveness to politicization.

Conclusion

What explains shifts in ECJ jurisprudence in general, and shifting citizenship case law in particular? We claim that the ECJ's more recent citizenship jurisprudence deviates from prior case law in a way that cannot sufficiently be explained solely by threats of legislative override or non-compliance – the most prominent mechanisms to influence the ECJ. An increasing, negative politicization of free movement and ‘welfare migration’ added to member states’ long-standing opposition (articulated in written observations and actual non-compliance). Our argument is supported by the closely parallel timing of the public debate's intensity and the most important rulings of the ECJ's turnaround, and by the way how the Court dealt with the increasing salience of the issue of cross-border welfare in its legal reasoning. Thus, the ECJ indeed was responsive to its environment, but this responsiveness was driven by more than just governmental threats. We cannot with certainty uncover the secret deliberations among the judges or the private reflections the individual judges have done. But resting upon a strong correlation and a plausibility probe we argue that the Court is sensitive to public mood – a trait that it shares with other high courts.

Acknowledgements

We would like to thank three anonymous reviewers and Daniel Naurin for their very helpful comments. The research for this article was part of a Norface-funded project, TransJudFare, Grant no. 462-13-009.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Michael Blauberger is Associate Professor at the University of Salzburg.

Anita Heindlmaier is Doctoral Researcher at the University of Salzburg.

Dion Kramer is Doctoral Reseacher at the Vrije Universiteit Amsterdam.

Dorte S. Martinsen is Professor at the University of Copenhagen.

Jessica Sampson Thierry is Doctoral Researcher at the University of Copenhagen.

Angelika Schenk is Doctoral Researcher at the University of Bremen.

Benjamin Werner is Postdoctoral Researcher at the University of Bremen.

ORCID

Michael Blauberger http://orcid.org/0000-0002-3581-400X

Dion Kramer http://orcid.org/0000-0002-4946-2762

Dorte Sindbjerg Martinsen http://orcid.org/0000-0002-2113-0950

Jessica Sampson Thierry http://orcid.org/0000-0003-2734-0219

Angelika Schenk http://orcid.org/0000-0002-8926-8112

Additional information

Funding

The research for this article was part of a Norface-funded project, TransJudFare, Grant no. 462-13-009.

Notes

1. Since we could not get access to all written observations and opinions of the Advocate General are sometimes selective, information on some observations is missing for cases after 2012 (see Annex I). A similar approach has been chosen by previous studies, e.g. by Carubba and Gabel (2015: 70–71).