Political and constitutional overrides: the case of the Court of Justice of European Union

ABSTRACT

Overrides are acts by politicians which undo or modify the impact of court decisions. In the field of judicial politics in general, and in studies of the Court of Justice of the European Union’s role in European integration in particular, the existence and effects of overrides have been intensely debated.

This article argues that much of this disagreement is driven by a lack of a common definition of the term at issue. It distinguishes between ‘political’ and ‘constitutional’ overrides, and defines the conditions under which either matters for judicial independence in the European Union and the Court's impact on European integration. This is followed by an investigation into the occurrence of overrides in the European Union. It is concluded that while political overrides, where the policy impact of court decisions are affected by subsequent EU legislation, have occurred, constitutional overrides, which challenge or revise the Court’s mandate, have not.

KEYWORDS:

• Court of Justice of the European Union
• EU law
• EU politics
• judicial politics
• override

Introduction

The independence of courts is of seminal importance to both national and international politics. On the domestic arena, courts are one of the pillars of constitutional democracy, serving as a check on the other branches of government. On the international arena, courts are one type of formal organization which can alleviate pressing international collective action problems, by making commitments among states credible (Aggarwal & Dupont, 1999; Garrett, 1992; Mansbridge, 2014). But for courts to exert either role, they have to be reasonably independent from the wishes of the politicians they are to check, or whose collective commitments they are to uphold.

Overrides are, in general terms, acts on part of politicians which affect (e.g., undo, limit, modify) the impact of court decisions. In theories of judicial politics, which study the interaction between law, courts and politics, overrides have been suggested to be a powerful tool by which politicians can influence the decisions of courts. They are potentially a threat to the independence of courts, and by extension to the rule of law, as the threat of override can lead to political considerations crowding out the role of the law in the process of adjudication (Ferejohn & Shipan, 1990; Ferejohn & Weingast, 1992; Marks, 1988; Segal, 1997). Recently, judicial independence has come under attack both in Europe and elsewhere (cf. Blauberger & Kelemen, 2017). To understand and protect the independence of courts, we must understand the interactions between politics and law that shape it.

While there is wide agreement that overrides can be powerful tools by which politicians can influence courts, there is disagreement regarding their viability in practice, especially in studies of the European Union (EU). In this article I will argue that much of this disagreement is conceptual rather than empirical, as scholars define the concept of override in very different ways. According to the narrowest definitions, overrides are hard to achieve, always constitute a threat to judicial authority, and are most often normatively problematic. According to the broadest, overrides are easier to achieve, and are only theorized to affect judicial behaviour under a specific set of circumstances. Absent these circumstances, overrides are less normatively problematic, and more attractive for democracy to correct unwanted judicial activism. This has made it difficult to reach a consensus regarding when and how override can occur, what effects they have, and when they are normatively important.

Section 1 of this article defines two types of overrides, the political and the constitutional, based on how previous theories have discussed ‘overrides’ and argued that they matter for judicial behaviour and the rule of law. In section 2, I investigate the occurrence of override where their existence has been most contested: The EU. In section 3, I then compare these overrides to the different definitions, and conclude with a discussion of what these overrides tell us about the relationship between EU member state governments and the Court of Justice of the EU (CJEU).

I conclude that while overrides broadly defined have happened in many of the cases here studied, they all belong to the category of ‘political overrides’. This does not in itself prove any theory of judicial politics, but show us that overrides, broadly defined, happen in the EU as well.

Political and constitutional overrides

In the most general sense, override occurs when politicians change the text of laws or treaties (or produces new ones) in order to undo or limit an unwanted implication of a court’s (or other adjudicative institution’s) decision. However, this general characterization no longer holds for all uses of the term in contemporary theory. These have parted ways in two different directions, following two different logics in their use of the term ‘override’. I call these political and constitutional overrides.

Political overrides are those which are defined by comparing the policy preferences, often in the light of broader ideological attitudes, of judges and politicians on the one hand, and actual and potential outcomes of court decisions and laws on the other (cf. Ferejohn & Weingast, 1992). I distinguish these from constitutional overrides, which constitute constitutional revisions of or challenges to a court’s mandate and by extension its authority (cf. Davies, 2016).

Political overrides

What I call ‘political overrides’ play an important role in Separation of Powers-theories in judicial politics. These theories study the interplay between courts, legislatures and executives, which all are assumed to have political preferences which they try to realize through policy (Ferejohn & Weingast, 1992; Marks, 1988; Segal, 1997). In this literature, an override is any decision by the legislature that changes the policy outcome of the court’s decision in some material way. The exact legal form of a political override is of a secondary nature. What is important is that judges are assumed to be policy seekers, who strive to enact their policy preferences through court decisions. Judges, and by extension courts, fear override since (a) they are invested in the policy outcomes of their decisions, and (b) overrides change these outcomes (Gely & Spiller, 1990, p. 267; Segal, 1997; Epstein & Knight, 1998, p. 598).¹

The literature on the Supreme Court of the US defines overrides in very broad terms, such as a law that ‘reverses’ (Gely & Spiller, 1990; Spiller & Gely, 1992; Meernik & Ignagni, 1997, p. 457), ‘replaces’ (Epstein & Knight, 1998, p. 82), or simply ‘changes’ (Marks, 1988, p. 2) the effects of a decision by the courts. Uribe et al. (2014, p. 14) consider any law that ‘ignored, overturned, modified, altered, undid, [corrected] or limited or reversed the effects of the Court case’ as an override; Stumpf (1965, p. 382) any ‘proposed congressional legislation the intent or effect, or part of the intent or effect, of which is to modify the legal result or impact, or perceived legal result or impact of a specific Supreme Court decision, or decisions’.

These overrides are often treated as having equal effects on the behaviour of judges, despite the definition being so broad. Scholars argue that the crucial comparison is not one between the actual court decision and the post-override situation, or even one between the situation before the decision and the post-override situation. Instead, what is argued to be crucial for the courts is the alternative cost. A core assumption in Separation of Powers-theories is that the judges are not only policy-seeking; they are also strategic and sophisticated agents. As courts strive to avoid override, they will therefore, it is argued, tailor their decisions to the preferences of the legislature in order to achieve a policy outcome that is as good as possible, i.e., one that cannot be overridden by the politicians, but still lies as close as possible to the judges’ preferred outcome. The crucial comparison after an override is the one between what could have been achieved and the post-override situation, i.e., the alternative cost of risking an override (Ferejohn & Weingast, 1992). It is secondary whether a law made things worse than they were before the court decision, or if the post-override court decision still improves on the status quo ante, from the judges’ perspective.

In simple terms, this literature argues that strategic sophisticated agents compare what they got to what they could have gotten, not to what they had. A political override is any change to a court decision which affects the court’s desired outcome, and what courts try to minimize is the difference between what they could have gotten and what they got post override.

Constitutional overrides

According to what I call the constitutional logic, an override must constitute an attempt by the legislature to revise the mandate of a court in order to diminish the court’s power, and thereby alter the constitutional order. By ‘constitutional’ I do not mean any change to a written constitution, nor exclude changes outside constitutions. Rather, constitutional overrides demand a ‘contest of wills’ among the different branches of government over the appropriate roles of the respective institution (Davies, 2016, p. 5). Revisions of a court’s mandate challenges that institution’s authority, in the sense that it diminishes the space of issues over which it has power, and hence, authority. For example, if a court interprets an article of a constitution or international treaty in an unwanted way, politicians can delete that article, or leave it but remove it from the court’s jurisdiction by other means (cf. Schaffer et al., 2013). A constitutional override constitutes a threat to a court and its judges even if they do not care about specific policies or have strong political preferences, as long as they care about the rule of law and their general authority in society (cf. Epstein & Knight, 1998, pp. 598–599; Gely & Spiller, 1990; Segal et al., 2011, p. 594; Uribe et al., 2014, p. 12).

Overrides in the European Union

Overrides were incorporated in theoretical discussion on the CJEU by authors who in the 1980s (Rasmussen, 1986) and early 1990s began to emphasize the importance of government preferences as a limitation (Garrett, 1992; Garrett & Weingast, 1993) to the CJEU’s alleged pro-European activism (cf. De Waele, 2010). Today, both proponents and opponents of the importance of override agree on the basic premises of the Separation of Powers model in studies of the CJEU, i.e., that judges are strategic policy seekers conscious of the political and institutional milieu that they operate in (Stone Sweet, 2004, p. 8).

Notwithstanding agreement on these fundamentals, some scholars have argued for the relevance of the threat of override as a systematic constraint on the CJEU (Larsson & Naurin, 2016), while others have long been more sceptical (Stone Sweet, 2004). The latter argue that the decision-making rules of the EU, coupled with the fact that the CJEU frequently interprets the treaties directly, effectively insulate the Court from override. In most instances, the passing of new legislation demands a qualified majority in the EU council, coupled with a majority in the European Parliament. As the preferences among the key decision-makers will often be far apart, the CJEU will always be protected by some veto-player which will block any attempt of override (cf. Kelemen, 2012, p. 46). The Court’s ‘zone of discretion’, i.e., the range of possible unoverrideable decisions, is ‘close to unlimited’ (Stone Sweet, 2004, p. 9). Moreover, in instances of direct treaty interpretation, the hurdles to decision-making become even higher. To override such a decision would (according to this view) require an amendment to the treaties themselves, a cumbersome process where each member state holds a veto. This has led scholars to argue that democratically elected politicians have lost control of a political-judicial system which asymmetrically favours deregulation over re-regulation (Scharpf, 2010).

Others go even further and dismiss ‘overrides’ as a myth, but these arguments primarily concern constitutional overrides, not political. According to Davies (Davies, 2016, p. 5), what I above termed political override, ‘is not true override’. As long as the legislature, in this case the EU Council and Parliament, confines its reactions to secondary law (i.e., do not change EU primary law, the treaties, directly) in ways which respect the CJEU’s interpretation of the treaties, they are not overriding the Court, but simply correcting secondary law in light of the new situation which arises post the Court’s judgement.

Crucially, those arguing for the importance of override tend to cite the Separation of Powers-literature and use broad definitions, such as ‘new EU legislation that counteract the effect of [CJEU] decisions’ (Garrett et al., 1998, p. 160), or amendments to the legal order that are taken as reactions to court decisions in general (Garrett, 1995, p. 172). Those arguing against this importance tend to use narrower definitions, such as Stone Sweet and Brunell (2012), who focus exclusively on ‘nullifications’.

Definitions and key differences

Based on the discussion above, I will define political override as ‘any legislative or treaty act which affects the policy impact of a court decision in some material way’; and Constitutional override as ‘any legislative or treaty act which contests and revises the mandate of a court’. Note that the categories of constitutional and political overrides are not mutually exclusive, rather they are hierarchically ordered: All constitutional overrides can also be political (they can affect a case outcome in some way), but not all political overrides can be constitutional.

In line with the original definitions of override discussed above, I limit mine to legislative and treaty-signing actions. Of course, executive action will also affect the future effects of principles given in specific court decisions. However, these are better conceptualized as concerning the implementation and compliance with the court decisions (or the laws concerned), which are (and should be) treated as distinct in most of the literature (cf. Rosenberg, 2008; Carrubba et al., 2008). The exact relationship between treaties, laws, court decisions and subsequent executive action can of course vary from system to system. However, override concern the meanings of the rules and not the degree to which they are upheld (i.e., not whether the rules are followed or not).

There are two key differences between political and constitutional overrides. Firstly, political overrides are interesting, i.e., matter, because they challenge political goals of judges, and will only affect judges, and thus matter theoretically, if judges have such goals. Constitutional overrides are interesting and matter because they challenge the courts as institutions in a democratic system characterized by a separation of powers.

From a constitutional perspective, there is not necessarily anything controversial about political overrides. These simply mean that (a) the representatives of the people codifies rules in e.g., laws; (b) the courts determine what these rules mean in any given legal conflict; and (c) the people’s representatives correct the meaning of the rule if their intents have not been realized. Such actions do not necessarily challenge the authority of a court, or undermine the rule of law (Brower & Schill, 2008, p. 495 ff.; Davies, 2016, p. 5; Roberts, 2010, p. 194 ff.). Changes to laws or treaties as a reaction to court decisions can even work to alleviate the ‘counter-majoritarian dilemma’, i.e., when courts rule against the wishes of the majority and thereby potentially undermine the norms of popular democracy (Dahl, 1957; Bickel, 1986; cf. Scharpf, 2010).

Secondly, a political override will naturally be much easier to achieve, given the breadth of the definition. Constitutional overrides, on the other hand, will most often require changes to a constitutional document, or in the case of the EU, the founding treaties. Here, the threshold to decision-making is higher, both in terms of the number of number of actors involved and the thresholds to decision-making (for example, unanimity among the governments in the EU).

This difference has a direct impact on the debates on overrides in the EU, given that so much scholarship has assumed that overrides will be blocked by the high threshold to treaty amendments. However, this threshold will not stop all possible political overrides, given that (a) political overrides are so broadly defined and (b) the broad range of possible implications an interpretative body can give a treaty article. Moreover, 66 per cent of CJEU cases do not contain treaty interpretations (Fjelstul, 2019, p. 16). In short, one should not rule out political override to a court decision just because a constitution, treaty or similar document is being interpreted, and definitely not in other cases.

Subcategories of political override

While I would argue that the definition of political override that I have presented here is perfectly in line with how the concept has been used in the previous literature, I would also argue that it is exceedingly broad for any purpose other than to test highly specific Separation-of-Powers theories. In these theories, it is assumed both that judges have political preferences which they want to realize through their decisions, and that they strategically monitor and anticipate the actions of the other actors. These are the assumptions which result in the conclusion that anything which changes the impact of a court’s decisions is viewed as a threat to the realization of these preferences. However, if we relax the assumptions about specific policy preferences, or sophisticated strategic behaviour, it becomes less reasonable to treat all political overrides in the same way. For example, a judge with vaguer ideological preferences and cruder strategies will in all likelihood treat a minor modification to a decision differently from a complete nullification.

Moreover, a widely defined concept also become harder to distinguish from other legislative reactions to court decisions, such as codifications, i.e., where the legislator enshrines a court decisions in legislation in order to avoid future legal uncertainty (Martinsen, 2015). A law which for example modifies parts of a decision will often be forced to codify the rest of the decision.

To address these concerns and make future uses of these concepts fruitful to wider fields of research, I distinguish between three subgroups of political overrides. Here, I focus on how an override changed the legal situation from the viewpoint of a hypothetical plaintiff in a situation similar to the one which gave rise to the case, following distinctions established in the previous literature between how various overrides affect case impacts. Eskridge (1991) offers, similarly to Uribe et al. (2014), three different categories of override: The first include any act that nullifies or ‘completely overrules the holding of a statutory interpretation decision’; the second refers to acts that ‘modifies the result of a decision in some material way’; and the third to acts that insulates an adjacent policy area from ‘the consequences of the decision, such that the same case would have been decided in the same way but subsequent cases would be decided differently’. The latter thus insulates other policy areas from the same type of court reasoning (Eskridge, 1991, note 1). We thus have three subcategories of political override: Nullification, modification and insulation.

Conceptually, the distinction between political and constitutional overrides should be valid beyond the EU or the USA. The two types of override are already frequently brought up in e.g., discussion on international trade- and investment treaties, but under different names. For example, (Ginsburg, 2005, pp. 661–665) distinguishes between ‘overrule’ by amendments of treaties, and interpretative statements meant to correct decisions by adjudicatory bodies such as NAFTA panels. The role given to interpretative statements in investment treaty arbitration arguably falls within the realm of political overrides, to the extent that they affect the concerned treaties’ impact in future situations (cf. Roberts, 2010; Brower & Schill, 2008, p. 494).

Identifying overrides in the EU

From previous research we already know that the member states of the EU have revised the treaties in response to CJEU decisions at least three times. In the second protocol to the Treaty of Maastricht, the member states limited the temporal scope of decision C-262/88 Barber, but left its impact otherwise intact. C-159/90 Grogan, which stated that the provision of abortions constituted a service within the meaning of the treaties, led to another protocol stating that nothing in the EU treaties should affect the Irish constitution’s ban on abortion. C-450/93 Kalanke undermined a German affirmative action programme, and led to article 141(4) of the Amsterdam treaty, which explicitly allowed such programmes (Castro-Montero et al., 2018; Conant, 2002, p. 235). These have been contested as overrides, but then using a definition which demands that overrides nullifies the impact of a court decision (Stone Sweet & Brunell, 2012). While this paper will not rehash the discussion on these cases in detail, I believe that both Kalanke and Barber can be readily categorized as political overrides, as they did not adjust the Court’s mandate, but target a specific policy implication. Grogan is more complicated, and could be argued to be a constitutional override as it limited the reach of EU law within a certain issue area in one member state.

Beyond these, no one has to my knowledge systematically investigated the occurrence of override in relation to a broader set of CJEU decisions. In this section, I investigate what happened to a sample of Court cases where override were most likely to happen. To select these, I use the Larsson and Naurin (2016) dataset on CJEU decisions, covering the years 1997–2008, which includes measurements of government opposition and CJEU decisions in regards to deeper legal integration, or ‘More Europe’. I use these to identify those cases where (a) the Court encountered the strongest member state resistance while (b) still ruling for ‘More Europe’. I limit myself to those cases where member states representing 30 per cent of the votes in the Council opposed more Europe while the CJEU ruled in favour of it. I chose vote shares and 30 per cent so as to capture cases which are either opposed by the big member states (.30 roughly equals the sum of Germany, France and the UK’s votes), a large group of smaller states, or a combination thereof. This gives a sample of 27 cases.

I then delve deeper into those cases where a legislative response which could be categorized as an override according to any of the above discussed definitions happened. For each case, I give a brief summary of the legal conflict and the subsequent legislation which had a material impact on the issues of the case. For a more detailed description of the sample selection process, see the appendix section A.I. For a full description of all cases, including more detailed descriptions of the legal issues involved, see the appendix section A.II.

There are two problems with my approach beyond those already mentioned. First of all, the threshold of 30 per cent was chosen so as to cut off the list of most resisted cases at a point where the number of cases became manageable for qualitative analysis, while still not limiting the sample too much, as most states do not participate in most cases, and high shares of the Council appearing in any case is rare (Larsson & Naurin, 2016). A higher threshold would have e.g., led to the omission of the override identified in Watts (see below), but a lower could of course have led to the identification of more overrides (see A.II for a discussion of borderline cases).

Secondly, in order to fully gauge the frequency and relative importance of override in the EU, more quantitative approaches will be required. This article is meant to combine an in-depth discussion of how to define overrides (and for what reasons), and a qualitative assessment of the existence of overrides in the EU. The exhaustive discussion of definitions and concept history, along with in-depth descriptions of all cases, are intended to minimize the subjectiveness of the results to a minimum. However, I have strived to be as transparent with difficult and questionable categorizations as possible, to help guide future endeavours to research override. In section 3, I offer my advice in regards to such research, e.g., how to identify and code overrides in larger set of cases.

Results

In total, my sample contains seven instances of overrides, including three cases where the legislative response nullified the impact of the Court’s decision. These are C-184/99 Grzelczyk and C-209/03 Bidar, which both concerned the rights of EU students residing in other member states, and C-372/04 Watts, which concerned the right to seek hospital care in other member states.

In Grzelczyk and Bidar, the Court, against substantial government opposition², extended the treaties’ provisions against discriminations in ways which opened up domestic systems of support to students from other member states. In the member states that the cases originated from (Belgium and the UK), such support was (before the judgments) either not given at all, or only given if the EU student had worked for a minimum period. In Directive 2004/38/EC, an exemption from the rules against discrimination is inserted through article 24(2) in regards to students who do not have a permanent right of residence. The directive is thus a clear nullification of Grzelczyk and Bidar, in the sense that both plaintiffs could have been denied aid under the new rules (see Martinsen (2011) for a discussion on how the directive both advanced and limited the impact of other related cases, and Blauberger and Schmidt (2017) for a discussion on subsequent case-law and implementation).

In Watts, the Court ruled that EU citizens have a right to seek hospital care in other member states and to have the costs of that care reimbursed, if the same care would be paid for by the system of their home state. Moreover, if the cost of the care in the host state exceeds that which would be relevant in the home state, the home state still has to reimburse the full amount. Finally, budgetary concerns relating to the financial stability of the system that had to bear the cost of this cross-border health care were not relevant, according to the Court.

Directive 2011/24/EU affected many of the points raised in Watts. First of all, the directive put a cap on the amount that could be reimbursed, at the level that would be reimbursed at home.³ I.e., if the care costs more abroad than at home, the patient will have to pay the difference.

The directive also added a possibility to limit the rules on reimbursement, based on ‘overriding reasons of general interest’⁴ and the budgetary planning requirements of the relevant national authority. In these terms, the directive thereby substantially increases the degree to which governments can use budgetary concerns to motivate limitations to the right to cross-border care (cf. Hatzopoulos & Hervey, 2013). The broader issue of whether there was an obligation derived from the treaties to fund hospital care in other member states was not addressed directly, but this was also the least contested issue of the case.⁵

It could be argued that the leeway of governments to refuse to reimburse cross-border care is not unconditional, and that the case should hence be classified as a modification, as the status of the plaintiff and others similarly situated is unclear in the post-override legal environment. However, I have classified Watts as a nullification, since the UK authorities could have invoked these new rules in order to deny the plaintiff the right to have her hospital care in France reimbursed. It is however admittedly a hard case to categorize.

A case which clearly has been overridden in a modifying manner is C-341/05 Laval. Here, the Court dealt with three issues originating from Sweden. The case concerned a conflict between a Latvian company which had posted workers to Sweden, and a Swedish trade union which tried to use collective action in order to force the company to sign a collective agreement. The first two issues concerned when collective agreements and the threat of collective action could be used against a service provider posting workers in a member state. The first of these issue concerned whether such action could be used to demand wages above the levels of a minimum wage; the second whether other conditions, such as working hours or vacation, could be demanded above the level laid down in national law.

The third issue concerned the Swedish ‘Lex Britannica’ legislation, which stipulated that while collective action could not be used to circumvent another Swedish collective agreement, it could be used against agreements from other member states.

Directive 2018/957 addresses a wide range of issues related to the rights and conditions of posted workers, but only addresses one of the points discussed above directly, namely that of wages. The directive replaces the reference to ‘minimum rates of pay’ to ‘remuneration’ (article 3c). It should thus be easier for unions to use collective action in order to ensure wages above minimum levels in the post-override situation. However, the other two issues have been left untouched, and hence the override has been classified as a modification, as the situation of future similarly situated plaintiffs has only been partly changed.

A clear example of insulation is C-285/98 Kreil, where the Court ruled that excluding women from the military falls under the scope of EU law and violates its rules on discrimination. Directive 2000/78/EC⁶ insulates rules against discrimination based on age from having similar effects on the armed services, by introducing a new explicit exception for the armed services when it came to age discrimination.

Another, though less clear and more peculiar, instance of insulation in the sample is C-147/06 SECAP, where the court ruled against an Italian law that automatically excluded tenders of public work contracts that fell below a certain price level. While many governments⁷ defended the Italian law, the Court struck it down with the support of the Commission, Slovakia and Lithuania, arguing that general principles of EU law as well as treaty provisions against discrimination prohibited laws which had the effect of putting contractors from other member states at risk of losing the competitive advantage that their lower prices might entail.⁸ Like Laval, the case concerned the conflict between regulation of wages, prices, working hours and the like (primarily in older member states) and the competitive advantage low wages, prices and long working hours might entail for market operators in other member states.

A subsequent directive⁹ addressed this theme directly, and increased the possibilities for member states to take into consideration social, environmental and other non-market concerns in public procurement, and the weight of the lowest cost criterion in selecting tenders was reduced (ETUC, 2014; van den Abeele, 2014).

In one case the override made the situation worse from the perspective of potential similar plaintiffs as compared to the pre-case situation (tentatively called “disintegration”). In C-413/01 Ninni-Orasche the Court ruled that even though an employment is of a very short duration, it still confers the EU-legal status of ‘worker’ on the person concerned. In directive 2004/38/EC, article 7.3 limits the conferral of worker-status to people who have worked for more than one year, or to people who have become involuntarily unemployed during the first year, in which case the worker-status must be granted for at least six months. The plaintiff, Mrs. Ninni-Orasche, would not have been able to claim ‘worker’ status in the post-override legal environment.

The governments who participated in the case¹⁰ initially had no complaints about the idea that even short durations of work could grant worker status. Where they disagreed with the Court was rather regarding how to define ‘voluntary unemployment’, as the plaintiff and others similarly situated would have lost the relevant rights if her unemployment would have been defined as such. But instead of adopting a common European definition of voluntary unemployment, the governments and the EP adopted Directive 2004/38/EC where new requirements, relating to the duration of the work-period, were introduced for the granting of worker status.

Before the judgment, even people who had worked for such a short time as the plaintiff could have been granted worker status on a case-by-case basis. However, when the Court started to undermine the governments’ control over this system, the governments instead made it impossible to obtain worker status in situations such as the one the plaintiff found herself in. Legal integration took a step back, even as compared to the status quo ante.

Discussion

As shown above, political overrides do happen, but constitutional ones are absent in this material. In many of the cases above, and most of the cases in the sample, the Court encountered resistance in regards to the scope of EU law, and by implication to its mandate. For example, Watts raised the issue of whether hospital treatment by the NHS constituted a provision of services in the meaning of the Treaties. However, no revisions of the Court’s mandate have taken place in the responses here studied. Below, I discuss the results in regards to how often overrides occur, their relationship to the treaties and normative issues, and conclude with a few take-aways for future research. Table 1 gives an overview of the number of overrides, along with the number of cases where other, non-override legislative responses occurred (discussed below). A.II includes a longer table with the names, numbers and categorizations for all cases in the sample.

Table 1. Responses to resisted CJEU decisions.

Political overrides:
Nullification	3^a
Modification	1^b
Insulation	2^c
Disintegration	1^d
Constitutional overrides	0
Other legislative responses	7^e
No response	13^f

^aC-184/99; C-209/03; C-372/04.

^bC-341/05.

^cC-285/98; C-147/06.

^dC-413/01.

^eC-175/98; C-110/01; C-322/01; C-9/02; C151/02; C-290/04; C-455/06.

^fSee appendix A.II for the rest of the cases.

In regards to whether overrides are rare or frequent events, no theory which predicts overrides give any strong claims in regards to how often they are expected to happen, which makes the question hard to answer. Moreover, it would be wrong to gauge the relevance of the threat of override based on how often they happen. Threats can be real even though their execution is rare (cf. Kelemen, 2012, p. 45). While perfect information on part of strategic judges would lead us to expect zero overrides, we do not know exactly how much information judges have (Larsson & Naurin, 2016). We do know that their information is not perfect, which should lead us to expect at least some overrides (seven out of 27, see Table 1), which is also what I find. To answer how often overrides happen, further and quantitative research needs to be conducted, which is discussed more below.

It should also be made clear that while seven cases led to an override, just as many led to other legislative reactions. For example, C-9/02 de Lasteyrie du Saillant gave rise to a lot of Commission activity, and what could be called a failed ‘codification’ of the principles of the case, using Martinsen’s (2015) terminology. An override of C-151/02 Jaeger was proposed but blocked by the European Parliament (Martinsen, 2015, pp. 103–118). In regards to Grzelcyk, subsequent directives codified the crucial principle that application for social assistance could not lead to an automatic withdrawal of a right of residence (Blauberger & Schmidt, 2017, p. 443, see also section 16 of A.II). Furthermore, C-175/98 Lirussi and Bizzaro, C-285/98 Kreil, C-110/01 Tennah-Durez and C-322/01 DocMorris all concerned provisions which were affected by subsequent legislation. These are listed as Other legislative responses in Table 1.

A key takeaway of this article is that even issues where the Court interprets the treaties directly can be the targets of successful political overrides. As discussed, treaty interpretations do not shield the CJEU from the kind of overrides discussed in Separation of Powers-theories. One could object that any legislation which goes against the Court’s statements regarding what a treaty article should imply in a given situation is a direct challenge to the Court’s authority, since primary law (treaties) are supreme to secondary law (e.g., directives). However, there is no reason to believe that the Court holds this view, as it has in many instances stated that the treaties are interpreted differently in situations where applicable secondary law exists as compared to where it does not. One example is Laval, where the Court explicitly pointed to a new directive as the reason for why it here changed course as compared to earlier cases, where it had interpreted similar issues in the light of the treaties directly (C-341/05 points 81-82, ff.). Moreover, several of the overrides here studies was the subject of subsequent adjudication, where the Court did not overturn them based on a violation of the treaties (cf. Martinsen, 2011; Blauberger & Schmidt, 2017).

Therefore, I argue that none of the overrides here presented should be seen as a challenge to the rule of law in the EU. Whether they affect judicial behaviour, in the sense that the Court modifies its decisions to try to avoid overrides such as these, depends on whether the Court holds extra-legal preferences in regards to specific policy or the reach of European legal integration. While research on the CJEU has long operated under the assumption that the Court has had a unified preference for more European integration, this assumption has increasingly come under scrutiny (Pollack, 2013, p. 1262 ff.).

To conclude, this article has argued that it is crucial to distinguish between political and constitutional overrides for both theoretical and normative reasons, as political overrides are easier to achieve; and their importance is based on assumptions of broad political preferences on part of the judges or arbitrators who make the judgments. Without these preferences, political overrides are not necessarily a normative problem. In a democracy, it is the right of the popular majority to correct policy in line with its own preferences, even after court decisions, as long as they respect the rule of law while doing so. Constitutional overrides have a much narrower definition, in turn resting on a much more narrowly focused idea of why they matter, i.e., as direct challenges to the constitutional order. In the EU, political overrides do happen, and these can be of substantial importance from a political and theoretical perspective, even when they respect the CJEU’s legal and constitutional authority.

While this study does not address overrides in other systems, scholars studying them should consider what kinds of overrides could be relevant, while taking the definitional issues here discussed seriously. As this article has revealed, whether scholars reach the conclusion that overrides exist or not depend a lot on how the concept is defined. But even in the EU, where the very idea of override has been most strongly contested, they do happen, which implies that the phenomenon should be taken seriously in other settings as well.

Finally, the Separation-of-Powers theories which garner support from these findings argue that member state preferences restricts Court’s scope of manoeuvre alongside the decision-making rules and legal architecture of the system, and not in place of it (Larsson & Naurin, 2016). While these theories are sometimes misunderstood as implying that the member states have the final say on all issues of importance in the EU, the classical conclusion of these theories was precisely the opposite. As stated by Ferejohn and Weingast (1992, p. 263), ‘if we can say nothing else with certainty, we can say that there is no ‘last word’ in politics’. An override can itself be interpreted by courts in ways which governments dislike, which might in turn trigger further political reactions, and so on. Override is one out of many potential important events which can shape the balance between law and politics, and neither the states nor the Court hold the last word.

Acknowledgements

I would like to thank Daniel Naurin, Mark Pollack, Jan Rovny, Taylor St John, Andreas Moberg, my colleagues at the Centre for European Research at the University of Gothenburg, and the editors and reviewers at the Journal of European Public Policy for very helpful comments and suggestions. The project has been partly funded from the Swedish Research Council (Vetenskapsrådet).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Olof Larsson

Olof Larsson is a Researcher at the Department of Political Science at the University of Gothenburg.

Notes

1 Attitudinalists argues that the Justices of the US Supreme Court primarily rule according to their political attitudes regardless of whether their decisions result in overrides or not (Segal, 1997).

2 Belgium, Denmark, Germany, France, the Netherlands, Austria, Finland, UK.

3 Article 7.4.

4 Directive 2011/24/EU article 3.

5 Only UK, Ireland and Finland and Commission opposed this, while France and Belgium supported it.

6 Directive 2000/78/EC.

7 Austria, the Netherlands, France, Germany and Italy.

8 C-147/06 point 26

9 Directive 2014/24/EU of the European Parliament and of the Council of 26 February 2014 on public procurement and repealing Directive 2004/18/EC.

10 Denmark, Germany, Austria, UK.