A Dialogue of Unequals – The European Court of Justice Reasserts National Courts’ Obligations under Article 267(3) TFEU

In a judgment delivered on 4 October 2018,¹ the European Court of Justice (‘the Court’) ruled for the first time that a failure by a national court adjudicating as the court of last instance to make a preliminary reference constitutes an infringement under the terms of Article 258 TFEU. This ruling arose from a decision by the French Conseil d’État not to make a second reference after an initial preliminary reference in the proceedings that had given rise to the Accor case.² The ruling was simultaneously groundbreaking and not entirely unexpected. Groundbreaking, in the sense that it is the first finding of such a violation in the context of an infringement proceeding. Expected, in the sense that the possibility of such a ruling had been well established and was, in reality, the logical outcome of a much older line of case law.³ There is a certain historical irony to the fact that the Court should have taken the final step in reasserting the limited scope of the acte clair exception established in CILFIT ⁴ in a case that involved the Conseil d’État. The concept of acte clair had, indeed, been constructed by the Conseil d’État itself in order to justify decisions not to refer certain questions of interpretation to the European Court of Justice.⁵ Although the circumstances of the case show that this was not a clear-cut refusal to comply with Article 267, the reaction of the President of the Litigation Section of the Conseil d’État, Jean-Denis Combrexelle,⁶ illustrates the broader issues at stake and the need for the Court to reassert its authority as the supreme court of the EU legal order as well as the obligations of national courts of last instance. As Judge Gervasoni of the General Court indicated in a reply to President Combrexelle, much of this conflict revolved around the persistent misunderstanding by national supreme courts as to the scope of the CILFIT exception and as to the meaning of ‘judicial dialogue’ in the European Union:⁷ a dialogue, yes, but between unequal partners when matters of EU law interpretation are concerned. The dialogue established between the European Court of Justice and national courts should not lead the latter – supreme courts, in particular – to overestimate their autonomy in interpreting EU law and deciding when to refer preliminary questions.

Background of the case

The case before the European Court of Justice is a new episode in a long saga which started in 2001 before the French administrative courts when the companies Accor and Rhodia challenged the French rules intended to avoid the economic double taxation of dividends. They were challenging the system of ‘avoir fiscal’ and ‘précompte’ which later ended in 2005 under Article 93 of the Finance Law of 2004.⁸ After the administration had rejected their claims, both the court of first instance⁹ and the court of appeal¹⁰ granted their requests on the basis of freedom of establishment and the free movement of capital but without having referred the issue to the Court.¹¹ The Conseil d’État quashed the appeal judgments and, before ruling on the facts, made a preliminary reference.¹² The Court ruled that the French regime was incompatible with Articles 49 and 63 TFEU insofar as it created a difference in treatment between dividends distributed by a resident subsidiary and those distributed by a non-resident subsidiary company.¹³ The Conseil d’État then ruled on both cases,¹⁴ establishing both evidentiary requirements for the reimbursement of advance payments made in breach of EU law and the amounts that could be claimed by the companies. In doing so, it ruled on an issue it had not referred to the Court but which had been decided a few weeks earlier in the Test Claimants case:¹⁵ the impact of freedom of establishment and the free movement of capital on the taxation of sub-subsidiaries established in other member states. In Test Claimants, the Court specified that the two freedoms precluded a member state that allows a resident company to avoid economic double taxation when it receives dividends from another resident company from refusing the same deduction when the resident company receives dividends from a non-resident company, even when foreign corporation tax has not or has not been wholly paid by the non-resident company itself but by its own direct or indirect subsidiaries. Following the opinion of its Rapporteur public,¹⁶ the Conseil chose to distinguish the case at hand from the British system at issue in Test Claimants, ruling that advance payments made by sub-subsidiaries did not have to be taken into account when determining the amount that should be reimbursed to the parent company. This understanding of EU law as interpreted in Test Claimants and Accor was, however, not quite as undisputed as the Rapporteur public had made it out to be and the companies subsequently sought to challenge the decision of the Conseil d’État.

The Commission received several complaints concerning both the scope of the reimbursements and the evidentiary requirements set by the Conseil d’État’s judgments.¹⁷ These formed the basis for an infringement procedure under Article 258 TFEU concerning suspected infringements related both to the incompatibility of the rulings with substantive rules of Union law and to a violation of an obligation to make a reference under Article 267 TFEU. The Commission was dissatisfied with the French authorities’ position in reply to the formal notice and the reasoned opinion and thus brought an action before the European Court of Justice on 10 July 2017.¹⁸

The decision of the european court of justice

The Commission’s application for a ruling finding that France had failed to fulfil its obligations under European Union law was based on four complaints. The first three are related to infringements caused by the way in which the Conseil d’État had set about implementing the first preliminary ruling without asking the Court for further clarification. The Commission, therefore, presented its disagreement with the Conseil d’État concerning the proper interpretation of the previous case law pertinent to solving the case at hand. The decision not to refer further questions in order to prevent that disagreement was the object of the fourth complaint based on a violation of Article 267(3) TFEU due to the decision not to make a reference. The Commission’s complaints thus raised the question of a court of last resort’s duties when faced with a new question that the Court has not yet answered but which its members think can be decided based on the first preliminary ruling.

The Court’s judgment clarifies the requirements set out in the Accor ruling. However, the most significant part of the judgment is the first finding of an infringement caused by the decision of a supreme court not to make a (second) preliminary reference.

The complaints

The first complaint alleged that the Conseil d’État had infringed Articles 49 and 63 TFEU in deciding that the taxation of sub-subsidiaries established in other member states should not be taken into account for purposes of reimbursing advance payments made by the parent company. In a domestic chain of interests, such a distribution of dividends would give rise to reimbursement, but not in cross-border cases. This was due to a peculiarity of the French tax regime¹⁹ by which it was impossible for parent companies to offset taxes paid by their sub-subsidiaries against their own taxes. The Commission argued that this constituted a difference in treatment.²⁰

The second complaint was aimed at the allegedly disproportionate evidentiary requirements laid down in the Conseil d’État’s judgments and was primarily based on the principle of equivalence.²¹ The Commission argued that the standards that applied to the documents that companies needed to provide and the amounts that could be taken into account for reimbursement were disproportionate and violated the principle of equivalence. France argued that providing adequate evidence of advance payment was a legitimate requirement for reimbursement²² and that the amount that could be claimed corresponded to the actual amount of the advance payment.²³

In the fourth complaint, the Commission alleged that the Conseil d’État had infringed Article 267(3) TFEU by not making a second preliminary reference before making its final judgments. The Conseil had an obligation to refer because it was ruling as a court of last resort and could not presume that the rules it was establishing were compatible with EU law. In a display of somewhat circular reasoning, the Commission argued that a further preliminary reference was needed due to persistent doubts about the proper interpretation of the previous case law and that these doubts were evidenced by the fact that the Commission had subsequently proved to have a different understanding of what the Accor ruling required.²⁴ In the absence of sufficient certainty about the law’s interpretation, the CILFIT exceptions could not be applied.²⁵ The finding of an infringement in relation to the first three complaints was therefore inherently linked to the fourth. France’s answer to this complaint was twofold: the difficulties faced by the Conseil in applying Union law after Accor were of a factual nature and it had good reason to believe that answers to questions of Union law could be ‘clearly inferred from the case law’ of the Court.²⁶ This was a clear reference to the ‘acte éclairé’ exception,²⁷ which – interestingly – avoided the issue of the ‘acte clair’, which was nevertheless discussed by the Court.

The Advocate General’s Opinion

Advocate General Wathelet²⁸ suggested that the Court reject the second and third complaints, as he felt that the Commission had not proved that the Conseil d’État’s rulings had led France to establish disproportionate evidentiary requirements that violated the principles of equivalence and effectiveness, nor that the cap on reimbursements set by the Conseil d’État was discriminatory.²⁹ However, the Advocate General followed the Commission’s reasoning in relation to its first and fourth complaints. Concerning the first complaint, he agreed with the French Government that member states are not under an obligation to adapt their own tax systems to those of other member states, although he did find that the failure to take into account the taxes levied on sub-subsidiaries constituted a difference in treatment. According to him, this was the appropriate reading of the rulings made in the Test Claimants case, which the Conseil d’État should not have distinguished from the present case in this respect.³⁰

Concerning the fourth complaint, the Advocate General noted this was the first time the Commission had introduced a complaint based on a single instance of violation of Article 267(3) TFEU.³¹ However, he did not find any major difficulties in accepting that such a claim could be made. He supported this by referring to previous case law concerning the possibility of national supreme courts causing infringements found under Article 258 proceedings³² as well as the Köbler ³³ principle of the liability of member states for violations of Union law.³⁴ He also relied on previous statements in the case law about the importance of the preliminary reference mechanism to preserve the uniformity of Union law and the integrity of the Court’s ‘fundamental mission’ to ‘ensure that in the interpretation and application of the Treaties the law is observed’. ³⁵

The Advocate General also insisted on the particular importance of the obligation to make a second reference. He established an interesting parallel between the referring court’s duty to comply with the Court’s interpretation of Union law and the general duty of states to comply with the Court’s judgments finding infringements under Article 260 TFEU.³⁶ The option to make other references in the same case³⁷ becomes an obligation when the appropriate interpretation of Union law remains uncertain.³⁸ After recalling the scope of the CILFIT exceptions, he found that they did not cover all of the Conseil d’État’s choices since it ‘could not be certain’³⁹ that its reasoning concerning sub-subsidiaries would seem as evident to the Court.

The judgment of the court

Although the Court fully agreed with its Advocate General,⁴⁰ it did not always follow his reasoning. The Court found no violation of the principles of effectiveness and equivalence in the evidentiary requirements⁴¹ and rejected the Commission’s claim that the cap on reimbursements was discriminatory, dismissing the Commission’s concerns regarding the risk of shortfall for the shareholders of distributing companies, since that situation was not at issue in the circumstances that led to the two judgments of the Conseil d’État.⁴² The Court did find that the choices made by the Conseil about the treatment of sub-subsidiaries created a difference in treatment between domestic chains of interests and cross-border dividend distributions.⁴³ It relied on its earlier case law, in particular the Test Claimants cases,⁴⁴ to establish that member states that had set up systems aiming to avoid the double taxation of dividends paid to residents by resident companies must ensure equal treatment of dividends paid by non-resident companies unless the differences are justified.⁴⁵ Compliance with Accor would have required putting an end to this discrimination, and the Conseil was wrong to depart from Test Claimants. ⁴⁶

The Court also found an infringement of Article 267(3) TFEU. The judgment refers to previous judgments indicating that national courts’ actions can cause infringements under Article 258,⁴⁷ restating the obligation when ruling as a court of last resort to refer questions of interpretation of the TFEU to the Court and the aim of ensuring the uniformity of Union law.⁴⁸ Perhaps surprisingly, the Court made no mention of the fact that this was a decision not to make a second reference and not a refusal to make any reference at all. This might have led the Court to judge the violation more severely, as Advocate General Wathelet suggested, or, alternatively, to grant the Conseil d’État more leeway.⁴⁹ It agreed that the Conseil d’État could not be certain that its reasoning concerning sub-subsidiaries would be ‘equally obvious to the Court’,⁵⁰ hence the acte clair exception could not apply, a position which is supported by the fact that the Court’s reasoning was, indeed, different. The existence of ‘reasonable doubt’⁵¹ as to the appropriate interpretation of Union law when the national court delivers a ruling against which no judicial remedy is available is enough to rule out the acte clair exception and to find a violation of Article 267(3).

Commentary

The Court’s judgment reaffirms the very limited scope of the CILFIT exceptions to Article 267(3) TFEU as well as the importance of the preliminary reference procedure to ensure the uniform interpretation of Union law across the member states. In this regard, it cannot be considered truly surprising: although this is the first time the Court has found such an infringement, the possibility of such a ruling being made has been evident for some time, especially since the judgment in Ferreira da Silva,⁵² and can be traced back all the way to Köbler and Commission v Italy in 2003, which could be considered the first attempt by the Court to compensate for the lack of any direct appeals mechanism by which it could control the application of European Union law by national supreme courts. Although the Köbler liability principle and the applicability of infringement procedures to judicial violations had been expected to function mainly as dissuasive control mechanisms, they were the first steps in a gradual development which has culminated in this judgment. It could, therefore, be said that the Court has only now completed the edifice it started constructing with its 2003 rulings. It is, however, interesting that this judgment should concern a decision not to refer made by the Conseil d’État, whose tumultuous relationship with EU law and the European Court of Justice is well-known. The members of the Conseil had apparently thought that they could decide on their own whether Test Claimants ⁵³ constituted binding precedent in the cases they had to decide.⁵⁴ The judgment thus provides a welcome reminder of national supreme courts’ duties under Article 267(3), not only to ensure the uniformity of Union law but also to guarantee adequate protection of individuals’ rights under the Treaties.

Confirming the responsibility of member states for their supreme courts’ actions

This judgment can be read as a further illustration of what Daniel Sarmiento has termed the European Court of Justice’s ‘constitutional mode’,⁵⁵ in that it is a further indication of the Court’s willingness to affirm its position as the supreme court of an increasingly federal judicial system.⁵⁶ Finding an infringement in a single decision not to refer by a national court ruling in last resort, as opposed to a consistent line of case law contrary to that of the Court, is certainly an important step. It is a reminder of national supreme courts’ constitutional duties within that legal order, as actors that play a crucial role in ensuring the uniformity of interpretation of Union law, not by interpreting it themselves but rather by allowing the Court to fulfil its mission. It is also another stage in the Court’s struggle to establish its position above the national supreme courts and to compensate for the lack of formal hierarchy in the Union’s judicial system. ‘Judicial dialogue’ within the preliminary reference procedure cannot, from the Court’s point of view, mean that national courts of last resort have the capacity to provide autonomous interpretations of EU law when confronted with new questions. Giving the Commission the ability to use the infringement procedure under Article 258 TFEU is certainly not an ideal solution if one wants to ensure that national courts comply with their obligations, but it is, unfortunately, one of two somewhat ill-suited solutions the European Court of Justice has come up with to compensate for the lack of a true appeals procedure,⁵⁷ the other being Köbler liability.

However, this is not a truly new development in Union law; rather, it is the first actual application of a rule that the Court has been establishing for some time. The fact that infringements of EU law can be caused by supreme courts has long been established by the case law, even though older rulings were less explicit about engaging with the judicial source of the infringement and the specific issue of regarding a ‘failure to refer’ as an infringement. Under the 258 procedure, in Commission v Italy ⁵⁸ the Court indirectly found the Corte di cassazione to be responsible for an infringement.⁵⁹ In that case, the judicial source of the infringement was, however, not explicitly identified; the Court found a violation in the failure of the Italian Republic to modify a piece of legislation which was being consistently interpreted by the courts in a way that was incompatible with EU law. Although the European Court of Justice’s ruling is therefore phrased in such a way as to avoid explicitly dealing with the specific nature of a judicial violation, the source of the infringement was clearly the national case law, as evidenced by the Commission and the Court’s references to the Corte di cassazione’s case law.⁶⁰ Moreover, Advocate General Geelhoed, in his Opinion in this case, presented the issue as raising the question of ‘the consequences which must follow from national case law which does not comply with [EU law]’ and refers to the Opinion of Advocate General Léger in Köbler as a useful guideline.⁶¹ This provides ample evidence that the ruling was, in fact, a ruling on a judicial violation of EU law.

In Commission v Spain,⁶² the Court found an infringement directly caused by a decision made by the Tribunal Supremo, but without explicitly dealing with the specificity of the judicial origin of the infringement. The Court avoided dealing with this issue, concentrating instead on the substantive violation at issue. Another interesting aspect of this judgment is that it could have been the first ruling on an infringement caused by a violation of Article 267(3), since the Tribunal Supremo had not made a reference to the Court before making its problematic decision. Spain argued that there was an ambiguity concerning the object of the complaint presented by the Commission, which could have been interpreted as being based on the decision not to refer.⁶³ The Commission denied that the procedure concerned a violation of what was then Article 234 TEC⁶⁴ so that the issue was not discussed further. The Spanish Government did, however, make an interesting point concerning the precise object of the complaint, as the decision not to refer by a court of last resort was clearly a significant contributing factor in the substantive violation at issue. Although the Commission had refused to engage with the issue of a violation of Article 267(3), the fact that this could have been part of the complaint and therefore of the judgment on infringement illustrates the inextricable link between substantive violations by national courts and decisions not to refer. It is difficult to establish a clear separation between violations of Union law caused by the interpretation of a specific rule by a national court and the fact that the same court decided not to refer a question concerning the interpretation of that rule – it can always be supposed that the appropriate interpretation could have been given by the Court, had a reference been made.

This link between violations of substantive norms of EU law and violation of the obligation to refer is evident in the case law concerning judicial violations of Union law in the context of the Köbler principle of liability. In Köbler, the finding that the Verwaltungsgerichtshof ought to have maintained its request for a preliminary ruling because it had an obligation to refer under what was then Article 234 TEC was an important element of the Court’s reasoning;⁶⁵ this led to a finding of a violation of Community law by the Austrian court, although the violation at issue was the free movement of workers, which the Court held was not manifest.⁶⁶ The representatives of the Republic of Austria were aware of the link, since they presented an argument to the effect that the conditions governing liability were not satisfied in regard to the Verwaltungsgerichtshof’s refusal to make a reference – not because the court had not violated the provision, but because Article 177 TEC was, according to them, not intended to confer rights on individuals. While the Court did not answer this argument at the time, its later case law shows that it would have disagreed: in the main proceedings at issue in Ferreira da Silva,⁶⁷ the claimants clearly based their claim for damages both on the Portuguese supreme court’s incorrect interpretation of a provision of EU law and on its related failure to comply with its obligation to make a reference under Article 267(3). Although the Court did not explicitly answer the question of whether a violation of Article 267(3) could itself form the basis of an action based on the Köbler principle, it did not give any indication that it could not. Moreover, Ferreira da Silva contains the first finding that a Court had violated its obligation to refer under Article 267(3) since Köbler.⁶⁸

The Commission v France judgment should, therefore, be read as a further stage in the Court’s efforts to establish some level of control over the misuse of the CILFIT exceptions by national courts of last resort rather than as a truly novel development. As evidenced by the link established as early as 2003 by Advocate General Geelhoed, the Köbler and Commission v Italy rulings were inextricably linked – two precedents attempting to establish some degree of verticality in a judicial system devoid of any strictly hierarchical relationship between national courts and EU courts. Both threats – i.e. damage claims before national courts and infringement procedures before the Court –were meant to act as a deterrent that would leave national courts, especially those at the helm of their national judicial systems, with little choice but to comply with EU law. Compliance with substantive norms of EU law often requires a preliminary reference in order to obtain the correct, or in any event authoritative, interpretation of EU law. Because of the structure of the EU judicial system, said threats cannot be directly enforced at the judicial level by direct hierarchical mechanisms linking national courts to the European Court of Justice but are addressed instead to the member state as a whole. Although explicit findings that a failure to refer constituted a violation of Article 267(3) TFEU had previously only appeared in the Köbler line of precedents, there was never any reason to suppose that the same finding could not be made if the sanction for the violation was not being pursued by individual claimants appearing before national courts but by the Commission itself.

However, one important element does indicate a shift in thinking. In Köbler, the fact that the appropriate interpretation of Union law remained unclear seemed to constitute a mitigating factor in evaluating the gravity of the violation of substantive law,⁶⁹ although a violation of Article 267(3) had been established. Here, the very fact that there were doubts and the Conseil did not refer a second question was enough to constitute an infringement under Article 258 TFEU. Some doubts were expressed by French scholars as early as 2009, following the first Conseil d’État ruling, indicating that the very issue which would cause an infringement in this case should be submitted to the European Court of Justice for a preliminary ruling.⁷⁰ However, the Rapporteur public in the 2009 case, while inviting the Conseil to make the first preliminary reference, did not really deal with the issue of the reimbursement of taxes in the case of sub-subsidiaries established in other member states,⁷¹ while the Rapporteur public in the 2012 case very clearly stated that she believed the Test Claimants ruling must be distinguished from the case at hand before the Conseil.⁷² This appraisal of European Union law was not based on any specific evidence, however, and the fact that the Rapporteur public was inviting the Conseil d’État to proceed to distinguishing – in the stare decisis sense – a particular precedent of the Court should in itself have led the Conseil d’État to consider making a reference. The scope of a specific precedent should, after all, be determined by further case law from its author. Other courts cannot be certain that their interpretation of precedent will be the one chosen by its author; Advocate General Wathelet makes precisely this point to justify finding an infringement in this case.⁷³ Thus, although the violation of the substantive rule at issue by the national court may appear less egregious than in other instances, the fact that the Conseil d’État thought it could determine the precise scope of a precedent from the European Court of Justice and decide that it was not applicable in this case was sufficient to find a violation of Article 267(3) TFEU.

If there has indeed been a shift towards a stricter approach to the obligation to refer, it can only be considered to have been gradual and has probably been influenced by growing concerns for the impact of refusals to refer preliminary questions on the right to effective judicial protection, as evidenced in the recent case law of the European Court of Human Rights,⁷⁴ which has certainly encouraged the European Court of Justice to adopt a stricter approach concerning the obligation to refer under Article 267(3) TFEU.

A necessary reassertion of national supreme courts’ duties under Article 267 TFEU

The European Court of Human Rights case law concerning violations caused by unjustified failures to refer preliminary references illustrates the importance of the mechanism for the protection of the rights of individuals. Indirect access to the Court is made all the more important by the strict limitations on direct access through the annulment procedure.⁷⁵ These requirements, in addition to the obligation to ensure the uniformity of the law across the Union, make it necessary to ensure that at least the national courts of last resort do in fact make references whenever the interpretation of Union law is unclear. The present judgment can be read as the last step in a gradual process whose latest development could probably better be understood as the Court seizing an opportunity granted by the European Commission’s complaint rather than by any intent to specifically target the Conseil d’État, rather than any other national court, at this specific time.

However, one must not forget that, as is the case for all courts, the European Court of Justice does not choose when to rule and on which issues, but is dependent on the cases brought before it by the parties. What the Court did choose to do, however, was to restate its consistently strict reading of the CILIFIT exceptions and to rule on this case in a manner which, although firmly based on previous case law, led to significant progress in the development of procedural law. This disagreement between national courts (supreme courts, in particular) and the Court concerning the scope of the CILFIT exceptions has long been a cause of concern. In the absence of a direct appeal mechanism, there is no doubt that the strict understanding of national courts’ obligation to refer is, combined with the ‘palliative’ mechanisms of infringement proceedings and Köbler liability, an attempt to establish a more hierarchical relationship with national courts. This is not to say that there must be a purely vertical relationship which could jeopardise the advances made through the more informal and horizontal relationships established via ‘judicial dialogue’. However, judicial dialogue should not result in national courts of last resort using their autonomy to make even more incorrect decisions that directly affect the effective protection of the rights granted to individuals under EU law, nor should it lead to even more inconsistencies in the interpretation of the law. The very fact that national courts are essential components of the Union’s judicial system and indispensable partners of the European Court of Justice also means that there must be some degree of oversight by the Court responsible for the uniform application and interpretation of Union law. This oversight is rendered more difficult if national courts of last instance fail to refer important questions of interpretation to the Court. If there is to be sufficient consistency in the application of the law but also sufficient protection of claimants’ rights, the relationship between the European and national levels of the EU judicial system cannot be horizontal but must contain certain vertical elements.

The dangers associated with national supreme courts using CILFIT to develop readings of Article 267(3) very different from the Court’s have long been discussed. In some cases, the CILFIT ruling seems to have been considered a carte blanche that allows national courts, especially supreme courts, to develop their own approach to the reference mechanism⁷⁶ and an encouragement to interpret EU law on their own if they do not feel a need to refer a question to the Court. This interpretation of the national supreme court’s role is apparent in the President of the Conseil d’État’s Litigation Section ‘response’ to the Court’s judgment.⁷⁷ He writes that national judges ‘were in charge of applying and interpreting both primary and secondary law […] while respecting the great principles defined by Luxembourg’⁷⁸ and that both institutional balance and wisdom dictate that supreme courts should not be restricted to ‘interpreting the obvious’.⁷⁹ The President’s position thus appears to be that national supreme courts such as the Conseil d’État have a duty to interpret Union law, and not only to comply with interpretations given by the European Court of Justice. While it is certainly possible to argue in favour of that position on the basis of the principle of horizontal cooperation – by which national supreme courts are considered partners cooperating with the Court rather than inferior courts simply applying its case law – or indeed on the basis of more pragmatic concerns that systematic references by national courts of last instance would lead to an excessive increase in the Court’s workload, this position has certainly never been explicitly shared by the Court, as evidenced by the strict criteria set out in CILFIT for the acte clair exception.⁸⁰ The Court has always stated that, although no direct hierarchical mechanism was available to enforce this, a court ruling in last resort could only decide not to refer if the interpretation of the rule was so obvious as to ‘leave no scope for any reasonable doubt’.⁸¹ Any questions on the interpretation could therefore only be answered by the Court itself.

It is nevertheless clear that a number of supreme courts such as the Conseil d’État have found it difficult to accept the changes that Union membership has wrought in their national legal systems as well as the existence of another, hierarchically superior court in the Union legal system that is capable of creating new principles and establishing interpretations which they are expected both to elicit and follow. It is well-known that the Conseil resisted the direct effect and primacy of Union law for a very long time.⁸² It also took 16 years before Francovich ⁸³ was accorded full effect in France.⁸⁴ It must be said that, over the past 10–15 years, under the influence of certain members and of Rapporteurs publics such as M. Guyomar, the Conseil d’État’s implementation of European Union law has considerably improved.⁸⁵ Moreover, in this case, there had already been a preliminary reference and the Conseil d’État had not simply ignored the Court’s case law since its Rapporteurs publics had both quoted the relevant rulings. Nevertheless, an excessively generous interpretation of the acte clair doctrine remains a problematic element in the Conseil d’État’s case law (and that of other national supreme courts).⁸⁶ Under the guise of equal judicial dialogue in the EU, this notion is used to avoid making references to the Court in cases where such a reference is necessary under a strict reading of the Court’s case law.

Although it is true that there is no formal hierarchy between the national courts and the European Court of Justice, there is a clear hierarchy in terms of the legitimacy of their respective interpretations of EU law. The Court, being the supreme court of the EU legal order and judicial system, gives authoritative rulings concerning the content and interpretation of norms within EU law. National judges contribute to the elaboration of this legal order but do not enjoy the same degree of legitimacy when making choices as to the appropriate interpretation of norms. Besides, considering the importance of the preliminary reference mechanism for the protection of fundamental rights, such refusals must be considered violations of the subjective rights individuals derive from Union law.⁸⁷ Furthermore, contrary to what President Combrexelle suggests, the present political context, in which greater defiance from national institutions may be feared, should certainly not lead the Court to give up on such requirements – precisely because they are necessary pre-conditions for a number of safeguards.⁸⁸ Clearly, while the use of infringement proceedings is probably not the most appropriate mechanism to establish a degree of control over national courts, it, with the Köbler principle, is the only suitable alternative available in the current judicial system. The criticisms made in the years that followed the Köbler judgment remain valid and the implementation of that principle in national legal orders has been difficult.⁸⁹ It has, however, been integrated into national legal systems, even those where no similar State liability for judicial actions existed.⁹⁰ More importantly, the Court certainly seems to intend it as a deterrent rather than a frequently-used tool affecting legal certainty. Similarly, infringement proceedings should probably not be expected to become a frequent mechanism for judicial review but rather another reminder to national supreme courts that they must comply with the obligation set out in Article 267(3) TFEU.

It is perhaps unfortunate, though, that this infringement concerns a case in which a first reference had indeed been made – a fact that the Court, unlike the Advocate General, does not address. The members of the Conseil likely thought that they had done their duty in referring the initial issue to the Court; the problem arose out of what they seem to have thought was their prerogative to interpret the Court’s case law in order to solve other issues that arose as a result of the first ruling. Advocate General Wathelet explicitly engaged with the specific issue of finding a violation of Article 267(3) in cases in which the Court has already ruled on one preliminary reference in the same case. He suggested that the obligation to refer is even stronger in such cases because it is then related to the proper implementation of the Court’s ruling under Article 260 TFEU insofar as this is necessary to determine the precise meaning and scope of the first preliminary ruling.⁹¹ The fact that the contentious point was not part of the initial reference, yet was necessary to be able to implement the Accor case law, had created a greater obligation to seek clarification from the Court.

From an opposing perspective, the fact that a national court has already made a preliminary reference in the context of the main proceedings at issue might be a mitigating factor insofar as the national court has shown its willingness to ask the Court for guidance. Moreover, in this case, the Conseil d’État had given clear indications that it was willing to engage with the Court’s case law and to implement the Accor judgment. Had the Court engaged with the specific issue of the second reference in the same proceedings, it might have provided details as to the possible distinctions between the gravity of different types of violation. How should a national court decide when it has received sufficient information not to make a second reference? How can it determine whether its questions following a preliminary ruling relate to further, unsolved issues of interpretation or to technical issues relating to the concrete implementation of the Court’s judgment in the national context? The Court seems to have rejected the idea of any distinction, in principle, between an utter refusal to refer during the main proceedings and a refusal to make a second reference. However, the Advocate General’s Opinion clearly indicated that the latter situation was a more serious violation of Article 267(3) TFEU. It is, therefore, regrettable that the Court did not engage with this aspect of the Opinion, nor in any other way with the fact that this was not a refusal to refer but a refusal to make a second reference in a single case. Further details are needed concerning the types of Article 267(3) violation that the Court thinks must always lead to infringement proceedings, as well as further clarification of the degree of autonomy a national court may exercise when implementing a preliminary ruling.

Rather than a response to a single instance of an egregious violation of Article 267(3), this case is an illustration of the ongoing problems experienced by courts (e.g. French administrative courts) in assimilating their role as members of the Union’s judicial system. Here, things might have run more smoothly if the lower courts had made preliminary references, allowing the Court to provide clarification at an earlier stage. There is no justification for the haphazard application of European Union law by the courts of first instance and the court of appeal in these cases,⁹² which should at the very least have applied the law with greater rigour and, in principle, should also have referred questions to the European Court of Justice since there were questions for which no answer had yet been provided. The absence of references at the level of the juges du fond led to considerable delays in obtaining the first preliminary ruling. Moreover, the Conseil d’État and other national supreme courts would do well to revise their interpretation of the acte clair and acte éclairé exceptions and exercise greater restraint in their own engagement with the Court’s case law. Even in the absence of particularly serious violations of European Union law, the decisions made at every stage of the French judicial system, in this case, illustrate the ongoing problems which the Court could legitimately seek to resolve.

The judgment delivered on 4 October 2018 adds more weight to the vertical aspect of judicial dialogue within the EU, a welcome development which at last reveals the full potential of 15-year-old case law. Though not a revolution but, rather, a logical evolution of the law, Commission v France marks an important step in the relationship between national courts and the European Court of Justice. It should not be read as an attack on judicial dialogue or on the competence of national courts to apply Union law; nor does it contradict CILFIT. Judicial dialogue and cooperation should remain at the heart of the Union’s judicial system, but it should be emphasised that the dialogue cannot entail a complete lack of judicial hierarchy if the legal order is to be coherent. The need to reinforce the limited control mechanisms that do exist in the EU judicial system is evidenced by the Conseil’s persistent attitude to preliminary references. National supreme courts, especially in the so-called ‘historical member states’, should know better than to let such situations arise. The (mis)application of EU law by the Conseil d’État and other French courts, in this case, is proof enough of the legitimacy of a ‘sword of Damocles’ hanging over national judges as they rule in last instance. Whether this new judgment will have greater dissuasive power than the previous precedents remains to be seen, however, and its full potential will only be revealed by future case law.