The Highest Dutch Courts and the Preliminary Ruling Procedure: Critically Obedient Interlocutors of the Court of Justice

Little is known about the motives of national courts to request a preliminary ruling from the Court of Justice of the EU (CJEU) or their satisfaction with and implementation of answers. This article aims to fill this empirical gap on the basis of an analysis of judgments complemented with interviews with judges of the highest courts in the Netherlands. This article shows that judges extensively use the procedure and follow its outcome almost without exception, despite some dissatisfaction. This discontent has surprisingly not affected the courts' willingness to refer in future. The findings also downplay the bureaucratic politics and judicial empowerment theses emphasising strategic motives to refer. Instead, legal&#8208;formalist considerations and the desire to contribute to the development of EU law explain most of the references of the Dutch Supreme Court. The decision (not) to refer of the three highest administrative courts is primarily based on practical and pragmatic considerations.

follow-up and satisfaction, the national court's follow-up judgment was compared with the requested CJEU ruling.
For this analysis use was also made of secondary literature and commentaries discussing the reasoning of the CJEU and/or follow-up by the referring court. Interviews played an important complementary role, because it is often difficult to distill particular reasons for referral or the 'true' appraisal of the CJEU answers from national court judgments. 14 Sixteen interviews were conducted with Dutch judges and legal secretaries of the three highest administrative courts. 15 Fifteen interviews were held with judges and Advocates-General (A-Gs) in the Supreme Court; seven in the civil chamber and seven in the tax chamber, while one interviewee served in both chambers. 16 In order to protect the anonymity of interviewees, their names and identities are not disclosed. 17 The reliance on interviews obviously has limitations. The reasons provided by interviewees are often ex post rationalisations or interviewees might be tempted to (un)consciously downplay particular reasons, because they are not in line with their self-image and professional attitude as a judge. In order to alleviate this problem, the interview data were complemented with the analysis of case law, extra-judicial writing of judges and secondary literature to triangulate the data as far as possible. 18 This article aims to fill an empirical gap in the literature. In addition to this academic relevance, this research also is practically relevant as well since it aims to assess whether allegations that the referral procedure is not working optimally and there is growing opposition to CJEU judgments are true. This article will contribute to the literature in the following ways and as such has a relevance going beyond the Netherlands. 19 Firstly, whereas the findings of this study cannot be transferred to other European jurisdictions, they show that we cannot merely base our understanding of the interaction between the CJEU and national courts on the actual judgments. Instead, we need to get 'into the heads and minds' of the key actors. In doing so, secondly, this article shows that that there are not only differences between EU Member States with respect to the use of the preliminary ruling procedure, but also within Member States between different courts and even within these courts between judges. 20 Thirdly, this article focuses on courts that have so far not received much in-depth attention, since the literature has primarily focused on constitutional courts and their interaction with the CJEU. 21 Fourthly, the findings complement a large part of the research to date, primarily of social science studies relying exclusively on quantitative research. 22 In doing so, this study demonstrates the nuance and the complexity that is often hidden in large-scale statistical research. There is a wide variety in the reasons for national courts to refer or not. The findings thus downplay dominant theoretical explanations, including the bureaucratic politics and judicial empowerment theses that mainly emphasise political strategic considerations of judges. This research shows that legal as well as practical and pragmatic reasons are equally, or possibly even more important (Section 5). Fifthly, this article shows that the high implementation rate of CJEU judgments by the referring courts conceals some dissatisfaction with those judgments (Section 3). Even in the Netherlands, where courts have traditionally been compliant interlocutors, there is quite some criticism as to the functioning of the procedure and the CJEU judgments as this introduction has also indicated (Section 4).
14 Besides a legal enumeration of the reasons why there is doubt as to the interpretation of EU law, the highest Dutch courts hardly include an explicit reflection on considerations that played a role in their decisions. 15 Two interviews at the ABRvS were held in September 2015, while the other 14 took place between December 2016 and August 2017. ABRvS (6), CRvB (5) and CBb (5). 16 Two interviews took pace in September 2016, while the other 13 were held between June and August 2018. 17 A number between 0 and 100 was randomly selected for the interviews. Note that no references are made in relation to specific cases to avoid indirect identification of the interviewees. 18 For more strategies, see Krommendijk, above, n. 13, 119-120. 19 The author and his colleague Jesse Claassen will conduct similar studies on states with a different practice of referral in the context of the earlier men-  Before delving into the findings of this research, it is important to sketch the legal context as well as a short background of the work of the four highest Dutch courts. Article 267 TFEU obliges courts or tribunals to refer to the CJEU when a question is raised about the interpretation or validity of EU law in a case in which 'there is no judicial remedy under national law' , provided that 'a decision on the question is necessary to enable it to give judgment'.
There are two exceptions to this obligation for the highest courts, which are commonly referred to as the Cilfit exceptions. The highest courts are not obliged to refer when the CJEU has 'already dealt with the point of law in question' (acte éclairé) or when 'the correct application of [EU] law may be so obvious as to leave no scope for any reasonable doubt' (acte clair). 23 The CJEU did not make it easy for national courts to find an acte clair, because courts 'must be convinced that the matter is equally obvious to the courts of the other Member States and to the Court of Justice '. 24 Before reaching that conclusion, they have to consider the characteristic features of EU law and compare, amongst others, different language versions of the provision(s) of EU law. 25 As will be discussed further below, there is much discussion in the literature as to the viability of the Cilfit test. 26 This is because the requirements for an acte clair seem more difficult to fulfil now than at the time Cilfit was rendered in 1982, because there are 28 (instead of 10) different legal systems and 24 (instead of 7) official working languages. In addition, nowadays EU law extends over much more legal fields. A-G Wahl formulated this problem in a lucid way: 'If one were to adhere to a rigid reading of the case-law, coming across a 'true' acte clair situation would, at best, seem just as likely as encountering a unicorn.' 27 The CJEU seemed to relax the Cilfit exceptions for the highest national courts in Van Dijk, because it held that the fact that other (national) courts ruled differently or did refer a question does not detract from the highest court's conclusion that the matter is clair. 28 Nonetheless, the CJEU decided on the same day in Ferreira da Silva that the Portuguese Supreme Court could not have determined an acte clair because of conflicting decisions of lower Portuguese courts and the fact that that matter frequently gave rise to difficulties of interpretation in various Member States. 29 The CJEU suggested that the Portuguese court's failure to request a preliminary ruling could constitute a breach of article 267 TFEU, (potentially) giving rise to Köbler state liability, leaving that matter for the referring court to consider. It was in Köbler that the CJEU determined that the principle of state liability also applies in case of infringements of EU law that stem from a decision of a national court of last instance. 30  Industry Appeals Tribunal has traditionally been a natural interlocutor of the CJEU, because it deals with economic public law and more than 80% of the legislation over which it has jurisdiction has European roots. 33 The other two highest administrative courts have also become prominent players at the EU level in recent years because of the Europeanisation of the fields of law over which they have jurisdiction, including for instance migration (Council of State) or social security (Central Appeals Tribunal). 34 The tax chamber of the Supreme Court is (one of) the biggest European players in the Netherlands with 36 references in the four year period 2013-2016, but also in comparison with other European tax judges (see Table 1). 35 The civil chamber is also increasingly confronted with EU law, especially in the fields of intellectual property (IP), consumer law, competition law and international private law. The criminal chamber is left out of this comparison, and will be dealt with separately at a later instance. 36 As will be further outlined below, the highest administrative courts are second-line courts of fact while the Supreme Court is the third instance cassation court that only examines questions of law. The Supreme Court also functions differently than, for example, the UK Supreme Court, which grants permission to appeal only when a case involves 'an arguable point of law of general public importance'. 37

| THE FULL IMPLEMENTATION OF CJEU JUDGMENTS
Nearly all requested answers of the CJEU are neatly followed by the referring court in the respective case submitted to the CJEU. During the analysis of the national court's follow-up judgment in the light of the requested CJEU judgment, no case was found in which a Dutch highest court clearly departed from the CJEU. During interviews, there were no judges that doubted the necessity of strictly complying with CJEU judgments. 38 Supreme Court judges held that CJEU judgments simply constitute law and cannot be disputed. 39 The Supreme Court and the other highest courts have never 'dared' to mention that the CJEU has been wrong or to choose a different direction than the CJEU. 40 The highest courts also had no problem with changing their own case law as a result of CJEU judgments. Interviewees referred to this as a 'fait accompli' and held that this is simply 'part of the game' . 41  is generally reluctant to refer in the field of criminal law. In addition, the limited number of references also stem from the fact that the impact of EU law on criminal law is minimal, even though that is changing in recent years with secondary law laying down a harmonized set of procedural rights of suspects.  4 | JUDGES' DISSATISFACTION WITH SOME CJEU JUDGMENTS Even though judges were generally satisfied with the CJEU and its judgments (Section 4.1), almost all judges and, especially highest administrative court judges and judges of the civil chamber, showed some discontent with particular judgments. They lamented that the CJEU did not answer the question asked or did so in an unclear way (Section 4.2), or the CJEU interpreted the national law and the facts of the case wrongly (Section 4.3).

| Cautious satisfaction with the CJEU and its judgments in general
Almost all highest court judges expressed their satisfaction with most, but certainly not all, of the answers of the CJEU. They noted in general that CJEU judgments are useful and clear. 47 The highest administrative court judges, especially those of the Council of State and Central Appeals Tribunal, were more reluctant and noted that the quality of CJEU judgments varies. 48 They were also more critical of the interaction with the CJEU and referred to the absence of a genuine dialogue. 49 One judge held that approximately 80% of the requested CJEU rulings were answered in a satisfactory way, while another noted that three out of ten cases were less satisfactory. 50  noted by interviewees. 53 One explanation for the difference between the highest administrative courts and the tax chamber mentioned by a Supreme Court judge is that highest administrative court judges have more ideas about the outcome of a case and the desired interpretation than Supreme Court judges, also because they generally 'think' more in political terms. With such an attitude, it is more confronting when the CJEU chooses a different direction.
This could result in more dissatisfaction. 54 Most judges, but especially those of the Supreme Court, were able to put their criticism in context. They, firstly, noted that criticism is normal and is something they are familiar with themselves. 55 Secondly, interviewees voiced their understanding for the difficult context in which the CJEU operates, with 28 different legal orders and 23 official languages as well as many different legal fields and specialisations. 56 One Supreme Court judge even expressed his 'admiration' for the CJEU, because his/her experience as a judge in the Benelux Court with three neighbouring countries and two languages showed how difficult the task of an international court is. 57 Within the highest administrative courts, there were more differences between judges. EU law-oriented judges had more understanding for the CJEU than national law-oriented career judges. 58 Some interviewees, especially those of the Supreme Court, also put part of the blame on themselves and argued that the quality of CJEU judgments depends on the order for reference of the referring court and the quality of the reasoning and analysis. 59 As will be argued below as well, deficient CJEU judgments could result from poor information provision by the referring court.
The interviews also revealed another difference between the Supreme Court and highest administrative courts. The highest administrative court judges primarily assessed CJEU judgments in terms of their ability to solve the case at hand.
They did not examine them from a more analytical perspective focused on whether the CJEU answered all questions satisfactorily and in line with its earlier jurisprudence. 60 Several interviewees of the Supreme Court, however, also adopted the latter perspective and primarily voiced concerns about the consistency of CJEU jurisprudence. 61 One interviewee, for example, held that the CJEU pays insufficient attention to the position of the judgment within the 'edifice' of its case law, something that the Supreme Court is very much concerned with also because of its responsibility for judicial law making. 62  One type of problematic CJEU judgments is those in which the CJEU does not answer the questions.
An interviewee of the civil chamber pointed to an older case about the registration of the melody of 'Für Elise' as a trade mark. According to this interviewee the CJEU delved into other issues than the question of the Supreme Court that was disqualified as irrelevant by the CJEU. 68 The interviewees of the tax chamber also pointed to an older case, Arens-Sikken. In this case about inheritance duties and overendowment debts, the CJEU considered that 'there is no need to answer' the second question about certain CJEU judgments, because it considered the question 'not relevant in the present case' . 69 Interviewees of the Council of State 70 and Central Appeals Tribunal 71 also discussed several questions that were dodged by the CJEU.
In Willems, a case dealing with the collection and storage of biometric data, the CJEU did not delve into the data protection directive 95/46 even though the Council of State had paid considerable attention to As mentioned earlier, an unsatisfactory answer can also stem from the order for reference and the fact that certain questions or aspects of the case are not formulated explicitly and accurately enough. Another problem occurs when the CJEU gives unfeasible directions to the referring court.
Judges at the Central Appeals Tribunal 'were at a loss' after van Delft because the CJEU basically entrusted it to 'delve into politics' and to ascertain whether an unjustified difference of treatment between residents and non-residents was made when the Health Care Insurance Act was negotiated. 75 In order to do so, the Tribunal invited several civil servants from the Ministry of Health who were involved in the drafting of the law in order to verify whether Art. 21 TFEU was sidelined in 'backrooms' during discussions between the Dutch government and insurance companies. 76 Tax judges of the Supreme Court mentioned Sopora as difficult to implement. The CJEU determined in this case that the tax advantage for persons working in the Netherlands and residing at a distance of more than 150 kilometres from the Dutch border was not in breach of EU law, unless 'those limits were set in such a way that that exemption systematically gives rise to a net overcompensation'. 77 The Supreme Court in essence had to examine what the effects of the tax advantage were at a macro level based on empirical data. These data did not exist, because they were simply not registered. What is more, the question was whether the Supreme Court should even do this since it is not a court of fact. Fortunately, the analysis of A-G Niessen provided sufficient guidelines for the Supreme Court to decide the case.
Judges also considered it problematic when the CJEU does not take their concerns seriously. In doing so, the CJEU fails to appreciate that the referral decision and the formulation of the order and the questions is based on a careful possible answers, but the CJEU seemed to suggest that the Tribunal wanted to 'push through the throat' a particular interpretation. 82 With respect to the earlier mentioned problem of unclear answers, some judges had the impression that the CJEU increasingly comes with judgments in which the CJEU only gives general guidelines. 83 Some civil chamber judges, for example, noted that a clear yes/no question was asked in Commerz. The question was whether or not the guarantees provided by the port authority are attributable to the municipality of Rotterdam when the sole director of the port authority acted on its own and against the authority's statutes. In its order for reference, the Supreme Court discussed two options: no imputability because the municipality was not involved in the adoption of the specific guarantee or attributability because the municipality determines in general the decision-making process. The CJEU did not choose either of the two options and left this for the referring court to decide 'in the light of all the relevant evidence' . Interviewees, however, noted that the Supreme Court had already identified all circumstances in its order.
There are also several vague CJEU judgments in highest administrative court cases in the field of migration and integration. 84 One good example is A., B., C. involving the question how courts can assess the credibility of a declared sexual orientation of an asylum seeker. According to some judges, the CJEU only mentioned what courts could not do in their assessment. 85 One judge even asked whether it was useful to refer in the end, because of the loss in time and the fact that the CJEU completely left it to the national court to solve the case on the basis of considerations that the Council of State had already identified and discussed before referral.
Another example is Martens about the Dutch three-out-of-six-years rule, which makes the continued grant of funding for higher education outside that State subject to the rule that the student applying for such funding has resided in that Member State for a period of at least three out of the six years preceding his enrolment. The CJEU only made clear that this rule is not consistent with EU law, but did not provide much clarity as to how and which factors the Central Appeals Tribunal should examine to find a genuine link with the Netherlands.
The previous analysis seems to put all the blame on the CJEU. Some judges, however, admitted that the deferential answers of the CJEU also stem from general or unclear question from the referring courts, as was mentioned the public' to the case in the main proceedings and concluded that it 'appears, subject to the checks to be made by the referring court' that the hyperlink to websites with leaked nude photographs constituted such a communication. 91 Tax chamber judges were also content with concrete judgments, that were criticised in the literature for being too interventionist. 92 They considered that such judgments are unavoidable when the interpretation could only lead to one outcome. 93 Some interviewees also held that the advantage of outcome cases is that they limit the discussion between the parties after the CJEU judgment and might do away with the need to organise another hearing. 94

| Incorrect reading of facts and national law
It happens, albeit in fewer cases, that the CJEU interprets the national legal framework wrongly.
The CJEU did not answer the third question in the social security case Franzen because it had misconstrued the Dutch law considering that the Central Appeals Tribunal is obliged to apply the hardship clause in order to remedy an unacceptable unfairness. 95 The CJEU failed to appreciate that this is not an obligation, but merely a possibility. In addition, it is not for the referring court to apply the clause, but the administration. Dutch judges were critical about the CJEU judgment and noted that it was 'annoying'. They lamented the lack of communication and the fact that the answer of the CJEU shows that the CJEU fails to appreciate the importance of the question at stake. Judges wondered whether there is an understanding of the Dutch social security system, also because this was the second time that the CJEU made a mistake in relation to the hardship clause. 96 One interviewee did not want to play a blame game but noted that the answers of CJEU caused problems and did not make it possible for the Supreme Court to decide the case. The tax chamber consequently decided to refer follow-up questions. One could debate whether only the CJEU is to be blamed, because the Central Appeals Tribunal had also provided minimal information on Dutch law. One Dutch judge recognised this as well and noted that they will pay extra attention to the formulation of the questions and the order for reference in future references.
The latter option of re-referral was also discussed recently in two joined cases on corporation tax and a single tax entity after a CJEU judgment based on a misunderstanding of Dutch tax law. 97 A-G Wattel noted that the Supreme Court could also be blamed for these 'blunders' since it did not ask the correct questions. He, nonetheless, advised against re-referral because the case could be solved based on CJEU case law. 98 Because the resolution of the cases was pending at the time of the interviews, judges were reluctant to speak about the joined cases, only noting that the issue at stake is wildly complicated.
In Van der Ham, the CJEU based itself on a different time period as a result of which the CJEU gave an interpretation of another Regulation than the Regulation about which the Council of State had asked questions. This mistake was not problematic, because the specific provision in the regulation was the same. 99 A related problem arises when the CJEU bases itself on the wrong facts. This is particularly difficult for a cassation court like the Supreme Court for whom the facts are established. Several interviewees noted that the CJEU should keep away from such factual determinations, also because it complicates subsequent decision making. 100 One older case of the civil chamber is Ten Kate Holding Musselkanaal in which the CJEU relied on the intervention of the Commission and the Dutch government. 101 The Supreme Court mentioned this problem in its follow-up judgment and stipulated that Dutch procedural law does not allow it to take into consideration the facts as determined by the CJEU. 102 One interviewee held that he/she was flabbergasted and noted that this judgment came as a complete surprise.
In Ladbrokes on games of chance via the internet, the Supreme Court had explicitly stated in its order for reference that it was established in cassation that the betting activities are limited in a consistent and systematic manner. 103 The CJEU, however, examined whether this indeed was the case and eventually concluded that it is not, contrary to the Supreme Court. 104 This finding resulted, according to one interviewee, in 'an enormous struggle' for the Supreme Court. It eventually ruled that this judgment is very much interwoven with factual assessments that are only subject to cassation to a limited extent. 105 The tax chamber was confronted with a similar problem in one customs case about the tariff classification of stand-alone music devices (Sonos). The CJEU itself had consulted the website in order to verify how the product was presented to consumers. 106 Several interviewees noted that the CJEU should not have done this, because facts are already established for a court of cassation and not subject to review. This determination did not prove problematic for the Supreme Court to deliver its final judgment.

| JUDGES' MOTIVES (NOT) TO REFER
One could imagine that when judges are not always satisfied with the requested CJEU judgments, this could discourage them from referring future cases. Surprisingly, the dissatisfaction does not seem to affect the courts' willingness to refer. property, but they argued that the Supreme Court tries to 'ignore this annoyance'. 108 This attitude also explains why the highest courts have been reluctant to ask follow-up questions because of a CJEU judgment. 109 This has only happened in a limited number of instances, discussed above. 110 This raises the question as to how it can be explained that the Dutch highest courts continue to refer cases to CJEU. Obviously, in the majority of cases a CJEU answer is considered necessary simply because there is uncertainty regarding the meaning of a particular provision, for example, because it is used in contradictory ways in EU rules or has not been interpreted by the CJEU before. This article, nonetheless, shows that this logic does not fully capture the referral practice while also concealing interesting differences between the Dutch highest courts and additional considerations that play a role. It will be shown that the decision to refer of the Supreme Court is informed by legalist considerations (Section 5.1), while highest administrative courts are more pragmatic and practical (Section 5.2). Especially the highest administrative courts also refer sometimes to obtain the CJEU's authority (Section 5.3).

| Legalist considerations: The supreme court as a loyal contributor to the development of EU law
Most Supreme Court judged mentioned legal-formalist considerations, relating to their obligation to refer under Article 267 TFEU as the primary reason behind their decision (not) to refer. 111 This shows that the Supreme Court feels responsible for the correct application of EU law. 112 Interviewed A-Gs and external observers confirmed this loyal and 'integer' approach of the Supreme Court. 113 As will be outlined in the next section, only a few highest administrative court judges mentioned the formalist reason that a case is referred with the idea of complying with the obligation to refer and rather pointed to a 'natural reluctance' to refer.
Closely related to the loyalty of the Supreme Court is the wish to contribute to the development of EU law. Several Supreme Court judges mentioned this as a general consideration behind the decision to refer. Supreme Court judges emphasised that it has the task of contributing to the development of EU law. 114 One Supreme Court judge, for example, noted that 'You are not here for yourself, but also for the rest of Europe'. 115 He/she mentioned that the idea is widely shared in the Supreme Court that it is important that EU law reaches its full potential. According to him/her, the delay of 1.5 years when referring should be taken for granted and is 'the price that is paid' for being part of the EU system. 116 Another interviewee referred to the 'sense of responsibility' and the role of the Supreme Court as a 'wheel in the gear chain' of the European legal order, while another stressed the importance of the uniformity of EU law and the need to avoid divergences ('one of the most severe sins'). 117 A few highest administrative court judges also made similar remarks about the need to contribute to the development of EU law, but noted at the same time that the primary purpose of the procedure is to solve a dispute. 118 As will be outlined in Section 5. Another judge proposed a strict application and mentioned, in line with Cilfit, that the Supreme Court should be 'convinced' that there is no doubt about the interpretation of EU law. He/she rejected the logic of some highest administrative court judges that no reference is needed when the question is 80% clair and argued that the Cilfit criterion is not whether a court is able to come up with a solution itself. 123 The Supreme Court hence prefers to play it safe. 124 The tax chamber is even more faithful to Cilfit than the civil chamber and is even criticised for referring too easily. One lower court tax judge, for example, mentioned that the CJEU has sent the tax chamber an implicit message 'Are you there already again?' with its judgments handed by a three judge formation without A-G Opinion. 125 Article 20 of the Statute of the CJEU stipulates that the latter happens when 'the case raises no new point of law'.
The (too) cautious Cilfit approach of the tax chamber is corroborated by data in the way in which the CJEU handled references from the highest Dutch courts (  A possible explanation as to why the tax chamber adheres to Cilfit so strictly is to prevent lower courts from being tempted to refer. Roughly two-thirds of all Dutch references come from the highest courts, whereas this has traditionally been the opposite for the majority of Member States where lower courts have been more active. 126 The explanation of preventing lower courts' referrals coincides with recent studies showing that the highest courts have 'reconquered' control from the lowest courts over the application of EU law and references to the CJEU. 127 The cautiousness of the tax chamber can be attributed specifically to the famous Van der Steen incident. The Supreme Court failed to refer the question as to whether a natural person carrying out all work in the name and on behalf of a company is himself a taxable person within the meaning of the Sixth Directive. The Amsterdam Court of Appeal subsequently referred this question to the CJEU. 128 In addition, the tax chamber is watched more closely by legal practice and academia than the highest administrative courts. It is not uncommon to find five commentaries in relation to all the steps in the reference procedure: the opinion of the A-G in the Supreme Court, the Supreme Court judgment in which a referral is made (or not), the opinion of the A-G in the CJEU, the CJEU judgment, an additional opinion of the A-G at the Supreme Court and the eventual judgment of the Supreme Court. By contrast, only a handful of judgments of the Dutch highest administrative courts are commented upon, while some judgments of the Council of State are not even published. The tax chamber is thus under more external pressure than the highest administrative courts.
In conclusion, the Supreme Court primarily adopts a legalist framework and primarily refers for legal-substantive reasons. The Supreme Court considers it important to contribute to the development of EU law and it adopts Cilfit strictly. This loyal attitude explains why the Supreme Court continues to refer, despite the earlier mentioned discontent with some CJEU judgments.

| Pragmatic and practical considerations: The highest administrative courts' natural reluctance to refer
By contrast, there is a natural reluctance to refer among highest administrative court judges. 129 This reluctance does not relate to the quality of CJEU judgments, but can be attributed to an idea that strict compliance with Article 267 TFEU would result in weekly references, especially in Europeanised fields such as migration. 130 The highest administrative court judges' first instinct is to solve a dispute and decide themselves. They held that it is not necessary to immediately refer when there is some doubt about the interpretation of EU law. 131 Several highest administrative court judges, for example, held that when the question is 75-80% clair, there is no need to refer. 132 Some judges thus proposed a 'lighter test' than Cilfit. 133 Interestingly, the Council of State has used the aforementioned CJEU judgment in Van Dijk, in which the CJEU seemed to loosen the Cilfit requirements, as an additional justification for non-referral in several migration cases. 134 Some highest administrative court judges even acknowledged that they (implicitly) 126 Lower courts made two tax law references in 2013-2016 and seven references in the field of customs. Six of these seven questions dealt with the valid-apply the less strict Köbler 'test' in order to avoid that a non-referral gives rise to state liability 'in the exceptional case where the court has manifestly infringed the applicable law'. 135 The highest administrative courts tend to examine whether the case can be solved on other grounds, preferably national grounds, before delving into the intricacies of EU law and examining whether the Cilfit exceptions are fulfilled or not. 136 By contrast, the Supreme Court studies all EU law aspects before asking the question whether a reference is indeed necessary. 137 One interviewee added that even when a case can be solved on other grounds, the Supreme Court still examines whether that is also desirable. 138 The divergence between the Supreme Court and highest administrative courts in the adherence to Cilfit can also be attributed to the judicial law-making function of the Supreme Court as a third instance court. The highest administrative courts are also courts of fact and often rule as second instance courts. 139 The reasonable reading of Cilfit also leaves more room for pragmatic and practical considerations. 140 This includes, for example, case-specific reasons that relate to the importance of the questions concerned or efficiency reasons concerning the consequences of referring in terms of the delay. The importance of the issue at stake plays an important role for highest administrative court judges. The decision (not) to refer was presented by highest administrative court judges as a balancing of competing considerations: the importance of the question versus the consequences of the delay in terms of cases to be put on hold. 141 Issues of minor importance are more easily decided upon by highest administrative courts without a referral. The same could also happen when too many cases are affected by a referral.
In such a case, highest administrative court judges do not find it desirable that many cases are put on hold for an uncertain period. 142 This is especially a consideration in the field of migration where there are often many, possibly hundreds of cases, in which the same question is relevant. 143 Interviewed highest administrative court judges acknowledged that justice would come to a standstill if every question of EU law about which there is doubt were immediately referred to the CJEU. 144 The current President of the Aliens Chamber of the Council of State likewise held that a responsible judge takes into consideration the consequences of such a delay. 145 The delay explains why the Council of State did not refer questions about the intensity of review of the credibility assessment of the asylum claim in relation to Article 46(3) of the Asylum Procedures Directive, even though it held that the matter was far from clair. 146 It decided the case itself, because a referral would mean that the Council of State 'could almost shut down' because a very large number of cases had to be put on hold. Nonetheless, when the stakes are high, this could also be (an additional) reason to refer for courts.
Interestingly, the civil chamber takes pragmatic and practical considerations into account in summary proceedings in which it is not obliged to refer. 147 The urgent character of summary proceedings does in principle not allow for a preliminary reference, even though most proceedings are no longer urgent once they reach the Supreme Court. 148 135 Interviews 10, 18; Case C-224/01, Köbler, ECLI:EU:C:2003:513, para. 53. Note that some A-Gs of the civil chamber also rely on this test in some opin-Interviewees, nonetheless, considered that there should be 'something special' with the case. 149 The civil chamber feels more compelled to refer when the case deals with a question in which there is hardly any CJEU jurisprudence or when few cases reach the Supreme Court in the main procedure. 150 One example where the latter happens frequently is procurement law. The referral in Connexxion was based on the idea that 'we have to grab the chance', otherwise such cases would hardly come before the CJEU. 151 By contrast, when the importance of the question is of limited practical relevance, the civil chamber is less eager to refer and 'dares' to decide itself. 152 The civil chamber also pays more attention to the wish of the parties in summary proceedings. An additional reason for the referral of Synthon was that both parties urged the Supreme Court to refer. 153 Overall, when courts (feel that they) have discretion (such as the highest administrative courts and the civil chamber in summary proceedings), they use pragmatic and practical considerations to make a decision which cases to refer or not. Such considerations act as some sort of filtering mechanism preventing those courts from referring on a weekly basis.

| The CJEU as a useful ally
Another explanation for the continued willingness of national courts to engage with the CJEU is that the CJEU can provide support to the national court. Especially in cases with considerable financial or political consequences, a referral can be helpful for the national court (Section 5.3.1). The authority of the CJEU can also be valuable for national courts internally vis-à-vis the legislature, executive or other branches of the judiciary (Section 5.3.2). The CJEU's authority can have the same function in a transnational context (Section 5.3.3).

| Assistance from the CJEU in 'big' cases out of precaution
Judges held that they are more cautious when the political and financial stakes are higher. Such consequences can offer an additional 'push' to refer. 154 The Trade and Industry Appeals Tribunal referred a case about the imposition of fine for market manipulation despite the idea among judges that the issue was clair, which was subsequently illustrated by the short CJEU judgment and the absence of an A-G Opinion. The Tribunal considered it a delicate matter from the perspective of legal certainty and was, therefore, extra cautious. 155 During interviews, judges and A-Gs were asked about Massar. This case dealt with the question whether the term 'inquiry' in the Directive on legal expenses insurance includes a procedure before the Employee Insurance Agency, in which the employer requests authorisation to dismiss an employee. The reason for questioning interviewees about this case is that the case at first sight seems rather simple from a legal point of view in the light of the jurisprudence of the CJEU. This is also because the CJEU handled the case in a three-judge formation without A-G opinion. The Supreme Court judgment explicitly referred to the considerable financial consequences of a positive answer to the question, also suggesting that other considerations played a role than purely legal ones. 156  box' . Other interviewees held that it was better to have the CJEU decide on the matter and 'take the consequences for its account', because the idea was that the insurance policy costs would rise because of a positive answer. One interviewee referred to 'an alibi'. One interviewee argued that given the financial consequences, the Supreme Court does not easily 'tap away' a decision with the risk of making the wrong choice that has implications for thousands of cases. Preliminary questions were thus wise in order to 'to make a potential time bomb harmless in advance' . There were, however, other interviewees that disagreed with this reading of Massar and simply held that there were doubts about the interpretation of EU law and noted that it was a disputed matter in relation to which one could reasonably differ.
Antroposana dealt with the question whether anthroposophical medicinal products could be marketed without authorisation (CJEU: no). According to the A-G in the Dutch Supreme Court, the answer to the question was clair. 157 An interviewee likewise argued that the Supreme Court primarily referred in the light of the consequences for pharmaceutical companies.
In ACI Adam, the CJEU was essentially asked whether the Dutch Law on Copyright was consistent with EU law (CJEU: no). This law did not distinguish between lawful and unlawful sources from which consumers download for private use. Again, the considerable implications in this case reinforced the Supreme Court's inclination to refer. 158

| Sword and leapfrog references: reliance on the CJEU's authority internally
The literature, especially earlier social science inspired studies, has primarily emphasised the internal function and relied on politico-strategic reasons for referral. Two approaches stand out. First, the bureaucratic politics model, suggesting that courts refer to solve struggles with other, often higher, courts and, hence, 'leapfrog' the national judicial hierarchy in order to seek support for their interpretation from the CJEU. 159 Second, the judicial empowerment thesis postulating that courts use the referral procedure as a 'sword' to force the legislature to amend legislation when they consider the law to breach EU law. 160 This research did not find much support for either thesis. These considerations only played a role in a number of highest administrative courts' cases. 161 The 'leapfrog' thesis primarily explains referral of lower courts challenging a highest court. Nonetheless, one case was found in which the Central Appeals Tribunal challenged its administrative counterpart, the Council of State, albeit in a rather indirect way.
In Chavez Vilchez, the Central Appeals Tribunal questioned the minimalist interpretation of Zambrano included in the Aliens Circular containing the policy rules as applied by the Immigration and Naturalisation Service. 162 The Circular only gave mothers a right of residence, derived from the right of residence of their children, when the father is not in a position to care for the child. 163 Some highest administrative court judges, albeit certainly not all, underscored the relevance of the what it wanted to say itself, but with more authority.
One could be critical about a strategic reading of several of these referrals. The decisions to refer in these cases can equally be explained from a purely legal perspective, as some judges did. Judges argued that when there is much sub- that the Supreme Court does not have 'a strategy', 'hidden agendas' or 'own interests' since it is a neutral organ that is 'too serious' for that. 171 Interviewees also held that the Supreme Court does not need the support of the CJEU vis-àvis the legislator to strike down national legislation, because it is 'man enough' to do so on its own given the constitutional setup and the independence of the judiciary. 172 The idea is also that the Supreme Court persuades based on its judgments. 173 Two explanations for the absence of politico-strategic consideration among Supreme Court judges were mentioned during interviews. 174 Firstly, some interviewees argued that the highest administrative court judges generally 'think' more in political terms and are more focused on the legislature and executive. 175 Secondly, it seems safe to say that politico-strategic reasons play a more important role in sensitive administrative law fields, such as migration. Emotions and moral or ethical considerations often play a bigger role in these fields than in an area such as tax law. Interviewees noted that tax cases are rather technical and 'only' deal with money. Such cases hardly ever relate to the essence of the rule of law and emotions or conscience issues hardly play a role. 176 One former asylum judge who made a move to tax law observed that the confrontational relationship between different levels in the judicial hierarchy and between the judiciary and the legislature is typical for asylum law. 177

| Reliance on the CJEU's authority transnationally
The CJEU's authority is not only sought in internal 'conflicts' with the legislature, executive or other courts. It is also used at times to solve a transnational conflict with other courts or to prevent those conflicts from arising, also with a view to ensuring the uniformity of EU law. This motive is most visible in the references of the tax chamber in customs cases. The question in these cases is frequently under which tariff heading of the Combined Nomenclature a particular good is to be classified. Such a reference does not mean that the Supreme Court does not 'dare' to decide the case itself, as one judge put it. 178 The problem is that when the Supreme Court answers the question itself and classifies the product, it could do so in a different way than courts in other EU Member States. When it chooses a more disadvantageous classification for the undertaking(s) concerned, this could have the consequence of disrupting trade flows and distorting competition. 179 The Supreme Court is thus more careful in customs cases and prefers referring to Luxembourg so that 'the whole of Europe knows where we stand' , thereby guaranteeing the uniform application of EU law. 180 Interviewees realised that the law-making character is limited in most of these cases and that they do not pertain to fundamental aspects of the EU legal order. 181 This explains, in their view, why most of these references are handled by the CJEU without A-G Opinion, as was concluded earlier. 182 The Central Appeals Tribunal also used the preliminary ruling procedure to solve a difference in opinion with a This case dealt with Mertens' right to unemployment benefits who had worked in Germany while living just across the border in the Netherlands. The Tribunal had tried to contact the German court, where Mertens had filed his appeal against the German refusal to grant unemployment benefits, with the idea of coming to a coordinated solution. The German judge, however, declined to do so and was unwilling to discuss individual pending cases because of privacy considerations. 184 According to some interviewees, the case was rather simple from a legal perspective. The Tribunal also stated in a rather straightforward way in its order for reference that it was obvious that Germany is obliged to grant the benefits. 185 The CJEU decided accordingly in a three-judge formation without an A-G opinion. It only needed 15 paragraphs in which it referred extensively to its previous case law, thereby suggesting that the matter was clair. The Central Appeals Tribunal thus received the desired CJEU authority.

| CONCLUSION
This article showed that the Dutch highest courts have implemented the requested CJEU judgments almost fully and automatically. This finding is remarkable because there has been more criticism than ever in recent years with respect to the preliminary ruling procedure, the CJEU and its case law. The conclusion is also surprising considering that  adhere strictly to Cilfit. While Supreme Court judges primarily include legal-formalist considerations in their decision (not) to refer, highest administrative court judges also take practical and pragmatic considerations into account, such as case-specific reasons relating to the importance of the question at stake as well as the delay caused by a referral (see Table 3). This approach was attributed to a natural reluctance to refer which is, however, not connected to the quality of CJEU judgments. The highest administrative courts' continued engagement with the CJEU was also attributed to the usefulness of the CJEU as an ally. The CJEU has provided support to national courts when decisions with considerable financial or political consequences needed to be made. In addition, the help of the CJEU has also been sought at times vis-à-vis the legislature, executive or other branches of the judiciary, as well as other national courts or administrations in a transnational context. The highest Dutch courts thus remain, for various reasons, loyal interlocutors of the CJEU that remain willing to continue sending questions and to implement CJEU judgments fully and automatically, despite some dissatisfaction.