FBF : On the Justiciability of Soft Law and Broadening the Discretion of EU Agencies

Before FBF was decided by the Court of Justice, it was described as a golden opportunity to transform a ‘wind of change’ into a ‘perfect storm’. In this metaphor, the wind represented different national legal systems becoming increasingly receptive to the judicial review of soft law, while the perfect storm would be the Court of Justice revisiting its own restrictive approach towards the justiciability of soft law.1

more important one from a constitutional perspective. While the Court in BNB found that a recommendation taken by the European Banking Authority was invalid, the Court's reasoning did not elaborate on either the examination of the legal nature of the recommendation or the question of the justiciability of soft law measures, unlike in FBF. The validity of the European Banking Authority's recommendation in BNB was merely incidental and the Court's attention focused on the validity of the national authority's decisions in light of the directive on deposit insurance schemes. The contribution of that case to the Court's case law on soft law is accordingly limited and has been largely subsumed within FBF. That difference in constitutional importance is reflected in the formation of the Court rendering the two judgments: while BNB was decided by the fourth chamber of the Court, FBF is a grand chamber ruling. Hence, we will focus on the judgment in FBF, while referring to BNB where appropriate.

L   
In the aftermath of the Global Financial Crisis, the EU overhauled its system of rules governing financial markets and the EU legislature established three European Supervisory Authorities, based on three similarly worded founding regulations, the ESAs Regulations. 9 The European Supervisory Authorities have typically been armed with two types of powers: quasi-normative powers; and powers of intervention. 10 Early on, the latter received most of the attention. In the ESMA (Short-selling) case, the United Kingdom even challenged the intervention powers granted to the European Securities Markets Authority by the short-selling regulation, before the Court of Justice. 11 By contrast, the presently discussed FBF case turns on the European Banking Authority's quasi-normative powers. 12  recommendations. This provision is also found in the same terms in Articles 16 of the other two ESAs Regulations. Article 16 is a procedural legal basis which in itself is insufficient for the European Banking Authority to adopt soft law. 13 Per the ESAs Regulations, guidelines and recommendations may only be adopted if they are also based on the substantive legislation -EU financial lawthat defines the European Banking Authority's scope of action. 14 Both guidelines and recommendations essentially share the same legal regime under Article 16. What distinguishes them is that recommendations are individually addressed to one or more specific competent authorities or market actors, while guidelines are addressed generally to all competent authorities or market actors. Both follow a 'comply or explain' regime under Article 16(3) of the EBA Regulation. Addressees 'shall make every effort to comply' with guidelines and recommendations and confirm whether they comply within two months following the issuance of a guideline or recommendation. If they refuse to comply, they must inform the European Banking Authority and give reasons therefor. The European Banking Authority, just like the other two European Supervisory Authorities, has a dedicated page on its website on which it lists the (non-)compliance of all national authorities. 15 For those authorities (but not for market actors), the 'comply or explain' regime is thus complemented and reinforced with a 'name and shame' mechanism.
In March 2016, the European Banking Authority issued the Guidelines on product oversight and governance arrangements of banking retail products (hereafter the Guidelines), which were challenged in the FBF case. 16 The Guidelines were mainly addressed to national competent authorities. Substantially, the Guidelines prescribe product governance requirements to be adhered to by financial and payment institutions (and to be monitored by national competent authorities) that sell financial products, governed by the Capital Requirements Directive, 17 the Payments Services Directive, 18 the E-Money 13 Similarly to Art. 294 TFEU, which the EU legislature cannot simply rely upon to adopt legislative acts.
Directive, 19 and the Mortgage Credit Directive, 20 to consumers. The relevant provisions of these directives thus constituted the substantive legal bases of the Guidelines.
The case itself essentially touches upon, as eloquently identified by the Advocate General in his Opinion, 21 whether the Guidelines related to product governance. This is important since, as the Advocate General noted, the specific provisions in the Directives constituting the substantive legal bases of the Guidelines relate to corporate, not product, governance. Corporate governance is not concerned with how products are designed and developed but instead requires control mechanisms at the organisation level, 'to have a transparent organizational structure, to determine clearly who is responsible for what, to have systems in place in order to resolve potential problems, and so on'. 22 This evidently raised the question whether the Guidelines at issue were adopted pursuant to a proper (substantive) empowerment.
In September 2017, the French competent authority, Autorité de Contrôle Prudentiel et de Résolution, published a notice (avis) on its website to state its compliance with the Guidelines. 23 According to this notice, the Guidelines would apply to credit institutions, payment institutions and e-money institutions under the supervision of the Autorité de Contrôle Prudentiel et de Résolution. The French banking lobby, Fédération bancaire française (FBF), challenged the notice before the Conseil d'État, requesting that it be annulled. FBF's main claim was that the Guidelines, on which the notice of the Autorité de Contrôle Prudentiel et de Résolution was based, were ultra vires, as its substantive legal bases do not provide requirements for product governance.
Entertaining doubts on the standing of FBF, the Conseil d'État referred three questions to the Court of Justice, 24 which can be summarised as follows: (1) whether the Guidelines can be challenged by means of an action for annulment and whether FBF would have legal standing to bring such an action; (2) in the negative, whether the Guidelines can be challenged through a preliminary

T O  A G B
The Opinion of Advocate General Bobek is of significant interest, as it takes an opposite view to the Court on the (constitutional) matters at hand. In this summary, we will focus mainly on four issues tackled by the Advocate General, those being: (i) the nature of the Guidelines; (ii) the validity of the Guidelines; (iii) the standard of review which should apply when reviewing the Guidelines; and (iv) the relationship between the action for annulment and the preliminary reference procedure specifically when acts such as the Guidelines are at issue.
The Advocate General found that the Conseil d'État's first two questions should not be addressed in the abstract but needed to be answered in light of the third question. As a result, in his Opinion he reordered the questions to start with the question on the validity of the Guidelines. Before looking into the validity, however, the genuine nature of the Guidelines needed to be established: were they genuine soft law or, rather, disguised hard law? That the Advocate General raises this question may at first seem remarkable, since in the beginning he affirmed that under the Court's Grimaldi case law, 25 it is already crystal clear that preliminary questions on both the interpretation and validity of soft law may be referred to the Court (even if the question in Grimaldi was only a question on interpretation). 26 Whether the Court's finding in Grimaldi was really as clear as the Advocate General suggested may be doubted, however. Advocate General Bobek implicitly recognised this when he later on referred to the Court's post-Grimaldi jurisprudence, in which it rejects the possibility to refer questions on validity pursuant to Article 267 TFEU or in which it rephrases such questions into questions on interpretation. 27 At this point, the reader perhaps wonders why the Advocate General presents his argument in such a contorted way. As we will try to make clear further on, this arguably has to do with the Advocate General's discontent on the Court's approach vis-à-vis soft law under Article 263 TFEU, which has spillover effects in the preliminary reference procedure.
Coming to the first issue, i.e. the nature of the Guidelines, the Advocate General took a nuanced approach. Worded in non-mandatory terms and lacking an obligation for national competent authorities to enforce them, the Guidelines 25 Opinion, supra n. 21, para. 32. 26 Ibid., para. 32. 27 Ibid., paras. 98-99. are prima facie non-binding. 28 However, by further examining the regime of the EBA Regulation's Article 16, the Advocate General recognised that the Guidelines 'can reasonably be perceived as inducing (or even effectively imposing) compliance'. 29 Nonetheless, Advocate General Bobek stopped short of recognising the (binding) legal effects of the Guidelines, given the inflexibility that the Court had previously shown in Belgium v Commission (in which Bobek also acted as the Advocate General), 30 and the application of that test in BNB on a similar soft law measure. 31 Instead, and rather grudgingly, he accepted that, by applying the Court's established test, the Guidelines were indeed non-binding.
Advocate General Bobek then turned to the second issue, i.e. the 'easier' question on the validity of the Guidelines. On this, Advocate General Bobek sided with the Commission in arguing that the European Banking Authority had exceeded its powers. Although the European Banking Authority is entitled to adopt guidelines to flesh out EU law, the Advocate General underlined the 'clear mismatch' between the focus of the Guidelines on product governance and the substantive legal bases relied upon. 32 The relevant provisions of the four directives noted above all relate to corporate governance, not product governance. As a result, the Advocate General found that the European Banking Authority did not have any competence to adopt the Guidelines, the latter being invalid. 33 The Advocate General even rejected the use of the Mortgage Credit Directive as a legal basis for the Guidelines, despite the Commission's suggestion that, differently from the other three directives, that one did provide an appropriate, however partial, legal basis for the Guidelines. According to the Advocate General, however, that Directive regulates the conduct of banks when deciding on granting (or not) mortgages to individual customers in individual cases which is still different from the Guidelines that deal with developing and monitoring financial products, such as mortgages. 34 The Advocate General then went on to consider the third issue which we flagged above, namely the judicial standard of review to be applied when reviewing soft law, like the Guidelines. Here he strongly rejected the notion that, 28 Ibid., para. 43. 29 Ibid., paras. 45-53. BT v Balgarska Narodna Banka (BNB), supra n. 8, para. 79. 32 Opinion, supra n. 21, para. 67. 33 Ibid., paras. 67-75. 34 Ibid., para. 71. because of its non-binding character, soft law should be subjected to a more lenient standard. 35 Instead, to safeguard against the risk of (ab)use of soft law to circumvent the allocation of competences under EU law, Advocate General Bobek suggested that a normal standard of judicial review should apply, 36 i.e. the same standard which the Court would apply when it reviews analogous hard law acts. Having found the Guidelines to be invalid, applying a normal standard of review, the only thing left for the Advocate General in addressing the third question was what the effects are of a finding of invalidity of soft law under Article 267 TFEU. Here the Advocate General was forced to effectively go back to one of his opening statements, namely that it was clear following Grimaldi that the Court could answer questions on the validity of soft law. As already noted, in its post-Grimaldi case law the Court had seemingly backtracked on this issue and, until BNB, had never actually invalidated soft law. As to whether the Court should pursue this post-Grimaldi case law, or honour its original statement in Grimaldi, the Advocate General opted for the latter, since the 'simplest, clearest and most honest answer' should be that the Guidelines are invalid in so far as the European Banking Authority exceeded its powers under its establishing Regulation. 37 Having answered the question on validity, Advocate General Bobek addressed the Conseil d'État's first two questions, which we have flagged above as the fourth issue, delving into the relationship between Articles 263 and 267 TFEU. The Advocate General interpreted the preliminary reference as expressing a concern with the 'TWD scenario' where an applicant, which has standing under Article 263 TFEU but fails to lodge an action for annulment against an act of EU law, is precluded from subsequently challenging the validity of the national measure implementing the EU act. 38 Yet, given the adherence of the Court of Justice to its strict approach set in Belgium v Commission, an act that genuinely consists of soft law is not reviewable in an action for annulment. In addition, the Advocate General highlighted that the applicability of TWD also depends on clearly having locus standi under Article 263 TFEU, a condition that could never be met for an act that is formally soft law but which the Court might find (under the test set out in Belgium v Commission) is binding in reality. 39 Pursuing his analysis, the Advocate General went as far as calling into question the relevance of Foto-Frost for soft law measures, since that line of case law should not be construed as imposing a duty for national courts to ask a preliminary 35 Ibid., paras. 81-83. 36 Ibid., paras. 86-94. 37 Ibid., para. 109. 38 Ibid., para. 112. 39 Ibid., paras. 116-118. In addition, the AG noted that it is doubtful whether an interest organisation like the FBF would fulfil the Plaumann criteria. question on the validity of a non-binding measure, given that neither uniformity of EU law nor effective judicial protection are at stake in such a case. 40 Noting that the logic behind both TWD and Foto-Frost starts from the premise that the remedies provided by Articles 263 and 267 TFEU constitute a single coherent system of legal protection, the Advocate General himself flags that his findingthat both TWD and Foto-Frost are of limited relevancemight lead to the conclusion that since soft law measures cannot be challenged under Article 263 TFEU, their validity should also not be reviewable under Article 267 TFEU, 41 which would of course go against Grimaldi and the Advocate General's earlier finding that it was clear that the Court could rule on the validity of the Guidelines. Here, Advocate General Bobek relies on both a textual and systemic argument to uphold Grimaldi. Textually, Article 267 TFEU refers to any and all acts of the EU institutions (bodies, offices and agencies), whereas Article 263 TFEU refers to acts having legal effects. 42 Secondly and systemically, precisely because both remedies form part of a complete system of legal protection, the Advocate General finds that they should be complementary rather than each other's exact mirror image. 43 Of course, dissociating Articles 263 and 267 TFEU as such also results in some drawbacks and challenges. The Advocate General discusses these as a veiled critique on the Court's refusal to follow his suggested solution in the Belgium v Commission case. In that case, the Court shut the door (even for privileged applicants) on challenging soft law measures directly, but this did not mean that the problem posed by soft law magically disappeared. Accepting that the legality of soft law may be reviewed under Article 267 TFEU, it then becomes necessary to ensure that Article 47 of the Charter is respected and that a complete system of legal protection is in place. The paradoxical result of this outcome, as noted by the Advocate General, is that privileged applicants are placed at a disadvantage as they would be unable to seek the annulment of a non-binding measure, 44 whereas that measure could be reviewed indirectly at the request of a non-privileged applicant.
The Advocate General concludes by highlighting a trilemma between Grimaldi, Foto-Frost and Belgium v Commission. The Advocate General finds that there are three options for the Court, each one requiring the Court to sacrifice one of the three precedents. While he is not entirely explicit on which elements of these rulings are mutually contradictory, the trilemma may arguably be understood as follows: Belgium v Commission implies that soft law cannot be challenged 40 Ibid., paras. 121-132. 41 Ibid., para. 134. 42 Ibid., para. 136. 43 Ibid., paras. 137-138. 44 Ibid., paras. 146-147. pursuant to Article 263 TFEU; it follows from Grimaldi that the Court can answer preliminary questions on soft law both regarding interpretation and validity; it follows from Foto-Frost that only the EU Courts can rule on the validity of EU law because the Court's monopoly under Article 263 TFEU should be mirrored in Article 267 TFEU. As a result, either Grimaldi would have to be sacrificed (by limiting preliminary ruling procedures to questions on interpretation) or Foto-Frost (by admitting that there is a degree of incoherence in the system of legal protection) or Belgium v Commission (by allowing soft law to be challenged pursuant to Article 263 TFEU). We will return to the question whether this was indeed a genuine trilemma but here it suffices to note that, unsurprisingly, Advocate General Bobek suggests that Commission v Belgium should ultimately be revisited. 45 T    C  J Unlike the Advocate General, the Court answered the questions as presented by the Conseil d'État. The Court started by examining the nature of the Guidelines to determine whether they could have been challenged via the annulment procedure. Here, the Court largely stuck to its approach as developed in Belgium v Commission. The Court emphasised that the Guidelines merely provided 'the European Banking Authority's view', 46 that they are formulated in non-mandatory terms ('should' instead of 'shall'), 47 and do not require compliance from its addressees. 48 Although the Court mentions the particularities of the Guidelines' legal regime, it largely ignored the potential binding effects of the European Banking Authority's soft law. Rather, the Court considered by analogy that, since guidelines and recommendations share the same legal regime under Article 16 of the EBA Regulation, and given that recommendations are not binding in accordance with Article 288 TFEU, guidelines also lack binding force. 49 By assimilating the Guidelines to Article 288 TFEU recommendations, the Court refuses to admit that the former can produce binding legal effects. 50 According to the Court, the EU legislature only intended to confer 'a power to exhort and persuade', a formula the Court also used in Belgium v Commission. 51  particularities of the European Banking Authority's different legal regimes for soft law measures. 52 Therefore, the Guidelines could not be challenged by means of an action for annulment. Yet, the lack of binding legal effects is not as such as to preclude the Court's jurisdiction under Article 267 TFEU, which, according to the Court, extends to all acts of EU law 'without exception'. 53 In this regard, the Court recalled that the strict admissibility criteria of Article 263(4) TFEU do not apply to preliminary references. The Court further restated its Inuit Tapiriit Kanatami case law regarding the need for admissibility rules before member states' judicial bodies to respect the right to effective judicial protection and guarantee the principles of effectiveness and equivalence. 54 To be admissible, a request for a preliminary reference on validity merely requires there to be a genuine dispute in which the validity of an act is called into question, even if that act has not been the object of an implementing measure against the concerned individual. 55 The Court therefore sees no admissibility issue when a national judge, seized by a professional federation, questions the validity of a non-binding act of EU law pursuant to Article 267 TFEU.
The Court then turned to the final question, regarding the validity of the Guidelines. Before delving into the specifics, the Court clarified the degree of intensity of review for a non-binding act. The lack of binding legal effect does not affect the standard of judicial review, which the Court qualifies itself as 'stringent'. 56 Surprisingly perhaps, the Court justified this by the potential effects of the Guidelines. 57 The stringent standard of review is also justified by both the duty, 52 BT v Balgarska Narodna Banka (BNB), supra n. 8, para. 79. While Art. 16 guidelines and recommendations fulfil a post-legislative function, the recommendations under Art. 17 are part of a (composite) enforcement procedure whereby the European Banking Authority first draws up a recommendation addressed to a national authority that misapplies EU law. If the national authority does not correct its behaviour in line with the recommendation, the Commission may adopt a formal opinion subsequent to which the European Banking Authority may adopt a (binding) decision requiring the necessary action to comply with EU law. Surprisingly, few authors discuss recommendations under Art Ibid., para. 70. This because they might be applied by national authorities to market operators or because when applying EU and national law to market operators, national authorities must take into account the guidelines.
first exposed in Grimaldi and reaffirmed also in BNB, 58 of national courts to take into consideration soft law measures when they are intended to supplement binding provisions of EU law, and the potential of uncontrolled soft law measures to undermine the allocation of powers of the EU legal order. 59 In BNB, the Court even established that individuals harmed by a breach of Union law 'must be able to rely' on a recommendation, such as the one taken by the European Banking Authority in this case under Article 17(3) of the ESAs Regulations, 'as a basis for establishing, before the competent national courts, the liability of the Member State concerned for the breach of Union law in question'. 60 Based on this understanding of the standard of judicial review, the Court started its examination of the Guidelines by framing the European Banking Authority's discretion. Echoing the ESMA judgment, the Court recalled that the 'EU legislature has precisely delineated the European Banking Authority's power to issue guidelines, on the basis of objective criteria' and that European Banking Authority's competence only goes as far as it is 'expressly provided for by the EU legislature'. 61 Interestingly, and unlike Advocate General Bobek, the Court did not thereby examine the substance of the Guidelines or the issue of the European Banking Authority's specific scope of action in the case at hand.
Instead, the Court first emphasised the generally broad mandate of the European Banking Authority, since Article 1(2) of the EBA Regulation allows the European Banking Authority to act 'within the scope of a series of acts listed in that provision, including all directives, regulations, and decisions based on those acts, and of any further legally binding EU act which confers tasks on the European Banking Authority'. In addition, the Court also highlighted Article 1(3), which provides that the European Banking Authority may also act: in the field of activities of credit institutions, financial conglomerates, investment firms, payment institutions and e-money institutions in relation to issues not directly covered by the legislative acts referred to in paragraph 2, including matters of corporate governance [ : : : ], provided that such actions are necessary to ensure the effective and consistent application of those acts. 62 This is highly relevant, since the Court subsequently relied on this provision as a deus ex machina to uphold the validity of the Guidelines.
The Court then concluded its review of the framework applicable to the European Banking Authority's soft law by looking into further relevant provisions 58 Grimaldi, supra n. 53, para 18; BT v Balgarska Narodna Banka (BNB), supra n. 8, para 80. 59 FBF, supra n. 46, paras. 71 and 72. 60 BT v Balgarska Narodna Banka (BNB), supra n. 8, para. 81. 61 FBF, supra n. 46, paras. 67 and 75. 62 Ibid., paras. 76 and 77 (emphasis added). of the EBA Regulation, including both the European Banking Authority's general objectives and the objectives that the power to issue guidelines ought to pursue. 63 This framework set the tone and effectively determined the outcome of the Court's judgment. The Court indeed announced that the validity of the Guidelines needs to be assessed in light of this more abstract framework rather than in light of the Guidelines' (more concrete) substantive legal bases, 64 as the Advocate General had done. Indirectly addressing the product governance versus corporate governance problem stressed by the Advocate General, the Court also relied on a teleological argument, by emphasising the ultimate purpose of the European Banking Authority's soft law power (i.e. ensuring a uniform and consistent application of EU law by the different national authorities). The Court noted that, 'nothing in Regulation No 1093/2010 [suggests] that measures relating to the design and marketing of products are excluded from that power, provided that those measures fall within the [European Banking Authority's] scope of action'. 65 This notion of 'the European Banking Authority's scope of action' is determined by the EBA Regulation's scope of application. The latter, in turn, is not simply limited to the EBA Regulation itself, since Article 1(2) and 1(3) of the Regulation contain cross references to an open-ended list of further EU legislation in the area of banking law, whereby the link between the European Banking Authority's normative output and the scope of application of the EBA Regulation is interpreted in a generous, teleological manner.
In its concrete assessment, the Court then first examined the content of the Guidelines in detail, 66 before it verified, 'whether the contested guidelines fall within the European Banking Authority's scope of action, as defined in Article 1(2) and (3) of [the EBA Regulation]'. 67 To do so, the Court looked into the substantive legal bases of the Guidelines, finding that the four directives 'must be regarded as constituting acts referred to in Article 1(2) of [EBA Regulation]'. 68 Having passed this hurdle, the Court subsequently juxtaposed the content of the Guidelines with the scope of application of the directives. 69 It may be recalled here that Advocate General Bobek noted that since the Guidelines dealt with product governance, they could not be considered to come within the scope of application of the directives, the relevant provisions of which deal with corporate governance. At this point however the Court presented its deus ex machina of Article Guidelines either come within the scope of application of the directives or are 'necessary to ensure the consistent and effective application of those directives'. 70 Thus, by reference to this provision, the Court introduces an alternative. It opens the door to a finding that the Guidelines may be valid, even if they do not, as such, fall within the scope of application of the directives.
On this basis, the Court proceeds to the examination, one by one, of the substantive legal bases referred to by the Guidelines. 71 For each legal basis, the Court accepts that requirements on product governance (the subject matter of the Guidelines) contribute to corporate governance (the legal bases' subject matter), but it does not go so far as to conclude that the Guidelines fall within the scope of its legal bases. Rather, the Court systematically concludes that the Guidelines 'may be considered as necessary to ensure the effective and consistent application' of the respective legal bases. 72 This remarkable reliance on the gap-filling function of Article 1(3) of EBA Regulation enables the Court to determine that the Guidelines 'must be regarded as falling within the European Banking Authority's scope of action'. 73 Finally, the Court returns to the procedural legal basis in Article 16 of EBA Regulation in order to verify whether the Guidelines fall within 'the specific framework laid down by the EU legislature for the exercise of [the European Banking Authority's] power to issue guidelines'. 74 Here the Court connects the Guidelines to the objectives specified in the EBA Regulation. Having stated that the Guidelines contribute to the objectives of consumer protection, depositor and investor protection, risk management by financial institution and the 'establishment of consistent, efficient and effective supervisory practices', the Court concludes that the European Banking Authority did not exceed its competences and that the Guidelines are therefore valid. 75 T    EU '  In our analysis in this section, we will focus on the clarifications brought regarding the discretion of EU agencies; the legal remedies available against (EU agencies') 70 Ibid. soft law and the repercussions of the case for the preliminary ruling procedure are then analysed in the sections that follow.
Before discussing the major clarifications brought by the FBF case, it is useful to recall the general constitutional framework governing EU agencies' empowerments. As discussed elsewhere in greater detail, 76 that framework largely boils down to two limbs: non-delegation and institutional balance. Subsequently we will add the EU agencies' reliance on soft law to the equation.

The non-delegation doctrine
A pillar of European administrative law, the non-delegation doctrinea misnomer by all accountshas been erected on the basis of the Court's Meroni jurisprudence. 77 This case dates back to the European Steel and Coal Community, and saw the Court invalidate a delegation of powers by the High Authority to two bodies of private law. The Court ruled in this case that delegated (executive) powers cannot entail a wide margin of discretion. 78 The policy discretion that cannot be delegated was defined as the ability to reconcile different objectives depending on the circumstance. Conversely, permissible delegations should be 'exactly defined and entirely controlled' by the delegating institution and should 'be subject to strict review in the light of objective criteria determined by the delegating authority'. To this day, Meroni remains a totem of European administrative law, despite the idiosyncrasies of the case, a changing context, and wholly different constitutional and legal frameworks. 79 In 1981, Romano completed the Meroni doctrine, as the Court ruled incompatible with the Treaties that a body established through secondary legislation be empowered to adopts acts 'having force of law' ('revêtant un caractère normatif'). 80 But it was only in 2014 that the Meroni doctrine was largely overhauled and clarified by the Court in the ESMA case, in which the European Securities and Markets Authority's powers of market intervention in exceptional circumstances were challenged by the UK. 81 In this judgment, the Court ruled that the European Securities and Markets Authority's exercise of its intervention powers is 'circumscribed by various conditions and criteria' of both substantive and UK v Parliament and Council, supra n. 11. procedural nature. 82 In sum, the Court (implicitly) admitted that discretionary powers had been delegated, but accepted the delegation because its exercise was sufficiently conditioned. 83 The Court also departed from Romano, as the Lisbon Treaty 'expressly permits Union bodies, offices and agencies to adopt acts of general application' and subsumed that jurisprudence under Meroni. 84 The outcome of ESMA has been a new standard applicable for the delegation of powers by the EU legislature to EU agencies: the powers delegated need to be 'precisely delineated and amenable to judicial review in the light of the objectives established by the delegating authority'. 85 Vesting agencies with 'a very large measure of discretion' remains, however, prohibited. 86 This judgment 'mellowed' the non-delegation doctrine, adapting it to the needs of EU administrative integration in the 21st century. At the same time, the judgment was criticised for insufficiently safeguarding the institutional balance when broadening the possibilities to empower EU agencies rather than the EU's default executive authority, the Commission, which also goes at the expense of the European Parliament's scrutiny over the EU executive. 87

The institutional balance
The institutional balance has been recognised by the Court of Justice as a principle of EU law which finds its normative foundation in Article 13(2) TEU. 88 Depending on the specific area of EU law, it assigns competences to specific institutions in order to realise the objectives of the EU in that area. 89 According to Le Bot, the telos of the principle is one of systemic protection. 90 Whenever EU agencies are empowered, the institutions therefore have to ensure that the agencies are not enabled to encroach on the (other) institutions' prerogatives. Because the Court in the original Meroni ruling also relied on the idea of 'balance of powers', the principle of institutional balance and the non-delegation doctrine are often amalgamated. Yet, both should be distinguished and they should be cumulatively met. 91 After all, a power may be precisely delineated (and therefore open to delegation) but still constitute a prerogative of one of the EU institutions. 92 Conversely, an insufficiently delineated power (that cannot therefore be delegated) might not pose institutional balance problems if it does not (explicitly or implicitly) encroach on the prerogatives of one of the EU institutions.

Adding soft law to the equation
The original non-delegation doctrine of Meroni set out above had a clear chilling effect on the delegation of powers in the EU in general, and on the delegation of powers to EU agencies specifically. 93 Of course, this chilling effect did not result purely from legality concerns but may have been equally inspired by political considerations. 94 Regardless of whether this was out of legal necessity or for reasons of political expediency, delegating rule-making powers to EU agencies was seen as impossible, even if this did not diminish the practical need for greater specialisation (and thus delegation to specialised bodies such as EU agencies). A way out of this conundrum was found by delegating important soft law powers to EU agencies. EU institutions and member states could then still benefit from agencies' expert input but, because no hard law powers were granted to these bodies, Meroni and Romano did not come into play. This may seem odd at first sight, since the use of soft law itself has in the past been criticised on constitutional grounds. 95 Indeed, in law, unlike in mathematics, two negatives do not make a positive. Still, the ambiguity on the constitutional limits to soft law combined with the ambiguity on the constitutional position of EU agencies did provide fertile ground for the agencification of the EU administration. 96 While some had expected that a relaxing of the Meroni doctrine in ESMA would pave the way for more straightforward delegations to EU agencies, the EU legislator has not sought the limits of the new ESMA doctrine but instead remained timid in empowering EU agencies with hard law powers. 97 Today, the qualitative agencification 98 of EU administration is in great part still channelled through the conferral of soft law powers to EU agencies.
The most important part of the normative output of EU agencies will therefore remain to consist of soft law, 99 including the adoption of preparatory acts, the socalled Regulatory Technical Standards and Implementing Technical Standards in the framework of the ESAs Regulations, that are enacted in the form of binding delegated and implementing acts upon the Commission's endorsement. 100 This is so for those agencies where ESMA has in principle opened a possibility for the legislature to entrust them with enforcement and rulemaking powers, 101 but evidently also for those agencies operating in fields where the EU has limited competence and where, as a result, the non-delegation doctrine is not the main constitutional hurdle. 102 All this of course means that it remains as critical as before to know with law; and third, which standard is relied upon (by the Court or any other body exercising the review) in scrutinising a contested soft law measure. On all three issues, the FBF judgment brings some useful clarification.

Constitutional limits to the adoption of soft law
What does FBF add to our knowledge on the legal limits of soft law? In the 2004 France v Commission case, the Court had already ruled that even if the contested measures are non-binding, like the Commission's Guidelines on Regulatory Cooperation and Transparency at issue, this did not exclude them from review, since 'determining the conditions under which such a measure may be adopted requires that the division of powers and the institutional balance established by the Treaty in the field [ : : : ] be duly taken into account'. 103 In more general terms, the Court thus confirmed that even when adopting soft law, institutions (and a fortiori subsidiary bodies) are required to respect the institutional balance. Conversely, in FBF, for the first time the Court confirmed that EU agencies are also required to respect the non-delegation doctrine of ESMA, even when adopting soft law: the EU legislature has precisely delineated the [European Banking Authority's] power to issue guidelines, on the basis of objective criteria, the exercise of that power must be amenable to stringent judicial review in the light of those objective criteria. The fact that the contested guidelines do not produce any binding legal effects [ : : : ] is not such as to affect the scope of that review. 104 In principle then, there is no difference between empowering an agency with soft law or hard law powers, and no constitutional objections can be 'circumvented' by having recourse to soft law rather than hard law powers. Whether this is indeed so in practice depends on the actual standard of judicial review adopted and the possibility to seize the EU Courts (cf below).

C EU '  
One remarkable aspect of the Court's finding that the same standard applies to soft law as it does to hard law is the fact that this assertion risks becoming a largely theoretic point. This is so because the Court, at the same time, maintains the position that soft law cannot be directly challenged before the EU judiciary: both hard and soft law are governed by the same constitutional principles but a direct review of whether those principles have indeed been respected in a specific case is only possible for hard law.
Before turning to the preliminary ruling procedure as the (only) avenue to have the Court of Justice review agency soft law measures, it is important to briefly flag other possible remedies, some of which are specific to the EU agencies or even the European Supervisory Authorities. While we will not further explore these remedies, it is useful to be aware of alternatives to judicial review. After all, as judge Jääskinen, writing in his academic capacity, observed, there are 'alternative ways of constitutionally legitimised control structures of the use of public powers' and 'judicial protection is a societally and economically scarce resource which cannot be light-heartedly allocated to cases that are better dealt with [by] other types of remedies or that do not deserve the attention of courts'. 105 One such remedy is available before the European Ombudsman (when a soft law act is vitiated by an instance of maladministration). Secondly, where EU agencies adopt soft law, one possible avenue which has not been relied on so far 106 is that of judicialised administrative review before the Boards of Appeal which are typically established within those agencies. 107 Thirdly, a more recent development specific to the European Supervisory Authorities is the introduction of an administrative review of soft law adopted by the European Supervisory Authorities before the Commission pursuant to Article 60a of the ESAs Regulations (introduced by the 2019 review). There are some doubts, however, about how this remedy may practically function. 108 In light of this, and after the Court's judgments in BNB and FBF, the most secure avenue becomes the preliminary reference procedure. 105 Jääskinen, supra n. 4, p. 361. 106 The General Court even confirmed that the Board of Appeal of the European energy regulator (ACER) is not competent to review opinions adopted by ACER.

The preliminary ruling procedure
Despite the above theoretic options to contest soft law, the Court's judgment in FBF means that to be offered a genuine remedy, applicants will have to convince national judges to refer questions on validity of soft law (adopted by the European Supervisory Authorities) to the Court of Justice. In this regard, it is useful to return to the trilemma identified by Advocate General Bobek. In light of Advocate General Bobek's trilemma, it seems that the Court in FBF partially sacrificed Foto-Frost, since it accepted an incoherence between Articles 263 and 267 TFEU. If we revisit Foto-Frost, however, we may doubt whether what was at issue in FBF really was a trilemma. In Foto-Frost, the Court simply noted that, just like under Article 263 TFEU, it should have the (only and) last word on the invalidity of EU acts under Article 267 TFEU. 109 That is arguably not the same as saying that the reviewable acts under both procedures must be identical, whereby accepting different sets of reviewable acts amounts to partially sacrificing Foto-Frost. As Advocate General Bobek himself noted, 110 accepting an incongruence between Articles 263 and 267 TFEU precisely results in both being complementary and resulting in a complete system of legal protection.
How should FBF be assessed, given that it requires litigants to go through the preliminary reference procedure, since, following Belgium v Commission, they cannot go through the action for annulment?
First there are a number of good reasons why claims should be channelled through Article 267 TFEU rather than Article 263 TFEU. 111 Some of these reasons overlap with the general reasons cited to restrict natural and legal persons' direct access to the EU Courts, while others are (especially) relevant when the validity of EU soft law is at issue. It thus seems advisable not to (over)burden the EU Courts with theoretic disputes on the validity of EU law which do not reflect a genuine practical need to have that law reviewed. When actual disputes arise, it is also helpful for the Court to be able to review the validity of EU law in the actual context in which it is applied. This goes for hard law and soft law alike, but especially relevant for the latter is that, while not denying the risks involved with reliance on soft law, it might lose a lot of its attractiveness and flexibility if it were open to exactly the same judicial review as hard law.
At the same time, FBF arguably also entails a number of risks. In this regard, a first broader repercussion was also briefly touched upon by the Advocate General. He noted that blocking direct access to the Courts and channelling everything through the preliminary reference procedure effectively conflicts with the current evolution of the judicial system at EU level. 112 Indeed, in terms of workload, it makes more sense to allow for a direct review of soft law measures, given the recent expansion of the General Court, rather than to process all these cases through the preliminary ruling procedure which end up on the docket of the Court of Justice. It may be recalled here that on the occasion of the last revision of the Court's statute in 2019, the Court itself proposed to pass on the less significant infringement proceedings to the General Court and to limit certain appeal proceedings before the Court against judgments of the General Court 113all this precisely to allow the Court to better manage its caseload, allowing it to focus on the most significant cases.
While this first ramification of FBF may of course indirectly affect the right of natural and legal persons to effective legal protection, a second repercussion puts an even more acute challenge to this right. The avenue to which the Court now directs applicants goes through the national judge. To anyone following EU law this harkens memories of the challenges posed by Plaumann and the Court's response in cases like UPA and Jego Quéré, as well as revealing a further challenge to the principle of procedural autonomy. Concretely, a natural or legal person taking issue with an EU soft law measure will only be able to indirectly challenge this measure before a national judge if there is a challengeable national measure. In most cases, however, EU soft law will be followed up on or implemented at the national level through further (national) soft law. While there is a 'wind of change' in different national legal systems to allow such challenges, it is fair to assume that this possibility will not exist in every national legal system and that in any event the requirements to be fulfilled by applicants will vary greatly between member states. Should these differences simply be accepted in light of national procedural autonomy? That is doubtful. While national procedural autonomy should be considered a legal principle, rather than a (temporary) state of affairs in EU law, it is still to be balanced with other principles such as that to effective legal protection. 114 In this regard, Arnull has noted that in the post-Lisbon era '[t]he venerable principles of national procedural autonomy, equivalence and effectiveness seem to have been absorbed into a more complex matrix of rules and principles which represent a considerable intrusion into fields formerly considered the 112 AG Bobek developed the same reasoning in another recent case where direct access of nonprivileged applicants to the General Court (challenging hard law) was at issue. See ECJ, Opinion of AG Bobek in Joint cases C-177/19 P, C-178/19 P and C-179/19 P, Germany v Paris, Brussels and Madrid, ECLI:EU:C:2021:476, para. 86. We would like to thank the anonymous reviewers for drawing our attention to this. 113  prerogative of the Member States'. 115 In line with UPA, Unibet, 116 and Article 19(1) TEU it is up to the member states to provide adequate remedies. As the Court refuses to revisit its Belgium v Commission ruling, it will be incumbent upon the national courts to safeguard effective judicial protection against EU soft law measures to fix any resulting lacunae in the EU judicial system in accordance with Article 19(1), second sentence TEU. 117 This is already a significant requirement imposed on those legal systems which have not (yet) accepted the reviewability of soft law endogenously. However, it will be most acute in those cases where there is no national measure to be contested. It may again be recalled here that the Conseil d'État could refer a question to the Court, since FBF challenged a notice of the Autorité de Contrôle Prudentiel et de Résolution taking over the Guidelines. This was (only) possible because the Guidelines had been addressed to national authorities. Focusing now specifically on the European Supervisory Authorities, some of the soft law which they adopt is only addressed to market operators or has no addressee as such (or generally will not give rise to a national measure). Examples of such measures are, inter alia, the opinions and Q&As codified in Articles 16a and 16b of the ESA Regulations after the 2019 revision. Mapping these possible ramifications of FBF underlines even more how, out of some of the different possible means to ensure sufficient legal protection against EU soft law acts (see above), channelling everything through the preliminary ruling procedure seems the most cumbersome of all.
T      -     Taken at face value, the considerations of the Court regarding the standard of review applicable to agencies' soft law support the idea that the exercise of powers by EU agencies, even those soft law powers, must be tightly controlled. Yet, the subsequent review carried out by the Court reveals a daunting gap between the 115 A. Arnull, 'Article 47 CFR and National Procedural Autonomy', 45 European Law Review (2020) p. 690. 116 While not as such at issue in that case, the Court did clarify in Unibet that EU law does not create new remedies in the national legal orders unless, in the  1629. We would like to thank the anonymous reviewers for drawing our attention to this. stated intent and the actual outcome. The latter indicates a permissible approach that undermines the principles it set in ESMA and significantly widens the discretion that EU agencies may legally exercise.
Prima facie, the standard of review set by the Court in this case appears commensurate with the non-delegation doctrine. Even though the Conseil d'État did not ask a question on that matter, Advocate General Bobek raised the issue in his Opinion, and it was subsequently addressed by the Court, which even seemed to outbid the Advocate General by stressing the need for 'stringent judicial review'. The justification of the Court is broadly in line with the underlying rationale of the non-delegation doctrine as revisited in ESMA. 118 By contrast with legally binding powers, the framework set by the EU legislature does not set extensive conditions for the issuance of guidelines by the European Supervisory Authorities. This makes it even more important to respect the limited conditions that are set.
Yet, the Court held back from applying this 'stringent judicial review' of the validity of the Guidelines. The Court did not apply the 'objective criteria' set by the EU legislature 119 to 'precisely delineate' the European Banking Authority's power to issue guidelines. 120 The fact that the legal bases of the Guidelines did not directly relate to product governance led national competent authorities in Germany and Austria to refuse their application. 121 The Court tacitly conceded the absence of such a direct link by resorting to the gap-filling provision of Article 1(3) of EBA Regulation. Yet, it did so by interpreting very extensively two conditions set by this provision. Article 1(3) requires that the European Banking Authority acts 'in relation to issues not directly covered by the legislative acts referred to in [Article 1(2)], including matters of corporate governance [ : : : ]'. As such, there must be some indirect link between the Guidelines and the provisions that it relies upon as legal basis. Furthermore, the second condition requires that 'such actions are necessary to ensure the effective and consistent application of those acts'.
The Court did not distinguish between those two conditions, although it laid out two series of arguments. As corporate governance is the object of three legal bases and is mentioned explicitly in Article 1(3), the Court used that reference by blending the concept of corporate governance with that of product governance 118  and oversight, in light of the objectives pursued by the European Banking Authority. Yet product governance and oversight, like any other form of product regulation or retail regulation, are absent from all corporate governance provisions referred to by the European Banking Authority as legal bases. Furthermore, product governance and oversight are instruments of investor protection, which is not the main objective pursued by these provisions. In contrast, and by way of example, one of the legal bases, Article 10(4) of the Payments Services Directive, refers to the prudential objective of 'sound and prudent management'.
The Guidelines on product governance and oversight requirements, as acknowledged by the Court, apply during the manufacturing and distribution process of financial products, which intervenes only after the licensing of an institution, unlike requirements of corporate governance. The distinction between product governance and corporate governance is most striking when comparing the Guidelines' legal bases with another directive, MiFID II, which does not fall within the European Banking Authority's scope of action, but within that of the European Securities and Markets Authority. 122 MiFID II indeed provides for a series of requirements related to product governance and oversight, 123 which gave rise to the European Securities and Markets Authority's own guidelines on product governance and oversight that are very close to the European Banking Authority's in the field of banking regulation. 124 The Court also linked product governance to the risk management obligations of financial institutions. While risk management is an essential part of prudential regulation, it is mainly concerned with the risks for financial institutions' safety and soundness. By contrast, the risks inherent to financial products manufactured and distributed are primarily borne by investors that buy those products from financial institutions, and not by the institutions themselves. The justification provided by the Court of the link between product governance and risk management is based essentially on the factual affirmations of the European Banking Authority and the Autorité de Contrôle Prudentiel et de Résolutionthe authors of the soft law in question. 125 The Court also justified the Guidelines in view of the European Supervisory Authorities' joint position on product oversight that the Guidelines 'directly implement'. 126 By doing so, the Court seems to admit that EU agencies can set the terms for the expansion of their own margin of discretion.
Finally, the Court considered that the vague and wide-ranging objectives that the European Banking Authority must pursue when undertaking its tasks constitute the 'objective criteria' amounting to a 'precisely delineated' framework set by the EU legislature. While this stands in sharp contrast with the Court's own claim that it conducts a 'stringent judicial review' of EU agencies' actions, it effectively boils down to the Court's established standard where it defers to EU institutions exercising technical judgements and only checks for manifest errors of assessment. 127 The loose approach of the Court to European Supervisory Authorities' discretion is even more striking when read in the context of the 2019 review of ESAs Regulations, which followed longstanding institutional concerns about the issue of legal bases for soft law. 128 On this occasion, the EU legislature narrowed down some of European Supervisory Authorities' powers, introduced proportionality and 'better regulation' requirements, and elaborated the European Supervisory Authorities' accountability obligations. 129 The EU legislature further reframed the European Supervisory Authorities' soft law powers to specify that guidelines should in principle (albeit not exclusively) be issued on the basis of legislative empowerments. 130 While the relevant legislative framework in FBF is the one predating the 2019 review, the Court's judgment in FBF sits uneasily with these new requirements.
Whether the Court's flexible application of its 'stringent review' signals a more fundamental reappraisal of the Court's case law is anyone's guess. On the one hand, the judgment cannot be dismissed as isolated or trivial. It dilutes ESMA's principles, which in turn had weakened the Meroni doctrine. The judgment was rendered by the Court's grand chamber. The extensive and original reasoning of the Court, insisting on Article 1(3), suggests a careful and considered drafting. And BNB, in which the Court struck down a recommendation taken by the European Banking Authority on the basis of Article 17(3) of the ESAs Regulations, cannot be understood as a counter-example, as the decision of the Court in this case was merely drawing the consequences on a previous ruling, are the heads of the national competent authorities. 135 On average, out of 113 guidelines and recommendations issued by the European Banking Authority, compliance reaches 94.6%, 136 which means that national competent authorities quasi-systematically comply with guidelines. 137 For the Guidelines at issue in FBF, they were followed in 23 Member States, and partly followed in 25 Member States. This compliance means that guidelines are in turn integrated into national rulebooks and applied to market actors by national competent authorities. 138 Endorsement by a national competent authority may, therefore, generate legally binding effects, although the Court overlooked those by focusing solely on the Guidelines' absence of legally binding effects.
The huge degree of compliance also sheds some light on the limits of the Meroni doctrine that derive from the governance of EU agencies like European Supervisory Authorities. In the absence of direct remedies, European Supervisory Authorities are free to expand the scope of their discretion through soft law, based on the common needs of the national competent authorities that govern them. In other words, national competent authorities may, through their control of EU agencies, decide what they can do at national level and extend their powers by exploiting the discretion left to EU agencies. It hardly needs to be stressed that in this light, the Court's permissible approach to discretion and apparent indifference towards the legal effects of soft law entails a risk for competence creep, legal certainty, and legal accountability. C FBF provides a series of useful, if controversial, answers to a few of the thorniest questions of EU constitutional, procedural and administrative law, from the scope of preliminary references to the application of the ESMA non-delegation doctrine in concrete cases.
On the first issue, FBF confirms what the Court had already suggested in Grimaldi: that national judges can refer questions on the validity of soft law under Article 267 TFEU. The Court does so even if such soft law cannot be challenged in direct actions, even by privileged parties. This clarification constitutes a doubleedged sword: it, helpfully, may ensure that there are no lacunae in legal protection but if it does, this will be at the expense of national procedural autonomy, since national judges will have to offer the flexibility that the Court is unwilling to grant under Article 263 TFEU. At the same time it will result in a mismatch between the action for annulment and the preliminary ruling procedure, channelling cases through the latter whereas the current judicial architecture of the EU and reasons of procedural efficiency argue in favour of channelling these cases through Article 263 TFEU proceedings. The Court therefore did not seize the 'golden opportunity' which we flagged at the beginning of this case note.
Similarly, while FBF is to be welcomed because it recognises that parties should be able to contest the validity of soft law, it does not (and, to be fair, could not) address the elephant in the room: what are the limits to the EU administration's recourse to soft law, or, put differently, how can we ensure that the EU administration does not become 'addicted' to soft law? This question is particularly important as the Court still seems to treat different categories of soft law measures, regardless of their author or legal effects, as one and the same.
FBF's second clarification deals more specifically with EU agencies and settles the question whether the same doctrine governing the delegation of hard law powers to EU agencies also applies to the delegation of soft law powers. The Court uses the same language to require that also the European Banking Authority's soft law powers are 'precisely delineated by objective criteria, whereby the exercise of that power must be amenable to stringent judicial review'.
However, although the Court stressed it would follow a 'stringent' standard of judicial review, the standard relied upon is in fact very permissive, allowing the European Banking Authority to effectively expand its mandate. That the ESMA doctrine is rather flexible was already clear from ESMA itself. But in that case, the Court, in abstract terms, verified the legality of an empowerment laid down in legislation. In FBF, the Court, in concrete terms, verified the legality of the exercise of an empowerment laid down in legislation, opting to set a low threshold to determine whether the European Banking Authority's action were necessary to ensure the proper application of EU legislation. By reading necessary as 'useful' rather than 'indispensable', the Court has opened the door to a broad interpretation by the EU agencies of their mandates. As a result, it is arguably converting the non-delegation doctrine into a discretion doctrine.