Belgium v Commission (Online gambling services - Protection of consumers and players : Opinion) [2017] EUECJ C-16/16P_O (12 December 2017)

BAILII [Home] [Databases] [World Law] [Multidatabase Search] [Help] [Feedback]

Court of Justice of the European Communities (including Court of First Instance Decisions)


You are here: BAILII >> Databases >> Court of Justice of the European Communities (including Court of First Instance Decisions) >> Belgium v Commission (Online gambling services - Protection of consumers and players : Opinion) [2017] EUECJ C-16/16P_O (12 December 2017)
URL: http://www.bailii.org/eu/cases/EUECJ/2018/C1616P_O.html
Cite as: [2017] EUECJ C-16/16P_O

[New search] [Contents list] [Context] [Printable version] [Help]


OPINION OF ADVOCATE GENERAL

BOBEK

delivered on 12 December 2017(1)

Case C16/16 P

Kingdom of Belgium

v

European Commission

(Appeal — Protection of consumers — Online gambling services — Protection of consumers and players of online gambling services and prevention of minors from gambling online — Recommendation of the Commission — Article 263 TFEU — Actionable act — Judicial review of soft law instruments — Non-binding acts producing legal effects — Acts that can reasonably be perceived as inducing compliance)






Table of contents


I. Introduction

II. Legal framework

A. Primary law

B. The Commission Recommendation

III. Facts and legal proceedings

IV. The order under appeal

V. Proceedings before the Court

VI. Assessment

A. Substance over form

1. ERTA

2. The ERTA test as applied by the General Court in the present case

3. The problematic elements of the ERTA test

(a) Internal issues

(1) Legal effects, binding force, or binding legal effect?

(2) What role for the drafter’s intent?

(b) External problems

(1) The rise of soft law

(2) Recommendations: no binding force but producing legal effects

(i) At the level of the EU

(ii) At the level of the Member States

4. Back to the roots: ERTA and legal effects

5. The application of the test to the present case

B. Form determines substance

1. A (full) exclusion: recommendation means recommendation

2. Substance or form?

3. The (potentially) necessary clarifications

VII. Conclusion


I.      Introduction

1.        One of the great debates that have shaped the (Anglo-American) legal philosophy over the last few decades is the exchange between Hart and Dworkin on the nature of the law and the structure of a legal system. In the late 1960s and 1970s, Dworkin’s critique of Hart’s Concept of Law (2) crystallised around several themes. One of Dworkin’s key propositions was that Hart’s understanding of a legal system is too narrow and too focused on legal rules, and omits another key element of any legal system: principles. (3)

2.        It is perhaps safe to assume that, notwithstanding its title, when adopting the ‘Recommendation on principles for the protection of consumers and players of online gambling services and for the prevention of minors from gambling online’ (the ‘Recommendation’), (4) the Commission did not intend to take sides in that theoretical debate. However, following an action for annulment brought by the Kingdom of Belgium against that Recommendation before the General Court, a nominally similar, but substantively somewhat different type of debate has been triggered: in EU law, for the purpose of judicial review, how do such principles differ from (binding, legal) rules? Furthermore, can a Commission Recommendation, an EU instrument explicitly excluded from judicial review under Article 263(1) TFEU, nonetheless be subject to an action for annulment under that provision?

3.        The General Court declared Belgium’s action inadmissible, (5) holding that it was not intended that the Recommendation have binding legal effects. The Kingdom of Belgium appealed against that decision to this Court.

4.        In this Opinion, my suggestion to the Court is essentially twofold: first, on the general level, in view of the changing legislative landscape of (not only) EU law, which is marked by a proliferation of various soft law instruments, access to the EU courts should be adapted in order to respond to those developments. In this sense, and as far as the theoretical pun permits, the approach should indeed become somewhat more Dworkinian, recognising the fact that there are norms generating significant legal effects that find themselves beyond the binary logic of binding/non-binding legal rules. Second, on the concrete level of the Recommendation at issue in the present case, a normative instrument that in the light of its logic, context, purpose and partially also language, can reasonably be seen as setting rules of behaviour, ought to be subject to judicial review, irrespective of the fact that it is somewhat disguised as a set of mere ‘principles’ in a recommendation.

II.    Legal framework

A.      Primary law

5.        By virtue of Article 4(3) TEU, the ‘Member States shall take any appropriate measure, general or particular, to ensure fulfilment of the obligations arising out of the Treaties or resulting from the acts of the institutions of the Union. The Member States shall facilitate the achievement of the Union’s tasks and refrain from any measure which could jeopardise the attainment of the Union’s objectives’.

6.        Article 263(1) TFEU provides: ‘The Court of Justice of the European Union shall review the legality of legislative acts, of acts of the Council, of the Commission and of the European Central bank, other than recommendations and opinions, and of acts of the European Parliament and of the European Council intended to produce legal effects vis-à-vis third parties. It shall also review the legality of acts of bodies, offices or agencies of the Union intended to produce legal effects vis-à-vis third parties.’

7.        Pursuant to Article 288 TFEU:

‘To exercise the Union’s competences, the institutions shall adopt regulations, directives, decisions, recommendations and opinions.

A regulation shall have general application. It shall be binding in its entirety and directly applicable in all Member States.

A directive shall be binding, as to the result to be achieved, upon each Member State to which it is addressed, but shall leave to the national authorities the choice of form and methods.

A decision shall be binding in its entirety. A decision which specifies those to whom it is addressed shall be binding only on them.

Recommendations and opinions shall have no binding force.’

8.        Article 292 TFEU reads as follows: ‘The Council shall adopt recommendations. It shall act on a proposal from the Commission in all cases where the Treaties provide that it shall adopt acts on a proposal from the Commission. It shall act unanimously in those areas in which unanimity is required for the adoption of a Union act. The Commission, and the European Central Bank in the specific cases provided for in the Treaties, shall adopt recommendations.’

9.        Article 296 TFEU states that:

‘Where the Treaties do not specify the type of act to be adopted, the institutions shall select it on a case-by-case basis, in compliance with the applicable procedures and with the principle of proportionality.

Legal acts shall state the reasons on which they are based and shall refer to any proposals, initiatives, recommendations, requests or opinions required by the Treaties.

When considering draft legislative acts, the European Parliament and the Council shall refrain from adopting acts not provided for by the relevant legislative procedure in the area in question.’

B.      The Commission Recommendation

10.      Pursuant to Recital 5 of the Commission Recommendation ‘in the absence of harmonisation at Union level, Member States are in principle free to set the objectives of their policy on games of chance and to define the level of protection sought for the purpose of protecting the health of consumers…’.

11.      Recital 8 sets out that ‘the rules and policies that Member States have introduced to pursue public interest objectives vary considerably. Action at Union level encourages Member States to provide a high level of protection throughout the Union.’.

12.      The aim of the Commission Recommendation is stated in Recital 9 as: ‘to safeguard the health of consumers and players and thus also to minimise eventual economic harm that may result from compulsive or excessive gambling. To that end, it recommends principles for a higher level of protection of consumers, players and minors as regards online gambling services. In preparing this Recommendation, the Commission has drawn from good practices in the Member States’.

13.      Recital 15 indicates that ‘it is appropriate to invite Member States to put forward rules providing consumers with information about online gambling…’.

14.      Recital 16 states that ‘where appropriate, the principles of this Recommendation should not only be addressed to operators but also to third parties, including so-called “affiliates”, who are allowed to promote online gambling services on behalf of the operator’.

15.      Recital 27 notes that ‘effective supervision is necessary for the appropriate protection of public interest objectives. Member States should designate competent authorities, lay down clear guidance for operators and provide easily accessible information for consumers, players and vulnerable groups including minors’.

16.      Recital 29 adds that ‘this Recommendation does not interfere with Directive 2005/29/EC of the European Parliament and of the Council and Council Directive 93/113/EEC’.

17.      Section I of the Recommendation sets out its purpose. Under paragraph 1, ‘Member States are recommended to achieve a high level of protection for consumers, players and minors through the adoption of principles for online gambling services and for responsible commercial communications of those services, in order to safeguard health and to also minimise the eventual economic harm that may result from compulsory or excessive gambling’. Paragraph 2 states that ‘this Recommendation does not interfere with the right of Member States to regulate gambling services’.

18.      Section III of the Recommendation lays down rather specific and detailed information requirements for operators’ gambling websites, notably as to the kind of information that should be displayed on their landing page and how many clicks it takes to access each piece of information. Section IV adds further requirements concerning minors.

19.      Section V concerns player registration and accounts. In particular, Paragraph 15 states that ‘Member States should ensure that a person is only permitted to participate in an online gambling service when registered as a player and holding an account with the operator’.

20.      Section VI addresses player activity and support. Section VIII covers commercial communication, Section IX sponsorship.

21.      Section XI deals with supervision. Paragraph 51 invites Member States ‘to designate competent gambling regulatory authorities when applying the principles laid down in this Recommendation to ensure and monitor in an independent manner effective compliance with national measures taken in support of the principles set out in this recommendation’.

22.      Section XII, the last section of the Recommendation, is on reporting. Its Paragraph 52 states that ‘Member States are invited to notify the Commission of any measures taken pursuant to this Recommendation by 19 January 2016 in order for the Commission to be able to evaluate the implementation of this Recommendation’.

23.      Under Paragraph 53, ‘Member States are invited to collect reliable annual data for statistical purposes on:

(a)      the applicable protection measures in particular the number of player accounts (opened and closed), the number of self-excluded players, those experiencing a gambling disorder and complaints by players;

(b)      commercial communication by category and by type of breaches of the principles;

Member States are invited to communicate this information to the Commission, for the first time by 19 July 2016.’

24.      Finally, Paragraph 54 states that ‘the Commission should evaluate the implementation of the Recommendation by 19 January 2017’.

III. Facts and legal proceedings

25.      In 2011, in a Green Paper ‘on online gambling in the internal market’, (6) the Commission identified the Member States’ common objectives on regulation of online gambling services. It also identified the key priority areas for EU action.

26.      In its Communication ‘Towards a comprehensive European Framework for online gambling’, adopted on 23 October 2012, (7) the Commission proposed a series of actions to respond to the regulatory, societal and technical challenges of online gambling. In particular, the Commission stated that, overall, it did not appear appropriate at this stage to propose sector-specific EU legislation for online gambling. In that Communication the Commission announced that it would present recommendations on the protection of consumers in the area of online gambling services, including the protection of minors, and on responsible commercial communication of online gambling services.

27.      The European Parliament, in its Resolution of 10 September 2013 on online gambling in the internal market, (8) urged the Commission to explore the possibility of interoperability between national self-exclusion registers. This would raise awareness about the risks of addiction to gambling and to consider compulsory third-party identification checks. The Parliament also called for online gambling operators to be obliged to provide information on regulatory authorities and warnings to minors and the use of self-restrictions on the gambling website. In addition, the Parliament advocated the setting out of common principles for responsible commercial communications. It recommended that commercial communications should contain clear warnings as to the consequences of compulsive gambling and the risks of gambling addiction. Commercial communications should be neither excessive nor displayed on content specifically targeted at minors, nor where there is a higher risk of targeting minors.

28.      The European Economic and Social Committee has also called on the Commission to intervene to improve consumer protection as regards online gambling, and to protect minors. (9)

29.      Within such a context, the European Commission adopted the impugned Recommendation on 14 July 2014 on the basis of Article 292 TFEU. It was published in full in the L series of the Official Journal of the European Union from 19 July 2014.

30.      The adoption of the Recommendation was accompanied by the Press Release (10) and a Memorandum. (11) That Memorandum presented a Commission Recommendation in the following terms: ‘A Recommendation is a non-binding instrument used by the European Commission to send a clear message to Member States as to what actions are expected to remedy a situation, while leaving sufficient flexibility at national level as to how to achieve this. By setting the objectives to be attained, it should act as a catalyst for the development of consistent principles to be applied throughout the European Union.’ As to the choice of the type of instrument, the Memorandum added that ‘there is no sector specific EU legislation in the online gambling services sector and it was not considered appropriate to propose such specific legislation. Moreover, a Commission Recommendation can be adopted immediately whereas proposals for legislation would have to be adopted by the EU’s Council of Ministers and the European Parliament which can take time’.

IV.    The order under appeal

31.      On 13 October 2014, the Kingdom of Belgium lodged an application for the annulment of the Recommendation in question before the General Court.

32.      By order of 27 October 2015, the General Court dismissed the action as inadmissible. (12) It held that, in the light of its wording, content and context, the Recommendation did not have and was not intended to have binding legal effect. As a result, it could not be classified as an act that can be reviewed as to its legality under Article 263 TFEU. (13)

33.      In its reasoning, the General Court first noted that the Recommendation was worded mainly in non-mandatory terms, despite some minor divergences between the language versions. (14)

34.      The General Court further held that the content of the Recommendation showed that the Commission had no intention to confer binding legal effects on it. (15) In particular, paragraphs 51 to 53 invite Member States to designate gambling regulatory authorities and to notify the Commission of any measures taken pursuant to the Recommendation. This does not impose an obligation on the Member States to apply the principles set out in that act. Furthermore, the Commission expressly stated that the Recommendation did not interfere with the Member States’ regulatory powers in this field: it merely invited Member States to comply with the principles laid down therein. (16)

35.      The General Court considered that the analysis of the wording and content of the Recommendation was confirmed by an analysis of its context. Discussions of the Council, the European Parliament and the Commission show that there was no intention to propose EU legislation in this field at that time. (17)

36.      The General Court added that publication in the L Series of the Official Journal rather than in the C Series could not on its own invalidate the conclusion that the Recommendation was not intended to have binding legal effect. (18) It also said that it cannot be inferred from the mere fact that recommendations, although not binding, may have legal effect, that they can be judicially challenged. To hold otherwise would run counter to Article 263 TFEU, according to which an action for annulment cannot be brought against recommendations, which do not have binding legal effect. (19)

37.      An alleged infringement by an EU institution of the principles of institutional balance, conferral of powers or of sincere cooperation cannot give rise to an exception to the admissibility rules governing actions for annulment laid down by the Treaty. (20) In particular, it does not follow from the fact that in infringement proceedings the Court may examine an act or action with no binding legal effect in the light of the Member States’ duty of sincere cooperation that the same must apply in an annulment action. (21)

38.      The General Court stated that the Recommendation does not lay down any rule or principle to harmonise the services market in the online gambling sector, contrary to the Appellant’s argument. This was clear in the light of paragraph 2 which expressly confirms the Member States’ regulatory powers in this area. (22) The Recommendation was not designed to limit the possibility for each Member State to determine, in accordance with its own preferences, what is required in order to ensure that moral, religious and cultural aspects are protected. (23)

V.      Proceedings before the Court

39.      By its appeal, the Kingdom of Belgium (the Appellant) asks this Court to set aside the General Court’s order, to declare admissible the action for annulment, to decide this case on the merits, to declare admissible the applications for intervention of the Hellenic Republic and the Portuguese Republic, (24) and to order the Commission to pay the costs.

40.      The Commission asks the Court to dismiss the appeal and to order the Appellant to pay the costs.

41.       In its appeal, the Appellant raises three grounds of appeal: (i) infringement of the principles of conferred competences, loyalty and institutional balance; (ii) the violation of the principle of loyalty and the disregard for the position of Member States as privileged applicants; and; (iii) the incorrect interpretation of the legal effect of the Recommendation vis-à-vis Belgium.

42.      In the first ground of appeal, the Appellant claims that the action should have been declared admissible because the General Court should have examined whether the Recommendation’s drafter had competence, instead of only examining whether it produces binding legal effect. More precisely, the General Court did not respect the principles of conferred competences, loyalty and institutional balance, as it failed to examine whether there was a substantive legal basis justifying the adoption of the Recommendation. Article 292 TFEU does not serve as an autonomous legal basis: it empowers both the Council and Commission to adopt recommendations, but to know which of the two is competent the substantive competences conferred on each of those institutions by the Treaties must be examined.

43.      Further, the Appellant claims that even a non-binding recommendation should still be amenable to judicial review to determine whether it complies with the abovementioned principles. By not allowing for judicial review, the General Court disregarded the principle of conferred competences. It also applied Article 263 TFEU in a manner that does not comply with established case-law: any measure adopted by the institutions should clearly state its legal basis. (25)

44.      According to the Appellant, the simple fact of the Commission adopting one of the legal instruments listed in Article 288 TFEU without such competence would in itself produce a legal effect because it would disturb the balance of competences between the EU and the Member States, and also between the EU’s institutions. Consequently, it should be reviewable by the Court. Compliance with those principles should be ascertained before assessing whether it is a ‘true’ recommendation. The Appellant submits that the actual legislative EU act must be adopted in accordance with EU law and respect the prerogatives of the other EU institutions and of the Member States.

45.      In its second ground of appeal, the Appellant criticises the fact that, basing itself on the difference between annulment proceedings and infringement proceedings, the General Court found the case-law setting out that acts with non-binding legal effects adopted in violation of the loyal cooperation requirement are amenable to judicial review in infringement proceedings to be irrelevant. (26) That goes fundamentally against the mutual character of the principle of loyalty. (27) This leads to the General Court preventing a Member State from accessing the Court in annulment actions whereas Member States are privileged applicants.

46.      In its third ground of appeal, the Appellant argues that the General Court has not correctly applied ERTA (28) and subsequent case-law in its examination of the Recommendation. It concludes that the legality of the Recommendation can be reviewed under Article 263 TFEU because it produces legal effects, at least vis-à-vis Belgium and Portugal.

47.      The Appellant claims that the General Court should have taken into consideration the fact that the Recommendation is worded in an imperative manner in several language versions, such as Portuguese, but also two out of the three official languages of Belgium, namely Dutch and German, in particular since recommendations must be taken into account by national courts. The General Court should not look at the Recommendation in a general way, but in a specific manner by determining whether it produces legal effects with regard to Belgium. Considering the imperative wording in Dutch and German, it is possible to conclude that there are ‘stronger’ legal effects vis-à-vis Belgium, compared to other language versions.

48.      In its defence, the Commission dismisses the Appellant’s arguments as to the conditions for admissibility of annulment proceedings against recommendations. It suggests that the Appellant’s arguments concern the merits of the case (in raising the principles of conferral, loyal cooperation, institutional balance, and legal basis) whereas the pending procedure is limited to admissibility.

49.      Turning to the mutual character of the principle of sincerity (second ground), the Commission states that it was wrong for the Kingdom of Belgium to assert that the General Court has created a procedural inequality in favour of the Commission in infringement proceedings and at the expense of the Member States in annulment proceedings. That conclusion of the General Court only implies that the legality of recommendations cannot be reviewed on the basis of Article 263 TFEU. Thus, there is no inequality in the application of Article 263 TFEU.

50.      The Commission considers that it is irrelevant that slightly different wording appears in one of the Member State’s official language versions. This is because an assessment of the legal effects produced by a recommendation, if any, should also be based on the instrument’s aim and context. According to the Commission, EU acts should be interpreted in an autonomous manner, independently of the domestic law.

51.      In its reply, the Appellant focused on the legal basis of the Recommendation, which it deems to be lacking. It insisted that a Commission Recommendation produces legal effects by virtue of its existence (albeit limited) which means it must be reviewable as to its legality. A Member State should be able to ask for a review of the validity of any EU act when it is not clear whether it was passed within the limits of the conferral principle, especially when there is no substantive legal basis. A mere reference to Article 292 TFEU is not enough. It does not fulfil the requirement for a substantive legal basis.

52.      In its rejoinder, the Commission argued that there is no lacuna in the procedures provided for by the Treaties. Recommendations are excluded from the scope of Article 263 TFEU. The only question is whether this Recommendation is a ‘true’ recommendation. The issue of the legal basis belongs to an assessment of the merits and should therefore be assessed only if the claim is admissible. In any event, the Commission did not consider that the Appellant had proved why a legal basis other than Article 292 TFEU would be necessary.

53.      In addition to their written submissions, the Kingdom of Belgium and the Commission presented oral argument at the hearing held on 26 June 2017.

VI.    Assessment

54.      In its appeal, the Appellant raised three grounds. I consider it appropriate to deal with the third ground of appeal first (by which the Appellant claims that the General Court erred in its assessment of the absence of legal effects of the contested Recommendation), essentially for two reasons. First, it is the thrust of the appeal before the Court. In one way or another, the third ground also touches on elements of the first and second ground. To some extent, in a case like the present one, elements of substantive assessment are already linked to and seep into the admissibility stage. Second, I am of the view that the Appellant's third ground of appeal is well-founded. The General Court erred in law: it incorrectly interpreted the effects of the Recommendation at issue, and thus it incorrectly assessed the admissibility of the application. (29)

55.      A recommendation is a ‘typical’ act of EU law, listed in Article 288 TFEU. In contrast to a potentially vast array of ‘atypical’ acts of EU institutions and bodies, namely those not listed, Article 288 TFEU sets out the characteristics of a recommendation - it shall have no binding force. In addition, the first indent of Article 263 TFEU clearly excludes recommendations from actions for annulment.

56.      In view of this legislative landscape in primary law, the extension of the ERTA (30) line of case-law, designed for ‘atypical’ acts, is perhaps not entirely automatic. Two approaches are therefore possible when dealing with potential actions for annulment against recommendations. First, there is the ‘substance over form’ approach, meaning that even in the case of typical acts, it is the analysis of the substance of the contested act that shall determine the admissibility of an action for annulment. If, in contrast to its title, the act is in fact something else than it says (for example that it is not a ‘true’ recommendation), it ought to be reviewable, irrespective of its appellation. Second, there is the ‘form determines substance’ approach, suggesting that a spade is a spade, even if it comes in a somewhat strange shape. But because and as long as there is ‘spade’ written on it, it shall be understood and interpreted as such.

57.      This Opinion is structured as follows. Section A sets out the first approach: I start by going back to the roots of the test in ERTA (A.1.), before turning to the subsequent case-law and the understanding of the test by the General Court in the present case (A.2.). Next, I shall explain why, in case of recommendations, that test is problematic on a number of levels (A.3.), before turning to its suggested readjustment (A.4.). I then demonstrate how such a more nuanced test would apply to the Recommendation at issue (A.5.).

58.      Section B starts by outlining the second approach (B.1.), before setting out further arguments why, in my view, that approach should not be embraced by the Court (B.2.). However, should the Court nevertheless wish to go down that road, I would invite it to provide at least several important clarifications as to the nature and effects of recommendations (B.3.).

A.      Substance over form

1.      ERTA

59.      In its first version from 1957, Article 173 of the EEC Treaty (later Article 230 EC, today Article 263 TFEU) stated that the legality of ‘legal acts of the Commission and Council, other than recommendations and opinions’ could be reviewed before the Court. It did not give a positive definition of the legal acts that could be reviewed. It was therefore for the Court to decide which acts were amenable to review: whether it was only those acts of the Commission or Council that were explicitly referred to as binding acts in the then Article 189 EEC (now Article 288 TFEU), namely regulations, directives, and decisions, or whether they also included ‘atypical acts’ adopted by those institutions, but not expressly mentioned in the Treaties.

60.      In its ERTA judgment, (31) which concerned minutes of the Council relating to the negotiation and conclusion of an international agreement, the Court laid down a test to determine whether or not an action for annulment of an act of the institutions is admissible (‘the ERTA test’). The Court held that under Article 173 EEC an act ‘open to review by the Court’ includes ‘all measures adopted by the institutions which are intended to have legal force … An action for annulment must therefore be available in the case of all measures adopted by the institutions, whatever their nature or form, which are intended to have legal effects’. (32)

61.      Thus, the original ERTA test was concerned with two elements: was there an act of the EU which is intended to have legal effects?

62.      In subsequent case-law, the Court has applied that test to a number of atypical acts such as internal Commission instructions or guidelines; (33) codes of conduct implementing a Council regulation; (34) communications; (35) information notes; (36) or letters. (37)

63.      Inspecting those decisions closely, the language of the test has not always been exactly the same. There is, however, one clear common theme: in relation to all those atypical acts, the Court has clearly held that the substance of an EU act shall prevail over its form when deciding on the admissibility of an action for annulment. (38) The actual name and form of the act are not conclusive in determining whether its legality can be reviewed or not.

2.      The ERTA test as applied by the General Court in the present case

64.      In the contested order the General Court stated that: ‘according to consistent case-law any measures adopted by the institutions, whatever their form, which are intended to have binding legal effects, are regarded as actionable measures within the meaning of Article 263 TFEU… Any act not producing binding legal effects, such as preparatory acts, confirmatory measures and implementing measures, mere recommendations and opinions and, in principle, internal instructions, falls outside the scope of the judicial review provided for in Article 263 TFEU… In the light of the case-law, in order to determine whether an act is capable of having legal effects and, therefore, whether an action for annulment under Article 263 TFEU can be brought against it, it is necessary to examine its wording and context, its substance and the intention of its author’. (39)

65.      The General Court then went on and applied the ERTA test to a recommendation, at least to my knowledge, for the first time. The General Court examined the wording, the context, the substance and the intention of the drafter. It held that the Recommendation did not have binding legal effect on its addressees. The General Court admitted that the principles set out in the Recommendation were very detailed. It held nonetheless that they had no ‘obligatory’ nature, as notably seen through the ‘invitation’-type wording of most language versions of the Recommendation. Consequently, by giving more weight to the wording than to the other factors, the General Court held that the application was inadmissible.

66.      In sum, the determining elements in the General Court’s analysis of the contested Recommendation appear to be the absence of binding legal effects of the Recommendation, which was determined by the intent of the Commission, as inferred primarily from the form of the act and its wording.

3.      The problematic elements of the ERTA test

67.      There are some problems that the ERTA test poses, if conceived of and applied in this way. They can be grouped in two categories: first, there are the issues internal to the test, its logic, conditions and their articulation, that come to the surface perhaps even more sharply when it is applied to a recommendation. Second, there are issues that could be called external ones. They relate to the fact that the ERTA test, over time effectively becoming narrower and narrower, is falling out of sync with the evolution of the EU’s normative landscape. In a world where various instruments of soft law are, in fact, becoming much more numerous and significant than in 1971, the conditions for standing and judicial review should react to such developments.

(a)    Internal issues

68.      There are two elements that stand out in this regard: the necessarily binding nature of an EU act in order to be able to review its legality (1) and the drafter’s intent as to its legal effects (2).

(1)    Legal effects, binding force, or binding legal effect?

69.      Paragraph 42 of the ERTA judgment (40) did not refer to measures intended to have binding legal effect, but only legal effects. The same appears to hold true of the expressions used in the other extant language versions of the time. (41)

70.      The shift in terminology from mere legal effects to ‘binding legal effects’ appeared in subsequent case-law. (42) That trend seems to be advancing further recently as the Court now conditions the possibility of judicial review of EU acts on whether they have binding legal effect. (43)

71.      Admittedly, it could be suggested that in spite of the use of the term ‘legal effects’, what the Court in fact meant in ERTA was ‘binding force’, although that proposition is, in view of the type of document reviewed in that case (Council minutes), not entirely convincing. Such a suggestion could in particular rely on the wording of Article 189 EEC, which back then already distinguished between binding acts (regulations, directives and decisions) and non-binding acts (recommendations and opinions). (44) Thus, even if the Court did not clearly construe Article 173 EEC in the light of the nomenclature laid down in Article 189 EEC, it is likely that that latter provision had an impact on the test.

72.      Be that as it may, it is also true that it is still a rather recent phenomenon that the Court seems to have become generally stricter by narrowing the scope of Article 263 TFEU to acts that have binding legal effect, adding to the wording of that provision, which limits itself to just legal effects (vis-à-vis third parties). However, in the absence of any clear discussion in the case-law on that distinction, and above all the evidence of an advised choice in one direction or the other, one may wonder whether the Court in fact wished to become stricter and narrower. It is nonetheless clear that neither the wording, nor the logic of ERTA necessarily implied binding in addition to just legal effects.

73.      This is not just a game of words. There is a considerable practical impact, as is evident in the present case. What exactly constitutes legal effects may be open to debate. The notion is, however, clearly quite broad, accommodating all types of impact on/in the law, its interpretation and application. By contrast, binding effect, a fortiori binding legal effect, represents a much, much narrower category.

74.      Traditionally, the binding force of the law connects with coercion. In case of failure to comply, enforcement and sanction may follow. In such a (pure positivist (45)) vision, the existence of a sanction is the defining element of binding force.

75.      Theoretical discussions aside, it is quite clear that if the yardstick of binding legal force were to be embraced, a number of acts that are likely to have significant legal effects on the behaviour of the addressees, but which are not in the traditional sense binding because they contain no direct or independent coercion mechanism, will escape review under the ERTA test, and ultimately under Article 263(1) TFEU. As shall be seen in due course, that is notably the case with the Recommendation at hand.

(2)    What role for the drafter’s intent?

76.      There is, second, a lack of internal clarity in the ERTA test as applied by the General Court: what is the precise role of the drafter’s intent for the purposes of determining whether an act shall be deemed to produce (binding) legal effects?

77.      ERTA, as well as the current wording of Article 263 TFEU, are based on the drafter’s intent. The use of the past tense (was intended to) would further imply that what is relevant is the ascertaining of past (historical) subjective intent of the drafter as it stood at the moment of the adoption of the act in question. Such an understanding would be arguably also in line with the general rules applicable to annulment actions. In the course of such actions, the contested EU acts must be assessed on the basis of the elements of fact and law existing at the time when the measure was adopted. (46)

78.      However, if intent of the drafter of the act is supposed always to be a subjective historical one, then in practice, no recommendation will ever be reviewable. The assessment of the nature and effects of a recommendation quickly falls into a loophole. Because the Commission had no intention of passing binding legislation, a recommendation was selected. Because the Commission selected a recommendation, its subjective intent was clearly for that instrument not to be binding. Because of such intent, effectively certified by the choice of the instrument, it can never be binding, irrespective of its content and wording, because the Commission had no intention of passing any binding legislation.

79.      In this way, the choice of the instrument will always pre-determine the context and the purpose of the measure, which is then likely to override any content and wording of it.

(b)    External problems

80.      Besides these logical problems, inherent in the articulation of the ERTA test embraced by the General Court when applied to recommendations, the same test arguably faces broader, external challenges. Two will be outlined in this section: first, there is the rise of various forms of soft law that strictly speaking do not have binding force but at the same time generate legal effects (1). Second, recommendations are in practice likely to generate a number of legal effects, often quite significant ones, on both the EU level as well as the national level (2).

(1)    The rise of soft law

81.      There is a wide array of instruments in (not only) EU law, under various names and forms (guidelines, communications, codes of conduct, notices, recommendations, opinions, interinstitutional agreements, conclusions, statements, resolutions and so on), that are generically referred to as ‘soft law’. They can be adopted in any field, at all possible stages of the decision-making process, whether that is early, upstream consultation of the stakeholders or downstream implementation of legislative acts. Thus, those instruments can equally be pre-legislative or post-legislative.

82.      There are perhaps two elements on which there is a general agreement in the otherwise very different approaches to such soft law instruments: first, soft law does not easily fit within the binary, black and white distinction between binding and non-binding legal effects. Second, in the past decade or two, it has been on the rise, becoming increasingly more frequent than before. (47)

83.      Consequently, the issue of soft law proliferation and (the absence of) the judicial review thereof have been discussed not only in scholarly literature, (48) but also by the institutions of the EU. (49)

84.      In addition, a number of Member States’ high jurisdictions have in the past years sought to address the same phenomenon on the national level. They have opened up judicial review so as to include acts that are not strictly speaking binding, thus effectively relaxing the admissibility criteria of annulment actions to ensure the right to an effective judicial protection. (50) That is for example the case when addressees can perceive the contested act as being binding on the basis of a set of elements, notably because they contain incentives, (51) or when their drafter disposes of the power to adopt sanctions, (52) or when it can have significant effects on the addressee. (53) The same, or in fact much more, is true of the common law courts which have been traditionally much more permissive than their Continental counterparts in admitting the judicial review of non-binding acts. In Ireland, for instance, the courts ensure the protection of fundamental rights even when the contested measure is not binding and has no concrete effect on the rights and obligations of the addressees. (54)

85.      Finally, the approach and the practice of the French Conseil d’Etat are worth singling out in this regard. First, the Conseil d’Etat mapped the ground so to speak in a comprehensive report which offered, inter alia, a definition of soft law. (55) Second, last year it also built on that study by devising a new judicial test focusing on economic effects and the existence of significant influence on the behaviour of the addresses of the instrument. (56)

86.      It appears that despite their diversity, both at the national as well as EU law levels, the various soft law instruments share the same key feature: they are not binding in the traditional sense. They are a type of imperfect norm: on the one hand, they clearly have the normative ambition of inducing compliance on the part of their addressees. On the other hand, no instruments of direct coercion are attached to them. Usually adopted in the wake of a process of consultation with the different stakeholders (a bottom-up approach), they may contain ‘mild obligations’ or ‘robust exhortations’ that are coined in terms of ‘invitation’.

(2)    Recommendations: no binding force but producing legal effects

87.      Recommendations generally fit such a description. In the Treaties, recommendations are only defined negatively: they have no binding force (Article 288 TFEU). Apart from that, the use and practice of recommendations is varied. (57) They usually present invitations to adopt certain behaviour, follow a policy, or rules that are considered appropriate by their drafter(s) in view of the aim pursued.

88.      However, while clearly described as not binding, recommendations can generate considerable legal effects, in the sense of inducing certain behaviour and modifying normative reality. They are likely to have an impact on the rights and obligations of their addressees and third parties. By way of illustration, a number of such effects shall be outlined in this section, on two distinct but interrelated levels: (i) the EU and (ii) the Member States.

(i)    At the level of the EU

89.      At the EU level, three types of legal effects of recommendations are worth highlighting: (i) reliance and legitimate expectations; (ii) their interpretative role; and (iii) the potential of recommendations to generate parallel sets of rules which pre-empt the legislative process and thus have an impact on the institutional balance.

90.      First, if an EU institution adopts recommendations as to how others are supposed to behave, it is perhaps fair to assume that should it become relevant, that institution can be expected to follow that same recommendation as to its own practices and behaviour. From this point of view, the legitimate expectation thus created is effectively analogous to other types of soft law that EU institutions or bodies generate and which is perceived as the (auto)limitation of the exercise of their own discretion in the future. (58)

91.      Second, recommendations are likely to be used in legal interpretation, in particular in order to give meaning to indeterminate legal notions contained in binding legislation. That is in particular (but certainly not exclusively) the case for post-legislative recommendations that are not adopted on the sole basis of Article 292 TFEU, but on the basis of a piece of secondary legislation, precisely in order to flesh out the legal notions set out therein. But pre-legislative recommendations may also fulfil the same function, with regard to either indeterminate legal notions in the Treaties, or for the purpose of the interpretation of another legal instrument, which overlaps rationae materiae with that recommendation. In this way, both types of recommendation can complement binding legislation.

92.      Third, in Grimaldi, the Court has already explained the circumstances in which recommendations may be taken: they ‘are generally adopted by the institutions of the Community when they do not have the power under the Treaty to adopt binding measures or when they consider that it is not appropriate to adopt more mandatory rules’. (59)

93.      What is perhaps the greatest strength of recommendations may also then be the greatest danger. They could be used as more than just tools for advancing policies that are politically (lack of consensus) or legally (no specific powers to that effect) gridlocked. They could also potentially be used as a tool to circumvent the same legislative processes.

94.      That creates two types of pre-emption: a short and a long-term one. The immediate problem of circumvention of the other institutions normally participating in the legislative process has already been recognised and discussed. (60) It is therefore clear that a recommendation may have an impact on institutional balance, (61) and so also on the separation of powers within the EU. Yet, if recommendations were excluded from a review of legality on the sole ground that they are not binding, the principle of institutional balance could never be upheld. (62)

95.      There is, however, another type of pre-emption that is likely to be present in particular for pre-legislative recommendations: the ability to articulate the norms before the actual legislative process takes place, which may even translate into unilateral pre-emption of the legislative process. It is not disputed that a recommendation has the ambition to induce compliance on the part of its addressees. Now if it is even partially successful, it will shape the range of conceivable (acceptable) normative solutions for the future. If, based on a recommendation, a number of EU institutions or Member States already comply, those actors will, in the legislative process that may potentially follow, naturally promote the legislative solution that they had already embraced. In this way, the soft law of today becomes the hard law of tomorrow.

(ii) At the level of the Member States

96.      There are at least three types of legal effects that recommendations have on the level of the Member States. Their exact scope would depend on how far the principle of loyal and sincere cooperation is pushed with regard to recommendations.

97.      The first and main type of obligations for recommendations that the Court has so far stated is the duty for national courts to take them into consideration in interpreting national law implementing those provisions. It is quite clear that recommendations cannot in themselves create rights upon which individuals may rely before a national court. (63) However, in Grimaldi, the Court also added that recommendations ‘cannot be regarded as having no legal effect. The national courts are bound to take recommendations into consideration in order to decide disputes submitted to them, in particular where they cast light on national measures adopted in order to implement them or where they are designed to implement binding Community provisions’. (64)

98.      It should be noted that, until now, such interpretative legal effects have been recognised by the Court only in the case of recommendations, thereby singling them out among other EU acts of non-binding nature. (65)

99.      However, what exactly does the obligation to take intoconsideration mean? There are a range of possible interpretations. A potential extreme of the spectrum is whether it would amount to a Von Colson (66) type of obligation of conform interpretation? In view of the formulation adopted in the judgment, it appears that the Court did not perhaps have the intention of going as far as imposing a duty on domestic courts to interpret national law in conformity with recommendations. (67) On another, imaginary extreme, to ‘take into consideration’ could also mean ‘have a look at’ and then choose to completely disregard.

100. There could be, in theory, an intermediate position: a national authority shall at least be required to state reasons when departing from a recommendation, without having a firm duty to interpret national law in conformity. That proposition has been made with regard to another type of soft law in the past. (68) It is possible to understand why such a seemingly intermediate position might have some traction: although the Court has so far only imposed such a duty on the drafter itself of rules of conduct laid down in internal measures or guidelines that have external effects, (69) it cannot be ruled out that that line of case-law could equally apply to recommendations, which are admittedly a more advanced and refined form of soft law since they are one of the ‘typical’ acts mentioned in Article 288 TFEU and since they must be taken into account by national courts. (70)

101. Why call such a position seemingly intermediate? For the simple reason that the obligation of stating reasons why a judge departs from a certain source necessarily means that that source is of binding nature. It is only the departure from mandatory sources which a judge is obliged to justify. (71) Thus, if the ‘obligation to take into consideration’ were to be interpreted as imposing a duty on national judges to justify and to explain why they did not follow a recommendation, that would eo ipso mean that such recommendations not only have ‘some legal effects’, but that they are in fact binding.

102. Second, what are the exact duties of national authorities vis-à-vis a recommendation? The wording of Article 4(3) TEU, in capturing the duty of loyal and sincere cooperation in the EU, is no doubt very comprehensive and potentially far-reaching. It could be suggested that since that provision refers only to ‘obligations’ and since recommendations are, as per Article 288 TFEU ‘non-binding’, they cannot by definition create any obligation and thus do not fall under Article 4(3) TEU at all.

103. I do not think that such an understanding of Article 4(3) TEU would truly reflect the interpretative approach the Court has, for quite some time, been taking to that provision. The duty of sincere and loyal cooperation tends to be applied on the level of principles, not necessarily always zooming in on a concrete and specific provision or a discrete legal obligation. (72)

104. Even if one were to assume that there is no positive obligation to implement a recommendation, can the same be said with regard to potential ‘milder’ obligations of the Member States, such as to take a recommendation into consideration when adopting legislation in the given field? That effect might perhaps be more visible for post-legislative recommendations, which are used to flesh out legal notions in binding legislation. Could it not be expected of a Member State, when implementing the original piece of EU legislation, to which the post-legislative recommendation is in a way ‘attached’, to implement it in the way further clarified in that recommendation? If not, what is the recommendation then for? If yes, then the considerable and real legal effects of a recommendation can hardly be disputed.

105. However, it could certainly be stated that there is no obligation to implement because there is no separate and distinct sanction for failing to do so. Even if one leaves aside the somewhat formalist understanding of ‘direct’ sanction being the key and defining element of binding force, (73) what about potential negative obligations incumbent on the Member States with regard to recommendations? At this stage this is certainly in the arena of conjecture and not an element of valid law, but if the blocking effect of directives applies to the period before their transposition period has lapsed, and in this period, Member States cannot adopt a measure liable seriously to compromise the result prescribed by a directive, (74) could the same logic not be applicable to a recommendation?

106. Third and final, what types of legal effects could recommendations produce in the context of national rules and procedures? In Grimaldi, the Court made clear that a preliminary reference for interpretation of a recommendation could be made. (75) The question remains whether a national court could ask the Court to assess the validity of a recommendation. To my knowledge, such a case has so far never occurred, but it would appear that in Grimaldi, the Court confirmed that such a reference would be possible. (76)

107. It would thus appear that the Court has clearly foreseen that a recommendation will produce legal effects on the national level. After all, it is meant to be taken into consideration in the Member States, whatever exactly that may entail. It might be added that in the past, the case-law of the Court has acknowledged and in fact reviewed, via a preliminary ruling procedure, several non-binding EU acts that did have had repercussions on the national level, recently including a press release of the European Central Bank. (77)

108. Thus, a recommendation, as other acts of EU law of seemingly non-binding nature, may be made subject to a request for a preliminary ruling, both on interpretation as well as validity. To my mind, that could hardly be otherwise, in a complete system of legal remedies. (78) The point of recommendations is to induce compliance. Imagine a Member State which, having acted in good faith and in the spirit of sincere and loyal cooperation, has transposed a recommendation into the national law. By an act of national legislation, that Member State established obligations for individuals on the national level. Now, if that national legislation is challenged before the national courts, it would be somewhat peculiar to refuse the review of what constitutes the substantive basis of that national legislation, namely, the EU recommendation,(79) with the somewhat formalistic excuse that what created those obligations was national legislation, not an EU law instrument, and that the Member State did so purely of its own volition.

4.      Back to the roots: ERTA and legal effects

109. The detailed discussion in the previous section had a twofold purpose: to demonstrate, first, the problems the ERTA test (as later gradually modified) faces when applied to recommendations (but, in my opinion, more broadly also when applied to other soft law instruments); and, second, that in spite of perhaps not being endowed with binding force in the traditional and rather narrow sense of the word, recommendations may produce significant legal effects, both on the EU as well as the national levels.

110. In my view therefore, the ERTA test as applied by the General Court is in need of some readjustment. My suggestion in this regard is quite simple: the test should move back to its origins, to the ERTA judgment, and also to the wording of Article 263(1) TFEU. Both of these refer to ‘legal effects’ and not to ‘binding legal effect’. With river beds as well as with case-law, it is sometimes necessary to clean the stream by removing the (verbal) sediments assembled over the years that make the law impossible to navigate.

111. Such a readjustment is not the revolution it seems to be at first glance. The basis of the test for assessing whether or not an EU law act produces legal effects vis-à-vis its addresses and/or third parties would remain the same: what shall be assessed is the text, context and purpose of the contested act. However, two clarifications to the way in which that test is to be applied are needed: first, what is to be assessed is the existence of legal effects only, not binding legal effect. Second, within the test, the stress would be on the content and context of the measure, not the mere text.

112. As already explained in the previous section, the dichotomy of binding and non-binding legal effects is of little analytical use for soft law. If the precondition for the existence of binding effects is direct enforcement mechanism and coercion, then by definition, soft law will never be binding, irrespective of the provisions it contains.

113. Instead, the assessment of the ability to generate legal effects, that is, to have an impact on/in the legal situation of its addressees, should focus on a different issue: could I, as a reasonable addressee, infer from the content, aim, general scheme and the overall context of a recommendation or, more generally, of a soft law instrument that I am expected to do something? Would I be likely to modify my behaviour accordingly, or is that act likely to impact on my legal position?

114. Next, within the three classical tenets of any statutory interpretation, text, context, and purpose, in cases of the assessment of a recommendation or other soft law instruments, the wording (notably denomination and verbal form) of an act should not outweigh its content, context and purpose in the framework of its assessment. The wording should even, through its essence, be rather secondary compared to substantial elements in the context of admissibility. If it were to be otherwise, the ‘inviting’ wording would necessarily lead to the exclusion of judicial review. It would in fact mean that form prevails over substance and, therefore, that no recommendations using vocabulary of ‘encouragement’ could ever be reviewed. Thus, a more delicate emphasis should perhaps be put on the logic, content, context and purpose.

115. Within such an assessment of context and purpose, three factors appear to be relevant in order to determine whether an EU act is likely to produce legal effects and can reasonably be expected to be complied with.

116. The first factor that should be taken into consideration is the degree of formalisation (does the EU measure take on the form of a legal act?) and of definitiveness of the measure (has it been adopted at the end, as the culmination of a consultation or, more generally, in a ‘soft-law making’ process?). In other words and considering both together, does the EU act in question appear to be rather like a finalised piece of legislation?

117. Regarding the format of a potentially challengeable act, it must appear to be like a legal text so that it could be reasonably perceived as producing legal effects. In this respect, an act will appear to be a legal act if, for instance, it is divided into articles or at least sections, if it is published in the Official Journal (certainly in the L series, where legislation is supposed to be published).

118. Regarding the definitiveness aspect, preparatory acts are likely to fall short on this account. The same logic already applies to preparatory acts in the context of the EU decision-making process. (80) It should a fortiori also hold for soft law-making processes. Such an exclusion from judicial review of preparatory acts appears all the more important in the context of soft law where the consultation process can entail the adoption of several acts.

119. The second factor relates to the content and overall purpose of the contested act: how precise are the ‘obligations’ contained therein? What is the general aim pursued? The more general and abstract EU acts are, the less likely that they will induce concrete, specific compliance in their addressees. If, on the other hand, the EU act features a number of specific and precise commitments, that element is certainly relevant. In addition, if the text has a clear harmonising purpose, it is even more likely to be perceived as likely to induce legal effects.

120. The third factor pertains to enforcement. Does the measure contain any clear and specific compliance, enforcement, or sanction mechanisms? Naturally, this does not aim only at direct enforcement, which is very unlikely to be present, but at indirect mechanisms or enforcement, both structural and institutional.

121. Amongst the structural compliance mechanisms might be a number of indirect mechanisms, such as reporting, notification, monitoring, and supervision. Elements of peer pressure might also be of relevance, such as the publishing of performance tables, reports involving public naming and shaming, and so on.

122. The institutional element is also of relevance: which institution adopted the instrument in question? Is it the same institution that in related or even the same fields of regulation is able to impose sanctions on the same addresses? (81) If that is indeed the case, then it is likely that the contested act will effectively induce compliance.

5.       The application of the test to the present case

123. Looking at the contested Recommendation through such lens, I am bound to conclude that overall that Recommendation goes considerably beyond what might be expected from a document that simply recommends certain principles. In this specific case, it can indeed be argued that that Recommendation is bound to produce legal effects and that reasonable addressees are likely to modify their behaviour in order to comply, at least partially, with the Recommendation.

124. If substance is to prevail over formal designation of an act as to the assessment of its true nature, then the analysis to be carried out is, having removed the formal title of the document, what the document appears to be from its wording, content, context and purpose.

125. Starting with the overall purpose of the Recommendation, its recitals (82) and the documents accompanying its adoption (83) state rather explicitly that the Recommendation aims at reaching a minimum degree of harmonisation inasmuch as it recommends principles for a higher level of protection of consumers, players and minors regarding online gambling services. At the same time, it also clear that it does so in a rather sensitive subject-matter, certainly from the point of view of a number of Member States. (84)

126. Next, there are several elements of content and context which are worth highlighting. First, the contested act is a highly structured text, and in appearance, a legal text. It has no fewer than 30 recitals. The Recommendation itself is divided into 12 numbered sections. It has been published in the L series of the Official Journal of the European Union.

127. Accordingly, the actual form of the Recommendation offers an impression that it is bound to produce legal effects. That is also confirmed by the fact that the Recommendation is not a preparatory act. It already marks the culmination of a consultation process since it fleshes out a previous Green Paper and a Communication of the Commission. Therefore, it clearly crystallises the position of the latter on the topic of consumer protection in online gambling services by making some very concrete recommendations to the Member States.

128. Second, the level of detail and precision of concrete provisions of the Recommendation is striking. Far from setting out mere ‘principles’, the Recommendation lays down rather clear and precise rules.

129. To give a few examples: in section III, the Recommendation sets out the detailed content of the information that should be displayed on the landing page of the operator’s gambling website and be accessible from all pages on the website. In section V, the Recommendation invites Member States to ensure that a person can only participate in an online gambling service when registered as a player and holding an account with the operator. The details of the player’s identity should be verified by the latter. Also, section VIII aims at regulating commercial communication. It notably precludes Member States from making certain statements, such as portraying gambling as socially attractive or suggesting that gambling can resolve personal problems.

130. Third, the Recommendation contains very detailed and comprehensive supervision and reporting ‘invitations’. Under section XI, Member States are ‘invited to designate competent gambling regulatory authorities when applying the principles laid down in this Recommendation to ensure and monitor in an independent manner effective compliance with national measures taken in support of the principles set out in this Recommendation’. Section XII concerns reporting. Under its provisions, not only are Member States invited to notify the Commission of any implementing measures, but also to collect annual data for statistical purposes before certain deadlines.

131. With regard to such reporting and supervision ‘invitations’, two observations are called for. First, it would appear that in the past and in general, the Court demonstrated a particular sensitivity towards inserting compliance or monitoring mechanisms into various atypical instruments. On at least two occasions, it annulled atypical Commission acts because they regulated reporting requirements in detail and arguably went beyond what was appropriate for that type of act. (85) Second, in the context of this particular Recommendation, it is interesting to note that a non-binding Recommendation that Member States are not obliged to implement still expects those Member States (or even only those who decided to accept the ‘invitation’) to designate, to monitor, to notify, to evaluate, to collect data, and to report back to the Commission by specific dates on all those (entirely voluntary) activities.

132. Paragraph 54 of the Recommendation then states that the Commission should evaluate the implementation (sic!) of the Recommendation by 19 January 2017. Yet, it was established at the hearing that the Commission has so far failed to do so because it still expects the reports of the Member States before drafting its own report. At this stage, it would seem to me that the degree of cognitive dissonance present in such propositions reaches the quality of an advanced Jedi mind trick.

133. Fourth, the Recommendation is designed to induce the Member States into adopting certain legislation and through that legislation to have an impact on gambling undertakings and players, which are its indirect addressees. Thus, of course it can be maintained that formally and in itself, it is not the Recommendation but the potential national legislation that will impact third party rights, but it is hard to deny that the effective source of the national legislation is to be found in that Recommendation. (86)

134. Fifth, perhaps in itself a marginal point but also highlighting the dissonance between the content and the title, it is interesting to note both Recital 29 and Paragraph 2 of the Recommendation. They state, respectively, that this Recommendation does not interfere with the (binding and valid) EU Directives and that it does not interfere with the right of Member States to regulate gambling services. Those statements simply beg the question that if the Recommendation were indeed a purely non-binding recommendation not intended to have any legal effects whatsoever, why any of this would be necessary to state explicitly. A true non-binding soft law instrument could never, by definition, interfere with binding and valid EU legislation, or with Member States’ competences.

135. In sum, already on these elements of purpose, content and context, if a reader were to be given that document with its title removed and were invited to read the text without knowing its title, it is safe to assume that he might think that he is reading a directive or, at some stages, even a regulation, but certainly a legislative document that seeks to impose clear and precise obligations and induce compliance.

136. It is only now that I turn to the wording of the Recommendation, more concretely to the specific text of the individual provisions (having dealt already in the previous points of this section with the level of detail and concreteness of those provisions). In the reasoning of the General Court wording seems to play a decisive role. However, for the reasons that I sought to explain in general in the previous section, (87) to my mind, the wording is significant, but not decisive. Furthermore, it should certainly not be decisive if in and of itself, it is actually inconclusive.

137. In its written as well as oral submissions, the Appellant contested the linguistic assessment carried out by the General Court in its order. In particular, the Appellant maintained that in two out of the three official languages of the Kingdom of Belgium, namely in Dutch and German, the wording of the provisions appears to be ‘stronger’ than in other language versions. Thus, as a logical consequence, the Recommendation would produce ‘stronger’ legal effects in Belgium.

138. This particular argument of the Appellant is unconvincing. Admittedly, some language versions could be perceived ‘more binding’ than others. That is notably the case with the German, Spanish, Dutch and Portuguese versions. Contrary to the affirmations of the General Court, the Polish or Czech versions might perhaps also be open to debate.

139. However, it is of little relevance that the Dutch and German versions seem to use a more imperative wording than others. The fact that they are official languages in Belgium does not give them more weight than any other language versions. There is the well-known principle of equal authenticity of all language versions of EU sources that to my mind ought to be applicable to recommendations as it is to any other measures adopted under Article 288 TFEU. It is established case-law that when language versions diverge, the provision in question must be interpreted by reference to the purpose and the general scheme of the rules of which it forms part. (88)

140. It is precisely such debatable linguistic comparisons that demonstrate and underline why the amount of turns of the phrase ‘are invited to’ instead of ‘shall’ is of only limited significance in the assessment of such a measure. In general, all language versions go in the same direction and allow for a clear conclusion as to the nature of the text without having to dwell in depth on the context and purpose of a measure. Or, they are at odds with one another and the text of a recommendation should not outweigh its context and purpose. It is, however, incorrect to state that there are discrepancies amongst the various language versions but then still maintain that because the majority of them state X, that ought to be the correct interpretation. In the system of equal authenticity of all languages, no languages can be ‘out-voted’ in interpretation. (89)

141. At the end of the day however, the minutiae examination of (non-) imperative use of language in some languages of the EU will always be inconclusive, in particular in the examination of legal effects of soft law instruments. For that particular type of assessment, context, system, and logic matter much more. There will inevitably be an enormous difference in the interpretation of the statement ‘I invite you to send me your comments on this matter in writing before noon this Friday’, depending on whether that statement is uttered by one’s boss, one’s research colleague, or one’s partner. Naturally it depends on the actual relationship in question, but it is quite likely that in the first case, the ‘invitation’ is in fact a command, the second one just a suggestion, and the third one a joke.

142. In sum, each of the individual elements of content and context in isolation could perhaps lead it still to be seen as a recommendation that is not intended to generate any legal effects. However, when taken together, the joint operation of these different elements in the context of this specific Recommendation, in addition taking into account the fact that it was adopted by the institution that is entrusted with the policing of the rules in the same area (namely the regulation of the internal market), leads me to the conclusion that it clearly seeks to produce legal effects and to induce compliance beyond mere policy suggestions.

143. For all these reasons, I consider that the Appellants’ third ground of appeal is well-founded. The General Court erred in law by incorrectly assessing the legal effects of the Recommendation in question, and consequently incorrectly declared the annulment action inadmissible.

B.      Form determines substance

144. In contrast to ‘atypical’ acts issued by EU institutions or bodies, for which the ERTA test was originally designed, recommendations are ‘typical’ acts, listed in Article 288 TFEU, the judicial review of which is expressly excluded under Article 263(1) TFEU. Should this fact play any role in the review of recommendations and its admissibility?

145. The argument in Part A of this Opinion was based on the premise that the ERTA test, albeit perhaps somewhat revised, is applicable to recommendations, as it is to any other form of soft law. The first part of this section outlines an alternative approach, which would put greater stress on the fact that recommendations are ‘typical’ acts, for which form should determine the interpretation of the substance (1). However, for a number of reasons, I would recommend to the Court to stay with the (modified) substance over form approach outlined in Part A of this Opinion (2). Should the Court nonetheless be of the view that in the specific case of recommendations, form should indeed determine the interpretation of the substance, then I briefly outline several important clarifications that would be called for (3).

1.      A (full) exclusion: recommendation means recommendation

146. The reasoning of the General Court started from the premise that the test that the Court coined in ERTA for ‘atypical’ acts of the EU institutions and bodies is also applicable to a ‘typical’ act, such as a recommendation. That starting point may be open to debate on two levels: normative and practical. Normatively, Article 288 TFEU clearly says that a recommendation is not binding. Then the first sentence of Article 263(1) TFEU explicitly excludes recommendations of the Commission from the scope of that provision and thereby from actions for annulment. Those two provisions read together clearly state that a recommendation cannot be binding and cannot be reviewed.

147. To that however add the statement of the Court in Grimaldi, (re)introducing the substantive assessment into the game: a recommendation is outside the review as long as it is ‘true recommendation’. (90) So by implication, there could be a ‘false recommendation’ which could be caught and reviewed. In spite of the rather clear wording of the Treaties, there is thus some support in the case-law of the Court to subject recommendations, even though they are ‘typical’ acts (the absence of their binding force being clearly stated in the Treaty) to the ERTA test. (91)

148. What then nonetheless happens on the practical level is that in the assessment of whether a document is a ‘true’ or a ‘false’ recommendation, the fact that the document is called a recommendation then inevitably ‘taints’ the assessment of its context and purpose. Yet again, a test which was designed for ‘atypical’ acts must be carried out in a way which to a great extent is ‘form-blind’, disregarding the title/cover of the document. Otherwise, one inevitably ends up with a certain kind of circular reasoning, in which form eventually determines the interpretation of the substance. (92)

149. This brings me to the possible alternative approach. It would entail, instead of the form of a ‘typical’ act being allowed to silently taint the interpretation of the nature of an ‘atypical’ act, bringing that distinction to its full logical conclusions: a ‘typical’ form implies ‘typical’ consequences, irrespective of the content. A recommendation will never have any binding force and should produce no legal effects. Period. There would be no need to carry out any further assessment of whether it is a ‘true’ or a ‘false’ recommendation. Recommendation means recommendation.

150. A parallel may in this respect be drawn with opinions under the ECSC Treaty: acts that were also expressly non-binding and non-reviewable. The Court confirmed, back in 1957, that those acts could not be reviewed. It made clear in particular that opinions only give guidance. They were seen as ‘merely advice given to undertakings. The latter thus remain free to pay regard to or ignore it but they must understand that in ignoring an adverse opinion, they accept the risks with which they are faced as the result of a situation which they themselves have helped to create (…). In other words, the freedom of decision and the responsibility of the undertakings remain, like those of the High Authority, unchanged’. (93) That very statement could actually be paraphrased in extenso in the case of recommendations.

2.      Substance or form?

151. There are two arguments that favour the formal approach to ‘typical’ legal acts: (i) the argument of legal certainty and foreseeability, and (ii) the argument of the need for certain legislative flexibility.

152. First, the formal approach remains faithful to the wording of Article 263 TFEU which since 1957 has always consistently and explicitly excluded recommendations and opinions from the scope of annulment actions.

153. Connected to that is reliance and expectations. Although almost always portrayed as an evil, formalism has also positive dimensions. It fosters legal certainty and foreseeability. What is called a spade should really be a spade, without the constant need for a substantive, contextual re-assessment of it.

154. That is even more valid with regard to ‘typical’, formalised sources. How far could the ‘substance over form’ way go? Should then other typical sources of EU law, such as a regulation and directive, also be assessed as to their genuine, actual content? Could they then also be potentially ‘reclassified’, because their title is out of sync with their content? In extreme cases, could such a reclassification lead even to the exclusion of judicial review? Could the Court declare an action for annulment against, for example a regulation, inadmissible because that regulation is so badly drafted that it is in fact not able to produce any (binding) legal effects?

155. Second, recommendations can be valid points of reference, inspiration and good practice. They may allow different solutions and ideas to be tested in order to determine which ones should be taken further and which one should be discarded. In this way, they could be seen as a sort of legislative laboratory. It is perhaps fair to acknowledge that if any recommendation could be attacked and called into question, such flexibility of a legislative laboratory would be lost. In particular the Commission could be effectively prevented from pursuing more informal forms of action to advance its agenda in the interest of the EU.

156. The answer to both of these arguments from the point of view of the ‘substance over form’ approach is not complicated: first, that approach is called for exactly when the formal confines of a ‘typical’ act are not respected. Thus, far from calling all typical acts into question, that approach is by its nature reserved to extreme scenarios. Second, the attractions of a flexible legislative laboratory should find clear limits in the principle of legality of public power and the principle of attributed competence. Even well-meant legislative laboratories can quickly turn into ‘legislation by stealth’.

157. By contrast, there are at least three arguments that militate in favour of the substantive approach, outlined in Part A: (i) substance over form being all around; (ii) the need to ensure effective judicial protection; and (iii) the overall coherence of legal remedies in EU law.

158. First, even if there are indeed some notable exceptions, the overall approach and mentality in EU law is simply a substantive one: in countless areas of EU law, what is being examined is the substance, the essence of a phenomenon, category or institution - not really its formal denomination or label. Form certainly has significance. However, in EU law, form constitutes a first approximation to the true nature of an act. It is not decisive.

159. Second, as already explained in detail above, (94) it is a hardly debatable fact that recommendations do produce a number of significant legal effects even if they may not strictly speaking be binding in the individual case. If that is indeed the case, effective legal protection must follow. It may be recalled that in the past, both originally in ERTA but then also in subsequent cases, the Court did not hesitate to acknowledge social and legal evolution and to close gaps thus generated in legal protection. (95) It has also been suggested that the advent of new and ‘softer’ modes of governance represents such a type of evolution. (96)

160. Third, the argument concerning the coherence of legal remedies in EU law is relevant at two levels: on the one hand the coherence between Article 263 TFEU (action for annulment) and Article 267 TFEU (the preliminary rulings procedure) and, on the other, the position of (non-)privileged applicants in those procedures.

161. It is established case-law that ‘the review of the legality of acts of the Union that the Court is to ensure under the Treaties relies … on two complementary judicial procedures. The FEU Treaty has established, by Articles 263 and 277, on the one hand, and Article 267, on the other, a complete system of legal remedies and procedures designed to ensure judicial review of the legality of European Union acts, and has entrusted such review to the Courts of the European Union… It is inherent in that complete system of legal remedies and procedures that persons bringing proceedings must, when an action is brought before a national court or tribunal, have the right to challenge the legality of provisions contained in European Union acts on which a decision or national measure adopted in respect of them is based…’. (97)

162. I wish to highlight two adjectives contained in that quotation: ‘complementary’ and ‘complete’. Somewhat paradoxically perhaps, that second adjective might be problematic for a privileged applicant, that is a Member State which is ‘invited’ to do something by a recommendation, and takes issues with that ‘invitation’, without there being, however, any case (yet) relating to that recommendation on the national level.

163. On the one hand, as already outlined, (98) a request for a preliminary ruling concerning the validity of a recommendation appears to be possible. If that is the case, it is difficult to understand why it should be more difficult for a privileged applicant to challenge an act directly before the Court on the basis of Article 263 TFEU than for a non-privileged applicant indirectly on the basis of Article 267 TFEU.

164. On the other hand, even if the overall logic of complementarity were to prevail, discouraging direct challenges to validity and seeking to channel them through the preliminary ruling procedure, the position of a Member State is not made much easier. Practically, how should a Member State go about it? Should it first implement the recommendation (which it did not want to in the first place) and then challenge it before its own national courts? Should it contrive a dispute?

165. Without wishing to re-open any of these issues,(99) suffice to underline that Member States are simply not individuals that, perhaps later on, might be obliged to comply with an EU measure after it has been applied to them by an individual decision, either on the national or EU law. It is the Member States that are invited to implement those rules in the first place. It would therefore be simply illogical to induce Member States to do something and, at the same time, to withdraw their possibility to bring an action before the Court. That would go against the interests of the good administration of justice as it would delay a possible action against a recommendation, but also against the interests of the drafter of the recommendation itself. Instead of being allowed to channel the potential conflict, solve it, and to move on, the Member State would simply be forced to refuse to cooperate and to wait until one of its own courts and possibly a court in another Member State makes a reference under Article 267 TFEU concerning the validity of the contested act. That makes very little practical sense to me.

3.      The (potentially) necessary clarifications

166. In sum, I see many more cogent reasons for inviting the Court to extend the modified ERTA test to include the potential judicial review of recommendations.

167. However, were the Court to decide to follow the route in which the form of a ‘typical’ act determines the perception and interpretation of its substance, without a separate examination of it being necessary, it would appear essential to clarify several points. Those necessary clarifications would effectively relate to the elements previously identified under actual legal effects of recommendations. (100) I would in particular underline three key elements: (i) the scope of the duty of loyal and sincere cooperation incumbent upon the Member States in relation to recommendations; (ii) the absence of pre-emptive effect of recommendations on the potential future legislative process on the level of the EU; and (iii) clarification of the scope of the Grimaldi obligation incumbent on national courts.

168. First, recommendations are neither binding, nor are they allowed to produce any legal effects. They accordingly cannot create any rights or obligations, for the Member States or for individuals. As far as the Member States are concerned, the principle of loyal and sincere cooperation cannot be used to start eroding that proposition, in whatever way. The Member States are fully entitled to entirely disregard the content of a recommendation without there being any possibility of direct or indirect sanctions. That holds not only for the concrete ‘obligations’ that Member States are encouraged to implement, but also for any reporting ‘invitations’. There can be neither any positive, nor negative obligations flowing from a recommendation. A recommendation also cannot be used to define a standard or an indeterminate legal notion that will then, after having been given content by that recommendation, be enforced against a Member State or an individual.

169. Second, a recommendation, certainly a pre-legislative one, is simply a unilateral, non-binding statement of the opinion of an institution. If it is ever to be followed up by any binding legislation, the legislative process must start with a clean slate. In particular, a recommendation cannot create a legislative ‘shortcut’ or a ‘pre-emption’ by excluding certain actors from the later legislative process, or effectively penalising some actors in the subsequent process because they had not already submitted their views, observations, data or reports on the recommendation and/or in the process of its ‘implementation’.

170. Last of all, there is Grimaldi. (101) If recommendations are non-binding, then they cannot by definition generate any obligation on the part of the national courts to take such non-binding guidance into consideration, a fortiori not to speak of any duty of conform interpretation. If the Court were to follow this line of a more formal approach to a ‘typical’ EU law act, it would be necessary to revisit Grimaldi in this regard and clearly state that there is no obligation to take a recommendation into consideration. The national courts may do so, if they consider it useful, but they are certainly not obliged to.

171. This would then mean that national courts should treat recommendations as any other permissible source of inspiration in the process of legal interpretation, such an academic commentary or a comparative argument. They may include it in their reasoning if they wish to, but they can also completely disregard it, with no duty to justify why.

VII. Conclusion

172. In the light of the foregoing, my conclusion is that the General Court erred in its assessment of the legal effects of the contested Recommendation. The Appellant’s third ground of appeal is therefore well-founded, without the need for separate examination of the first and second grounds of appeal. Accordingly, to the extent that it declared the action inadmissible, the General Court’s order should be set aside.

173. In accordance with the first paragraph of Article 61 of the Statute of the Court of Justice of the European Union, the Court may also, after quashing the decision of the General Court, itself give judgment in the matter, where the state of proceedings so permits.

174. In the present case, the Court is not in a position, at this stage of the proceedings, to give judgment on the substance of the action brought before the General Court. As the General Court declared the action inadmissible, only a very limited and rather indirect discussion concerning the merits of the case took place before that jurisdiction. In addition, for the same reasons, no other interveners were allowed to present their observations. (102) If the Court were to declare the action for annulment admissible, it is likely that those interveners, and potentially also others, would be interested in submitting their observations.

175. I consider, however, that the Court has all the necessary material before it to be able to give a ruling rejecting the preliminary plea of inadmissibility raised by the Commission at first instance. In the interests of efficiency and economy of procedure, I propose that the Court take that route, declaring the action admissible and referring the case back to the General Court for a decision on merits.

176. I therefore propose that the Court:

–        set aside the order of the General Court of the European Union in Case T–721/14 and find admissible the Appellant’s action for annulment in that case;

–        refer the case back to the General Court for a decision on the merits;

–        order that costs be reserved.


1      Original language: English.


2      Hart, H. L. A., The Concept of Law, 2nd ed (with a postscript), Clarendon Press, Oxford, 1997 (first edition published 1961).


3      Dworkin, R., Taking Rights Seriously (New Impression with a Reply to Critics), London, Duckworth, 1987 (first edition published 1977), p. 22 and ff.


4      Commission Recommendation 2014/478/EU of 14 July 2014 (OJ 2014 L 214, p. 38). Emphasis added.


5      Order of 27 October 2015, Belgium v Commission (T‑721/14, EU:T:2015:829).


6      COM (2011) 128 final.


7      COM (2012) 596 final.


8      P7_TA(2013)0348.


9      2012/2322(INI).


10      IP/14/828 of 14 July 2014, available online at: http://europa.eu/rapid/press-release_IP-14-828_en.htm.


11      MEMO/14/484 of 14 July 2014, online at: http://europa.eu/rapid/press-release_MEMO-14-484_en.htm.


12      Order of 27 October 2015, Belgium v Commission (T‑721/14, EU:T:2015:829).


13      Paragraph 37 of the order under appeal.


14      Paragraph 21 et seq. of the order under appeal.


15      Paragraph 29 of the order under appeal.


16      Paragraphs 32 to 35 of the order under appeal.


17      Paragraph 36 of the order under appeal.


18      Paragraphs 38 to 40 of the order under appeal.


19      Paragraphs 42 to 48 of the order under appeal.


20      Paragraphs 51 to 52 of the order under appeal.


21      Paragraphs 54 to 55 of the order under appeal.


22      Paragraph 64 of the order under appeal.


23      Paragraph 68 of the order under appeal.


24      The Portuguese Republic and the Hellenic Republic applied to intervene in support of the Appellant before the General Court. However, as it was rejecting the action as inadmissible, the General Court stated that there was no need to rule on those applications to intervene (paragraph 86 of the order under appeal).


25      The Appellant relies on Opinion 2/00 to stress the constitutional importance of the choice of the appropriate legal basis (Opinion 2/00 (Cartagena Protocol on Biosafety) of 6 December 2001, EU:C:2001:664, paragraph 5). See also judgment of 1 October 2009, Commission v Council (C‑370/07, EU:C:2009:590, paragraph 47).


26      Judgments of 12 February 2009, Commission v Greece(C‑45/07, EU:C:2009:81), and of 20 April 2010, Commission v Sweden (C‑246/07, EU:C:2010:203).


27      Judgment of 16 October 2003, Ireland v Commission (C‑339/00, EU:C:2003:545, paragraph 71).


28      Judgment of 31 March 1971, Commission v Council (22/70, EU:C:1971:32).


29      To my mind, the interpretation of legal effects (or the absence thereof) of a (potential) source of EU law, such as a recommendation, is a pure point of law question, thus subject to full review on the appellate level. Iura (item ‘ius mollis’) novit Curia.


30      Judgment of 31 March 1971, Commission v Council (22/70, EU:C:1971:32).


31      Judgment of 31 March 1971, Commission v Council (22/70, EU:C:1971:32).


32      Paragraphs 39 and 42 of that judgment (emphasis added). That wording referring to ‘acts producing legal effects’, was then incorporated into Article 173 EEC by the Treaty of Maastricht.


33      See judgments of 9 October 1990, France v Commission (C‑366/88, EU:C:1990:348), and of 6 April 2000, Spain v Commission (C‑443/97, EU:C:2000:190).


34      See judgment of 13 November 1991, France v Commission (C‑303/90, EU:C:1991:424).


35      See judgments of 16 June 1993, France v Commission(C‑325/91, EU:C:1993:245) and of 20 March 1997, France v Commission (C‑57/95, EU:C:1997:164).


36      See judgment of 1 December 2005, Italy v Commission (C‑301/03, EU:C:2005:727), together with the elucidating Opinion of Advocate General Jacobs (C-301/03, EU:C:2005:550, points 70 seq.).


37      See judgment of 5 October 1999, Netherlands v Commission (C‑308/95, EU:C:1999:477).


38      See already, regarding the acts of the High Authority, judgment of 10 December 1957, Société des usines à tubes de la Sarre v High Authority (1/57 and 14/57, EU:C:1957:13).


39      Paragraphs 16 to 18 of the order under appeal (emphasis added).


40      Above, footnote 31.


41      In French, ‘qui visent à produire des effets de droit’; in German, „Rechtswirkungen zu erzeugen” ; in Italian, ‘che miri a produrre effetti giuridici’; in Dutch, ‘die beogen rechtsgevolgen teweeg te brengen’.


42      See for instance order of 17 May 1989, Italy v Commission (151/88, EU:C:1989:201, paragraph 21). See also judgment of 5 October 1999, Netherlands v Commission (C-308/95, EU:C:1999:477, paragraph 30). However, referring to mere ‘legal effects’, see for instance judgment of 1 December 2005, Italy v Commission (C-301/03, EU:C:2005:727, paragraphs 22 to 24).


43      See judgments of 13 October 2011, Deutsche Post and Germany v Commission (C‑463/10 P and C‑475/10 P, EU:C:2011:656, paragraph 36), and judgment of 13 February 2014, Hungary v Commission (C‑31/13 P, EU:C:2014:70, paragraph 54).


44      See judgment of 31 March 1971, Commission v Council (22/70, EU:C:1971:32, paragraph 39), in the light of Opinion of Advocate General Dutheillet de Lamothe in Commission v Council (22/70, not published, EU:C:1971:23, p. 287) stating that ‘Article 173 and Article 189 form a coherent whole’. See also judgment of 23 April 1986, Les Verts v Parliament (294/83, EU:C:1986:166, paragraph 24).


45      In this way much more Kelsenian than the already quoted Hartian tradition, since the latter is much more ‘sociological’. Hart accepted that a legal rule may be binding not only because it was enacted in conformity with some secondary rule that states that the rule shall be binding, but also because a group of people accepts that norm as standard of their behaviour, generating internal social pressure within the group to comply – see Hart, H. L. A., The Concept of Law, 2nd ed (with a postscript), Clarendon Press, Oxford, 1997, Chapters V and VI.


46      See for instance judgments of 7 February 1979, France v Commission(15/76 and 16/76, EU:C:1979:29, paragraph 7), and of 29 April 2004, Germany v Commission (C‑277/00, EU:C:2004:238, paragraph 39).


47      Judging from the amount of scholarly literature and attention dedicated to the topic.


48      Generally on soft law in the EU context, see for instance Wellens, K.C. and Borchardt, G.M., ‘Soft Law in European Community law’, European Law Review 14, 1989, p. 267; Klabbers, J., ‘Informal Instruments before the European Court of Justice’, Common Market Law Review 31, 1994, p. 997; Senden, L., Soft Law in European Community Law, Hart Publishing, Oxford and Portland Oregon, 2004; Schwarze, J., ‘Soft Law im Recht der Europäischen Union’, Europarecht, 2011, p. 3; Scott, J., ‘In Legal Limbo: Post-Legislative Guidance as a Challenge for European Administrative Law’, Common Market Law Review 48, 2011, p. 329; Knauff, M., ‘Europäisches Soft Law als Gegenstand des Vorabentscheidungsverfahrens’, Europarecht, 2011, p. 735; Stefan, O., Soft Law in Court. Competition Law, State Aid and the Court of Justice of the European Union, Kluwer, Alphen aan den Rijn, 2013; Bertrand, B., ‘Les enjeux de la soft law dans l’Union européenne’, Revue de l’Union européenne, 2014, p. 73.


49      See notably European Parliament Study, ‘Checks and Balances of soft EU rule-making’, Directorate General for Internal Policies, Policy Department C, Citizens’ rights and constitutional affairs, 2012 (PE 462.433), in particular pp. 54 to 58.


50      For instance, the German Bundesverwaltungsgericht, judgment of 15 November 2010 – 19 BV 10.871 (concerning administrative circulars setting emission rates in environmental matters).


51      See Swedish Supreme Administrative Court, 24 May 1996, case 2904-1994 (I) (RA 1996 ref 43).


52      See for instance in relation to administrative circulars the Belgian Conseil d’Etat, judgment of 237/674 of 16 March 2017; see also, about views (prises de position) or reports adopted by independent administrative authorities empowered to adopt sanctions, the French Conseil d’Etat, judgment of 17 November 2010, Syndicat français des ostéopathes, n° 332 771; and of 11 October 2012, Société Casino Guichard-Perrachon, n°357193.


53      For instance in Sweden public and private bodies may challenge the legality of an administrative act when they have perceived it as binding and have acted in accordance with it (Supreme Administrative Court, 10 February 2004, case 2696-03 (RA 2004 ref 8), about an ’information’ likely to have real effects on the personal and economic situation of the addressee).


54      See for instance, about a legal action against an administrative report in the context of planning procedures, High Court, De Burca v Wicklow County Manager (2009) IEHE 54; also about guidelines of the Irish Competition Authority, see High Court, Law Society of Ireland v Competition Authority (2006) 2 IR 262.


55      In its annual report of 2013, it defined soft law as the set of instruments fulfilling the three following criteria: 1) they must have as an object to modify or guide the behaviour of their addressees by leading to their adherence; 2) they do not create in themselves rights or obligations for their addresses; 3) they represent, by their content and the way that they are structured, a degree of formalisation and structure which brings them closer to looking like rules of law (Conseil d’Etat, Etude annuelle 2013 - Le droit souple, La Documentation française, 2013, pp. 61 to 63).


56      Conseil d’Etat, judgments of 21 March 2016, Numericable, n° 390023 and of 21 March 2016, Société Fairvesta International GmBH, n° 368082, respectively about a prise de position of the French Competition Authority and press releases of the French Financial Market Authority.


57      For early, by now classic accounts see e.g. Morand, C., ‘Les recommandations, les résolutions et les avis du droit communautaire’, Cahiers de droit européen, 1970, p. 523; Soldatos, P., Vandersanden, G., ‘La recommendation, source indirecte du rapprochement des legislations nationales dans le cadre de la Communauté économique européenne’, in De Ripainsel-Landy, D. et al., Les instruments de rapprochement des législations dans la Communauté économique européenne, Editions de l’Université de Bruxelles, Bruxelles, 1976, p. 94.


58      See judgments of 29 September 2011, Arkemav Commission (C‑520/09 P, EU:C:2011:619, paragraph 88), and of 28 June 2005, Dansk Rørindustri and Others v Commission (C‑189/02 P, C‑202/02 P, C‑205/02 P to C‑208/02 P and C‑213/02 P, EU:C:2005:408, paragraphs 209 to 211). But see, for a more nuanced approach with regard to notices in competition law, judgment of 13 December 2012, Expedia (C‑226/11, EU:C:2012:795, paragraph 29).


59      Judgment of 13 December 1989, Grimaldi (C‑322/88, EU:C:1989:646, paragraph 13).


60      The Better Lawmaking initiative has put emphasis on this problematic dimension of soft law. See notably Framework Agreement on relations between the European Parliament and the European Commission (OJ 2010, L 304, p. 47). Paragraph 43 states in particular: ‘in areas where Parliament is usually involved in the legislative process, the Commission shall use soft law, where appropriate and on a duly justified basis after having given Parliament the opportunity to express its views. The Commission shall provide a detailed explanation to Parliament on how its views have been taken into account when it adopts its proposal’.


61      It might be recalled that it was the conferral of powers and institutional balance that also led the Court to its judgment of 23 April 1986, Les Verts v Parliament (294/83, EU:C:1986:166, paragraph 25).


62      That is incidentally the argument raised by the Appellant in its second ground of appeal. From a certain point of view, it is true that subjecting the admissibility of the annulment action to the existence of binding legal effects (as opposed to mere legal effects) prevents ensuring that the drafter of the contested act has acted within the realm of its competence. It then opens the issue of whether or not soft law instruments, including a recommendation, can only be adopted within the sphere of competence attributed to the Union and to the institution in question. But, not without a certain whiff of Catch-22, since such binding legal effects do not exist, their review via an action of annulment is not admissible.


63      See judgment of 13 December 1989, Grimaldi (C‑322/88, EU:C:1989:646, paragraph 16). See also judgments of 21 January 1993, Deutsche Shell (C‑188/91, EU:C:1993:24, paragraph 18); of 11 September 2003, Altair Chimica (C‑207/01, EU:C:2003:451, paragraph 41) and of 18 March 2010, Alassini and Others(C‑317/08 to C‑320/08, EU:C:2010:146, paragraph 40).


64      Judgment of 13 December 1989, Grimaldi (C‑322/88, EU:C:1989:646, paragraph 18).


65      The Court has notably declined to recognise that legal effect for Commission notices (judgment of 13 December 2012, Expedia (C‑226/11, EU:C:2012:795, paragraph 31).


66      Judgment of 10 April 1984, von Colson and Kamann(14/83, EU:C:1984:153).


67      As successively developed and consolidated in, for example, judgments of 5 October 2004, Pfeiffer and Others (C‑397/01 to C‑403/01, EU:C:2004:584, paragraphs 114 to 115); of 4 July 2006, Adeneler and Others (C‑212/04, EU:C:2006:443, paragraphs 108 to 109); and of 15 April 2008, Impact (C‑268/06, EU:C:2008:223, paragraphs 99 to 101).


68      See Opinion of Advocate General Kokott in Expedia(C‑226/11, EU:C:2012:544, points 38 to 39), which considered that national authorities and courts must take due account of the Commission’s competition policy notices.  That notably entailed in that particular case that those authorities and courts must consider the Commission’s assessment, as set out in those notices, of what constitutes an appreciable restriction of competition and must give reasons which can be judicially reviewed for any divergences.


69      See judgments of 28 June 2005, Dansk Rørindustri and Others v Commission(C‑189/02 P, C‑202/02 P, C‑205/02 P to C‑208/02 P and C‑213/02 P, EU:C:2005:408, paragraphs 209 to 211), and of 29 September 2011, Arkemav Commission (C‑520/09 P, EU:C:2011:619, paragraph 88).


70      See judgment of 13 December 2012, Expedia (C‑226/11, EU:C:2012:795, paragraph 26).


71      As opposed to permissible or persuasive sources - see for example Peczenik, A. On Law and Reason, Kluwer, Dordrecht, 1989, pp. 319 ff.


72      See, by analogy, how loyal cooperation has been used to create obligations in connection with direct effect, Member State liability for breaches of EU law or the adoption of sanctions (respectively, by way of illustration, judgments of 16 December 1976, Comet (45/76, EU:C:1976:191, paragraph 12); of 2 February 1977, Amsterdam Bulb (50/76, EU:C:1977:13, paragraph 32); and of 19 November 1991, Francovich and Others (C‑6/90 and C‑9/90, EU:C:1991:428, paragraph 36).


73      As already suggested above (footnote 45), such narrow view does not even prevail in a number of streams of positivist legal theory. Besides, the same understanding is quite remote from reality of how ‘non-binding wishes’ expressed by one and the same regulator who can also, the next day, issue binding legislation and sanctions in the same or connected area, will be understood and perceived by their addressees. Thus, apart from direct sanctions, there can be also indirect sanctions, certainly in cases of repeat players on both sides (one and the same regulator and the same group of addressees). It might be recalled that similar logic led to some higher national jurisdictions to subject the ‘non-binding’ acts of such type of regulators to judicial review (above, footnote 52).


74      Starting with the judgment of 18 December 1997, Inter-EnvironnementWallonie (C‑129/96, EU:C:1997:628).


75      ‘It is settled case-law that the fact that a measure of Community law has no binding effect does not preclude the Court from ruling on its interpretation in proceedings for a preliminary ruling under Article 177’ (judgments of 13 December 1989, Grimaldi (C‑322/88, EU:C:1989:646, paragraph 9), and of 21 January 1993, Deutsche Shell (C‑188/91, EU:C:1993:24, paragraph 18 and the case-law cited)).


76      ‘…unlike Article 173 of the EEC Treaty, which excludes review by the Court of acts in the nature of recommendations, Article 177 confers on the Court jurisdiction to give a preliminary rulingon the validity and interpretation of all acts of the institutions of the Community without exception’ - judgment of 13 December 1989, Grimaldi (C‑322/88, EU:C:1989:646, paragraph 8) – emphasis added.


77      Judgment of 16 June 2015, Gauweiler and Others (C‑62/14, EU:C:2015:400). The Court did not examine the admissibility of the action against a press release as such. But see Opinion of Advocate General Cruz Villalón in Gauweiler and Others (C‑62/14, EU:C:2015:7, point 70 et seq.).


78      See for instance judgments of 23 April 1986, Les Verts v Parliament (294/83, EU:C:1986:166, paragraph 23), and of 28 March 2017, Rosneft (C‑72/15, EU:C:2017:236, paragraph 66).


79      In particular in cases in which the national challenge to the national transposition merely mirrors the same potential issue of the recommendation itself – see in this regard, by analogy, judgment of 22 June 2010, Melki and Abdeli(C‑188/10 and C‑189/10, EU:C:2010:363, paragraphs 54 to 55).


80      Judgment of 12 September 2006, Reynolds Tobacco and Others v Commission (C‑131/03 P, EU:C:2006:541, paragraph 55 and the case-law cited).


81      See also above, footnote 73.


82      Recitals 8 and 9 quoted above in points 11 and 12 of this Opinion.


83      Above, point 30 of this Opinion.


84      See for instance judgment of 22 January 2015, Stanley International Betting and Stanleybet Malta (C‑463/13, EU:C:2015:25, paragraph 51).


85      Judgment of 13 November 1991, France v Commission (C‑303/90, EU:C:1991:424, paragraphs 20 to 25), where a code of conduct implementing a Council regulation requested information from Member States with a certain frequency and through certain means; judgment of 16 June 1993, France v Commission (C‑325/91, EU:C:1993:245, paragraphs 22 to 23), where a Commission communication giving flesh to a directive imposed extra obligations, such as annual reporting of financial data to the Commission at a certain date.


86      See also above, point 102 to 105. See also points 97 to 98 and the potential of the Recommendation to shape the interpretation of existing national rules adopted in the same area and dealing with the same subject-matter.


87      Above, point 114.


88      See, for example, judgment of 26 April 2012, DR and TV2 Danmark(C‑510/10, EU:C:2012:244, paragraph 45 and the case-law cited).


89      Including, for that matter, also the extreme situations in which it appears rather clearly that there is a mistake in translation in just one language version of the EU measure – see e.g. judgment of 19 April 2007, Profisa (C‑63/06, EU:C:2007:233).


90      Judgment of 13 December 1989, Grimaldi (C‑322/88, EU:C:1989:646, paragraph 16).


91      Which then led the General Court to state that ‘The mere fact that the contested recommendation is formally designated as a recommendation and was adopted on the basis of Article 292 TFEU cannot automatically rule out its classification as a challengeable act’ (paragraph 20 of the order under appeal).


92      Outlined in detail above, points 77 to 79.


93      Judgment of 10 December 1957, Société des usines à tubes de la Sarre v High Authority (1/57 and 14/57, EU:C:1957:13) Rec. p. 115.


94      Above, points 87 to 108.


95      For a notable example, see e.g. judgment of 23 April 1986, Les Vertsv Parliament(294/83, EU:C:1986:166, paragraph 24).


96      Above, points 81 to 86.


97      Judgment of 28 March 2017, Rosneft (C‑72/15, EU:C:2017:236, paragraphs 66 to 67 and the case-law cited). See also Opinion 1/09 (Agreement creating a Unified Patent Litigation System) of 8 March 2011 (EU:C:2011:123, paragraph 70).


98      Above, points 106 to 108.


99      Settled by the judgment of 1 April 2004, Commission v Jégo-Quéré (C‑263/02 P, EU:C:2004:210).


100      Above, points 87 to 108.


101      Above, points 97 to 101.


102      Above, footnote 24.

© European Union
The source of this judgment is the Europa web site. The information on this site is subject to a information found here: Important legal notice. This electronic version is not authentic and is subject to amendment.


BAILII: Copyright Policy | Disclaimers | Privacy Policy | Feedback | Donate to BAILII
URL: http://www.bailii.org/eu/cases/EUECJ/2018/C1616P_O.html