A. Admissibility of the appeal
40.
Belgium, Magnetrol International, Soudal, Esko-Graphics and Wabco Europe consider the Commission’s appeal to be inadmissible in its entirety. First, they find fault with the forms of order sought by the Commission. Second, they consider that the Commission seeks a reassessment of the facts of the case. Third, they assert that the Commission has not explained in how far the General Court erred in law.
1.
The forms of order sought by the Commission
41.
Magnetrol International, Soudal, Esko-Graphics and Wabco Europe take the view that the forms of order sought by the Commission are inadmissible. The Commission seeks only that part of the judgment which burdened it to be set aside. However, the contested part of the operative part, by which the General Court annulled the contested decision, is indivisible.
42.
Under Article 169(1) of the Rules of Procedure of the Court of Justice, an appeal must seek to have set aside, in whole or in part, the decision of the General Court as set out in the operative part of that decision.
43.
The Commission claims that the Court should ‘set aside [the judgment under appeal] in so far as it holds that [the contested decision] erroneously classified the “excess profit” system as a scheme within the meaning of Article 1(d) of Regulation 2015/1589’. In that regard, it is unambiguously clear from the form of order sought by the Commission in its appeal that it seeks to have set aside the operative part by which the General Court annulled the contested decision. This corresponds to the setting aside in full of the General Court’s decision by which it upheld Belgium and Magnetrol International’s actions in their entirety. This is because the General Court did not dismiss those actions in part.
44.
Since it is clear from this that the appeal seeks to have the judgment under appeal set aside,(14) the forms of order sought by the Commission are admissible in this respect.
2.
The allegation that the Commission seeks a reassessment of the facts established by the General Court
45.
Belgium, Soudal and Esko-Graphics also contend that the Commission seeks a reassessment of the facts by the Court of Justice. They submit that, by its appeal, the Commission is ultimately seeking to substitute the statement of reasons of the contested decision.
46.
It should be recalled, as follows from the second subparagraph of Article 256(1) TFEU and from the first paragraph of Article 58 of the Statute of the Court of Justice of the European Union, that an appeal lies on points of law only. The General Court thus has exclusive jurisdiction to find and appraise the relevant facts and to assess the evidence. The appraisal of those facts and the assessment of that evidence thus do not, save where the facts and evidence are distorted, constitute a point of law which is subject, as such, to review by the Court of Justice on appeal.(15)
47.
However, the Commission does not confine itself to criticising the General Court’s finding and appraisal of the relevant facts. On the contrary, it submits on several occasions that the General Court distorted the facts. Aside from that, the Commission does not confine itself to complaints at the factual level but, as I will explain below,(16) also alleges errors of law.
3.
The adequate reasoning in the appeal
48.
Finally, Belgium, Soudal and Esko-Graphics complain that the Commission failed to provide reasons explaining how the General Court misinterpreted Article 1(d) of Regulation 2015/1589.
49.
According to settled case-law, it follows from, inter alia, Article 168(1)(d) and Article 169(2) of the Rules of Procedure of the Court of Justice that an appeal must indicate precisely the contested elements of the decision which the appellant seeks to have set aside and the legal arguments specifically advanced in support of the appeal. An appeal which merely repeats or reproduces verbatim the pleas in law and arguments previously submitted to the General Court does not satisfy the requirements relating to the duty to state reasons under those provisions.(17)
50.
However, the Commission criticises certain parts of the judgment under appeal and refers to specific paragraphs in that regard.(18) Moreover, it explains how, in its view, the General Court erred in law in its application of the individual conditions of Article 1(d) of Regulation 2015/1589.
51.
The appeal therefore has adequate reasoning.
4.
Conclusion regarding the admissibility of the appeal
52.
In conclusion, the appeal is admissible.
B. Merits of the appeal
53.
The Commission’s appeal consists of one ground of appeal, alleging that the General Court erred in law in its interpretation of the concept of an aid scheme within the meaning of Article 1(d) of Regulation 2015/1589 and, moreover, distorted the contested decision by not setting out the conditions for the existence of an aid scheme.
54.
The General Court takes the view that the Commission erroneously considered that the systematic practice of the Belgian authorities of adjusting profits, which goes beyond the scope of Article 185(2)(b) of the CIR 92, constituted an aid scheme (paragraph 135 of the judgment under appeal). The Commission did not review all the advance tax rulings issued, but only a sample of them. The General Court takes the view that the Commission therefore did not prove that the Belgian tax authorities had followed a systematic approach in all the advance tax rulings (paragraph 134 of the judgment under appeal).
55.
Pursuant to Article 1(d) of Regulation 2015/1589, an aid scheme is any act on the basis of which, without further implementing measures being required, individual aid awards may be made to undertakings defined within the act in a general and abstract manner.
56.
According to Article 1(e) of that regulation, ‘individual aid’ means aid that is not awarded on the basis of an aid scheme and notifiable awards of aid on the basis of an aid scheme.
57.
Accordingly, an aid scheme within the meaning of Article 1(d) of Regulation 2015/1589 has three conditions: First, it must be an act. Second, the individual aid must be granted without further implementing measures. Third, the undertakings to which the aid is granted must be defined in the act in a general and abstract manner. These conditions must be met cumulatively.
58.
The General Court concluded that none of those conditions had been met. The Commission takes the view that the General Court therefore misinterpreted all three conditions of Article 1(d) (first to third parts of the ground of appeal). It is only if this is true of all three conditions that the Commission’s appeal can ultimately be successful.
59.
On the other hand, the fourth and final part of the ground of appeal, according to which the General Court also disregarded the ratio legis of Article 1(d) of Regulation 2015/1589, is not in actual fact an independent part of the ground of appeal. Rather, the ratio legis of a provision must be taken into consideration in the interpretation of each of its conditions. This part will therefore be examined together with the other three parts.
1.
The first condition of an ‘act’ (first part of the ground of appeal)
60.
By the first part of its ground of appeal, the Commission alleges that the General Court misinterpreted, in paragraph 78 et seq. of the judgment under appeal, the first condition of Article 1(d) of Regulation 2015/1589, namely the concept of an act. Furthermore, it distorted recitals 94 to 110 of the contested decision by considering, in paragraph 94 of the judgment under appeal, that only the acts listed in recital 99 constituted the basis of the scheme in question.
61.
In recital 99 of the contested decision, the Commission stated that Article 185(2)(b) of the CIR 92, the Memorandum to the Law of 21 June 2004, the Circular of 4 July 2006 and the replies by the Minister of Finance to parliamentary questions on the application of Article 185(2)(b) of the CIR 92 constitute the acts on the basis of which the excess profit exemption is awarded.
62.
The first part of the ground of appeal is in turn broken down into two arguments. According to the Commission, first, contrary to the view of the General Court, the concept of ‘act’ within the meaning of Article 1(d) of Regulation 2015/1589 can also encompass a consistent administrative practice (see Section (a) below). Second, the General Court failed to have regard to the fact that the Commission had also adequately demonstrated that consistent administrative practice (see Section (b) below).
(a)
The concept of ‘act’
63.
The Commission takes the view that the General Court interpreted the concept of ‘act’ too narrowly. Contrary to the view taken by the General Court, the concept of ‘act’ could also encompass a consistent administrative practice.
64.
The concept of ‘act’ within the meaning of Article 1(d) of Regulation 2015/1589 is to be given a broad interpretation.(19) While only a few language versions indicate that a legal act may be necessary,(20) it appears to be sufficient, according to most language versions,(21) that a legal consequence is established(22) – whether by law or otherwise.
65.
The practical effectiveness of State aid control also militates in favour of a broad interpretation. While point (e) concerns individual cases, point (d) covers a large number of similar cases. The effectiveness of the Commission’s work would be jeopardised if Member States were able to prevent an abstract aid scheme from being reviewed by moving it from the statutory to the administrative level. The Commission would then have to deal with all decisions individually, even if they are similar.
66.
By contrast, the Commission’s argument appears to be based on an incomplete reading of the judgment under appeal. In a first step, the General Court examined, in paragraph 80 et seq. of its judgment, whether the legal acts identified in recital 99 of the contested decision constituted the basis for the downward adjustments granted. The General Court found this not to be the case in paragraph 96 of the judgment, stating that they did not contain the constituent elements of the scheme complained of by the Commission. In a second step, the General Court then examined, in paragraph 121 et seq., whether the Commission’s argument alleging the existence of a consistent administrative practice calls that conclusion into question. The General Court refers to such a consistent administrative practice as a ‘systematic approach’.
67.
It is true that, in paragraphs 79 and 122 of the judgment under appeal, the General Court rightly refers to the judgment in Germany and Pleuger Worthington v Commission. According to that judgment, in cases where a legal act establishing an aid scheme has not been identified, the Commission may rely on a set of circumstances which taken as a whole indicate the existence of an aid scheme.(23)Contrary to the views taken by some of the parties involved, however, the Pleuger judgment(24) does not demonstrate that an administrative practice can constitute an aid scheme only if a legal act has not been identified. It is true that in that case the Court of Justice only had to rule on a situation in which it was not possible to identify a legal act on which the aid scheme was based. However, in doing so, the Court of Justice did not comment on cases in which a legal act is uniformly applied by the administration in a particular way and in a way that goes beyond its wording. Accordingly, in paragraph 123 of the judgment under appeal, the General Court does not in fact rule out the possibility that the Commission may conclude that there is an aid scheme where the characteristics of a systematic approach meet the requirements set out in Article 1(d) of Regulation 2015/1589. Rather, the General Court merely stated, in paragraph 124, that the Commission had not demonstrated a consistent administrative practice by virtue of such a systematic approach.
68.
This argument raised by the Commission in the first part of its ground of appeal therefore comes to nothing.
(b)
Proof of the consistent administrative practice
69.
The decisive question is therefore whether the General Court erred in law – as the Commission complains – by imposing excessive requirements on proving the existence of a consistent administrative practice. In that regard, the General Court held, in paragraph 134 of its judgment, that the Commission had not proved in the contested decision that a systematic approach on the part of the Belgian tax authorities existed and that it was followed in all of the advance rulings.
70.
As stated above,(25) the assessment of the facts by the General Court does not in principle constitute a question of law which is subject, as such, to review by the Court of Justice. However, when the General Court has established or assessed the facts, the Court of Justice has jurisdiction, under Article 256 TFEU, to review the legal characterisation of those facts by the General Court and the legal conclusions which it has drawn from them.(26)
71.
Specifically, the General Court takes the view that the Commission had not demonstrated to the requisite legal standard the existence of a systematic approach (paragraph 126 of the judgment under appeal). The Commission did not explain either the choice of its sample or why it had been considered to be representative of all of the advance rulings (paragraph 127 of the judgment under appeal). Examples had been used to illustrate all of the advance rulings, without an explanation as to why they were chosen and why they were representative (paragraph 128 of the judgment under appeal).
72.
As explained below, the Commission may confine itself to examining samples for the purposes of proving the existence of a scheme (see Section 1 below). It is therefore necessary to examine whether the General Court erred in law in its objection to the Commission’s selection of the sample (see Section 2 below).
(1) Proof of the existence of the aid scheme by means of a sample
73.
The Commission typically objects to advance tax rulings as individual aid.(27) However, this does not preclude the Commission from being able to prove the existence of a consistent administrative practice on the part of the tax authorities of a Member State in the case of advance tax rulings also. To do so, however, the Commission must demonstrate that the tax authorities use a systematic approach.
74.
An administrative practice is a practice that is, to some degree, of a consistent and general nature.(28) Such a practice is consistent if it has been developed in such a way as to give the impression that cases in a certain category are always treated in this way.
75.
By its very nature, proving the existence of a consistent administrative practice is associated with a certain degree of ambiguity, particularly if the State has not adopted any administrative provisions in that regard. Nevertheless, the mere appearance of an administrative practice is not sufficient. Rather, in cases where a law exists, the Commission must prove that the administrative practice has consistently gone beyond the mere exercise of the powers provided by that law. Otherwise, the law itself would constitute the aid scheme.
76.
In the present case, the Commission complains that the Belgian tax authorities consistently misapplied Article 185(2)(b) of the CIR 92. This is because, while that provision allows an adjustment of profits in respect of agreements between two undertakings of the same group, the Belgian tax authorities would have granted the exemption irrespective of such agreements.
77.
According to the statements of the General Court, it was sufficient if the profits were linked to a new situation, such as a reorganisation leading to the relocation of a central entrepreneur to Belgium, the creation of jobs, or the making of investments.(29) The Belgian authorities even advertised – as graphically shown by the Commission once again at the hearing – the possibility of advance tax rulings making a downward adjustment to profits (sometimes also referred to as excess profit exemption).(30)
78.
However, if, on the face of it, a large number of rulings appear to use the same approach and reasoning, this serves as an indication of consistent administrative practice.
79.
Thus, the Commission does not have to examine all the contested rulings individually, but may also base its proof of a consistent administrative practice on a sample.(31) However, the Commission must provide justification that the selection of its sample is representative. The sample must be sufficiently relevant for the specific case.(32) To that end, the sample could be selected purely at random or else in such a way that the total mass can be deduced as reliably as possible from a certain partial survey.
80.
In the case of the latter method, all the parameters relevant to the selection of the sample must be clear from the contested decision. Any reasons put forward by the Commission in its appeal to explain how the sample is representative would have been provided after the fact and therefore cannot be taken into consideration.
(2) Proof of a consistent administrative practice in the present case
81.
The question is therefore whether the General Court was right to find, in paragraph 126 of the judgment under appeal, that the Commission had not demonstrated to the requisite standard the existence of a consistent administrative practice in the contested decision.
82.
In paragraph 127 of the judgment under appeal, the General Court states that the Commission examined a sample of 22 of the 66 advance rulings concerned. However, the Commission did not explain either the choice of that sample or why it had been considered to be representative of all of the advance rulings. The Commission merely confined itself to covering, with the decisions it examined, the period of the decisions at issue.
83.
In addition, in paragraph 128 of the judgment under appeal, the General Court criticises the lack of detail in the reasons why six advance rulings from that sample were described briefly in the contested decision. The General Court states that the contested decision also gives no explanation as to why those 6 examples are sufficiently representative of all 66 advance rulings.
84.
Contrary to the view taken by the General Court, the contested decision cannot be challenged in so far as the six exemplary tax rulings are merely – in the words of the General Court – ‘examples capable of illustrating [all of the advance rulings]’.(33) This is because illustrative examples are not the reasons underpinning a decision; they merely serve to illustrate it.
85.
However, as regards the reasons underpinning the contested decision, the Commission’s selection must be representative, as stated above. To that end, all the parameters that led to the selection of the sample must be taken into account. Those parameters must be subjected to a global assessment.(34)
86.
In the present case, the Commission initially examined 22 of the 66 advance rulings concerned, meaning that the sample comprised a third of the rulings. There can be no objection to this number. Furthermore, all the rulings came from the same issuer, namely the Belgian Ruling Commission.
87.
In addition, all 22 advance rulings chosen and examined included profit adjustments in favour of applicants which were part of a multinational group. As a result, the Belgian tax authorities carried out what the Commission refers to as taxation on fictitious profit in respect of certain undertakings, which does not result from Article 185(2) of the CIR 92.
88.
Finally, the Commission selected decisions from 2004, 2007, 2010 and 2013. Although that information is apparent only from recital 3 of the contested decision, that recital does not appear in the section of the contested decision dealing with the assessment of the measure, but is confined to a description of the procedure.
89.
However, it is clear from the table prominently displayed in recital 59 of the contested decision that those 22 rulings in the sample represent all the rulings of the years selected. As no rulings were issued in 2004, the Commission used 2005, the first year of rulings of this kind. In that regard, the General Court rightly documented, in paragraph 127 of the judgment under appeal, the Commission’s statements according to which its examination of those years would cover rulings issued at the beginning, middle and end of the period.
90.
Consequently, the Commission explained in the contested decision that the sample was representative overall and therefore sufficient to demonstrate a consistent administrative practice.
91.
The fact that individual requests were withdrawn after the preliminary examination phase does nothing to change this. According to paragraph 112 of the judgment under appeal, only around half of the requests ultimately resulted in an advance ruling in 2014, for instance. No ruling was issued in relation to the other half of the requests.
92.
However, this is irrelevant, since, on the one hand, the Commission did not actually include 2014 in its sample. On the other hand, the Belgian authorities informed the Commission ‘that they have never refused any request for a ruling to benefit from the Excess Profit exemption since the contested scheme’s introduction’.(35) Against this background, the Commission was entitled to assume that it covered the entire tax ruling practice of the Belgian tax authorities.
93.
In the context of State aid control, it should be noted in this regard that the Commission is dependent on the cooperation of the Member State concerned. It may send requests for information to the Member State in accordance with Articles 5, 12 and 20 of Regulation 2015/1589. The Commission can rely on the accuracy and completeness of the information provided by the Member State. It does not have to carry out any investigations that go beyond that.
94.
Consequently, the Commission did sufficiently explain, in the contested decision, the choice of the sample and why it had been considered to be representative of all of the advance rulings. It is therefore not necessary to consider, in any greater detail, the Commission’s argument that a mere reference to the opening decision in the contested decision is sufficient to satisfy the requirements of evidence.(36)
95.
In conclusion, the General Court therefore erred in law by wrongly classifying the sample as not sufficiently representative to the requisite legal standard, and not sufficient to demonstrate a consistent administrative practice.
(c)
Conclusion regarding the first part of the ground of appeal
96.
The first part of the ground of appeal is therefore well founded.
2.
The second condition ‘without further implementing measures’ (second part of the ground of appeal)
97.
By the second part of its ground of appeal, the Commission alleges that the General Court misinterpreted the second condition of Article 1(d) of Regulation 2015/1589, namely that no further implementing measures are necessary.
98.
In paragraph 120 of the judgment under appeal, the General Court comes to the conclusion that the Commission wrongly concluded that the Belgian excess profit exemption at issue does not require further implementing measures.
99.
Previously, in paragraph 86 of the judgment under appeal, the General Court correctly defined the condition for the existence of an aid scheme within the meaning of Article 1(d) of Regulation 2015/1589 according to which individual aid must be granted without further implementing measures being adopted. For that reason, it is necessary that the essential elements of the aid scheme in question emerge from the provisions identified as the basis for the scheme.
100.
However, the General Court’s error of law specified above(37) also extends to this part of the ground of appeal. Since the General Court considered that the consistent administrative practice of the Belgian tax authorities had not been sufficiently demonstrated, it examined the condition regarding the absence of implementing measures only in relation to the legal basis for profit adjustments in Article 185(2)(b) of the CIR 92.
101.
If the aid scheme is based on a law, its application by the administration traditionally constitutes a possible further implementing measure. However, such further implementing measures do not exist if individual aid is granted solely on the basis of the law without the administration having any individual decision-making power.
102.
The General Court is in principle correct to assume, in paragraph 87 of the judgment under appeal, that the national authorities cannot have any ‘margin of discretion’. Rather, their power must be limited to technical application. It is only in this way that the existence of further implementing measures is precluded. The decisive question is therefore whether the authorities have real decision-making leeway or only bound decision-making competence. The fact that a request by the taxpayer is necessary is irrelevant to the need for further implementing measures, as correctly pointed out by the General Court in paragraph 100 of the judgment under appeal.
103.
If, however, as is the case here, a consistent administrative practice constitutes the act, there are generally no further implementing measures, since the consistent administrative practice already consists of a set of measures for granting individual aid.
104.
In the case of a consistent administrative practice, a further implementing measure could simply amount to the individual administrator being granted, within the framework of that practice, decision-making powers enabling him or her to deviate from the treatment actually practised.
105.
This is not the case here. In so far as the Belgian tax authorities, going beyond the wording of Article 185(2) of the CIR 92, carried out a comparative analysis of the profit with an undertaking not forming part of a group, an adjustment was always made to the lower comparative profit of an undertaking not forming part of a group. As this adjustment was made using the same method without exception, the authority did not have any independent, individual decision-making leeway.
106.
In particular, there is nothing in the present case to indicate that the Belgian tax authorities could make different profit adjustments if two identical group undertakings apply for an advance ruling because those authorities consider that one of those undertakings should receive a more favourable treatment than the other.
107.
In that regard, the General Court erred in law in finding that further implementing measures were necessary in this case. It follows that the second part of the ground of appeal is therefore also well founded.
3.
The third condition regarding the general and abstract definition of beneficiaries in the act (third part of the ground of appeal)
108.
By the third part of the ground of appeal, the Commission complains that the General Court misinterpreted the third condition of Article 1(d) of Regulation 2015/1589, namely the general and abstract definition of beneficiaries, and distorted recitals 66, 102 and 103, 109, 139 and 140 of the contested decision, by concluding that further implementing measures were necessary to define the beneficiaries of the excess profit exemption.
109.
The General Court addresses the definition of ‘beneficiaries’ in paragraphs 114 to 119 of the judgment under appeal. It concludes in paragraph 119 of the judgment under appeal that further implementing measures necessarily have to be taken in order to define such beneficiaries. According to paragraph 115 of the judgment under appeal, the beneficiaries cannot be identified on the sole basis of Article 185(2)(b) of the CIR 92. According to paragraph 116 of the judgment under appeal, the beneficiaries of the scheme as per recital 102 of the contested decision that were identified by the Commission correspond to a much more specific category than that defined in the law. However, the other legal acts identified by the Commission did not provide any additional details (paragraph 117 of the judgment under appeal).
110.
The General Court erred in law also in this regard. It is true that, in paragraph 115 of the judgment under appeal, the General Court does in principle apply the correct standard to the general and abstract definition of beneficiaries within the meaning of Article 1(d) of Regulation 2015/1589. Such a definition is general and abstract if the beneficiaries can be identified on the sole basis of the act, without further implementing measures.
111.
In the same paragraph, the General Court also correctly took the view that, pursuant to Article 185(2)(b) of the CIR 92, the use of the tax scheme is limited to entities that form part of ‘a multinational group of associated companies’. In that respect, the identically worded recital 109 of the contested decision has not been distorted.
112.
However, in paragraph 116 of the judgment under appeal, the General Court fails to recognise that the detailed description of the beneficiaries in recital 102 of the contested decision is itself part of the aid scheme.(38) The fact that the beneficiaries of the scheme identified by the Commission differ from those referred to in Article 185(2) of the CIR 92 is simply due to the fact that the Commission did not classify Article 185(2) of the CIR 92 as an aid scheme, but rather classified the consistent administrative practice as an aid scheme.
113.
Therefore, the General Court incorrectly classified the definition of beneficiaries in the act – in this case the consistent administrative practice – as not being general and abstract. The third part of the ground of appeal is therefore also well founded.
4.
Conclusion regarding the merits of the appeal
114.
In conclusion, the General Court incorrectly assumed that the conditions of Article 1(d) of Regulation 2015/1589 were not met in the present case. On the contrary, in the contested decision the Commission sufficiently demonstrated that the Belgian practice of making downward adjustments of the profits of undertakings forming part of a multinational group constitutes an aid scheme within the meaning of Article 1(d) of Regulation 2015/1589. The appeal is therefore well founded.
D. Inadmissibility of the cross-appeal
118.
It is also necessary to examine whether the cross-appeal brought by Belgium is admissible.
119.
Pursuant to the first sentence of the second paragraph of Article 56 of the Statute of the Court of Justice of the European Union, an appeal may be brought by any party which has been unsuccessful, in whole or in part, in its submissions. In the present case, the form of order sought by Belgium seeks annulment of the decision of the General Court in so far as the latter rejected Belgium’s first plea in law.
120.
Although, under Article 178(1) of the Rules of Procedure, forms of order sought in cross-appeals may ‘seek to have set aside, in whole or in part, the decision of the General Court’, the Court of Justice has held that it is a basic principle applying to appeals that an appeal must be directed against the operative part of the General Court’s decision and may not merely seek the amendment of some of the grounds of that decision.(40)
121.
This is also in line with the wording of the general rule on forms of order sought in appeals in Article 169(1) of the Rules of Procedure. According to that rule, an appeal that does not seek to have the judgment under appeal, that is to say the operative part thereof, set aside, even in part, but merely to amend some of the grounds of that judgment, is inadmissible.(41)
122.
The rationale behind this is that every appellant must have an interest in bringing an appeal. This follows already from the first sentence of the second paragraph of Article 56 of the Statute of the Court of Justice of the European Union and also applies to cross-appeals.
123.
That is not the case here. This is because the Commission’s appeal will either be dismissed, and the annulment of the contested decision will therefore become final – fully in line with Belgium’s aims – or the Court of Justice will refer the case back to the General Court. The Court of Justice will not rule on Belgium’s considerations concerning the tax jurisdiction of the Member States until a further appeal.
124.
Belgium’s cross-appeal is therefore inadmissible.