C-459/24

OPINION OF ADVOCATE GENERAL BIONDI delivered on 21 May 2026 (1) Case C‑459/24 P Ideella föreningen Svenska Bankföreningen med firma Svenska Bankföreningen, Näringsverksamhet v European Commission ( Appeal – State aid – Swedish tax law – Risk tax – Decision of the European Commission not to raise objections – Selectivity – Objective of the measure ) I.      Introduction 1.        When it comes to State aid and taxation, the issue of selectivity inevitably ‘presents considerable difficulties’, to quote Advocate General Kokott’s prime example of the art of understatement. (2) 2.        It is indeed self-evident that taxation is not merely an instrument for raising revenue to finance public expenditure. It also serves a range of broader objectives, including income redistribution, the promotion of international competitiveness, debt management, and, regrettably, at times the conferral of undue advantages on particular categories of taxpayers. In that context, the application of selectivity as a means of upholding the principle of non-discrimination becomes particularly important –and inherently challenging –when it comes to national regulations, which by their very nature involve making distinctions and exercising discretion. 3.        These difficulties should not however unduly impede the operation of well-established principles of EU State aid law. As early as the judgment in Italy v Commission – to the best of my recollection, the first case addressing a fiscal measure in the context of State aid – the Court articulated two foundational tenets of State aid control. First, the concept of aid is objective in nature and does not vary according to the sphere of national regulatory competence in which it arises. Second, the existence of national competence in a given domain, including taxation, does not insulate its exercise from scrutiny for compatibility with EU law. (3) 4.        Such findings are not merely of historical or doctrinal interest. They retain direct and significant relevance for the present appeal, raising the question whether a tax that applies exclusively to nine credit institutions, in a competitive market comprising more than 100 operators, is selective under Article 107(1) TFEU. 5.        Ideella föreningen Svenska Bankföreningen med firma Svenska Bankföreningen, Näringsverksamhet (‘the appellant’), a Swedish association of bankers, which represents its 31 members in matters of common interest, both nationally and internationally, seeks to have the judgment of 17 April 2024, Svenska Bankföreningen and Länsförsäkringar Bank v Commission (4) (‘the judgment under appeal) set aside. By that judgment, the General Court dismissed the action, brought by the appellant and one of its members, Länsförsäkringar Bank AB, for the annulment of European Commission Decision COM(2021) 8637 final of 24 November 2021 on State aid SA.56348 (2021/N) – Sweden: Swedish tax on credit institutions (‘the decision at issue’). II.    Background to the dispute, the decision at issue and the judgment under appeal 6.        In accordance with Article 108(3) TFEU, on 3 September 2021 the Kingdom of Sweden notified the Commission of a draft law concerning a risk tax on credit institutions, as well as draft amendments to the Law on foreign tax credit (together, ‘the draft law’). (5) The Kingdom of Sweden was nevertheless of the opinion that the tax introduced (‘the risk tax’) did not fulfil the selectivity criteria laid down in Article 107(1) TFEU and that it did not therefore constitute State aid. The draft law was adopted, and the law entered into force on 1 January 2022. 7.        Based on paragraphs 5 to 9 of the judgment under appeal, the content of the draft law can be summarised as follows for the purpose of the present appeal. 8.        Under Section 1 of point 2.1 of the draft law, credit institutions are to be liable to pay the tax to the State. 9.        Under Section 4 of point 2.1 of the draft law, Swedish credit institutions are to be liable to risk tax if the sum of their liabilities at the beginning of the tax year exceeds the threshold limit laid down by the draft law. (6) In the reasons for the draft law relating to that section, reference was made to Swedish law for the purposes of defining credit institutions, namely Swedish banks and Swedish credit market companies, as well as foreign banking companies and foreign credit companies. As a result, nine credit institutions were liable to pay the tax (paragraphs 5 and 6 of the judgment under appeal). 10.      In Section 5 of point 2.1 of the draft law, the threshold limit is set at 150 billion kronor (SEK) (approximately EUR 13.3 billion) for tax years commencing in 2022. For tax years commencing in 2023 or later, the threshold limit amounts to SEK 150 billion multiplied by a figure stating the ratio between the price base amount in 2022 and the price base amount for the year in which the tax year in question began, expressed as a percentage with two decimals rounded down plus two percentage points (paragraph 7 of the judgment under appeal). 11.      Section 6 of point 2.1 lays down specific provisions governing situations where a credit institution is part of a group of credit institutions (paragraph 8 of the judgment under appeal). 12.      Section 9 of point 2.1 of the draft law sets the tax rate as 0.05% of the sum of the liabilities of the credit institutions liable for the tax. Point 2.2 of the draft law provides that the rate is to be 0.06% for the 2023 tax year (paragraph 9 of the judgment under appeal). 13.      In the contested decision, adopted after a preliminary examination pursuant to Article 4(3) of Regulation (EU) 2015/1589, (7) the Commission concluded that the tax did not constitute State aid within the meaning of Article 107(1) TFEU, on the ground that it did not satisfy the selectivity criterion set out in that provision. 14.      The appellant and one of its members (together, ‘the applicants at first instance’) brought an action for annulment of the decision at issue before the General Court. They relied on a single plea in law alleging that their procedural rights had been infringed. They argued that, in its analysis of the risk tax, the Commission should have encountered serious difficulties and initiated a formal investigation procedure according to Article 4(4) of Regulation 2015/1589, giving them the possibility to make their views known. In that context, they contested the Commission’s assessments made as regards the selective nature of the tax. The Kingdom of Sweden intervened in support of the Commission. 15.      By the judgment under appeal, the General Court found that the applicants at first instance had not established that the Commission should have had doubts as to the classification of the risk tax within the meaning of Article 107(1) TFEU, which should have led it to initiate the formal investigation procedure. Consequently, it dismissed the action in its entirety and ordered them to pay the costs. III. Analysis A.      Preliminary remarks 16.      In line with the principles set out in point 3 above, the Court has consistently emphasised that actions taken by Member States in areas not harmonised by EU law remain within the scope of the provisions of the FEU Treaty regarding the monitoring of State aid. Member States are therefore required to refrain from enacting any tax measure that could constitute State aid incompatible with the internal market. (8) This is not just an empty formula, but rather represents a specific manifestation of a broader and well-established principle: national competence is constrained when it affects the exercise or the application of EU law. 17.      State aid rules must thus not be regarded as a separate ‘system’. Rather, they are an essential component of the EU internal market governance. Indeed, while the Commission does not need to perform a detailed assessment of every EU rule potentially affected by a national measure, the Court has repeatedly affirmed that State aid that is in breach of EU law or its general principles cannot be considered compatible with the internal market. (9) 18.      In the present case, the Swedish risk tax not only has the clear potential to affect the application of State aid law, but also visibly impinges upon an area of EU law that has already been harmonised – to a considerable extent – namely, the EU banking regulatory framework, aimed at preserving the stability of the financial system (‘the EU banking regulatory framework’). 19.      The EU banking regulatory framework comprises, inter alia,(10) the bank recovery and resolution framework, that is to say, the Bank Recovery and Resolution Directive (11) (‘the BRRD’), the prudential requirements framework, namely the Capital Requirements Regulation (12) (‘the CRR’) and the Capital Requirements Directive (13) (‘the CRD’), and the deposit guarantee scheme framework, which consists of the Deposit Guarantee Schemes Directive (14) (‘the DSGD’). It also comprises Regulation (EU) No 806/2014, which applies to credit institutions established in the Banking Union. (15) 20.      The CRR, together with the CRD, forms the prudential capital requirements framework, (16) governing, in particular, the prudential rules for credit institutions. After the 2008 global financial crises, the EU legislature found it necessary to safeguard the banking environment in the European Union by imposing capital requirements on credit institutions, proportionate to the risks addressed. (17) As regards the DGSD, it is intended to reinforce the internal market by facilitating the freedom of establishment and the freedom to provide financial services, while simultaneously enhancing banking system stability and ensuring the protection of depositors. (18) With the BRRD, the legislature intended to provide authorities with a credible set of tools to intervene sufficiently early and quickly in an unsound or failing institution so as to ensure the continuity of the institution’s critical financial and economic functions, while minimising the impact of an institution’s failure on the economy and financial system. (19) 21.      This long detour into the EU banking regulatory framework only to say that, in my view, although none of its provisions expressly limits the right of the Member States to adopt measures of a fiscal nature such as the risk tax, the adoption of such measures cannot undermine the objectives pursued by that framework. (20) It seems to me that, despite the Member States’ autonomy in fiscal matters, the comprehensiveness of EU banking regulatory framework should have warranted particular care on the part of the Commission in examining a national tax measure liable to interfere with those rules, including at the stage of assessing its selectivity. 22.      That being said, I recall that, notwithstanding the particular challenges posed by national fiscal measures, the assessment of selectivity under Article 107(1) TFEU must invariably be situated within the well-established three-step test developed in the Court’s case-law. (21) That test functions uniformly across various national measures, irrespective of the specific national competences involved, and is sufficiently adaptable to accommodate the peculiarities of fiscal measures. 23.      According to the three-step selectivity test, in order to classify a national tax measure as ‘selective’, the Commission must begin by identifying the reference system, that is, the ‘normal’ tax system applicable in the Member State concerned, and demonstrate, as a second step, that the tax measure at issue is a derogation from that reference system, in so far as it differentiates between operators who, in the light of the objective pursued by that system, are in a comparable factual and legal situation. The concept of ‘State aid’ does not, however, cover measures that differentiate between undertakings which, in the light of the objective pursued by the legal regime concerned, are in a comparable factual and legal situation, and are, therefore, a priori selective, where the Member State concerned is able to demonstrate, as a third step, that that differentiation is justified, in the sense that it flows from the nature or general structure of the system of which those measures form part. (22) 24.      The Court has pointed out that the determination of the reference system within the first step of the selectivity test is of particular importance in the case of tax measures, since the existence of an economic advantage for the purposes of Article 107(1) TFEU may be established only when compared with ‘normal’ taxation. (23) In that respect, the Court has clarified that, where the tax measure in question is inseparable from the general tax system of the Member State concerned, reference must be made to that system. On the other hand, where it appears that that measure is clearly severable from that general system, it cannot be ruled out that the reference system to be taken into account may be more limited than that general system, or even that it may equate to the measure itself, where the latter appears as a rule having its own legal logic and it is not possible to identify a consistent body of rules external to that measure. (24) 25.      It is of course for the Member State to decide the characteristics constituting each tax. These includes the tax rate, tax base, calculation methods, taxable events, and thresholds. The Court clarified in a recent judgment that exemptions should also be considered as such. (25) Those characteristics, in principle, define the reference system or the ‘normal’ tax regime, for the purposes of analysing the condition of selectivity (26). This is, however, without prejudice to the possibility of finding that the reference framework itself, as it results from national law, is incompatible with the EU law on State aid, because the tax system at issue has been configured according to manifestly discriminatory parameters intended to circumvent that law. (27) B.      The appeal 26.      The appellant raises four grounds in support of its appeal. 27.      The first ground of appeal, the first limb of the second ground of appeal and the fourth ground of appeal, are directed against paragraphs 39 to 43 of the judgement under appeal, in which the General Court rejected the arguments raised by the applicants at first instance concerning the determination of the objective of the risk tax. I shall therefore examine them together. 1.      The first ground of appeal, the first limb of the second ground of appeal and the fourth ground of appeal 28.      By its first ground of appeal, the appellant claims that the General Court erroneously defined the objective of the risk tax and therefore erred in law in the application of Article 107(1) TFEU as regards the selectivity criterion. The appellant argues that the General Court determined the objective of the risk tax to be to tax ‘credit institutions that pose a systemic risk’, whereas Sweden, in the draft law, and the Commission, in the decision at issue, identified that objective as taxing ‘large’ and not ‘systemically important’ credit institutions. In the appellant’s view, this error in law vitiates the entire analysis of the reference system and is, in itself, sufficient to justify setting aside the judgment under appeal. 29.      By the first limb of its second ground of appeal, the appellant claims that the Commission, and consequently the General Court, accepted the often opaque and vague reasoning of the Kingdom of Sweden explaining the objective and rationale for the risk tax. It argues that that lack of precision should have prompted the Commission to open a formal investigation procedure under Article 108(2) TFEU in order to better understand the measure. According to the appellant, the judgment under appeal is vitiated by an error of law for that reason, regardless of whether the first ground of appeal is upheld by the Court. 30.      By its fourth ground of appeal, the appellant contends that the General Court distorted facts and evidence by incorrectly determining the objective of the tax at issue in paragraphs 40 and 41 of the judgment under appeal. The appellant refers to its arguments developed in support of its first ground of appeal. 31.      I will then assess the merits of the appellant’s arguments after first addressing the appropriate standard of review to be applied to the General Court’s findings on the objective of the risk tax. (a)    The standard of review to be applied  to the General Court’s findings regarding  the objective of the risk tax 32.      The appellant claims that the General Court incorrectly interpreted the section of the draft law in which the Swedish legislature explained the reasons for introducing the risk tax and its objectives. The Commission argues that, for such a claim to be admissible before the Court of Justice, the appellant must demonstrate a distortion of facts. 33.      In that respect, it must be borne in mind that, according to the second subparagraph of Article 256(1) TFEU, an appeal against a decision given by the General Court is to be on points of law only and must be made ‘under the conditions and within the limits laid down by the Statute’. In a list setting out the grounds that may be relied upon in that context, the first paragraph of Article 58 of the Statute of the Court of Justice of the European Union states that an appeal may be based on infringement of European Union law by the General Court. (28) 34.      It is true that, with respect to the assessment in the context of an appeal of the General Court’s findings on national law – which, in the field of State aid, constitute findings of fact – in principle, the Court of Justice has jurisdiction only to determine whether that law was distorted. (29) 35.      However, the Court of Justice cannot be deprived of the possibility of reviewing whether such assessments themselves constitute an infringement of EU law. (30) In paragraph 78 of the judgment of 5 December 2023, Luxembourg and Others v Commission, (31) the Court clarified that ‘the arguments aimed at calling into question the choice of reference [system] or its meaning in the first step of the analysis of the existence of a selective advantage are admissible, since that analysis derives from a legal classification of national law on the basis of a provision of EU law’. 36.      The Commission argues that those clarifications are confined to the identification of the reference system in the first step of the selectivity test and do not apply to the identification of the objective of a tax to be assessed under State aid rules. That argument cannot be accepted. 37.      First, the Court’s finding in paragraph 78 of Luxembourg and Others v Commission clearly applies to the overall analysis of selectivity, and therefore to each of its three steps. 38.      Second, the limitation alleged by the Commission is not supported by the logic underlying the Court’s reasoning, as set out in paragraph 79 of the judgment in Luxembourg and Others v Commission. In that paragraph, the Court stated that ‘to concede that [it] is not in a position to determine whether the General Court made no error of law when it endorsed the definition of the relevant reference framework, the interpretation thereof and the application thereof as the decisive parameter for the purpose of examining whether there was a selective advantage would be tantamount to accepting that the General Court may have infringed a provision of primary EU law, namely Article 107(1) TFEU, without any possibility of that infringement being found in an appeal, which would contravene the second subparagraph of Article 256(1) TFEU’. (32) 39.      In line with the reasoning underlying paragraphs 78 and 79 of the judgment in Luxembourg and Others v Commission, I therefore believe that the Court of Justice has jurisdiction to review the General Court’s findings on national law concerning the definition of the objective of a tax measure for the purpose of the analysis of selectivity under Article 107(1) TFEU. 40.      It follows that the examination of the appellant’s arguments in the context of its first ground of appeal should not be limited to the question whether evidence of distortion of the relevant section of the draft law has been provided. (b)    The first ground of appeal 41.      In order to examine the arguments of the appellant, I shall consider, in turn, the definition of the objective of the risk tax as adopted in the judgment under appeal, in the decision at issue and in the draft law. (1)    The objective of the risk tax in the judgment under appeal 42.      In paragraph 41 of the judgment under appeal, the General Court found that it was apparent from the draft law that the objective of the risk tax was ‘to strengthen public finances by improving them and keeping public debt at a low level in order to provide room to cope with future financial crises by requiring the tax to be paid by large credit institutions, the failure or serious disruption of which would, on an individual basis and because of their size and importance for the functioning of the financial system, present a systemic risk and would have a very negative impact on that system and on the economy in general, thus causing significant indirect costs for society’. The same very extended definition appears in paragraphs 58, 80, 81, 93 and 120 of the judgment under appeal. While the General Court further elaborated, in the judgment under appeal, that ‘the [risk] tax is not intended to prevent or address the risks that credit institutions represent’, (33) the focus is clearly on the ‘systemic risk’ that liable credit institutions may pose ‘for the functioning of the financial system’ due to their ‘size and importance’. (2)    The objective of the risk tax in the decision at issue 43.      In recitals 6 to 8 of the decision at issue, the risk tax is defined as an ‘annual tax applicable to large credit institution operating in Sweden’, the objective of which is ‘to strengthen public finances with contributions from large credit institutions potentially creating significant indirect costs to the society, thereby creating scope to cope with such indirect costs future financial crises may entail’. The persons liable to the risk tax are described as ‘large credit institutions’, that, ‘given their size and importance, risk causing significant indirect costs to the society in the event of a financial crisis’. There is no reference to the concepts of ‘systemic risk’ or of ‘systemically important’ credit institutions in the definition of the objective of the risk tax set out in those recitals. 44.      However, in assessing whether the choice of the taxable persons and the metric used to identify those persons revealed manifestly discriminatory elements, the Commission states that large credit institutions may be  ‘systemically important’ and that the level of liabilities is ‘a metric suitable to represent the systemic risk that may be caused by a credit institution’ as well as a ‘relevant indicator of both a credit institution’s importance and of its contribution to the risk its failure may pose’. (34) It is true that, in making those assessments, the Commission noted that the objective of the risk tax was to target credit institutions liable to create significant indirect costs. (35) Nevertheless, other recitals of the decision at issue are less clear-cut. While the Commission never openly affirms that the objective of the Swedish legislature was to tax credit institutions that pose a systemic risk to the financial system, some of its assessments seem hardly reconcilable with a different definition of that objective, (36) showing a substantial lack of clarity on that point. (3)    The objective of the risk tax in the draft law 45.      The explanations provided by the Swedish legislature in the draft law reveal an even greater lack of clarity regarding the actual objective of the rationale of the risk tax. 46.      It appears from the preparatory works leading to the adoption of that tax, included in the General Court’s file, (37) that the plan to introduce a tax on the financial sector dates back to 2015. (38) A proposal for increasing taxation of credit institutions which, in the event of a financial crisis, risk causing significant indirect costs to society materialised in 2020, when the Swedish Minister for Finance presented a memorandum entitled ‘Risk tax for certain credit institutions’, which was revised in 2021 and subsequently notified to the Commission. 47.      Section 5.1 of the draft law, detailing the reasons for the legislative proposal of a risk tax, explains that large institutions are such a vital part of the financial system that their failure, or serious disruption of their activities, would have negative consequences for the financial system or the economy. According to that section, those institutions are ‘systemically important’ and their failure constitutes a ‘systemic risk’, defined as a threat to the financial stability. (39) It is explained that the financial system is vulnerable and that, despite the regulations governing credit institutions – such as those concerning capital adequacy and resolution – and supervision of credit institutions, it is still possible that financial crises will arise in the future. The draft law emphasises that financial crises can be very costly and have negative consequences for households, businesses and the public sector over an extended period. The reduced economic growth in the aftermath of a financial crisis in turn gives rise to indirect costs in the economy. (40) Those consequences make it reasonable, according to Section 5.1 of the draft law, to increase taxation on credit institutions which, in the event of a financial crisis, risk causing significant indirect costs to society. That section of the draft law further points out that the risk that, in a financial crisis scenario, an institution might generate indirect costs for society, also varies according to the institution’s size and importance. In this sense, small institutions do not affect macroeconomic developments to the same extent as large ones. Therefore, credit institutions which, due to their size, importance for the functioning of the financial market and macroeconomic developments, ‘risk causing significant indirect costs in case of a financial crisis, should pay an additional tax’. 48.      Section 6 of the draft law, containing the Swedish authorities’ assessment regarding the compatibility of the proposed risk tax with EU State aid law, states that that tax aims to compensate for the risk that certain credit institutions pose to the macroeconomic conditions in Sweden, due to their size and financial position. Those credit institutions are ‘significant’ and are hence to be taxed. The analysis under that section further outlines that in accordance with the EU banking regulatory framework, the degree of significance of a credit institution’s impact on the financial system depends on its size and complexity, as well as the scope of its operations. According to the memorandum, the credit institutions to be taxed are only institutions which pose a potential risk of significant indirect costs to the society, due to their size. It is reiterated that the purpose of the tax is to tax credit institutions which, in the event of a financial crisis, risk causing significant indirect costs to society. 49.      Despite the ambiguity of the draft law concerning the actual purpose of the risk tax, particularly with regard to the choice of persons liable to it, I believe it is appropriate to make the following remarks. 50.      First, the Swedish authorities refer to ‘systemically important’ credit institutions only in Section 5.1 of the draft law. (41) The rest of the draft law refers to the ‘significance’ of credit institutions or, more generally, to their ‘size and importance’. 51.      Second, the aforementioned section of the draft law addresses the concepts of ‘systemically important credit institutions’ and ‘systemic risk’ in a manner that echoes the corresponding definitions in the EU banking regulatory framework. (42) However, those concepts are referred to solely in order to distinguish ‘large credit institutions’ from ‘small credit institutions’, emphasising the importance of the former for the financial system compared to the latter. In particular, the adjectives ‘systemically important’, ‘significant’ and ‘large’ are used interchangeably by the Swedish authorities. 52.      Third, the criterion used throughout the draft law to determine which credit institutions are subject to the risk tax refers to their ‘size and importance’ to the functioning of the financial system, rather than to whether they pose a systemic risk. 53.      Fourth, the draft law repeatedly emphasises that the objective of the risk tax is to tax credit institutions which pose a risk of causing significant ‘indirect costs’ for society in Sweden in the event of a financial crisis. (4)    Intermediate conclusion 54.      Despite the lack of clarity in the draft law and certain inconsistencies in the decision at issue, it can be concluded that, as regards the persons liable, the objective of the risk tax set out in those acts was not to tax credit institutions posing a systemic risk to the financial system – contrary to what the General Court stated in paragraph 41 of the judgment under appeal and reiterated throughout its analysis – but to tax credit institutions exceeding a predetermined liabilities threshold, referred to as ‘large credit institution’. 55.      The endeavors by the Commission and Sweden, in their respective written observations, to demonstrate that the General Court was consistent with the decision at issue and the draft law in its reconstruction of the objective of the risk tax are unconvincing and, indeed, clearly reveal the difficulty of demonstrating such consistency. (43) 56.      In an attempt to clarify the ambiguities in the draft law, the General Court combined some of the concepts mentioned therein to formulate its own definition of the purpose of the risk tax, which does not correspond to that envisaged by the Swedish legislature. As I shall demonstrate later, this enabled the General Court to employ a somewhat circular line of reasoning in rejecting the arguments put forward by the applicants at first instance, relying, alternatively, on one or the other of those concepts. 57.      For the reasons set out above, I am of the view that, in determining the objective of the risk tax, the General Court misinterpreted the relevant national law, thereby erroneously defining that objective. (5)    Conclusion on the first ground of appeal 58.      In the light of the foregoing, I propose that the first ground of appeal be upheld. (c)    The first limb of the second ground of appeal 59.      As already mentioned, by the first limb of its second ground of appeal, the appellant contends that the General Court erred in law in finding that the determination of the objective of the risk tax did not raise serious difficulties warranting the opening of a formal investigation procedure under Article 108(2) TFEU. The appellant claims that, in so concluding, the General Court did not properly compare the grounds of the decision at issue with the information available when the Commission took its decision. 60.      I note that, according to a settled case-law, reflected in Article 4(4) of Regulation 2015/1589, if the Commission finds, following a preliminary examination, that doubts are raised as to the qualification of the State measure in question as ‘aid’ within the meaning of Article 107(1) TFEU or, if classified as aid, as to its compatibility with the internal market, the Commission is under a duty to initiate the procedure under Article 108(2) TFEU, without having any discretion in that regard. (44) 61.      It is therefore for the EU Courts, when ruling on an application for annulment of a Commission decision not to raise objections, to determine whether the assessment of the information and evidence which the Commission had at its disposal during the preliminary investigation phase of the national measure at issue should objectively have raised doubts as to the categorisation of that measure as aid, given that such doubts must lead to the initiation of a formal investigation procedure. (45) 62.      In the present case, the applicants at first instance criticised the opacity of the objective of the risk tax as defined by the Swedish authorities and endorsed by the Commission. They argued that that objective was (i) biased in itself, since it was too narrow and singled out a non-homogeneous category of undertakings, namely ‘large credit institutions’ defined by the size of their liabilities; (ii) vague, in that it relied on unclear concepts such as ‘significant indirect costs’, and (iii) inconsistent because the draft law did not contain any rules on how the revenue from the risk tax would be used. 63.      In paragraphs 39 to 43 of the judgment under appeal, rather than addressing those arguments in the light of the objective of the risk tax as assessed in the decision at issue, the General Court reformulated that objective on the basis of its own interpretation of the draft law, endeavouring to clarify and reconcile the various statements contained therein. In so doing, the General Court failed to evaluate, as it was called upon to do, whether, in the light of the arguments submitted by the appellants, the Commission should have had doubts in endorsing the objective of the risk tax as described by the Swedish authorities. 64.      The appellant is therefore correct in arguing that the General Court erred in law by failing to assess whether the definition of the objective of the risk tax raised serious difficulties, such as in justifying the initiation of the formal investigation procedure. 65.      Moreover, the General Court did not address at all the argument concerning the biased ‘scope’ of the objective of the risk tax as endorsed by the Commission. The applicants at first instance had submitted, in essence, that all credit institutions are likely to create indirect costs in the event of a financial crisis and should therefore be regarded as equally concerned by a tax the purpose of which was to cope with such costs. The fact that a very small number of credit institutions among all the market players had been identified upstream and included in the definition of the objective of the risk tax was therefore likely to distort the whole analysis on selectivity and should have raised doubts as to the correctness of that definition. The General Court rejected that argument, merely noting that the identification of ‘large credit institutions’ as the taxpayers was consistent with the objective of the risk tax as redefined in the judgment under appeal, thereby failing to address the core of that argument. 66.      Yet the issue raised by the appellant was an important one. 67.      Special-purpose taxes, such as the one discussed in the present appeal, which exist outside the broader taxation system, require a focused analysis, in which the second step of the three-step selectivity test recalled in point 23 above becomes central. Selectivity must be assessed according to precise criteria that ensure ad hoc taxation can be recognised as State aid even in the absence of a clear general reference system, while simultaneously preventing mere departures from such a system from automatically being deemed selective. 68.      In that context, correctly defining the objective of a stand-alone tax – against which the reference system in the first step of the selectivity analysis is assessed and the comparability test under the second step is carried out – is essential and even more crucial in the case of asymmetric taxes, (46) such as the risk tax, where the differentiation in tax liability need to be justified on the basis of the overall aim of the measure. 69.      Lastly, the appellant is, in my view, also correct to criticise the opacity and vagueness of the objective of the risk tax as defined by the Swedish authorities and endorsed by the Commission. On that point, I refer back to the analysis concerning the first ground of appeal, where I pointed out the lack of clarity of the draft law and the inconsistencies in the decision at issue concerning the determination of that objective. 70.      In the light of the above, the first limb of the second ground of appeal must, in my view, also be upheld, regardless of the outcome of the first ground of appeal. (d)    The fourth ground of appeal 71.      By its fourth ground of appeal, the appellant claims a distortion of facts and evidence on the basis of the same arguments raised in support of the first ground of appeal. In the light of the Court’s settled case-law, the distortion must be obvious from the documents in the case file without there being any need to carry out a new assessment of the facts and the evidence. (47) 72.      As stated in points 32 to 40 above, the appellant did not have to demonstrate a distortion of the relevant national law in order to challenge the definition of the objective of the risk tax in the judgment under appeal. However, by reformulating the objective of the risk tax and failing to acknowledge the lack of clarity of the draft law, as explained in the analysis of the first ground of appeal, the General Court committed, in my view, such a distortion. It also misinterpreted the decision at issue. (48) (e)    Conclusion on the first ground of appeal, on the fourth ground of appeal and on the first limb of the second ground of appeal 73.      For the reasons set out above, I am of the view that the judgment under appeal is vitiated by the errors in law alleged by the appellant in its first and fourth grounds of appeal as well as in the first limb of its second ground of appeal. Given that the correct determination of the objective of a tax measure is central to the overall analysis of its selectivity under Article 107(1) TFEU, those errors must, in my view, lead to the judgment being set aside. 74.      I shall proceed, below, to carry out an analysis of the remainder of the appeal, in the event that the Court does not endorse my proposal. 2.      The second ground of appeal 75.      By its second ground of appeal, the appellant alleges that the General Court erred in law in the application of Articles 107 and 108 TFEU in finding that the applicants at first instance did not demonstrate that the Commission should have encountered serious difficulties in its assessment of the risk tax leading it to open a formal investigation procedure. 76.      The second ground of appeal is divided into three limbs. The first limb concerns the determination of the objective of the risk tax. That limb has already been examined. The second and third limbs concern respectively the assessment of the reference system of the risk tax and the assessment of the existence of a derogation to that system. 77.      Before addressing those second and third limbs, I observe that, while some of the arguments developed in support of the second ground of appeal are based on the premiss that the Court considers the objective of the risk tax to be correctly defined as taxing systemic risk, this is not the case for all of them. In fact, most of those arguments, as I shall illustrate, ultimately criticise the reasoning of the General Court, regardless of how the objective of the risk tax is defined. (a)    The second limb of the second ground of appeal 78.      By the second limb of its second ground of appeal the appellant contends that the General Court erred in law in finding, in paragraphs 44 to 99 of the judgment under appeal, that the Commission was not faced with serious difficulties in assessing the reference system of the risk tax. 79.      In the decision at issue, the Commission defined the reference system for the purpose of analysing the condition of selectivity as being limited to the risk tax, namely a special-purpose risk tax, based on the following elements: (i) a tax base relying on the credit institutions’ liabilities; (ii) a tax rate of 0.05% for the fiscal year starting in 2022; (iii) taxable persons, defined as credit institutions established in Sweden or carrying out business operations through a branch in Sweden; (iv) the existence of a threshold and (v) a consolidation mechanism for intra-group situations when calculating the threshold and the tax base (see recital 38 of the decision at issue). 80.      The appellant raises three complaints in support of the second limb of its second ground of appeal, concerning respectively the choice of the tax base, the persons liable to the tax and the threshold of the tax. (1)    The tax base 81.      The appellant claims that, if the objective of the risk tax were to tax the systemic risk, as stated by the General Court, total liabilities do not constitute a good indicator, since the identification of systemic risk is a complex assessment and cannot be done based on a single parameter. The appellant also contends that the General Court merely accepted the assessment carried out by the Commission in this respect without carrying out a proper review. 82.      I note that, in the decision at issue, the Commission considered that the sum of liabilities as a metric for defining which credit institutions should be subject to the risk tax was inherent in the reference system, was consistent with its objective and did not reveal a manifestly discriminatory element in the design of the tax. (49) 83.      The applicants at first instance argued that liabilities are not associated with risk, unlike assets and that the same applies to the size of credit institutions. They added that indirect costs – or the risk of such costs – are not directly proportional to a credit institution’s liabilities, a fact which had been brought to the Commission’s attention. Accordingly, the Commission had not properly examined whether the tax base chosen by the Kingdom of Sweden introduced a manifestly discriminatory parameter into the reference system. They also submitted that, contrary to what is required by recent case-law on tax measures and selectivity, the notified measure should have either impose a significant tax burden on the entire tax base of undertakings or imposed no tax burden at all. (50) 84.      The General Court, relying on the judgments of 16 March 2021, Commission v Poland (51) and Commission v Hungary (52), held that EU law does not preclude non-progressive taxation from being based on the aggregate sum of the liabilities of credit institutions and that the fact that there are more relevant or more precise indicators is irrelevant in matters of State aid, since EU law on that matter seeks only to remove the selective advantages from which certain undertakings might benefit to the detriment of others which are placed in a comparable situation (paragraph 57 of the judgment under appeal). The General Court added that the objective of the tax was not to prevent or address the risks that credit institutions represent, implying that, for that reason, the Swedish authorities were not required to adopt metrics specifically designed to target those risks (paragraph 58 of the judgment under appeal). Lastly, the General Court took into consideration the scenario where credit institutions own each other’s bonds to come to the conclusion that the higher the level of liabilities, the greater the risk to the financial system in so far as the credit institution may not be able to meet its liabilities. On the basis of that scenario, it concluded that a condition relating to the level of liabilities in order to distinguish between credit institutions according to whether they have a greater or lesser impact on the financial system was consistent with the objective pursued. 85.      The arguments raised by the appellant against paragraphs 57, 58 and 59 of the judgment under appeal are in my view well founded for the reasons set out below. 86.      In the first place, the similarities between the present case and the cases which gave rise to the judgments in Commission v Poland and Commission v Hungary are scarce. 87.      First, those rulings relate very specifically to turnover taxes, where the burden fell uniformly but progressively on gross sales. The risk tax is very different. Liability for the tax is triggered once the threshold has been reached and there is no variation mechanism in the rate applied. (53) Second, the taxes at issue in the judgments in Commission v Poland and in Commission v Hungary did not provide for exemptions from liability, whereas the risk tax does. (54) Third, turnover and liabilities are two indicators that cannot be compared: the former is an expression of the taxable person’s ability to pay, whereas the latter is a metric for calculating its level of indebtedness. Fourth, no question of interference with an existing EU regulatory framework existed in the cases that led to those judgments. 88.      In these circumstances, a mere reference to the judgments in Commission v Poland and in Commission v Hungary and to the broad discretion enjoyed by Member States was not enough to justify the General Court’s conclusion that the Commission had not encountered serious difficulties in assessing the metric chosen by the Swedish legislature and approved by the Commission as the tax base for the risk tax. 89.      By merely stating in paragraph 57 of the judgment under appeal – paraphrasing the judgments in Commission v Poland and in Commission v Hungary – that ‘EU law does not preclude non-progressive taxation from being based on the aggregate sum of the liabilities of credit institutions’, the General Court declined, as the appellant correctly claims, to carry out a proper review of the existence of such serious difficulties. Such an analysis would have entailed an examination of whether, on the basis of the information available during the preliminary proceedings and in the light of the specific characteristics of the risk tax, the Commission could conclude, without raising doubts, that the aggregate sum of the liabilities as a tax base was consistent with the objective of that tax, however it is defined, and was a legitimate expression of Sweden’s tax sovereignty. 90.      A thorough analysis was all the more necessary in the present case. As I have already mentioned, when a tax measure risks interfering with existing EU legal regulatory frameworks – as is the case with the risk tax, regardless of how its objective is defined – its implications for their proper functioning must be carefully considered. While taxation aimed at the eventuality that the preventive policies established by EU law, such as those under the EU banking regulatory framework, fail to avoid a financial crisis is not prohibited per se, the exercise of Member States’ fiscal competence in this domain must nonetheless be examined in the light of the regulatory context in which it is embedded. 91.      In the second place, regarding the General Court’s statement that the risk tax was not intended to prevent or address systemic risk, it suffices to note that an asymmetric tax such as the risk tax, imposing a fiscal burden on only a small number of (systemically important/large) credit institutions on the basis of the level of their liabilities, may encourage liable credit institutions to keep that level low. 92.      In the third place, as regards paragraph 59 of the judgment under appeal, I note that the General Court merely replicated what is mentioned in section 5.1 of the draft law where the Swedish legislature illustrated a typical scenario of financial interconnectedness between institutions and its negative impact on financial stability. However, it is not clear how, on the sole basis of that scenario, the General Court came to the conclusion that, the level of liabilities was a reliable metric in order to ‘distinguish between credit institutions according to whether they have a greater or lesser impact on the financial system’. The tax base of the risk tax as specified in the draft law does not explicitly take into account the level of interconnectedness through bonds. Moreover, it does not appear from the draft law, as the appellant correctly observes, that the Swedish legislature relied on the premiss that the credit institutions subject to the risk tax owned each other’s bonds to any significant extent. Lastly, the Commission itself did not rely on interconnectedness in order to assess the tax base in the decision at issue. 93.      In the light of the foregoing, I am of the view that the first complaint in the second limb of the second ground of appeal should be upheld. (2)    The persons liable to the risk tax 94.      The second complaint in the second limb of the second ground of appeal is directed against paragraphs 73 to 82 of the judgment under appeal, where the General Court concluded that the appellant had not established that the Commission should have experienced serious difficulties in the assessment of the persons liable to the risk tax. 95.      The appellant raises four claims in support of its complaint. 96.      First, the appellant challenges paragraph 73 of that judgment, in which the General Court rejected the argument of the applicants at first instance that there was not a perfect correlation between the list of credit institutions subject to the risk tax and the list of institutions of systemic importance identified by the Riksgäldskontoret (National Debt Office, Sweden). The appellant points out that nine credit institutions are subject to the risk tax, of which two are not deemed systemically important for the purpose of the resolution framework as transposed in Sweden, and that three banks which are deemed systemically important, are not subject to the risk tax. According to the appellant, those inconsistencies between the two lists, of which the Commission was aware during the preliminary procedure, clearly show the incoherence between the persons liable to the risk tax and the objective of the tax, as defined by the General Court. 97.      Contrary to what the Commission argues, that claim is not purely factual, but calls into question a legal characterisation of the facts by the General Court and is therefore admissible. 98.      In paragraphs 73 to 75 of the judgment under appeal, the General Court found, in essence, that the two lists of, respectively, systemically important institutions under the BRRD and credit institutions liable to the risk tax were drawn up on the basis of different criteria. The General Court also emphasised the differences between the aims pursued by the risk tax and those pursued by the EU banking regulatory framework. In particular, it pointed out that the purpose of the BRRD was to minimise significant direct costs and not indirect costs and that the CRR addressed direct costs by ensuring that credit institutions do not fail. As regards the DGSD, the General Court stated that its aim was to avoid depositors losing their deposits in the event that credit institutions fail. Lastly, the General Court considered that the stress tests carried out by the European Banking Authority (EBA) (55) in 2021, on which the applicants at first instance had relied, (56) concerned the risk of a credit institution’s failure. 99.      Given that the risk tax is based on a financial indicator (the credit institution’s debt) and aims to address, through taxation, the risks associated with financial crises – whether these relate to risks to financial stability or the risk of incurring indirect costs as a result of such crises – the General Court’s reasoning is, in my view, unsatisfactory. The fact that the two lists – one of credit institutions subject to that tax and the other of systemically important institutions under the EU banking regulatory framework – are drawn up on the basis of different criteria is self-evident and cannot justify the discrepancies highlighted by the applicants at first instance. That is all the more so since those discrepancies relate to the obligation to contribute to a fund specifically aimed at dealing with risks linked to a bank’s failure. (57) 100. Similarly, the point highlighted by the General Court in paragraph 74 of the judgment under appeal, namely that the risk tax does not apply to credit institutions representing a greater risk of failure but to credit institutions whose failure may give rise to significant indirect costs, is not decisive. The mere fact that the objectives of the risk tax and of the BRRD do not fully coincide – even if they are, at the very least, complementary – does not mean that potential interference can be ruled out without further inquiry, in a situation where there is a partial and asymmetrical overlap between the credit institutions subject to contributions under those two instruments. 101. Moreover, the references made by the General Court to the EU banking regulatory framework appear partial. For instance, when referring to the consequences caused by the failure of a bank, the BRRD and the CRR do not specifically refer to direct or indirect costs. As for the DGSD, it expressly refers, in recital 3 to the ‘costs of the failure of a credit institution to the economy as a whole …’. It is not obvious that the purpose of those legal instruments is limited ‘to minimis[ing] significant direct costs for society’, as the General Court stated in paragraphs 74 and 75 of the judgment under appeal, especially considering the vague definition of ‘indirect costs’ in the draft law. 102. In these circumstances, the Kingdom of Sweden's assertion that the draft law clearly states that the risk tax proposal is not intended to replace the current EU regulatory framework is of little consequence. The relevant EU bank resolution and prudential requirements frameworks were part of the national reference system that the General Court should have taken into account in order to review the Commission’s assessment of the coherence of the choice made by the Swedish legislature as regards the persons liable to the risk tax. 103. The first claim raised by the appellant is therefore, in my view, well founded. 104. Second, the appellant submits that the General Court erred in law in paragraph 79 of the judgment under appeal when it rejected the argument of the applicants at first instance that one of the credit institutions liable to the tax lends only to Swedish municipalities and thus does not give rise to any indirect costs. The General Court noted that, according to the case-law, in the case of an aid scheme, the Commission may merely consider the general characteristics of the scheme at issue, without being required to examine each particular case in which that scheme applies. 105. Such an argument pertains, in essence, to the intensity of the analysis carried out by the Commission. It is certainly true that, according to the Court’s case-law (58), the Commission is not required to conduct an analysis of the aid granted in individual cases under an aid scheme. However, one might wonder whether that case-law, which concerned cases involving fiscal schemes applicable to hundreds or thousands of taxpayers, is relevant to the present case. The risk tax applies to nine credit institutions. It appears that the Commission was aware of the very limited number of liable persons during the preliminary investigation procedure (or in any case, it should have been aware), most of which are also covered by the EU banking regulatory framework. I observe, however, that the appellant does not seem to challenge the finding of the General Court, at the end of paragraph 79 of the judgment under appeal, according to which the applicants at first instance had not disputed the arguments of the Commission and of the Kingdom of Sweden that the credit institution at issue was not completely exempted from any risk of failure. The second claim raised by the appellant could therefore be ineffective. 106. Third, the appellant challenges paragraph 80 of the judgment under appeal, in which the General Court rejected the argument of the applicants at first instance that, since all credit institutions generate indirect costs, a distinction between credit institutions for the purposes of liability to the tax was not mandated. 107. In that respect also, the reasoning of the General Court appears unsatisfactory. (59) The risk tax was introduced to strengthen public finances by increasing fiscal revenue to cope with indirect costs that future financial crises may entail. In principle, all credit institutions likely to generate such costs are concerned. Exempting some of them from taxation appears contradictory and reflects the intention of the Swedish legislature to place the burden of all indirect costs that might be generated by future financial crises on a select few institutions. Such a differentiation between credit institutions – pointed out by the applicants at first instance and bypassed by the General Court by referring alternatively to the absence of systemic risk posed by untaxed credit institutions or to the capacity of taxed institutions to generate significant indirect costs – raises serious concerns about the selectivity of the risk tax and needed, in my view, to be carefully investigated. 108. In paragraph 80 of the judgment under appeal, the General Court, referring to the judgment of 26 April 2018, ANGED, (60) added that ‘it cannot be disputed that the impact on the financial system of credit institutions depends largely on the size of those institutions and the level of their liabilities’. In that regard, I note, first of all, that the General Court referred to the ‘size’ of credit institutions and the ‘level of their liabilities’ as two distinct criteria on which the Swedish legislature relied. However, the ‘size’ of the ‘large credit institutions’ subject to the tax is determined solely by the level of their liabilities. That level is therefore the sole indicator considered. As regards the reference by the General Court to the judgment in ANGED I, I must confess that I find the analogy between the environmental impact of retail establishments and the risk posed by the level of liabilities of credit institutions to the financial system and the whole economy to be rather far-fetched. That being said, in the case giving rise to the judgment in ANGED I, the Court of Justice found that there was a direct link between the size of the sales area and the effects on the environment. (61) In the present case, such a direct link between the level of liabilities of credit institutions subject to the risk tax and their ability to create indirect costs to society is not clearly established, especially when the risk of those institutions failing is not taken into account. 109. For the above reasons, I am inclined to the view that the third claim raised by the appellant is well founded. 110. Lastly, the appellant challenges paragraph 81 of the judgment under appeal, in which the General Court found that the applicants at first instance had not called into question ‘the ability of large credit institutions alone to cause, on an individual basis, by their failure, a systemic risk, to have a very negative impact on the financial system and on the economy in general and to generate significant indirect costs for society’. The appellant contends, in essence, that it follows from the reasoning of the General Court that the Swedish legislature may impose a tax on systemic risk, whereby systemically important credit institutions are treated dissimilarly. 111. For the same reasons already discussed above, I am of the view that the appellant’s fourth claim should be upheld, regardless of how the objective of the risk tax is defined. 112. In the light of the above, the second complaint in the second limb of the second ground of appeal is well founded. (3)    The threshold 113. By the third complaint in the second limb of its second ground of appeal, the appellant argues that the General Court erred in law, in assessing, in paragraphs 89 to 98 of the judgment under appeal, the threshold for liability to the risk tax and concluding, in paragraph 99, in the light of the determined objective of the risk tax, that the applicants at first instance had not established that the Commission should have encountered serious difficulties in the assessment of that threshold. 114. In paragraphs 89 to 91 of the judgment under appeal, the General Court, relying on its own judgments (62) upheld by the Court of Justice in the judgments in Commission v Poland and in Commission v Hungary, held that the Kingdom of Sweden could not be prevented, on the one hand, from introducing a tax with a tax threshold and, on the other hand, from establishing a variation mechanism that goes as far as to exempt credit institutions below that threshold, provided that those elements do not run counter to the objective of the tax. Relying on paragraph 43 of the judgment of 26 April 2018, ANGED (63), the General Court also pointed out that the determination of the tax threshold and of the methods for calculating the basis of assessment comes within the discretion of the national legislature and is based, in addition, on technical, complex assessments that the EU Courts have only limited powers to review. On the basis of those principles, it examined whether the threshold fixed by the Swedish legislature ran counter to the objective of the risk tax or was manifestly discriminatory, and came to the conclusion that that was not the case. 115. The appellant challenges, in my view correctly, the relevance of the case-law relied on by the General Court and the correctness of the conclusion drawn by it. 116. I have already mentioned in paragraph 87 of the present Opinion that the fiscal measures at issue in the cases which led to the  judgments in Commission v Poland and Commission v Hungary judgments differed considerably from the risk tax. What was at stake in the cases giving rise to those judgments was the progressive nature of the fiscal measures in question and, more specifically the difference in the average tax rate resulting from the progressivity of the rates. The risk tax does not provide for any ‘variation mechanism’ once the threshold is exceeded and liability for the tax is triggered. 117. The fiscal measures at issue in the judgments in Commission v Poland and in Commission v Hungary provided for an exemption in respect of the portion of turnover below a certain amount. (64). However, all the undertakings concerned benefited from that exemption in respect of the part of their turnover which did not exceed the ceiling corresponding to the exempted band. The Commission had therefore not objected to that element of the fiscal measures in question. (65) In the same vein, in the judgments in ANGED I and II, the mechanism of the taxes provided for general exemptions, whereby no undertakings paid the tax under a certain threshold. (66) 118. On the contrary, in the case of the risk tax, there is no such a general exemption. The tax rate is applicable to the sum of all the liabilities of the credit institutions subject to the tax, and not only to the liabilities exceeding the threshold triggering the application of the tax. In other words, liability for the risk tax also extends below the dividing line between ‘large’ or ‘systemically important’ institutions that are subject to the tax and credit institutions whose size, in terms of their level of liabilities, justifies their exclusion from the tax. 119. That should have raised major concerns about the selective nature of the threshold and the methods used for calculating the taxable amount as well as their actual consistency with the objective of the risk, however it is defined. 120. For the sake of completeness, it is worth recalling that in the judgment in Prezydent Miasta Mielca, (67) the Court clarified that a general and abstract exemption to which a direct tax is subject cannot, as a general rule, confer a selective advantage for the purposes of Article 107(1) TFEU, in so far as it is presumed to be inherent in the ‘normal’ tax regime. However, ‘by way of derogation’ from that rule, ‘a general and abstract exemption to which a direct tax is subject cannot be regarded as falling within the “normal” tax regime where the conditions established by the relevant legislation for benefiting from that exemption are connected, in law or in fact, with one or more specific characteristics of the only category of undertakings capable of benefiting therefrom, those characteristics being inextricably linked to the nature of those undertakings or the nature of their activities’ (68). The fact that only such a consistent category of undertakings is capable of benefiting from a tax exemption is such as to substantiate the potentially discriminatory and anticompetitive nature of that exemption, even if the reference framework itself has not been configured according to manifestly discriminatory parameters. 121. Regardless of whether or not the clarifications provided by the judgment in Prezydent Miasta Mielca apply in cases of asymmetric taxes like the risk tax, in the present case the credit institutions not liable for the risk tax form part of a consistent category of undertakings identified on the basis of specific characteristics inherent in the nature of their activity (credit institutions with a level of liabilities below the applicable threshold) and fall within the derogation to the rule established by that judgment. 122. For the reasons set out above, the third complaint in the second limb of the second ground of appeal should in my view also be upheld. (4)    Conclusions on the second limb of the second ground of appeal 123. In the light of the foregoing reasons, I am of the view that the second limb of the second ground of appeal is well-founded. (b)    The third limb of the second ground of appeal 124. The appellant contends that the General Court erred in law in paragraphs 116 to 127 of the judgment under appeal when it did not find that the Commission should have encountered serious difficulties in the assessment of the existence of a derogation from the reference system. 125. In those paragraphs, the General Court considered that the applicants had not demonstrated the existence of a body of consistent evidence capable of showing that the credit institutions whose liabilities exceeded the threshold were, in the light of the objective of the tax as described in paragraph 41 of the judgment under appeal, in a factual and legal situation comparable to that of credit institutions whose liabilities did not exceed that threshold. 126. In support of that conclusion, the General Court relied almost exclusively on the grounds on which it had previously rejected the arguments of the applicants at first instance on the reference system. For that reason, in the context of the third limb of the second ground of appeal, the appellant reiterates its complaints challenging those grounds. 127. In those circumstances, I, in turn, refer to the reasons set out in points 78 to 122 above to support my conclusion that the General Court erred in finding that the different aspects of the risk tax’s reference system did not raise doubts concerning their discriminatory and therefore selective nature. As I have already mentioned, the overall aim pursued by the Swedish legislature – namely, to strengthen public finances in order to cope with the costs incurred by society in the event of future financial crises – does not appear to justify, without raising doubts, the imposition of an asymmetrical tax burden on credit institutions which, albeit to varying degrees, can all contribute to those costs. 128. That being said, it is worth noting that the reasoning of the General Court, namely in paragraph 119 of the judgment under appeal, where it dismissed the arguments of the applicants at first instance by referring to its own findings that the reference system of the risk tax had not been designed in a manifestly discriminatory manner, raises concerns, as it may affect the correct application of the second step of the selectivity test. 129. As I have already mentioned, the second step, in which the derogatory nature of the tax measure and its selectivity prima facie is assessed, is essential in the case of ad hoc taxation. In the judgment under appeal, it appears that such an assessment has been largely incorporated into the evaluation of the inherently selective nature of the relevant system, under the first step. However, first, this latter evaluation is based on the quantitatively different criterion of ‘manifest discrimination’. Second, the Court has made clear that even in cases where the sole purpose of a tax is to generate tax revenue intended to achieve a specific scope, the second step of the selectivity test has to be carried out in full. (69) 130. In the light of the foregoing, the third limb of the second ground of appeal should be upheld. (c)    Conclusions on the second ground of appeal 131. It follows from all the foregoing considerations that, in my view, the second ground of appeal must be upheld. 3.      The third ground of appeal 132. By its third ground of appeal, challenging paragraphs 40 to 43, 80 to 83, 93, 94 and 121 of the judgment under appeal, the appellant contends that the General Court failed to consider the arguments presented at first instance and therefore infringed its obligation to state reasons. 133. I note that in the context of an appeal, the purpose of review by the Court of Justice is, inter alia, to consider whether the General Court addressed, to the requisite legal standard, all the arguments raised by the appellant. A plea alleging that the General Court failed to respond to arguments relied on at first instance amounts essentially to pleading a breach of the obligation to state reasons. (70) 134. The appellant claims, first, that in paragraphs 42, 43, 81, 93 and 121 of the judgment under appeal, the General Court erroneously held that the applicants at first instance had not demonstrated that the failure of credit institutions whose total liabilities are below the threshold presented a systemic risk. The appellant further contends that the General Court failed to consider that the applicants at first instance had specifically given the example of the only bank in Sweden that was bailed out during the 2008 financial crisis, which had liabilities well below the threshold. They had already referred to the fact that three banks with liabilities below the threshold are deemed systemically important according to the EU banking regulatory framework. 135. In that respect, I note that, according to settled case-law the obligation to state reasons does not require the General Court to provide an account which follows exhaustively and one by one all the arguments put forward by the parties to the case. The General Court’s reasoning may therefore be implicit on condition that it enables the persons concerned to know why it has not upheld their arguments and provides the Court of Justice with sufficient material for it to exercise its power of review. (71) The mere fact that the General Court did not explicitly reject the arguments of the applicants at first instance referred to in point 134 above does not mean that it did not take those arguments into account in its assessment. By stating that the appellants had not demonstrated that the failure of credit institutions whose total liabilities are below the threshold presented a systemic risk, the General Court made clear, in the exercise of its exclusive jurisdiction to find and appraise the relevant facts and to examine the evidence, that the evidence submitted by the applicant at first instance was insufficient for such demonstration. That claim must therefore be rejected. 136. Second, the appellant argues that, in paragraph 83 of the judgment under appeal, the General Court misconstrued the argument put forward by the applicants at first instance, namely that large credit institutions would pay for all indirect costs in the event of an economic downturn. By not addressing that argument, the General Court failed to address the core argument that all banks would cause indirect costs, but only some would pay for them. 137. I am inclined to agree with the appellant that that argument played a central role in the line of reasoning of the applicants at first instance and that it was not properly addressed by the General Court. 138. Lastly, by incorrectly determining the objective of the risk tax, the General Court failed to correctly address several arguments put forward by the applicants at first instance and thereby infringed its duty to state reasons. Most importantly, as regards the persons liable to the tax and the threshold, the General Court failed in paragraphs 80, 81, 93, 94 and 121 to correctly assess the arguments that all credit institutions risk causing indirect costs to society, and therefore a distinction between large and small credit institutions is not mandated. 139. In that respect also, I am of the view that the appellant is right. The General Court’s errors in determining the objective of the risk tax influenced its entire reasoning, including how it assessed the relevance and merits of the arguments of the applicants at first instance and the reasons for rejecting them. 140. On the basis of the foregoing, the third ground of appeal is partly well founded. 4.      Conclusion on the appeal 141. For the reasons set out above, I conclude that the appeal must be upheld in its entirety and the judgment under appeal set aside. C.      The action before the General Court 142. In accordance with Article 61 of the Statute of the Court of Justice of the European Union, the Court of Justice may, where it has quashed the decision of the General Court, itself give final judgment in the matter, where the state of the proceedings so permits. 143. In that regard, I consider that, if the Court finds that the judgment under appeal should be set aside, it has all the information necessary to give a ruling on the single plea at first instance, alleging that, in the analysis of the selectivity of the risk tax, the Commission should have experienced serious difficulties. 144. It is apparent from the considerations set out in the analysis of the appeal that, in my view, that plea must be upheld and the decision at issue annulled. D.      Costs 145. According to Article 184(2) of the Rules of Procedure of the Court of Justice, where the appeal is well founded and the Court itself gives final judgment in the case, the Court is to make a decision as to the costs. 146. I propose that the Court order the Commission to bear its own costs and to pay the costs incurred by Ideella föreningen Svenska Bankföreningen med firma Svenska Bankföreningen, Näringsverksamhet relating to the present appeal and order the Kingdom of Sweden to bear its own costs, pursuant to Article 138(1) and Article 140(1) of the Rules of Procedure, which apply to the procedure on appeal by virtue of Article 184(1) of those rules. IV.    Conclusion 147. In the light of the foregoing considerations, I propose that the Court should: –        set aside the judgment of the General Court of the European Union of 17 April 2024, Svenska Bankföreningen and Länsförsäkringar Bank v Commission (T‑112/22, EU:T:2024:250); –        annul European Commission Decision COM(2021) 8637 final of 24 November 2021 on State aid SA.56348 (2021/N) – Sweden: Swedish tax on credit institutions; –        order the Commission to bear its own costs and to pay the costs incurred by Ideella föreningen Svenska Bankföreningen med firma Svenska Bankföreningen, Näringsverksamhet relating to the present appeal and order the Kingdom of Sweden to bear its own costs. 1      Original language: English. 2      Opinion of Advocate General Kokott in ANGED, C‑233/16, EU:C:2017:852, point 76. 3      Judgment of 2 July 1974, Italy v Commission (173/73, EU:C:1974:71, paragraph 13 of the English-language version). 4      T‑112/22, EU:T:2024:250. 5      Lagrådsremissen Riskskatt för kreditinstitut (Referral to the Council on Legislation relating to a risk tax on credit institutions; ‘the legislative proposal’), referred to in the judgment under appeal as the draft law. 6      Foreign credit institutions are to be liable to that risk tax if, at the beginning of the tax year, they have liabilities that are attributable to business operations carried out from a Swedish branch and the sum of those liabilities exceeds the threshold limit laid down by the draft law. 7      Council Regulation of 13 July 2015 laying down detailed rules for the application of Article 108 [TFEU] (OJ 2015 L 248, p. 9). 8      Judgments of 10 September 2024, Commission v Ireland and Others (C‑465/20 P, EU:C:2024:724, paragraph 73) and of 29 April 2025, Prezydent Miasta Mielca (C‑453/23, EU:C:2025:285, paragraph 41, ‘the judgment in Prezydent Miasta Mielca’). 9      Judgment of 22 September 2020, Austria v Commission (C‑594/18 P, EU:C:2020:742, paragraph 44 and the case-law cited). 10      The list is not exhaustive as it includes only the EU legislation relevant for the present case. 11      Directive 2014/59/EU of the European Parliament and of the Council of 15 May 2014 establishing a framework for the recovery and resolution of credit institutions and investment firms and amending Council Directive 82/891/EEC, and Directives 2001/24/EC, 2002/47/EC, 2004/25/EC, 2005/56/EC, 2007/36/EC, 2011/35/EU, 2012/30/EU and 2013/36/EU, and Regulations (EU) No 1093/2010 and (EU) No 648/2012, of the European Parliament and of the Council (OJ 2014 L 173, p. 190), as recently amended by Directive (EU) 2024/1179 of the European Parliament and of the Council of 11 April 2024 (OJ L, 2024/1174). 12      Regulation (EU) No 575/2013 of the European Parliament and of the Council of 26 June 2013 on prudential requirements for credit institutions and investment firms and amending Regulation (EU) No 648/2012 (OJ 2013 L 176, p. 1), as recently amended by Regulation (EU) 2025/2088 of the European Parliament and of the Council of 8 October 2025 (OJ L, 2025/2088). 13      Directive 2013/36/EU of the European Parliament and of the Council of 26 June 2013 on access to the activity of credit institutions and the prudential supervision of credit institutions and investment firms, amending Directive 2002/87/EC and repealing Directives 2006/48/EC and 2006/49/EC (OJ 2013 L 176, p. 338). 14      Directive 2014/49/EU of the European Parliament and of the Council of 16 April 2014 on deposit guarantee schemes (OJ 2014 L 173, p. 149). 15      Regulation of the European Parliament and of the Council of 15 July 2014 establishing uniform rules and a uniform procedure for the resolution of credit institutions and certain investment firms in the framework of a Single Resolution Mechanism and a Single Resolution Fund and amending Regulation (EU) No 1093/2010 (OJ 2014 L 225, p. 1) (‘the SRMR’). SRMR also applies in respect of banks established in Member States whose currency is not the euro which have established a close cooperation (see recital 15 of the SRMR). Sweden does not participate in the SRM. 16      As stated in recital 5 of the CRR. 17      See in particular recitals 32 and 72 of the CRR. 18      See recital 3 of the DGSD. 19      See recital 5 of the BRRD. 20      See, by analogy, judgment of 16 April 2026, Nitrogénművek (C‑519/24, EU:C:2026:297, paragraph 33). 21      See, inter alia, judgment of 8 September 2011, Paint Graphos and Others (C‑78/08 to C‑80/08, EU:C:2011:550, paragraph 49 and the case-law cited). 22      See, inter alia, judgment of 10 September 2024, Commission v Ireland and Others (C‑465/20 P, EU:C:2024:724, paragraph 76). 23      Judgment of 5 December 2023, Luxembourg and Others v Commission (C‑451/21 P and C‑454/21 P, EU:C:2023:948, paragraph 108). 24      Judgment of 6 October 2021, World Duty Free Group and Spain v Commission (C‑51/19 P and C‑64/19 P, EU:C:2021:793, paragraph 63). 25      Paragraph 49 in the judgment in Prezydent Miasta Mielca. 26      See judgments of 16 March 2021, Commission v Poland (C‑562/19 P, EU:C:2021:201, paragraphs 38 and 39) and of 16 March 2021, Commission v Hungary (C‑596/19 P, EU:C:2021:202, paragraphs 44 and 45). 27      See the judgment in Prezydent Miasta Mielca, paragraph 53, referring to judgment of 15 November 2011, Commission and Spain v Government of Gibraltar and United Kingdom (C‑106/09 P and C‑107/09 P, EU:C:2011:732). 28      See judgment of 5 July 2011, Edwin v OHIM (C‑263/09 P, EU:C:2011:452, paragraph 46). 29      See, to that effect, judgment of 8 November 2022, Fiat Chrysler Finance Europe v Commission (C‑885/19 P and C‑898/19 P, EU:C:2022:859, paragraph 82 and the case-law cited). 30      See judgment of 5 December 2023, ‘the judgment in Luxembourg and Others v Commission’, (C‑451/21 P and C‑454/21 P, EU:C:2023:948, paragraph 77). 31      C‑451/21 P and C‑454/21 P, EU:C:2023:948. 32      According to the second subparagraph of Article 256(1) TFEU, an appeal to the Court of Justice against decisions given by the General Court must be made ‘under the conditions and within the limits laid down by the Statute’. Article 58 of the Statute of the Court of Justice of the European Union, states that an appeal may be based on infringement of European Union law by the General Court; see paragraph 76 of the judgment in Luxembourg and Others v Commission. 33      See paragraph 58 of the judgment under appeal. 34      See recitals 49 and 50 of the decision at issue. 35      See recital 51 of the decision at issue. 36      See, for example, recital 68 of the decision at issue, in which the Commission held that financial institutions other than credit institutions have a lower capacity to generate systemic risks and, ultimately, significant indirect costs and are therefore not in a similar factual and legal situation to credit institutions in light of the objective of the risk tax. 37      This is the legislative proposal, produced by the appellant before the General Court. 38      Initially, the idea was to reduce the tax advantage that the financial sector could be assumed to gain as a result of the exemption of the supply of financial services from value added tax. 39      As concerns systemically important credit institutions, the draft law further explains that ‘it is difficult to determine exactly where the line should be drawn as to whether an institution is systemically important or not. Factors that play a role in this respect are the size of the institution, its significance for the national economy, its complexity and its interconnection with other institutions. Because players in the financial system hold each other’s bonds, a problem arising in one part of the system can quickly spread to other parts and, as a result, threaten financial stability’. 40      According to the draft law, those indirect costs may take the form of, for example, higher expenditure on unemployment insurance and other social security systems or lower tax revenues due to poorer economic performance. 41      The locution ‘systemically important’ to indicate the credit institutions liable to the proposed risk tax was used by some of the stakeholders consulted, whose opinions are reported in the draft law. 42      For instance, under the CRD ‘systemically important institutions’ are defined as ‘an EU parent institution, an EU parent financial holding company, an EU parent mixed financial holding company or an institution the failure or malfunction of which could lead to systemic risk’ (see point 30 of Article 3(1) (30) thereof). The same directive defines the ‘systemic risk’ as ‘a risk of disruption in the financial system with the potential to have serious negative consequences for the financial system and the real economy (see point 10 of Article 3(1) thereof). 43      In order to justify the reference to ‘systemic risk’ in the judgment under appeal, the Commission submits that large credit institutions present a ‘certain systemic risk’, that ‘the size of a credit institution and its overall systemic importance are interrelated’, that ‘a credit institution which holds a large portfolio of liabilities is usually also systemically important’ and that, when a credit institution is classified as large because of its total liabilities, ‘the aspects of “systemic importance” and “systemic risk” are never far away’. However, it clarifies that ‘large’ credit institutions are not necessarily ‘systemically important’ under bank resolution and prudential supervision frameworks. In turn, Sweden affirms, in a quite apodictical manner, that, regardless of the terms used by the General Court, the objective as determined by that court does not differ from the one identified by the Commission. 44      See, to that effect, inter alia, judgment of 22 December 2008, British Aggregates v Commission (C‑487/06 P, EU:C:2008:757, paragraph 113 and the case-law cited). 45      Judgments of 5 September 2024, Slovenia v Commission (C‑447/22 P, EU:C:2024:678, paragraphs 50 and 51) and of 6 October 2021, Scandlines Danmark and Scandlines Deutschland v Commission (C‑174/19 P and C‑175/19 P, EU:C:2021:801, paragraph 67 and the case-law cited). 46      In case of asymmetric tax liability, the aid may arise from the fact that competitors of the taxed economic operators are not subject to that tax, see to that effect judgment of 7 September 2006, Laboratoires Boiron (C‑526/04, EU:C:2006:528, paragraph 34). 47      Judgment of 6 November 2018, Scuola Elementare Maria Montessori v Commission, Commission v Scuola Elementare Maria Montessori and Commission v Ferracci (C‑622/16 P to C‑624/16 P, EU:C:2018:873, paragraph 86 and the case-law cited). 48      Pleas seeking to call into question the General Court’s understanding of the Commission’s contested decision are questions which raise points of law that may be subject to review by the Court of Justice in an appeal; see judgment of 10 September 2024, Commission v Ireland and Others (C‑465/20 P, EU:C:2024:724, paragraphs 111 and 112). 49      See recital 50 of the decision at issue. The Commission relied on the judgment of 16 March 2021, Commission v Hungary (C‑596/19 P, EU:C:2021:202, paragraph 48). 50      See paragraph 54 of the judgment under appeal. 51      C‑562/19 P, EU:C:2021:201, ‘the judgment in Commission v Poland’, paragraph 41. 52      C‑596/19 P, EU:C:2021:202, ‘the judgment in Commission v Hungary’, paragraph 47. 53      The Kingdom of Sweden explains, in its response, that it was decided that a flat-rate method would be applied in order avoid the need for the Skatteverket (Tax Agency, Sweden) to assess the risk posed by each credit institution at a given time. 54      Only credit institutions whose total liabilities are above the threshold are required to pay, while institutions whose total liabilities remains below the same threshold are exempted. See also points 116 and 117 below. 55      Regulation (EU) No 1093/2010 of the European Parliament and of the Council of 24 November 2010 establishing a European Supervisory Authority (European Banking Authority), amending Decision No 716/2009/EC and repealing Commission Decision 2009/78/EC (OJ 2010 L 331, p. 12) as recently amended by Regulation (EU) 2025/2088 of the European Parliament and of the Council of 8 October 2025 (OJ 2025 L 2088, p. 1). That regulation mandates the EBA to monitor and assess market developments in the area of its competence, including stress tests, and to develop criteria for the identification and measurement of systemic risk and an adequate stress-testing regime (see recital 43, Articles 23 and 32 thereof). 56      The applicants at first instance had submitted that the five Swedish banks which were subject to those tests obtained excellent results. 57      I refer to the resolution fund as provided in the BRRD. 58      Judgment of 4 March 2021, Commission v Fútbol Club Barcelona (C‑362/19 P, EU:C:2021:169, paragraph 65). 59      After recalling the objective of the risk tax as defined in paragraph 41 of the judgment under appeal, the General Court stated that, even if all credit institutions may give rise to certain indirect costs, in that they may contribute to generating those costs, this does not mean that, in the event of failure, all credit institutions are likely to give rise to the same consequences for the financial system and the economy in general. 60      C‑233/16, ‘the judgment in ANGED I’, EU:C:2018:280, paragraph 53. 61      According to the Court ‘the larger the sales area, the higher the attendance of the public, which results in greater adverse effects on the environment’; see paragraph 53 of the ANGED I judgment. 62      Judgment of 16 May 2019, Poland v Commission (T‑836/16 and T‑624/17, EU:T:2019:338, paragraph 89) and of 27 June 2019, Hungary v Commission, (T‑20/17, EU:T:2019:448, paragraph 101). 63      C‑236/16 and C‑237/16, EU:C:2018:291 (the ‘ANGED II judgment’). That judgment concerns a Spanish regional tax on large retail establishments similar to that at issue in the case which gave rise to the ANGED I judgment. 64      These were the ‘exemptions’ to which the General Court referred in the judgments Poland v Commission and Hungary v Commission as mentioned in footnote n. 62, relied on by the General Court in paragraph 89 of the judgment under appeal. 65      See judgments of 16 March 2021, Commission v Poland (C‑562/19 P, EU:C:2021:201, paragraph 35) and of 16 March 2021, Commission v Hungary (C‑596/19 P, EU:C:2021:202, paragraph 41). 66      The first 2 000 m2 or 2 500 m2 were not taxed. Therefore, no undertakings paid the tax on areas below those thresholds; see paragraph 8 of both judgments. Some establishments pursuing specific businesses were also exempted. 67      Paragraph 50. 68      Ibidem, paragraph 54. 69      See judgments of 4 June 2015, Kernkraftwerke Lippe-Ems (C‑5/14, EU:C:2015:354, paragraphs 78 and 79) and of 7 November 2019, UNESA and Others (C‑105/18 to C‑113/18, EU:C:2019:935, paragraphs 64 to 67). 70      Judgment of 28 September 2023, Changmao Biochemical Engineering v Commission (C‑123/21 P, EU:C:2023:708, paragraph 185). 71      Judgment of 13 February 2025, Commission and Others v Carpatair (C‑244/23 P to C‑246/23 P, EU:C:2025:87, paragraph 67 and the case-law cited).