Progressive turnover-based tax systems and EU state aid law: Case C-562/19 P Commission v. Poland and Case C-596/19 P Commission v. Hungary

In Commission v. Poland (C-562/19) and Commission v. Hungary (C-596/19) the Court of Justice of the European Union ruled that progressive tax systems based on turnover do not by definition provide selective advantages to undertakings with lower turnovers in violation of EU state aid law. The European Commission had declared a Polish tax on retailers and a Hungarian tax on advertisement incompatible with Article 107(1) TFEU because the progressive, turnover-based taxes favoured undertakings with smaller turnovers over those with larger turnovers. The General Court annulled both Commission decisions because such advantages were inherent to the content and objectives of the general tax system, which was for Poland and Hungary to define. The Court of Justice dismissed the appeals by the Commission, affirming that Member States are free, in line with their fiscal autonomy, to opt for a progressive and/or turnover-based tax system. While turnover-based corporate taxation may have market-distortive effects, the Court was right to dismiss the Commission's appeals. The principles of fiscal autonomy and legal certainty require an assessment of selectivity in light of Member States’ own definition of the content and objectives of their tax systems.


Introduction
Does the progressive taxation of companies based on their turnover provide a selective advantage to companies with a smaller turnover in violation of EU state aid law? This is the central question of two judgments of the Court of Justice of the European Union (CJEU) concerning a Polish tax on the retail goods sector and a Hungarian tax on advertisements. 1 The European Commission had concluded that both tax measures constituted state aid incompatible with the internal market. The General Court (GC), however, annulled both decisions on the ground that the Commission had erred in its application of Article 107(1) TFEU. 2 On appeal by the Commission, the CJEU upheld the GC's judgment.
These judgments provide an important clarification of the application of EU state aid law to fiscal measures. The CJEU confirmed that the selectivity of a tax measurewhich is central to the question of whether a measure constitutes state aidmust be assessed in light of the 'normal' tax system applicable in the Member State. It is principally for the Member States to decide on the characteristics of their 'normal' tax system in accordance with their fiscal autonomy. To this end, the Court concluded that only manifestly discriminatory tax systems exceed Member States' discretion. This leaves Member States with more leeway to emphasize, for instance, redistributive purposes in their tax systems, even if they result in de facto advantages to certain undertakings.
In this case note, we first introduce the relationship between progressive, in particular turnoverbased tax systems and EU state aid law. We then discuss the factual and legal background of both cases, the decisions of the Commission and the GC, the Opinions of Advocate General Kokott, and the judgments of the CJEU, after which we will offer some comments on the reasoning of the Court and its legal implications.

Progressive tax systems based on turnover and EU state aid law
A. Progressive turnover-based tax systems and EU law A progressive tax system is a system under which tax liability increases as the taxable amount increases. Typically, progressive tax systems operate by segmenting taxable income into tax brackets that progress from lower to successively higher rates. It has, however, become increasingly more common for states inside and outside the EU to implement progressive tax rates based on the turnover of undertakings rather than on their taxable income. 3 Recent initiatives include a Commission proposal for a turnover-based Digital Services Tax. 4 This turnover-based approach deviates from the original philosophy of progressive taxation, since a higher turnover does not necessarily translate to higher taxable income and, therefore, a higher ability to pay. Turnover-based progressive tax systems may create a number of problems with regard to EU law. As Szudoczky and Károlyi note, turnover-based tax measures in general are imposed in certain sectors, thereby potentially favouring undertakings in other, adjacent sectors of the economy. 5 More specifically, progressive turnover-based tax systems favour undertakings with smaller turnovers over undertakings with larger turnovers, which may distort competition among them. 6 Such systems may also be designed to benefit specific companies over others. 7 In two earlier cases, Vodafone Magyarország 8 and Tesco-Global Áruházak, 9 the CJEU was asked to rule on the compatibility of two Hungarian progressive, turnover-based taxes on telecommunications operators and retailers with EU state aid law and the freedom of establishment. The complainants had argued that the progressive scale of these taxes was discriminatory because almost all companies that fell within the highest tax brackets were foreign companies. 10 While the Court declared the preliminary reference on state aid inadmissible, it noted in the context of the freedom of establishment that 'Member States are free, given the current state of harmonisation of EU tax law, to establish the system of taxation that they deem the most appropriate'. 11 Furthermore, it held that '[t]he fact that the greater part of such a special tax is borne by taxable persons owned by natural persons or legal persons of other Member States cannot be such as to merit, by itself, categorisation as discrimination'. 12 Commission v. Poland and Commission v. Hungary confirm the applicability of these principles to EU state aid law. Before turning to the factual and legal background of these cases, we will briefly outline the EU state aid framework and the selectivity criterion as it applies to national fiscal measures.

B. EU state aid law and the selectivity criterion in respect of tax measures
In order for a national measure to constitute state aid in the sense of Article 107(1) TFEU, the measure must (i) involve the grant of an economic advantage, financed by the State or through State resources; (ii) affect trade between Member States; (iii) confer a selective advantage on certain undertakings or the production of certain goods; and (iv) distort or threaten to distort competition. 13 The criterion that the measure confers a selective advantage on certain undertakings not only refers to measures that are formally reserved for certain undertakings (de jure selectivity), but also includes measures that are de facto selective, because conditions or barriers imposed by Member States prevent certain undertakings from benefiting from an otherwise general measure. 14 Recent years have seen a growing body of case law on the possible selectivity of national tax measures that appear general in nature but may de facto be more advantageous to certain economic operators. 15 In accordance with case law and soft law, the selectivity of tax measures is assessed on the basis of a three-step test. 16 The first step is the identification of the 'reference system' or 'normal' tax system, which the Commission defines as 'a consistent set of rules that generally applyon the basis of objective criteriato all undertakings falling within its scope as defined by its objective'. 17 In a second step, the prima facie selectivity is assessed, by determining whether there is a differentiation between economic operators that are in a comparable factual and legal situation in light of the objective of the reference system. 18 Thirdly, a prima facie selective measure can be justified in the light of the nature or general scheme of the reference system. 19 The identification of the reference system has proved to be a crucial part of this analysis. If the 'normal' tax system is defined as a flat-rate tax, progressive rates of the system or functional equivalents such as minimum thresholds for taxable amounts will provide prima facie selective advantages to certain undertakings. If, by contrast, the 'normal' tax system is defined as a progressive tax system, progressivity will not be selective even though the tax system de facto favours those falling within the lower tax rate brackets. The extent of the discretion of Member States to define their 'reference system' is central to both Commission v. Poland and Commission v. Hungary.

A. Commission v. Poland
Commission v. Poland centres on a Polish progressive tax on the retail sector based on turnover, which Poland adopted on 6 July 2016. 20 According to this measure, all retailers, regardless of legal status, were to pay this newly introduced tax, with taxation ranging from 0% on monthly turnover under 17 million złoty (PLN), 0.8% on the portion of monthly turnover between PLN 17 million and PLN 170 million, and 1.4% on the monthly turnover exceeding PLN 170 million. 21 The Commission concluded that the progressive tax rate structure was selective because it departed from the objective of the reference system, which according to the Commission was to tax the turnover of all retail operators. 22 Consequently, according to the Commission 'all undertakings deriving turnover from the sale of goods to natural persons should be considered to be in a similar factual and legal situation'. 23 The progressive nature of the tax system was not deemed part of the reference system. 24 According to the Commission, Poland had exceeded its discretion to define the content and purpose of its 'normal' tax system because the progressive tax rate structure was 'specifically designed to favour smaller retailers over larger ones by applying different tax rates […] thereby subjecting undertakings with lower turnover to a lower average effective tax rate than undertakings with a higher turnover, which also tend to be foreign-owned'. 25 In defining the reference system differently from Poland itself, the Commission relied on the CJEU's Gibraltar judgment. That case dealt with a proposal by the government of Gibraltar to replace the taxation of company profits with a payroll tax and a business property occupation tax. Although the two new tax rules were generally applicable, the Court held that although 'a different tax burden resulting from the application of a "general" tax regime is not sufficient on its own to establish the selectivity of taxation': 26 the fact that offshore companies are not taxed is not a random consequence of the regime at issue, but the inevitable consequence of the fact that the bases of assessment are specifically designed so that offshore companies, which by their nature have no employees and do not occupy business premises, have no tax base under the bases of assessment adopted in the proposed tax reform. 27 According to the Commission, the Gibraltar judgment allowed it to define the 'real' reference system of the Polish tax system. 28 In light of this reference system, it is clear that the progressive tax structure favours retailers with lower turnovers over retailers with larger turnovers. The selective nature of the progressive tax structure is inevitable once the appropriate reference system is defined as a single rate tax. 29 Poland appealed the Commission's final decision declaring the tax incompatible with the internal market. 30 It had also appealed the Commission's earlier decision to initiate the formal investigation procedure laid down in Article 108(2) TFEU. 31 Such a decision must be taken if the Commission, after a preliminary examination, is not convinced that the measure is compatible with the internal market. 32 In this decision, the Commission had also ordered Poland to suspend the tax. 33 The appeals were joined for the purpose of the oral procedure as well as the decision. 34 The GC annulled both Commission decisions. With respect to the final decision declaring the advertisement tax incompatible with the internal market, the GC held that the Commission had erred in the identification of the 'normal' tax system and in proving selectivity. It disagreed with the Commission's claims that the 'normal' tax system would consist of a flat-rate tax, and that the objective of the Polish tax was to tax the turnover of all undertakings in the relevant 24 sector. 35 In this regard, the GC simply concluded that '[t]here is nothing in the file to indicate that the Polish legislature had that intention', 36 and that the Commission erred by 'selecting an objective of the tax on retail trade that was different to the one put forward by the Polish authorities'. 37 The Polish law clearly had a redistributive purpose in addition to the purpose of taxing turnover in general. 38 Regardless of the Commission's error in identifying this reference system, a redistributive purpose could in any event justify the progressivity of a turnover tax. 39 As regards the decision to initiate the formal investigation procedure, the GC emphasized that such a decision can be justified as long as the initial classification of a measure as state aid is the result of legitimate doubts. 40 Judicial review is limited to the question of whether the Commission made a manifest error of assessment. 41 However, the information available to the Commission at the time of the decision to initiate the procedure made clear that the classification as state aid had to be dismissed. 42 According to the GC, the Commission had based its decision on the principled view that progressive, turnover-based taxation is inherently selective, and not on the basis of legitimate doubts about the legality of the advertisement tax in light of the available evidence. 43

B. Commission v. Hungary
Commission v. Hungary dealt with the Hungarian Advertisement Tax Act, pursuant to which broadcasters or publishers of advertisements, including print media and audiovisual media, were taxed at a progressive rate on annual turnover. 44 The initial law entered into force on 15 August 2014 and comprised six progressive brackets with tax rates between 0% (for the part of the turnover below 0.5 billion forint (HUF)) and 40% (for the part of the turnover above HUF 20 billion). 45 Furthermore, under the Tax Advertisement Act, companies could deduct from their 2014 tax base 50% of their losses carried forward from previous years, provided that they did not make a profit in 2013. 46 According to then Vice President of the Commission Neelie Kroes, the tax clearly targeted at international media conglomerate RTL, which would be the only company facing the highest tax rate of 40%. 47 RTL indeed filed a complaint at the European Commission and started reporting extensively on alleged corruption within Fidesz and the Orbán government. 48  Advertisement Tax Act was amended and the six progressive brackets were replaced with a twobracket system comprising a 0% rate on the part of the turnover below HUF 100 million and a 5.3% rate on the part of the turnover exceeding HUF 100 million. 50 The Commission identified the reference system as 'a special advertisement tax on turnover derived from the provision of advertising services, i.e. the full remuneration received by publishers for the publication of advertisements, without deduction of any costs'. 51 It concluded that 'the progressive tax structure introduced by the Act appears deliberately designed by Hungary to favour certain undertakings over others'. 52 Once again, the Commission referred to the Gibraltar judgment, 53 noting that selectivity could not be justified by the progressive nature of the tax system and its redistributive purpose, because turnover is not a good proxy for the ability to pay of undertakings. 54 As regards the deduction of losses carried forward, the Commission concluded that this artificially distinguished between undertakings which are in a comparable legal and factual situation, namely undertakings which made a profit in 2013 and undertakings which did not make a profit in that year. 55 The Commission separately discussed the consequences of the 2015 amendment which substituted the six-bracket system with a dual-rate system. It concluded that the 0% rate on turnover below HUF 100 million continued to provide an exemption for companies with a turnover lower than that amount, and progressivity for companies with a higher turnover. According to the Commission, the system therefore remained selective and unjustified. 56 On appeal, the GC annulled the Commission's decision with a reasoning largely similar to the one in the Polish case. It held that the selectivity criterion must be analysed in light of the objective of the 'normal' taxation system, 57 that this normal taxation system must be determined on the basis of the objectives of the national tax system itself, 58 and that the redistributive aspect of the Hungarian progressive tax was an inherent part of the system as such. 59 The GC emphasized that advantages for certain undertakings which result from the straightforward application of the reference system are not selective advantages for the purpose of EU state aid law.

C. The Opinions of AG Kokott
The Commission appealed both of the GC's judgments. AG Kokott, who had also been assigned the abovementioned cases Vodafone and Tesco-Global, delivered two separate but similar Opinions, in which she advised the Court to dismiss the Commission's appeals. AG Kokott's Opinions took a slightly different angle than the GC's judgments, in that her analysis regarding selectivity relied more explicitly on the fiscal autonomy of the Member States.
To this end, she recalled that Vodafone and Tesco-Global had confirmed that the freedom of establishment does not preclude progressive taxation based on turnover. 61 AG Kokott argued that the same principles apply in the context of state aid law as well, 62 in line with her analysis in her Opinions in Vodafone and Tesco-Global. 63 Consequently, she dismissed the Commission's arguments about the progressive nature of the system on the simple ground that there is no rule of EU law that prevents Member States from favouring a progressive system over a flat-rate system. 64 AG Kokott subsequently argued that the Gibraltar judgment merely provides an exception to this general principle. Gibraltar advanced a review of the consistency of the national tax system in order to prevent Member States 'from abusing their general tax law in order to grant advantages to individual undertakings in circumvention of the rules on State aid'. 65 According to AG Kokott, the Commission had not demonstrated such an abuse in the present cases.

The judgments of the Court of Justice
The Court of Justice followed the Opinions of AG Kokott and dismissed the Commission's appeals. It held that Member States have discretion to establish a tax system they deem appropriate in accordance with their fiscal autonomy. 66 According to well-established case law, however, it noted that such 'retained powers' of the Member States must be exercised in accordance with their obligations under EU law. 67 EU law does not, however, preclude either progressive tax systems in general or turnover-based progressive tax systems specifically. The CJEU held to this end that the fact that progressive tax rates are common in the taxation of natural persons does not mean that EU state aid law precludes progressive taxation of legal persons. 68 EU state aid law also does not prohibit progressive taxation from being based on turnover because turnover is 'a criterion of differentiation that is neutral and a relevant indicator of the taxable person's ability to pay'. 69 The Court rejected the Commission's statement that profits are a more relevant or more precise indicator than turnover, since EU state aid law is not concerned with the 'best' way to design a tax system as long as the system does not provide a selective advantage to certain undertakings. 70 Accordingly, the Court concluded that it was for Poland and Hungary to establish the 'normal' tax regime which constitutes the reference system for the purpose of state aid law. 71 The choice for a particular reference system and its characteristics -flat or progressive, turnover-based, or profitbasedis subject to judicial review, but only incompatible with Article 107(1) TFEU if it is 'designed in a manifestly discriminatory manner, with the aim of circumventing the requirements of EU law on State aid'. 72 In the Hungarian case, the Court held that the progressive nature of the tax system was an inherent part of the reference system and not a deviation from a 5.3% flat-rate turnover tax. 73 The Court distinguished the case from Gibraltar on the ground that the tax system in the latter case was manifestly discriminatory by design because the tax criteria favoured certain offshore companies inconsistently with the objective of creating a general tax. 74 The Commission had not demonstrated such discriminatory design in this case. 75 Whether Hungary's initial six-band system with rates between 0% and 50% would also have survived the Court's 'manifest discrimination' test remains unclear, a point to which we will return below.
Regarding the partial deductibility of losses carried forward, the CJEU concluded that that mechanism centred on the distinction between undertakings which did not make a profit in 2013 and which could deduct losses carried forward from previous years, and undertakings which did make a profit in 2013. 76 According to the CJEU, this distinction is an objective one in light of the advertisement tax's redistributive purpose. 77 Since undertakings making a loss in 2013 had a lesser ability to pay in 2014, the Hungarian legislature was allowed to draw this distinction even though the undertakings which would benefit from this mechanism could already be identified by the time the tax law was introduced. 78 In the Polish case, the CJEU concluded similarly that the Commission had not demonstrated that the three-rate system of 0%, 0.8% and 1.4% was manifestly discriminatory. 79 The progressive scale of the tax measure was an integral part of the reference system and its redistributive purpose. 80 By contrast, the Commission had wrongly relied on an 'incomplete and fictitious reference system'. 81 This conclusion dismantled the Commission's entire case because if progressivity is deemed an integral part of the reference system, the Polish retail tax cannot be deemed even prima facie selective. 82 The Commission had also appealed the GC's annulment of the former's decision to initiate the formal investigation procedure pursuant Article 108(2) TFEU. According to the Commission, the GC had wrongly reviewed the decision to initiate the formal investigation procedure to the same degree as in respect of the final decision, instead of applying a limited review in respect of a manifest error of assessment. 83 However, the CJEU ruled that the GC had not erred in law by concluding that the Commission was required to undertake a sufficient examination of the information provided by the Member State before initiating a formal investigation procedure, particularly if the Member State contends that the measure does not constitute aid. 84 The GC was right to conclude that the Commission's preliminary conclusions as to the selectivity of the measure were manifestly incorrect. 85

Comment
As noted in the introduction, progressive, turnover-based taxation can have several marketdistorting effects, for instance because it favours undertakings with smaller turnovers over undertakings with larger turnovers. 86 Furthermore, as regards turnover taxation of digital companies, the OECD concluded in a recent report that such taxation is 'likely to generate some economic distortions, double taxation, increased uncertainty and complexity, and associated compliance costs for businesses operating cross-border'. 87 From a legal perspective, however, the GC and the CJEU in our view rightly decided in favour of Poland and Hungary because it is for the Member States to decide on the appropriate reference system and its economic merits.
Consequently, the judgments demonstrate how fiscal autonomy and legal form significantly constrain the ability of state aid law to remove market distortions caused by fiscal measures. The remainder of our comments below will focus on two dimensions of the judgments and underlying issues: firstly, the choice for the appropriate reference system in light of fiscal autonomy, and secondly, the relevance of legal certainty in state aid cases in light of the possible under-inclusiveness of the three-step framework to assess selectivity.
A. The appropriate reference system and fiscal autonomy That fact that turnover-based progressive tax systems can have market-distorting effects is problematic from an effects-based perspective to Article 107 TFEU insofar as this provision aims to create a level playing field among competitors. 88 Furthermore, the Commission aims to employ 'a refined economic approach […] to ensure a proper and more transparent evaluation of the distortions to competition and trade associated with state aid measures'. 89  measures cannot, by themselves, determine their selective nature. Given the three-step test of the selectivity criterion, the 'reference system' is defined by the content and objectives of the national tax system itself. If that tax system is generally and inherently progressive, it is not selective. This is expressly confirmed by the present judgments, where the CJEU observed that tax advantages 'resulting from a general measure applicable without distinction to all economic operators does not constitute [state] aid'. 90 In both Commission v. Poland and Commission v. Hungary, the Commission's arguments centre on the distortive effects of the progressive tax rate. This suggests that the Commission applied a 'backwards' reasoning in respect of the three-step test: since the progressive tax rate advantaged smaller undertakings with lower turnovers and disadvantaged larger undertakings with higher turnovers, the tax system was selective. Starting with the conclusion, however, required the Commission to second-guess the 'best' or 'normal' tax system in respect of taxation of companies, that is, a flat-rate system. It also required the Commission to claim that certain economic theories allegedly prove that progressive tax rates are unsuitable for taxing legal persons, 91 and that turnover is not a good proxy for ability to pay. 92 We agree with the CJEU that this reasoning is legally unconvincing. Nothing in EU law requires Member States to opt for a flat-rate tax as opposed to a progressive rate system. Likewise, nothing in EU law requires Member States to opt for the 'best' proxy of ability to pay. Moreover, as AG Kokott noted, the Commission's proposal for a Digital Services Tax adopts a turnover-based, progressive approach as well, which further weakened the Commission's arguments against Poland and Hungary. 93 While the Commission relied heavily on the Gibraltar judgment, the analogy is not very convincing, at least in the Polish case, because the Gibraltar tax reform involved a tax which was obviously selective by design. The Polish tax system, by contrast, involved a straightforward example of a progressive tax system, which is increasingly common and reflects arguably legitimate political choices.
As regards the Hungarian case, the two-rate system as such also seems far removed from a Gibraltar-type, 'selective-by-design' tax system. However, the fact that companies could only deduct losses carried forward if they did not make a profit in 2013 seems more suspicious insofar as RTL, allegedly the main target of the initial, six-bracket tax proposal, precisely did not make a loss in that year. This raises the suspicion that the limited possibility to deduct losses carried forward in 2014 was designed to exclude RTL, especially in the context of the initial sixbracket tax proposal in which RTL would pay for 80% of all tax revenue. 94 Nevertheless, the CJEU ruled that the fact that the undertakings which would (not) benefit from the deductibility could be identified from the start, did not 'in itself' make the measure selective. 95  is understandable: the opposite conclusion would make all tax measures for which it is possible to identify the companies that de facto benefit from them prima facie selective. On the other hand, this part of the judgment arguably facilitates Member States to design their tax systems in such a way that they can de facto favour certain undertakings that are more likely to benefit from the 'normal' tax system than others.
The key question left unanswered by the Hungarian case is whether the initial proposal for a sixbracket progressive tax also would have survived the CJEU's scrutiny. In our view it is more likely that that tax measure qualifies as manifestly discriminatory and, therefore, a selective measure granting unlawful state aid. This is because of the steep progressivity of that proposal, the fact that RTL would be the only undertaking falling in the 40% top tax rate, and the increase of that top rate to 50% after RTL's negative media coverage about the Hungarian government and the Fidesz party.
However, while the CJEU did not discuss this earlier proposal, the GC did reject the Commission's conclusion that the selective nature of the initial proposal could be inferred from the fact that only one undertaking was subject to the 30% and 40% tax brackets, and paid approximately 80% of the tax advances. 96 The GC concluded that the Commission's reasoning was solely based on the inherent nature of progressive taxation. 97 Furthermore, the CJEU's reasoning does seem to entail a robust presumption in favour of the legality of progressive, turnover-based taxation, which may be difficult to rebut. 98 A successful case against a steeply progressive tax system such as Hungary's initial proposal would therefore require strong emphasis, supported by concrete data, on the manifest discrimination inherent to the very design of the tax system. Mere distortive effects will, in any case, not suffice.
In general, the CJEU's deference to Member States in defining the reference system is convincing in light of the principle of fiscal autonomy, which the Court expressly invoked to justify its conclusions. This approach also fits the assessment of national tax measures under EU internal market law. While tax measures are subject to the free movement rules, the free movement of goods grants Member States a large margin of discretion in regard to determining their own taxation policies as long as they do not directly or indirectly discriminate against imports. 99 In recent case law on the free movement of services and capital as well as the freedom of establishment, the Court has also been reluctant to conceive of non-discriminatory tax measures as restrictions to free movement even though they are seemingly capable of hindering trade. 100 This deferential approach is also in line with unanimity requirements to harmonization of indirect taxes, 101 and direct taxes. 102 According to well-established case law, Member States should exercise their retained powers in accordance with their obligations under EU law. 103 Accordingly, fiscal autonomy obviously does not mean that tax measures are exempt from state aid law. The application of the selectivity criterion in EU state aid law to fiscal measures can, however, defer to political choices of the Member States. The CJEU's cautious approach with regard to progressive taxation and redistributive purposes is, in our view, responsible and legally convincing.
B. The relevance of legal certainty and the limits of EU state aid law The effects-based approach used by the Commission to infer the selectivity of the Polish and Hungarian tax measures from the advantages they generated for certain undertakings also would have likely resulted in considerable legal uncertainty concerning the legality of national tax measures. The way in which the Commission reasoned from the effects of the progressive tax system towards establishing its selectivity is problematic in this regard for two reasons.
Firstly, the Commission second-guessed the reference system of the Polish and Hungarian tax systems by arguing that a flat-rate system would be the 'normal' manner of corporate taxation. Extrapolating this approach would mean that the Commission, by relying on (some) economic theory, 104 acquires a considerable discretion to evaluate all national tax systems to the extent that they have some redistributive effects or are otherwise deemed economically unsound.
Secondly, one of the main purposes of legal tests such as the three-part selectivity test is to avoid looking at market distorting effects directly, by applying legal rules that indirectly contribute to the objective(s) of EU state aid law. Effects-based reasoning, therefore, can have negative consequences for legal certainty because it is usually unclear which effects are deemed dispositive. 105 Article 107(1) TFEU itself does not indicate how much redistribution, or how much progressivity, could be compatible with EU state aid law, nor does any other rule of EU law. Adding to the significance of fiscal autonomy, the structured test applicable to selectivity improves legal certainty by specifying which market-distorting effectsnamely those relative to the reference system defined by the Member Stateare relevant. In this regard, the CJEU's judgments rightly reaffirmed the importance of legal form.
The inevitable cost of legal certainty, however, is that it leads to both false positive and false negative outcomes. In state aid law, this may mean that national measures, although they de facto provide selective advantages to some undertakings, are not caught by Article 107(1) TFEU because of the manner in which the Member State defined its reference system. In turn, such selective effects undoubtedly have market-distortive effects, which run counter to the objectives of Article 107 TFEU and the policy objectives of the Commission. 106 It seems that the Commission tried to use the Gibraltar judgment in order to interpret the selectivity criterion more broadly, even though this entailed an artificial interpretation of the 'reference system' step. It might be that the Commission used these two progressive tax systems as test cases to establish the boundaries of the selectivity criterion. In any event, the CJEU's judgments significantly constrain the Commission's ability to use EU state aid law to create a genuine level playing field. 107 In conclusion, as long as it is primarily for the Member States to determine the way in which they organize their tax systems and define their 'normal' taxation system, the selectivity criterion will not catch all tax systems that have de facto selective effects. This under-inclusiveness of the selectivity criterion in tax cases works as a constitutional limit to the ability of EU state aid law to correct market distortions caused by Member State measures. 108

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship and/or publication of this article.