The preliminary ruling of the Court of Justice of the European Union in the Google Ireland case turned on the compatibility with the rules on free movement of some of the administrative arrangements put in place by Hungary in order to administer its controversial advertisement tax (namely, the obligation to register and the penalties attached to the failure to comply with that obligation). The preliminary ruling offers some interesting insights on the way in which the Court assesses the compatibility with the freedom to provide services of national administrative arrangements aimed at ensuring the effective collection of taxes. This is a topical issue in the context of the recent efforts made by Member States to tax the digital economy more effectively.
On 3 March 2020, the Grand Chamber of the Court of Justice issued a preliminary ruling on the compatibility with the free movement rules of the administrative design of the Hungarian advertisement tax. The controversial elements of the tax were the registration obligation imposed upon digital companies that publish advertisements in Hungarian and the harshness of the administrative fines attached to that obligation. After reviewing the background of the case (Section 2) and the preliminary ruling (Section 3), this case note analyses the way in which the Court handled the clash between the administrative arrangements aimed at ensuring the effective collection of the tax and the free movement of services (Section 4). In its preliminary ruling, the Court did not address the compatibility of the Hungarian advertisement tax’s substantive legal framework with the rules on free movement. However, on the same day in which the case was decided, the Court rendered two additional rulings on the compatibility with Union law of two other similar taxes adopted by Hungary. The case note briefly discusses both cases (Section 5) and the implications that these developments may have from the perspective of the digital service taxes recently implemented by other Member States (Section 6).
The Hungarian advertisement tax was introduced in 2014. The tax was meant to be levied upon the net turnover generated by the broadcasting or publication of advertisements in Hungary. Newspapers, media outlets, and other traditional billposters were all subject to the tax, but so were tech companies active in the digital advertising business. Indeed, the Act on the advertisement tax provided that the tax would also be levied upon revenue generated by the publication of paid advertisements on the internet where the advertisements are mainly in Hungarian or on webpages written in that language. From a Union law perspective, two different aspects of the tax proved to be controversial.
The first one was the design of its progressive structure, which de facto imposed a heavier burden on foreign than on domestic taxpayers. The original design of the Hungarian advertisement tax included six different bands, with rates ranging from 0 to 50%. However, only providers belonging to foreign corporate groups fell within the scope of the bands subject to the highest rates. Furthermore, while progressivity is standard in direct taxes based on profits, it is not so common in turnover taxes. This is because revenue or turnover are traditionally not deemed to be good indicators to assess the taxpayers’ ability to pay, since they do not reflect the costs incurred in the income-generating activities.W Schön, ‘International Tax Coordination for a Second-Best World (part I.)’ (2009) 1 World Tax Journal 67, 74. Taxing revenues rather than profits can generate situations where loss-incurring businesses are liable to pay the tax, and to do so at a potentially higher rate than profitable entities generating smaller revenues.
The peculiar design of the Hungarian advertisement tax drew the attention of the Commission, which censured it on the ground that it favoured small over large advertising companies and was therefore contrary to the rules on State aid.Commission Decision (EU) 2017/329 of 4 November 2016 on the measure SA.39235 (2015/C) (ex 2015/NN) implemented by Hungary on the taxation of advertisement turnover  OJ L49/36, para 60. However, the General Court subsequently annulled the recovery decision of the Commission, holding that there was no reason turnover taxes had to be based on a single rate in order to escape the long shadow of the rules on State aid.Case T-20/17 Hungary v Commission, EU:T:2019:448. For a critical analysis of the Commission decision and of the General Court’s ruling, see R Szudoczky and B Károlyi, ‘The Troubled Story of the Hungarian Advertisement Tax: How (Not) to Design a Progressive Turnover Tax’ (2020) 48.1 Intertax 46. On appeal, the Court of Justice upheld the judgment of the General Court.Case C-596/19 P Commission v Hungary, EU:C:2021:202. For an early comment of this judgement, see S Moreno González, ‘Progressive Turnover Taxes and EU State aid law: Green light for digital services taxes?’ (EU Law Live, 19 March 2021) <https://eulawlive.com/op-ed-progressive-turnover-taxes-and-eu-state-aid-law-green-light-for-digital-services-taxes-by-saturnina-moreno-gonzalez/> accessed 8 April 2021.
The other controversial aspect of the Hungarian advertisement tax concerned the administrative arrangements made in order to ensure its effective collection from digital service companies with no significant physical or legal presence in Hungary. More specifically, the Act on the advertisement tax made it compulsory to register with the National Tax Authority for anybody publishing online advertisements in Hungarian or on webpages written in that language, unless already registered for the purposes of other taxes. Furthermore, the Act laid down a very severe system of administrative sanctions in case of failure to comply with the registration obligation, which consisted of an initial fine of HUF 10 million [€30.100], which would be tripled on a daily basis until it reached the sum of HUF 1 billion [€3,1 million]. Finally, taxpayers subject to these fines were not given the opportunity to contest them at the administrative stage, their only remedy being to start judicial proceedings.
The dispute at the origins of the Google Ireland case revolved around the fines the Hungarian Tax Authority imposed upon Google Ireland Limited in 2016. Google had been carrying out activities that were subject to the Hungarian advertisement tax ever since it entered into force. However, Google had not registered with the Hungarian Tax Authority despite the fact that it did not have any legal presence in the country. The Hungarian Tax Authority noticed the infringement and started sanctioning proceedings against Google. Given the exponential growth of the fine attached to the infringement of the registration obligation, it only took a few days before the Hungarian Tax Authority issued a decision imposing on Google Ireland the maximum fine allowed under the Act [€3,1 million].
It was in the context of the action for judicial review brought against that fine that the Administrative Court of Budapest submitted a preliminary reference pursuant to Article 267 TFEU.Consolidated Version of the Treaty on the Functioning of the European Union  OJ C202/49. The preliminary reference included seven questions. These essentially turned on the conformity with the freedom to provide services under Article 56 TFEU of three different aspects of the Hungarian advertisement tax outlined above: the registration obligation, the system of sanctions attached to that obligation, and the remedies available against those sanctions.
In her Opinion of 12 September 2019, Advocate General Kokott defended – as a preliminary point – the compatibility of the Hungarian advertisement tax itself with Union law. Although this was not at stake in the case, she made the point that neither the Union legislation on indirect taxation nor the rules on free movement prevent Member States from putting in place this type of taxes.Case C-482/18 Google Ireland Limited v Nemzeti Adó- és Vámhivatal Kiemelt Adó- és Vámigazgatósága, EU:C:2020:141, Opinion of AG Kokott, paras 29-56. It is against this background that she addressed the three issues outlined earlier.
She easily disposed of the registration obligation because it involved no discrimination against foreign providers (since Hungarian providers were also subject to that obligation) and because, even if it were a restriction on the freedom to provide services, it would be justified on account of the objective to ensure the effective collection of taxes.ibid, paras 57-62. However, as regards the system of sanctions attached to the registration obligation for foreign providers, she argued that it was indirectly discriminatory because it was harsher than the sanctions foreseen under the general rules on taxation to which residents are subject. At the justification stage, she also relied on its harshness to argue that it was disproportionate and hence unjustifiable.ibid, paras 63-100. Finally, she saw no justification whatsoever for the difference in remedies foreseen under the general rules on taxation to which residents are subject (which included administrative and judicial remedies) and under the Act on the advertisement tax (which made no room for administrative remedies).ibid, paras 101-107.
Consequently, Advocate General Kokott invited the Court to rule that both ‘the specific manner in which the Hungarian Law on the taxation of advertisements imposes coercive measures on undertakings established outside Hungary’ and ‘the limitations of the possibilities for legal redress with regard to the very high coercive penalty payments in connection with the Hungarian tax on advertisements’ constituted an unjustified restriction of the freedom to provide services.ibid, para 108.
On 3 March 2020, the Grand Chamber of the Court of Justice delivered its preliminary ruling. Unlike the Advocate General, the Court did not address the compatibility of the tax on advertisements itself with Union law.Case C-482/18 Google Ireland Limited v Nemzeti Adó- és Vámhivatal Kiemelt Adó- és Vámigazgatósága, EU:C:2020:141, para 24.
The Court ruled that the obligation to register did not amount to a restriction on the freedom to provide services enshrined in Article 56 TFEU, notwithstanding the fact that service providers established in Hungary are not subject to that obligation. In reaching that conclusion, the Court relied on three considerations: foreign providers could also escape the registration obligation if they were already registered for the purposes of other taxes; forcing already registered national providers to register again would be a useless formality; and there were no reasons to believe that the administrative burden of the registration obligation imposed upon non-registered providers was heavier than the registration requirements for other taxes.ibid, paras 24-36.
As far as the sanctions are concerned, the Court reached the opposite result. The starting point of the Court was the observation that, despite the seemingly neutral character of the severe system of sanctions foreseen for the failure to comply with the registration obligation, national and resident corporations were actually shielded from it. This is because resident corporations are always registered for the purposes of other taxes.ibid, para 41. The Court therefore looked at the sanctions they could incur should they fail to respect analogous registration obligations foreseen under the general rules on taxation to which residents are subject. These sanctions were considerably lower than the ones laid down by the Act on the advertisement tax, which is why the Court held that the contested measure constituted a restriction on the freedom to provide services of Article 56 TFEU.ibid, para 44.
The Court then considered the justification for the restriction. After admitting that the need to ensure the effective collection of taxes constitutes a legitimate overriding reason in the public interest,ibid, para 47. the Court went on to assess its proportionality. The Court concluded that the system of sanctions was ‘suitable’ but not ‘necessary’ to achieve the objective of ensuring the effective collection of taxes for the following reasons: there was no connection between the exponential increase in the fine and the amount of the tax whose payment was at stake;ibid, para 50. the exponential increase took place within days, without providing the taxpayer with sufficient time to comply;ibid, para 51. and, finally, even if the Hungarian Tax Authority had the power to reduce the amount of the fine, the purely discretionary character of this power could not save the lack of proportionality of the measure.ibid, para 53.
Consequently, the Court concluded that the system of sanctions attached to the registration obligation for the purpose of the Hungarian tax on advertisements was contrary to Article 56 TFEU. This conclusion made it unnecessary for the Court to consider the remedial aspects of the tax.ibid, para 55.
As already noted, the Google Ireland case does not deal with the compatibility of the Hungarian tax on advertising itself with Union law. The referring court did not raise this question and, unlike the Advocate General, the Grand Chamber deemed it unnecessary to address it. It follows that the case only deals with the institutional aspects of the Hungarian advertisement tax – namely, the registration obligation that it imposes upon unregistered taxpayers and the sanctions foreseen in order to promote compliance with that obligation.
It could be tempting to downplay the importance of this issue as a peripheral problem, one of little relevance from the perspective of the broader debate on the taxation of the digital economy. However, there is more to this issue than what would appear at first glance. This is because the institutional design of the Hungarian advertisement tax was clearly motivated by the specific difficulties that the collection of taxes raises in the digital economy. The reason Hungary included a specific registration obligation within the Act on the advertisement tax was that the tax targeted an economic activity that can be carried out without any actual physical or legal presence in Hungary. Similarly, the Act attached such a harsh system of sanctions on the registration obligation because of the belief that digital advertising companies without any presence in Hungary would be tempted to resist that obligation and that, given their size and deep pockets, only a harsh system of sanctions could have some deterrent effect.
If one looks at the case from this angle, it becomes a test on the limits that the free movement rules impose upon Member States in their attempts to shore up their taxing authority vis-à-vis foreign digital companies. In this sense, the preliminary ruling confirms that Member States have a broad margin of manoeuvre when it comes to imposing formal obligations such as the registration obligation, even if they represent an additional administrative burden. However, it also confirms that in Union law ‘all roads lead to proportionality’,B van Leeuwen, ‘Rethinking the Structure of Free Movement Law: the Centralisation of Proportionality in the Internal Market’ (2017) 10 European Journal of Legal Studies 235, 260. and that the compatibility of the instrumental sanctions attached to those formalities ultimately depends on their conformity with this principle.
As is well known, the Court normally uses a two-step version of the proportionality test when Member States invoke a public interest reason to justify free movement restrictions.T Marzal, ‘From Hercules to Pareto: Of bathos, proportionality, and EU law’ (2017) 15 International Journal of Constitutional Law 621, 631-633 and W Sauter, ‘Proportionality in EU Law: A Balancing Act? (2012-2013) 15 Cambridge Yearbook of European Legal Studies 439, 461. The purpose of the first step (suitability) is to verify that the contested measure actually serves the legitimate purpose alleged by the Member State. In this case, Hungary relied on the effectiveness of fiscal supervision and the effective collection of taxes in order to justify the system of sanctions attached to the registration obligation. As it had already done in previous cases, the Court accepted that the deterrent effect of administrative fines makes them suitable to promote compliance with this type of obligations. The purpose of the second step (necessity) is to verify that the contested measure is the least restrictive means to achieve the goal. This is where Hungary’s defence failed, because the Court deemed the system of sanctions foreseen in the Act on the advertisement tax to be excessive.
It is important to note that, in free movement cases where national administrative sanctions are at stake, the proportionality test does not limit itself to considering the balance between the fine amounts and the gravity of the infringement that they seek to punish. Needless to say, the fact that a sanction is out of proportion with the gravity of the underlying infringement may be a strong sign that there are less restrictive means to achieve the same result.The relationship between the sanction and the gravity of the offence may also be relevant from the perspective of the ‘requirements of good governance’ that the Court has infused into the proportionality test: see C Barnard, The Substantive Law of the EU. The Four Freedoms (4th edn, Oxford University Press 2013) 535-536 and S Prechal, ‘Free Movement and Procedural Requirements: Proportionality Reconsidered’ (2008) 35 Legal Issues of Economic Integration 201. However, there are other factors to consider. On the one hand, it is not possible to adopt a myopic view and to focus exclusively on the specific rule protected by each sanction. This is because seemingly minor rules may play an important role within the broader regulatory framework to which they belong. Arguably, this is true of formal duties such as the registration obligation that was at stake in the Google Ireland case. The registration obligation may seem trivial, but it serves a crucial purpose, which is to provide Member States with the information they need to effectively exercise their tax authority.
On the other hand, it is important to bear in mind that the retributive dimension of administrative sanctions is not necessarily their main function. The main purpose of administrative sanctions is to ensure compliance with the rules that they serve. Consequently, it is not possible to assess the proportionality of sanctions without somehow considering the circumstances in which they project themselves and the optimum level of deterrence required in those circumstances. This means that, in the Google Ireland case, the difficulties raised by the collection of service taxes in the digital context could be relevant to assess the proportionality of the sanctions attached to the registration obligation. In her Opinion, Advocate General Kokott hints at this idea when she discusses the amounts of the fines foreseen under the Hungarian Act on advertisements: ‘Lower amounts would be a less onerous means, but would not be equally appropriate as they would reduce the financial pressure’ and hence call into question ‘the objective of ensuring effective and uniform taxation.’Case C-482/18 Google Ireland Limited v Nemzeti Adó- és Vámhivatal Kiemelt Adó- és Vámigazgatósága, EU:C:2020:141, Opinion of AG Kokott, paras 88-89.
It is true that, in the preliminary ruling, the Court censures the Hungarian system of sanctions without a fully-fledged analysis of these issues. However, this is probably because some of the features of the contested sanctioning scheme were so draconian that they could not be justified under any circumstances (in particular, the automatic and exponential increase of the fine’s amount, which left no time for good-faith infringers to comply). The point to note is that the Court took issue with these features, but not with the existence of a specific sanctioning regime for non-resident taxpayers. Indeed, in order to assess the suitability and necessity of the sanctions foreseen under the Act on the advertisement tax, the Court did not give any weight to the fact that they were considerably higher than the sanctions foreseen for the same type of infringements under the general law on taxation. This asymmetry was obviously relevant to assess the discriminatory character of the scheme and to establish that it constitutes a prima facie restriction on the rules on free movement. However, it was not fatal at the justification stage, because the whole point of the justification stage is to verify whether there are good reasons behind the asymmetry.
This observation is important, because it goes to show that Union law does not prevent Member States from adapting the institutional design of digital taxes to the special challenges that they raise. The effective collection of digital taxes may require a tailor-made set of institutional arrangements and Union law does not object to that, so long as the principle of proportionality is respected. Less drastic or draconian arrangements may be acceptable even if they primarily or exclusively target digital companies established abroad. However, a more ‘proportionate’ case will have to reach the Court before we can get more detailed guidance on the topic.
On the same day in which the Google Ireland case was decided, the Court of Justice rendered two additional preliminary rulings on the compatibility with Union law of the substantive framework of two other Hungarian turnover taxes: the tax on turnover of telecommunication operators, which was at stake in Vodafone,Case C-75/18 Vodafone Magyarország, EU:C:2020:139. and the tax on turnover of store retail trade operators, which was at stake in Tesco.Case C-323/18 Tesco-Global Áruházak, EU:C:2020:140. Like the advertisement tax, these two turnover taxes also featured a progressive rate structure, as a result of which all companies in the lowest tax bracket (0% tax rate) were Hungarian, while most – if not all – of the companies in the highest tax brackets (subject to a 6,5% and 2,5% top marginal rate) belonged to foreign corporate groups.R Szudoczky and B Károlyi, ‘The Troubled Story of the Hungarian Advertisement Tax: How (Not) to Design a Progressive Turnover Tax’ (2020) 48.1 Intertax 46, 57.
The preliminary references at the origin of Vodafone and Tesco went to the core of the problem. They essentially asked the Court whether turnover taxes imposing most or all of their burden on undertakings controlled by individuals or entities established in other Member States were precluded by Article 49 TFEU on freedom of establishment and Article 54 TFEU on equal treatment.
However, the Court reformulated the inquiry. Instead of adhering to and starting its analysis from the postulation that all of the taxable entities in the lower brackets are domestic and those in the upper brackets are foreign, the Court framed the inquiry around the issue of progressivity of the tax rates. Specifically, the Court stated that
the referring court considers that it is necessary in order to resolve the dispute in the main proceedings, to determine whether the progressive scale, using bands, of the special tax may constitute, in itself […] indirect discrimination vis-à-vis taxable persons that are controlled by natural persons or legal persons of other Member States, who bear the actual tax burden, and, therefore, be contrary to Articles 49 and 54 TFEU.Case C-75/18 Vodafone Magyarország, EU:C:2020:139, para 36 and Case C-323/18 Tesco-Global Áruházak, EU:C:2020:140, para 48.
By redefining the case as one challenging the progressive nature of the tax – rather than the discriminatory crafting of thresholds – the Court rapidly came to the conclusion that such progressivity was admissible. Indeed, the Court recognized that ‘Member States are free […] to establish the system of taxation that they deem the most appropriate, and consequently the application of progressive taxation falls within the discretion of each Member State’.Case C-75/18 Vodafone Magyarország, EU:C:2020:139, para 49 and Case C-323/18 Tesco-Global Áruházak, EU:C:2020:140, para 69. It also recognized that
in that context, and contrary to what is maintained by the Commission, progressive taxation may be based on turnover, since, on the one hand, the amount of turnover constitutes a criterion of differentiation that is neutral and, on the other, turnover constitutes a relevant indicator of a taxable person’s ability to pay.Case C-75/18 Vodafone Magyarország, EU:C:2020:139, para 50 and Case C-323/18 Tesco-Global Áruházak, EU:C:2020:140, para 70.
Finally, the Court concluded that ‘by means of the application of a progressive scale based on turnover, the aim of that law is to impose a tax on taxable persons who have an ability to pay that exceeds the general obligation to pay tax’,Case C-75/18 Vodafone Magyarország, EU:C:2020:139, para 51 and Case C-323/18 Tesco-Global Áruházak, EU:C:2020:140, para 71. and that ‘the fact that the greater part of such a special tax is borne by taxable persons [from] other Member States cannot be such as to merit, by itself, categorisation as discrimination. […] Accordingly, that situation is an indicator that is fortuitous, if not a matter of chance’.Case C-75/18 Vodafone Magyarország, EU:C:2020:139, para 52 and Case C-323/18 Tesco-Global Áruházak, EU:C:2020:140, para 72.
By focusing exclusively on the legitimacy of progressivity through apparently neutral turnover thresholds, the Court missed the opportunity to conduct an in-depth and complete analysis of the concrete effects of this type of tax design. In other words, the Court avoided the real issue at stake, which is whether the use of a supposedly neutral criterion such as the turnover threshold may in fact lead to indirect discrimination on the basis of nationality.L Parada, ‘Vodafone Magyarország. Special Hungarian tax on turnover of telecommunications operators is compatible with EU law’ (2020) 10 Highlights and Insights on European Taxation 32. In so doing, the Court partially deviated from the reasoning advanced by Advocate General Kokott, who had argued that the assessment of indirect discrimination should have been based on a combination of quantitative and qualitative criteria, and that the intention of the legislature should have been taken into consideration.Case C-75/18 Vodafone Magyarország, EU:C:2020:139, Opinion of AG Kokott, paras 63, 74, 78-81, 83-85.
The Hungarian advertisement tax is just one of the numerous new taxes that Member States have adopted in recent years, with the aim of tackling the ‘base erosion and profit shifting’ (BEPS) issue and to collect an amount of tax revenue from the so-called digital economy that is consistent with the market size of such industry.For a general illustration of the BEPS issue, see OECD, ‘Addressing Base Erosion and Profit Shifting’ (OECD Publishing 2013). For a more detailed analysis of the BEPS issue with specific regard to the digital economy, see OECD, ‘Tax Challenges Arising from Digitalization – Interim Report 2018: Inclusive Framework on BEPS, OECD/G20 Base Erosion and Profit Shifting Project’ (OECD Publishing 2018). The Commission has tried to articulate a coordinated response to this issue. However, its 2018 proposal for the adoption of a common digital service tax (DST) did not gather the unanimous support of Member States, and was therefore placed on hold.Commission, ‘Proposal for a Council Directive on the common system of a digital services tax on revenues resulting from the provision of certain digital services’ (Proposal) COM/2018/0148 final. As a result, various Member States have unilaterally set in place taxes of this sort.
As of March 2021, Austria, France, Italy, Poland, and Spain have adopted a DST, while other Member States – such as Belgium, the Czech Republic, and Slovakia – have published proposals for the adoption of a DST.E Asen, ‘What European OECD Countries Are Doing about Digital Services Taxes’ (Tax Foundation, 25 March 2021) <https://taxfoundation.org/digital-tax-europe-2020/> accessed 8 April 2021. Although these unilaterally adopted levies differ in their respective scope, rate, and administrative arrangements, most of them mirror the essential features of the Commission’s proposal. In particular, all of these levies are imposed on the gross revenues – net of VAT – originating from enumerated types of digital services rendered by businesses meeting certain quantitative requirements.
The Google Ireland case and the other Hungarian cases discussed above are relevant from this perspective, because the compatibility of DSTs with Union law is likely to be questioned in the near future. The preliminary ruling of the Court in Google Ireland suggests that the registration duties imposed by Member States on foreign service providers are unlikely to raise any problems. The challenge lies in the design of the sanctions attached to those duties, which shall not run afoul of the proportionality requirements so cherished by the Court, even if some Member States may feel like they are in a David vs Goliath position vis-à-vis large tech corporate groups.
The preliminary rulings in the Vodafone and Tesco cases could suggest that the Court will refrain from striking down these types of taxes on discrimination grounds. However, the Hungarian turnover taxes analyzed in those cases are not identical to all the DSTs adopted or proposed thus far by Member States. In particular, most DSTs do not provide for a system of progressive rates, but rather for an exception threshold and a single flat rate applicable to any taxable base. Moreover, while the applicable rate of the Hungarian taxes was determined using in-country turnover, the applicability threshold of DSTs refers not only to the amount of domestic revenue, but also to worldwide group revenue.ibid. This may appear to be a distinct source of national discrimination that the Court should not ignore and for which additional justifications may be necessary. Consequently, the reasoning of the Court in the Hungarian cases cannot be automatically transplanted to DST cases.
Thus, the preliminary rulings of the Court in Google Ireland and in the other Hungarian tax cases cannot be used to make reliable forecasts on the compatibility of the substantive aspects of DSTs with Union law – and, in particular, on the treatment that their quantitative thresholds deserve from a free movement perspective. In any event, a preliminary reference on DSTs could grant the Court a renewed opportunity to provide clear guidance and standards for the assessment of covert tax discrimination through turnover taxes.For a broad and in-depth discussion of the compatibility of DSTs with Union law, see R Mason and L Parada, ‘The Legality of Digital Taxes in Europe’ (2020) 40 Virginia Tax Review 175.