A7 Statliga stöd

Följande bidrag är författat på engelska.

In case C-547/23 PAutonomous Region of Madeira v European Commission, the Tenth Chamber of the CJEU delivered a judgment concerning a previously approved and long-established (since 1987) State aid scheme. The case concerned a regional operating aid scheme, Regime III of the Madeira Free Zone (MFZ), whose implementation ultimately led the Commission to find that unlawful aid had been granted, with significant legal consequences for Portugal, and in particular for the Autonomous Region of Madeira.1

This commentary focuses on the Commission decision SA.21259, the subsequent annulment proceedings before the General Court, and, finally, the appeal before the CJEU and its judgment, which concluded the judicial discussion and confirmed that the Portuguese authorities had deviated from the conditions laid down in the original approval of the MFZ Regime III.2 As a result, the measure constituted new aid, thereby breaching the notification procedure and standing still clause of Article 108(3) TFEU, which led to the order to recover the aid from beneficiary companies and abolish further implementation of Regime III.

Background

The MFZ was designed to promote regional development and economic diversification, compensating for the structural disadvantages inherent to an outermost region of the Union.3 To describe the MFZ Regime III scheme, I rely primarily on the Commission State aid decision SA.21259 that declares the previously approved aid scheme as incompatible with the internal market, ordering Portugal to abolish it and recover the unlawful aid granted to beneficiaries.4

The MFZ Regime III scheme comprised the following areas: the International Business Centre of Madeira (IBCM), the International Shipping Register (MAR), and the Industrial Free Trade Zone (IFTZ). It was designed as a tiered, employment-based tax incentive, in which the fiscal advantage to beneficiaries was quantitatively capped depending on the number of jobs created and maintained in Madeira. The scheme tied the maximum taxable base ceilings and corresponding maximum aid amounts to verifiable employment positions created and maintained in each fiscal year, thereby ensuring the measure’s contribution to regional development concerning the labor outcomes it generated.5

The regime was centered on the number of jobs created and maintained in Madeira to justify its compatibility with the internal market by proportionating economic contributions locally. Accordingly, the table below summarizes the employment threshold and the associated ceilings for taxable bases and maximum aid amounts under the 2007 and 2013 versions of the MFZ Regime III as authorized by the Commission:6

Under the 2007 and 2013 frameworks, the State aid schemes reduced the corporate income tax rates on profits of activities “effectively and materially performed in Madeira”: 3% (from 2007 to 2009), 4% (from 2010 to 2012), and 5% (from 2013 to 2020), alongside exemptions from certain municipal, local and property transfer taxes for business set up in MFZ, up to the aid ceilings linked to job creation shown in the table above.7 Companies operating in the industrial MFZ could obtain an additional 50% corporate tax reduction if they satisfied at least two of the following qualitative criteria related to technological innovation:8

1. modernization of the regional economic fabric through technological innovations relating to products, manufacturing, or business methods.

2. diversification of the regional economy, particularly through the introduction of new activities with high added value.

3. employment of highly qualified human resources.

4. improvement of environmental conditions; and (e) creation of at least 15 new jobs to be kept for a minimum of five years.

Access to the scheme was limited to specific economic sectors defined by the NACE Rev. 1.1 classification, covering only selected activities: agriculture and animal production; fisheries aquaculture and related services; manufacturing industry; production and distribution of electricity, gas and water; wholesale trade; transport and communication; activities relating to immovable property; leasing and services to companies; higher education and/or adult education; other collective services activities.9

Procedural aspects from the perspective of the EU State aid control system

This case illustrates a more dynamic understanding of the EU State aid control system, notably, regarding compliance with an authorization decision as a continuous obligation and the possibility that deviations in implementation may retroactively render previously authorized aid unlawful.

From a State aid control system perspective, the MFZ Regime III constituted authorized aid.10 The authorization decision adopted by the Commission under Article 108(3) TFEU constitutes an autonomous act of EU law, whose wording defines the normative framework within which the approved scheme may operate. As a decision addressed to a Member State, it is binding in its entirety and produces specific legal effects within the State aid control system.11 The conditions of the authorization decision must therefore be interpreted in accordance with EU law, and not by reference to the internal logic of national tax legislation. While national law governs the structure of the domestic measure, it cannot determine the scope or meaning of the compatibility conditions laid down in the approval decision. Once the scheme is authorized, the decision itself becomes the legal benchmark for assessing compliance.

The Commission concluded that Portugal had deviated substantially from the approved regime, thereby unlawfully introducing new aid within the meaning of Article 108(3) TFEU without notification to the Commission and observance of the standing still obligation.12 This situation is conceptually closer to the notion of misuse of aid defined in Article 1(g) of Regulation 1589/2015, where the approved aiding scheme remained in force, but its implementation departed from the parameters that defined its compatibility. As a result, what Portugal implemented was not the approved Regime III, but another aid scheme not reviewed or approved by the Commission.

Why is this discussion relevant?

If the regime substantially applied by Portugal had been recognized by the General Court as corresponding to the one approved by the Commission, and therefore not deviating from that regime, the beneficiary companies would have been protected by the principles of legal certainty, legitimate expectations, good administration, and proportionality.13 The recovery order would likely have been annulled on that basis, given that the Commission is also legally bound by the decisions it issues.14 However, those principles can only be invoked to prevent aid recovery and abolition—the decision’s annulment—if the merits of the discussion support the view that it was the Commission’s interpretation of the approved MFZ Regime III that deviated from the regime, rather than Portugal’s implementation.

In the next part of this commentary, I focus on the merits of this discussion. However, it should be recalled that this case is an appeal, limited to points of law pursuant to Article 256(1) TFEU and Article 58 of the Statute of the CJEU. The Court was therefore not called upon to reassess the factual finding of the General Court, but only to determine whether the General Court had erred in law or distorted evidence when it rejected Madeira’s annulment proceedings. The judgment must be read against this procedural limitation.

Finally, the following pleas raised on appeal were excluded from this commentary. Madeira’s contention that the measure did not constitute State aid per se is an argument difficult to reconcile with the scheme’s long-standing approval, operation, and monitoring, as well as with its very design as a geographically targeted preferential tax regime called Madeira Free (trade) Zone.15 The Commission’s objection to Madeira’s locus standi, as the General Court dismissed that plea based on settled case-law under Article 263(4) TFEU.16 Madeira’s allegation that the decision breached legal certainty, legitimate expectations, good administration, and proportionality, and that recovery was either impossible or time-barred, was rejected because the CJEU did not consider that the General Court erred in law or distorted facts in upholding the Commission’s decision in question.17 Although most of these grounds of appeal do not alter the substantive assessment, the judgment nevertheless reaffirms a systemic point of importance: prolonged administrative practice for several decades or the reliance by beneficiaries on the scheme cannot generate legitimate expectations where implementation of an approved scheme departs from the conditions laid down in the authorization decision. Undertakings benefiting from an aid scheme therefore assume the risk that advantages granted outside the scope of the approval may be subject to recovery, notwithstanding the duration or apparent stability of the regime in practice.

Controversial aspects of Regime III (formal regime approved versus substantial implementation)

The controversial aspect of this case and the merit of the discussion stem from two main issues. First, the geographical origin of eligible profits. Second, the methodology and rules used by the authorities for accounting jobs.

Concerning the geographical origin of eligible profits, the Commission concluded that tax benefits were granted to undertakings that did not effectively and materially perform their activities in Madeira and thereby fell outside the scope of the geographical limits of the authorized aid.18

In both 2007 and 2013 approval decisions, the reduced corporate income tax rate was expressly limited to “profits resulting from activities effectively and materially performed in Madeira.”19 However, during the 2015 monitoring exercise and subsequent 2016 investigation, the Commission identified cases in which undertakings registered under the MFZ scheme had applied the reduced rate to profits derived from activities not genuinely carried out in Madeira.20

Portugal confirmed that, under its interpretation, companies with their head offices or place of effective management in Madeira were subject to taxation on their worldwide income, irrespective of where the underlying economic inputs, transactions, or value-creating activities occurred, provided that the income was linked to authorized activities and recorded by the Madeira-based entity.21 Portugal explained that it required a system of separate accounting to distinguish between income generated within Portuguese territory (taxed at the ordinary rate) and income generated outside Portuguese territory or within the MFZ (eligible for a reduced rate).22 Moreover, it sustained that in practice, this territorial distinction did not require that substantive economic activity giving rise to the profits be physically confined to Madeira. Rather, income from transactions with non-residents—or even from operations geographically external to MFZ—could benefit from the preferential regime if attributed to an entity established in the MFZ.23

Based on the view that NACE sectors authorized in the scheme were predominantly internationally oriented/concerned activities, Portugal argued that an interpretation restricting eligible profits to those geographically generated in Madeira would undermine the very rationale of outermost-region aid, which was to attract investments and internationally mobile activities. Moreover, it argued that the Commission’s restrictive interpretation would also be inconsistent with the fundamental freedoms of the internal market, and that the 2013 increase in the ceiling reflected the need to enhance MFZ competitiveness in relation to other jurisdictions.24

Portugal also maintained that the requirement of activities being effectively and materially performed in MFZ was satisfied where the undertaking had a genuine establishment in the region. That is, comprising appropriate premises, staff, resources, and an effective decision-making place, even if not all operational functions or revenue-generating transactions physically occurred there. The authorities relied on self-assessment through annual tax declarations, subject to ex post verification and tax inspections in cases of doubt, supplemented by administrative cooperation and information exchange.25 Consequently, Portugal viewed that the approved aiding scheme did not require that substantive economic activity giving rise to the profits was physically arising in the MFZ.

The controversy crystallized around the interpretation of the substantive link between aided profits and territory. The Commission adopted the view that regional operating aid must be geographically circumscribed and directly connected to value-creating activities actually performed in the MFZ.

Concerning the methodology and rules used by the authorities for accounting jobs, the Commission considered that the Portuguese authorities relied on formal or declaratory employment figures instead of assessing whether the activities carried out in the region generated genuine and substantial job creation.26 In particular, the Commission questioned whether the employment thresholds—used to determine the applicable tax base ceilings and corresponding aid intensity—reflected real economic activity performed in MFZ or merely the formal registration of employees linked to entities benefiting from the regime.

Portugal defended its approach by invoking consistency with national labor law. According to the Portuguese authorities, the identification and qualification of employment relationships followed the criteria of the national Labor Code, which recognizes employment based on contractual subordination, remuneration, and organizational integration, rather than the precise geographical location where tasks are performed.27 From this perspective, employees legally hired by an undertaking genuinely established in MFZ and performing functions connected to that undertaking’s activity—such as management, coordination, administrative support, or decision-making—should be regarded as meeting the condition for the aid as regionally created and maintained jobs. Even where certain operational or revenue-generating functions occur elsewhere.28 Consequently, Portugal defended its methodology as coherent and aligned with its national labor law, which should give logic to the authorized aid scheme.

From the Commission’s perspective, the objective of the regional operating aid in outermost regions is not only to attract capital or corporate presence, but also to foster tangible socio-economic development, in this case through employment. Consequently, the mere existence of employment contracts or nominal jobs counts (formal proof of such job creation and maintenance) could not suffice to meet that objective, notably where the underlying economic activity and the effective labor demand were located outside the MFZ.29 This raised concerns that the aid calculation mechanism allowed undertakings to access a higher aid ceiling without producing a proportional employment increase in Madeira, thereby undermining the proportionality and necessity requirements inherent in the State aid control system that classifies aid as compatible with the internal market.

The controversy in this second merit discussion ultimately mirrors the broader tension already evident in relation to profit and labor allocation: a formal establishment-based conception of regional linkage (Portugal’s view) versus a substantive activity-based interpretation (Commission’s view). Under Portugal’s approach, the existence of a genuine establishment in Madeira—together with legally recognized employment relationships under national labor law—was sufficient to justify both the attribution of profits to the region and the counting of jobs for aid intensity (as shown in the table above). By contrast, the Commission required a palpable connection between aided profits, the counted employment, and economic activity materiality carried out in MFZ. The decisive question, therefore, is which of these competing logics prevails under EU State aid law.

The CJEU examined whether the Commission was entitled to conclude that the Portuguese implementation of Regime III departed from the conditions authorized in the 2007 and 2013 State aid decisions in three respects: (i) the requirement that eligible profits derive from activities “effectively and materially carried out in Madeira”; (ii) the method used to calculate the number of jobs created or maintained in the region; and (iii) the effectiveness of the fiscal controls designed to verify compliance with those requirements.30

First, the Court upheld the Commission’s interpretation that only profits stemming from activities genuinely performed in MFZ could benefit from the tax reduction.31 This reading followed from the ordinary meaning of the wording (linguistic interpretation) and the administrative context of the authorizing decisions, and the restrictive interpretation governing regional operating aid under Article 107(3)(a) TFEU. Activities carried out outside the region could therefore not be covered, irrespective of the beneficiary’s registration in MFZ.

Second, the Court confirmed that the Portuguese method for counting jobs failed to ensure that the declared employment was real, permanent, and objectively verifiable. The incompatibility did not stem from the absence of specific EU calculation methods as such, but from the inability of the national approach to guarantee effective control over compliance with the employment condition.32

Third, the Court agreed that Portugal’s tax monitoring system was ineffective, since fiscal controls were based on an interpretation of the profit-origin and employment requirements that already diverged from the authorizing decisions. Consequently, even extensive audits or accounting obligations could not demonstrate proper implementation of the scheme.33

Concluding remarks

The CJEU ruling is neither surprising nor disruptive. By dismissing the two substantive pleas on appeal, the Court essentially confirms the logic apparent in the Commission’s decision. The State aid control system is built upon the foundation that the assessment of a measure takes into account both its formal configuration and its substantive effects (the so-called effect principle).34 Where both dimensions diverge, it is the substantive effects—or the absence thereof—that guide the interpretation of the applicable rules, particularly in a clear case of State aid within the meaning of Article 107(1) TFEU, which by definition affects the functioning of the internal market.

Against this background, reliance on national law—such as the national Labor Code—to give normative coherence to a compatible aid scheme designed to promote job creation is necessarily limited. While domestic labor law may legitimately safeguard individual rights in the provision of services, it cannot redefine the substantive condition under which regional operating aid was authorized. As consistently held by the CJEU, derogations under Article 107(3) TFEU must be interpreted restrictively in the light of the inherently distortive nature of the aid.35 Accordingly, the protective reach of the Portuguese Labor Code could not justify the inclusion of employment physically carried out outside the MFZ, nor could profits generated beyond its territorial scope be brought within the preferential regime. Even if such allocations might be in line with Portugal’s general system, the involvement of the MFZ as an approved aid scheme subjects the regime to the specific framework and conditions defined in the Commission’s authorization decision.

Permitting profits and employment with no substantive connection to the MFZ to benefit from the scheme would reduce the approved conditions to merely formal requirements, undermining the effects-based logic of the State aid control system discussed above. It would also risk transforming the regime into a de facto distortive “tax paradise” within the Union, detached from the regional development objective that justified its compatibility in the first place. A Commission authorization decision does not merely endorse a measure; it establishes a structured framework of control. The operative logic of the scheme is consequently the one articulated in the approval decision itself, and its implementation must unfold through a restrictive interpretation of the mechanisms deemed compatible, ensuring alignment—both formally and substantially—with the objective pursued.

In conclusion, this case illustrates with particular clarity how the obligation to comply with a Commission authorization decision may generate tension between national law and the EU State aid control system. Although the classification of a measure as State aid under Article 107(1) TFEU requires the Commission to assess it in the light of national law and its inherent logic, the review of compliance with an approved aid scheme operates within a distinct normative framework. Once authorized, the regime must be interpreted and implemented strictly in accordance with the conditions and objectives that justified its compatibility with the internal market. To ensure that both its formal configuration and its substantive effects remain aligned with the rationale underpinning the exemption from the prohibition of State aid.

This conclusion does not affect the specific constitutional status of Madeira as an outermost region under Article 349 TFEU. While this provision recognizes that structural, economic, and social constraints affect such regions and may justify the adoption of State aid measures, it does not alter the strict compliance with the conditions set out in the aid. That outermost region classification simply provides the link to apply Article 107(3)(a) TFEU—thus, it does not allow deviation from the parameters defined in the authorization decision itself. In the present case, the MFZ was approved as a regional operating aid aimed at promoting economic development in Madeira through the creation and maintenance of employment via reduced corporate taxation. That objective cannot be satisfied by a formal headcount requirement but presupposes a genuine substantive nexus between the preferential tax treatment and the effective generation of economic activity and labor within MFZ. The judgment therefore confirms that compliance with an authorization decision constitutes a continuous obligation, designed to preserve the structural link between the aid’s operation and its approved purpose.

In case C-776/23 PEuropean Commission versus Kingdom of Spain, the Eighth Chamber of the CJEU delivered a judgment annulling the Commission’s State aid decision SA.35550, which considered the Spanish tax regime—permitting resident companies to deduct through amortization financial goodwill arising from acquisitions of shareholdings in non-resident companies under Article 12(5) of the Spanish Corporate Tax Law (TRLIS)—as unlawful aid.36 The Commission ordered Spain to recover the unlawful aid and abolish the regime.37 Similar to the previous commentary, I base this case analysis on the decision SA.35550, its annulment proceeding judgement, and its appeal judgement to provide a better analysis of the ruling delivered by the CJEU.38

Background

Before the adoption of decision SA.35550 under scrutiny, the Commission had already examined Article 12(5) TRLIS (hereafter, the goodwill regime) in two earlier State aid decisions, where the tax amortization of financial goodwill for foreign shareholding acquisitions was considered to constitute incompatible aid.39 The ensuing annulment actions before the General Court and subsequent appeals before the CJEU led to a significant judicial clarification of the goodwill regime and Spain’s legislative development of that system in compliance with the rule forbidding State aid (Article 107(1) TFEU). Following the CJEU setting aside of the General Court’s initial judgements concerning those previous decisions and the remand proceedings, the EU judicature finally endorsed the Commission’s approach, thereby consolidating the analytical framework applicable to Article 12(5) TRLIS, in particular by confirming that a tax measure may be selective even when it is formally open to all undertakings, if its conditions operate as a derogation from the ordinary corporate tax system and thereby favor, substantially, a defined category of cross-border acquisitions.40

Against this background, SA.35550 did not emerge in isolation but formed part of a clear progressive refinement strategy of Spain, in which the compatibility of the Spanish goodwill regime was shaped through successive administrative decisions, annulment proceedings, and appellate review. Spain introduced a new binding administrative interpretation in 2012 of the aid scheme (object of the previous two State aid decisions), which, when applied by Spanish authorities, had the effect of extending the deduction to indirect acquisitions achieved through intermediate holding companies, even where no economic activity generating goodwill existed at the holding level.41 The Commission’s State aid decision SA.35550 viewed the new administrative interpretation as extending the scope of Article 12(5) TRLIS, and since it was not notified pursuant to Article 108(3) TFEU before its implementation, it was treated as a non-notified aid.42

The goodwill regime

Article 12(5) TRLIS allowed a Spanish resident company to amortize financial goodwill arising from the acquisition of at least 5% in a non-resident company, provided the conditions of Article 21 TRLIS were met.43 Financial goodwill corresponded to the portion of the purchase price exceeding book value that could not be allocated to identifiable assets, and it was deductible up to 1/20 per year.44 Although the statutory wording of Article 12(5) TRLIS did not expressly distinguish between direct and indirect acquisitions, the initial consistent administrative practice of the Spanish tax administration (DGT) and the Economic and Administrative Court (TEAC) limited the deduction to direct acquisitions of operating companies.45 The rationale was that goodwill can only arise in entities carrying out an economic activity, whereas holding companies merely own shares and do not themselves generate goodwill.

In 2012, the DGT and the TEAC, departing from previous interpretation and adopting a new interpretative view, extended the deduction to indirect acquisition resulting from the purchase of a non-resident holding company.46 This interpretation relied on: (i) references in Article 21 TRLIS to direct and indirect shareholdings; (ii) the internalization objective of Article 12(5); (iii) perceived indications in the Commission’s earlier decisions; (iv) consideration of the informational requirement under Article 15 of the Implementing Regulation of the Corporate Tax Act; and (v) neutrality.47 However, the Spanish specialized High Court (Audiencia Nacional) in 2014 rejected this new interpretation and reaffirmed the original restrictive approach, holding that goodwill can only arise in operating companies that conduct economic activity.48

Against this background, Spain sought to annul the Commission’s State aid decision, particularly because the Commission adopted the view that the new interpretation constituted new aid, not previously notified and incompatible with the internal market.49

The appeal

First and foremost, the appeal did not reopen the discussion concerning the qualification of Article 12(5) TRLIS concerning the goodwill regime as constituting State aid under Article 107(1) TFEU—since this discussion was already consolidated in the previous decisions trialed by the EU Courts.50 Instead, the appeal shifted the focus of the discussion to a structurally distinct question within the boundaries of the EU State aid control system. By treating the new administrative legal binding interpretation as qualifying as “new aid” implemented without prior notification and in breach of Article 108(3) of the TFEU, the Commission sought to detach specifically indirect acquisition from the scope of earlier negative decisions (Decisions 2011/5 and 2011/282) and to subject them to full recovery under that Article. As a result, the central question discussed in the appeal was whether the Commission lawfully adopted the decision SA.35550, reinterpreting the material scope of the previous decisions concerning the goodwill regime in a manner that effectively narrowed the protective carve-outs previously granted.

Concerning the scope of earlier decisions, the Commission argued that the General Court misconstrued the scope of the Commission’s first and second decisions concerning the goodwill regime, by treating them as covering both direct and indirect acquisitions.51 In the Commission’s view, those decisions had to be read in light of the administrative description provided by Spain in 2007, according to which the goodwill regime was applied only to direct acquisitions; any references to indirect acquisitions would merely reflect the cross-reference to Article 21 TRLIS rather than an assessment of indirect structures.52

The CJEU rejected that argument by anchoring the analysis in the principle of legal certainty, which it considered particularly stringent in the context of negative decisions ordering cessation and recovery.53 Given that the operative carve-outs in the two decisions before SA.35550 expressly referred to “rights held directly and indirectly”, the Court held that the meaning of those acts was straightforward and could not be neutralized by contextual elements.54 As a consequence, the Commission was not free, through a subsequent decision, to treat indirect acquisitions as falling outside the scope of the earlier decisions without undermining predictability and the stability of legal relationships generated by its own acts.

As discussed in the previous commentary on the Madeira Free Zone, a Commission State aid decision does not merely assess compatibility or incompatibility of an aid measure with the internal market. It establishes the normative framework within which both Member States and the Commission must operate. The decision delineates the legally permissible scope of implementation—grounded in the principle of legality—thereby framing the Member State’s discretion, while simultaneously fixing the boundaries of the Commission’s subsequent control.

In the present case, the decisive element is that the earlier decisions expressly referred to “direct and indirect” acquisitions in defining the scope of the recovery exemptions.55 A feature explicitly incorporated into the operative reasoning of a negative State aid decision cannot subsequently be reclassified as a deviation from the scheme. Unlike MFZ litigation—where new aid stemmed from an implementation that substantially diverged from the conditions laid down in the approval decision—here the Commission sought to narrow, ex post, the material scope of its own prior acts. Such an approach would undermine the principle of legal certainty and predictability in the State aid control system, as the parties concerned operate based on the normative framework established by the Commission itself.

The Court’s reasoning therefore reinforces a structural symmetry inherent in the EU State aid control that is not always articulated with equal clarity: just as Member States are bound by the terms and conditions set out in a Commission State aid decision, the Commission itself is bound by the legal contours it has defined therein. That symmetry operates as a constraint on both sides of the enforcement relationship. However, as illustrated by the MFZ litigation and by this third iteration of the goodwill regime, such symmetry can generate tensions at the implementation stage. Where the scope of a decision is not drafted with sufficient precision, or where subsequent administrative developments interact ambiguously with earlier findings, the boundaries between permissible implementation and unlawful modification may become contested. The present judgment demonstrates that, in situations of such tension, the principles of legal certainty and legitimate expectations function as a decisive stabilizing legal mechanism.

Concerning the binding administrative interpretation effects, the Commission invoked the effect principle that grounds the interpretation of State aid rules, contending that a binding administrative interpretation may, in practice, broaden the scope of a regime that was narrowly circumcised, and therefore constituted new aid.56 On that premise, which the CJEU confirmed could have such an effect,57 the Commission sought to characterize the new interpretation as a substantial modification, triggering an autonomous notification obligation under Article 108(3) TFEU.

However, the CJEU did not need to resolve that issue in the abstract. The current analysis of the goodwill regime, following the Madeira Free Zone commentary, illustrates another aspect of the State aid control system that was not perceptible in the previous (MFZ) case. Once the General Court held the view that indirect acquisitions were already encompassed by the previous two decisions, the Commission’s premise that the “new” administrative interpretation expanded the scheme beyond what had previously been decided could not stand.58 The CJEU dismissed it as ineffective because it could not affect the legality assessment once legal certainty fixed the scope of the earlier decisions.

Concerning the General Court’s error in law when interpreting the principle of legitimate expectations, the Commission argued, inter alia, that such expectations cannot be grounded in uncertain future developments and should not be extended to indirect acquisitions allegedly not covered at the time of the earlier decisions. In substance, the Commission sought to decouple the legitimate expectations carve-outs from indirect structures, thereby supporting full recovery under SA.35550.59 The Court again treated the ground as ineffective since the previous two decisions themselves explicitly recognized legitimate expectations in relation to rights held “directly or indirectly,” any alleged error in the General Court’s auxiliary reasoning could not vitiate the annulment of SA.35550.60

Concluding remarks

Ultimately, the appeal in this case discussed a more structural question: whether the Commission could reinterpret, through a subsequent decision, the material scope of its own earlier negative decisions in a manner that altered the legal consequences previously attached to the scheme. The Court’s answer was unequivocal. Where the operative part and reasoning of a State aid decision expressly refer to “direct and indirect” acquisitions, that formulation fixes the legal framework within which both implementation and enforcement must occur.

The first ground of appeal was therefore decisive. Once the Court anchored the analysis in the principle of legal certainty and confirmed that the wording of the two prior decisions was clear, the remaining grounds became secondary. The debate over whether a binding administrative interpretation may broaden the scope of a scheme—and thereby constitute new aid—was rendered legally irrelevant, albeit possible. Similarly, the discussion on legitimate expectations could not alter the outcome, since those expectations had already been explicitly recognized in respect of indirect acquisitions in the earlier decisions themselves.

The broader significance of this judgment lies in its institutional message. EU State aid control operates within a dynamic enforcement environment, often involving successive decisions concerning the same national measure. Yet that dynamism is bounded by the principle that Commission decisions generate normative stability. The Commission may refine its assessment of compatibility over time, but it cannot, ex post, contract the scope of protections or exceptions it has previously formulated in clear terms. In this sense, the judgment confirms that legal certainty functions not only as a shield for beneficiaries and Member States, but also as a “discipline” on the Commission’s own enforcement strategy.

If the MFZ litigation illustrated the consequences of Member State implementation departing from the conditions of an authorization decision, the present case demonstrates the reverse: the Commission, too, is bound by the legal architecture it constructs. That structural symmetry, while capable of generating tension in complex and evolving tax regimes, is essential for preserving predictability and coherence within the EU State aid system.

In case C-452/23Prezydent Miasta Mielca v Rzecznik Małych i Średnich Przedsiębiorców, the Grand Chamber of the CJEU delivered a preliminary ruling on a municipal property tax regime under the Polish Law on local taxes and charges (LLTC). The signal that this case deserves attention lies in that opening fact alone: it was decided by the Grand Chamber, an institutional choice that normally indicates that a question of particular doctrinal or systemic importance is at stake.

However, considering the type of tax involved and the substance of the ruling ultimately delivered, it is not immediately apparent that the case required clarification at the highest judicial formation. On the surface, the dispute concerns a rather formal and straightforward property tax exemption, hardly the kind of complex fiscal engineering that has characterized earlier landmark preliminary rulings in fiscal aid. And yet, the Grand Chamber stepped in. As I read and re-read the judgment while reflecting on how to structure this commentary, I kept asking myself: What is really going on here? I could not identify a doctrinal breakthrough or a critical selectivity question of such magnitude that would justify the Grand Chamber’s involvement.

Instead, this case seems to confirm perceptions I developed in my doctoral thesis about an imbalanced interplay. My reading of this case suggests a shift in the State aid control system—as the arena in which the interplay between the EU and Member States unfolds—concerning the evolving boundaries of fiscal autonomy and State aid control. What is at stake here is not a property tax exemption, but a recalibration of the boundary between fiscal autonomy and State aid control. For that reason, I must warn the reader: this commentary will not be entirely conventional. It contains some heterodox reflections on what this judgment represents within the broader trajectory of fiscal State aid law.

Background

Under Article 2(1) LLTC, property tax is levied on (1) land, (2) buildings or parts thereof, and (3) structures or parts thereof connected with carrying on an economic activity.61 Article 2(2) LLTC excludes agricultural land and forests from the property tax scope, except when they are used for carrying on an economic activity.62 Article 2(3)(4) LLTC exempts land used for road lanes or public roads, unless the land is connected with carrying on an economic activity other than maintaining public roads or operating toll motorways.63 The taxable amount is determined in Article 4(1) LLTC as follows: (1) for land, the surface area; (2) for buildings or parts thereof, the usable floor area; and (3) for structures or parts thereof, connected with economic activity, their values as defined by income-tax rules and depreciation bases.64

The part of the scheme that gave rise to the Supreme Administrative Court of Poland referring the matter as a preliminary ruling to the CJEU concerned Article 7(1)(1)(a) LLTC, which provides an exemption from property tax for “land, buildings and structures forming part of railway infrastructure” within the meaning of the rules on rail transport provided that they are “made available to rail carriers.”65

Main proceedings

Company E—owner of an individual railway siding—intended to make the siding available to a rail carrier and therefore applied for an advance tax ruling (ATR) confirming that it could benefit from the exemption laid down in Article 7(1) (1)(a) LLTC. The mayor refused the request, taking the view that granting the exemption would amount to conferring unnotified State aid.66 Company E challenged the ATR before the Regional Administrative Court, which upheld the mayor’s position.67 Company E then lodged an appeal on a point of law before the Supreme Administrative Court, which decided to stay the proceedings and refer the following questions to the CJEU:68

1. In the light of Article 107(1) [TFEU], does the grant by a Member State of tax relief addressed to all operators, such as that provided for in Article 7(1)(1)(a) of [the Law on local taxes and charges], consisting in an exemption from tax on immovable property for land, buildings and structures forming part of railway infrastructure within the meaning of the provisions on rail transport, which is made available to rail-transport operators, distort or threaten to distort competition?

2. If the answer to Question 1 is in the affirmative, is an operator which has availed itself of the tax exemption pursuant to the abovementioned provision of national law, introduced without following the required procedure, as laid down in Article 108(3) [TFEU], [read] in conjunction with Article 2 of [Regulation 2015/1589], required to pay the outstanding tax, plus interest?’

Answers to the questions

The Court declared the second question inadmissible on the ground that it was hypothetical, since Company E had not in fact benefited from the exemption, citing case-law to support its conclusion.69 Notwithstanding that dismissal, it is evident that, had the exemption been found to constitute unlawful aid implemented in breach of Article 108(3) TFEU, recovery with interest would, as a general rule, be required pursuant to Article 16(1) and (2) of Regulation 2015/1589.

As regards the first question, the Court reformulated it to address not only the distortion of the competition condition, expressly mentioned by the referring Court, but also the existence of a selective advantage.70 In doing so, it directed the analysis toward what is typically the decisive issue in fiscal State aid cases: whether the measure confers a selective tax advantage within the meaning of Article 107(1) TFEU—favors certain undertakings or the production of certain goods. This reframing put selectivity as the crux of the judgment (as usual).

Between paragraphs 36 and 80, however, the reasoning reads more as a structured restatement of the Court’s selectivity doctrine in direct taxation. The Grand Chamber appears to be telling a story beyond the Polish property tax, giving the judgment a somewhat unusual texture (what is the broader narrative the Grand Chamber is constructing?). Although Article 107(1) TFEU sets out four cumulative conditions—so that failure to satisfy one normally ends the State aid assessment—the Court nonetheless outlines the distorting criterion even after indicating that selectivity is not fulfilled. The discussion thus serves an explanatory, almost systematizing function, so that Member States understand a potential shift in the trajectory of State aid control.

Before this case, formal exemptions would easily meet the selective advantage condition where even one objectively comparable beneficiary was excluded from the regime.71 After this case, however, we need to be cautious with that presumption and reflect carefully on the underlying message the Grand Chamber is conveying. I explore it below through the analysis of the paragraphs that, in my view, deserve particular attention.

Starting with paragraph 50, after recalling that the characteristics constituting the tax define the reference framework—such as the taxable event and even exemptions—the Court introduces what appears to be a novel presumption: “a general and abstract exemption to which a direct tax is subject” is normally not to be classified as State aid.72 Such an exemption is presumed to be inherent in the normal tax regime and, as a rule, not to confer a selective advantage.

Paragraph 51 anchors that presumption explicitly in Member States fiscal autonomy. Direct taxation may legitimately pursue objectives of general interest—budgetary or otherwise—and those objectives may form part of the internal logic of the reference framework. Let us look more closely at paragraphs 50–51:

(50) Given that, in principle, the characteristics constituting the tax define the reference framework in the light of which the examination of the condition of selectivity must be carried out, a general and abstract exemption to which a direct tax is subject, such as the exemption at issue in the main proceedings, is normally not to be classified as ‘State aid’. In so far as that exemption is presumed to be inherent in the ‘normal’ tax regime, it cannot, as a general rule, confer a selective advantage for the purposes of Article 107(1) TFEU.

(51) Such a finding is derived from the autonomy which the Member States are recognised as having in the area of direct taxation, as has been recalled in paragraph 48 of the present judgment, as that autonomy means that those States have the possibility of making use of the tax classifications, and in particular of the tax exemptions, which they consider the most suitable for achieving the objectives of general interest pursued by those States, whether or not those objectives are tax-related. As was stated by the Advocate General, in essence, in point 33 of her Opinion, in the context of their fiscal autonomy, Member States may legitimately pursue, through direct taxation, in addition to a purely budgetary objective, one or more other objectives which, as the case may be, constitute, when taken together, the objective of the relevant reference framework.

For the moment, I set aside paragraph 52 and will return to it after completing the selective analysis, rather than following the Grand Chamber’s non-linear approach. Paragraphs 53 to 58 then qualify and refine the presumption introduced in paragraph 50.

First, the Court clarifies that even if a general and abstract exemption is normally part of the reference framework, the framework itself may still be incompatible with the EU State aid law, when it has been using manifestly discriminatory parameters intended to circumvent that EU law.73

Second, a general and abstract exemption will fall outside the “normal” regime—and may therefore be selective—if its conditions are linked in law or in fact to specific characteristics that define a consistent and defined category of undertakings.74 The Court exemplifies this by explaining the situation where the exemption effectively benefits only undertakings with a particular capital structure, operating in a specific sector or region, of a certain size, or with particular resource characteristics.75 The Court’s exemplification raises more doubt than clarifies the matter, since it did not bridge the discussion to how the property rail lines differ from the circumstances outlined in the LLTC.

By contrast, the Court explains that an exemption remains within the normal tax regime where the conditions for obtaining it are not intrinsically linked, in law or in fact, to the specific nature or activity of a defined and coherent category of undertakings. In such circumstances, the eligibility criteria are considered competition-neutral: the mere fact that some undertakings satisfy the conditions, while others do not, is not in itself relevant for State aid purposes.76

It becomes clearer what the Court intends when it acknowledges that every exemption subject to conditions will, by definition, benefit only those able to meet them. However, this structural limitation does not automatically render a measure selective. Selectivity does not arise merely because only a subset of undertakings qualifies.77 Paragraphs 56 and 57 of the ruling thus function as the Court’s general statement on the architecture of tax law, as a general measure (not selective).

The Court goes on to clarify that exemptions based on objective open-ended criteria—such as recruitment policies, or environmental measures—are not selective per se, provided those criteria do not, in practice, confine the benefit to a predetermined consistent category of undertakings.78 Read together (paragraphs 50 to 58), the Court is delineating what does not constitute selectivity (State aid), but rather falls within the scope of general measures “protected” by Member States’ autonomy.

What remains problematic, however, is the Court’s persistent abstraction. Across paragraphs 50 to 60, the Grand Chamber constructs a carefully layered doctrinal framework yet postpones any meaningful engagement with the concrete features of the Polish property tax. The reasoning operates at a high level of generality—defining what selectivity is not—without clearly testing those principles against the specific condition at issue: ownership of railway infrastructure attached to immovable property. The bridge between doctrine and factual design is largely implicit, and the national Court will have to puzzle to solve when applying this ruling to the main proceedings. This analytical distance makes the ruling appear methodologically elaborate, but substantively superficial.

The Court then observes in paragraphs 59 and 60:

(59) Furthermore, the fact that a tax exemption is granted irrespective of whether or not the persons subject to the tax to which it relates carry out an economic activity is an indicator that exemption falls within the reference framework.

(60) It should also be added that, even where it has to be regarded as not falling within the reference framework, a general and abstract exemption to which a direct tax is subject does not automatically become selective as a result. In such a situation, it is necessary to verify, as was indicated by the Advocate General, in essence, in point 28 of her Opinion, and as is apparent from paragraph 44 of the present judgment, whether the undertakings benefiting from that exemption are, in the light of the objective pursued by that reference framework, in a factual and legal situation comparable to that of undertakings not benefiting therefrom. If that is the case, the exemption will be regarded as selective, unless it can be shown that the differentiation between undertakings deriving therefrom flows from the nature or general structure of the system of which that exemption forms part.

The Court not only introduces terminology it has not consistently used before (such as “general and abstract exemption”), but also repeatedly constructs and deconstructs its own analytical framework without clearly anchoring it in the LLTR regime. Paragraphs 59 and 60, therefore, continue the discussion of what may or may not be selective, leaving the reader with more questions than answers. Nevertheless, it becomes apparent that the Court is preparing the ground for a finding of non-selectivity.

In paragraphs 61 to 74, the Court finally applies its doctrinal discussion on selectivity to the Polish exemption laid down in Article 7(1)(1)(a) LLTC. However, some uncertainty remains as to how the national court is expected to interpret and operationalize that exemption not clearly analyzed by the Court.

The Court identifies the relevant “normal” regime as the general Polish property tax system, reasoning that it applies broadly to all entities owning or holding immovable property and defines the taxable event, tax base, taxpayers, and rates.79 In other words, the exemption must be assessed within the overall structure and internal logic of the property tax as a whole and not in isolation.

It then examines whether the exemption forms part of that framework or derogates from it, based on the logic established in para. 54: an exemption remains part of the normal regime unless its conditions are linked to specific characteristics that define a coherent and closed category of undertakings.80 The condition at issue is straightforward: the owner must possess railway infrastructure that is made available to rail carriers.81 The Court observes that this criterion does not appear to be tied to the intrinsic nature, activity, legal form, size, or sector of particular undertakings.82 In this sense, any taxpayer, whether or not engaged in economic activity, could qualify. The potential beneficiaries, therefore, form a heterogeneous and open group rather than a consistent category.83 On that basis, and subject to verification by the national Court, the Court concludes that the criterion is thus structurally neutral and that the exemption forms part of the reference framework itself.84

Following this reasoning, any owner of a railway infrastructure could, in principle, benefit from the exemption. However, the Court remains silent on whether, in practice, the exemption is effectively available only to railway undertakings or closely related sectors—as it is inferred from Company E’s refusal at the ATR stage. Although the judgement repeatedly states that the exemption forms part of a reference framework, it does not clearly explain how Article 7(1)(1)(a) LLTC is intrinsically embedded—de jure and de facto—in the structure and logic of the Polish property tax system.

The Court then observes that the LLTC pursues not only a budgetary objective but also an environmental and transport-policy objective—namely, encouraging the use of rail infrastructure and promoting modal shift away from road transport.85 This part of the ruling is particularly relevant because the Court expressly states that the exemption “pursues an objective of an environmental nature, intended to encourage the undertakings concerned to restore disused railway sidings and to use rail transport, which does not entail carbon dioxide (CO2) emissions and provides greater safety than road transport.”86

Had the Court applied, by analogy, the same environmental reasoning developed in the Adria-Wien Pipeline, it might have framed the issue differently. It could have emphasized that “the ecological considerations underlying the national legislation at issue” are not dependent on who owns the property as such, but rather on whether the property owner makes the rail structure available to rail carriers so that it can be restored and used for more climate-efficient transport. From that perspective, the exemption would function as an instrument to ensure the environmental effectiveness of the LLTC, linking the tax measure objective directly to the exemption mechanism itself.87

Instead, the Court confines itself to clarifying that, within the scope of Member State fiscal autonomy, such non-budgetary objectives may legitimately form part of the internal logic of the tax system.88 It further concludes that there is no indication that the regime was designed using manifestly discriminatory parameters intended to circumvent State aid law.89 Taken together, these elements lead the Court to find that the exemption does not confer a selective advantage within the meaning of Article 107(1) TFEU.90

Conclusion about the Grand Chamber selective assessment

I respectfully disagree with the Court, particularly because the ruling provides limited concrete guidance—analysis—to the national Court on how to resolve the dispute in the main proceedings. The overall reading of the judgment suggests that the Grand Chamber considers the exemption to be non-selective simply because it is formally open to any owner of railway infrastructure, even though, in practice, other property owners seem to be able to equally fulfil those conditions (the case of Company E).

The denial to Company E suggests that the exemption’s formal neutrality may not reflect its practical operation aligned with the stated environmental objectives. As a result, Company E will likely remain without benefitting from the exemption, even though it appears capable of meeting conditions closely aligned, in my view, with the environmental purpose invoked by the Court itself. Ultimately, it will be for the national Court to assess the State aid conditions of the LLTC and apply this ruling to the main proceeding.

It remains difficult to reconcile the level of abstraction in the ruling with the practical effects of the exemption in the main proceedings. I will not dwell on the distortion of competition condition, as it comprises the weakest limb of the State aid analysis, given that a mere threat of distortion suffices to satisfy that condition under Article 107(1) TFEU. Now, I return to paragraph 52, which sits somewhat awkwardly positioned within the middle of the Court’s selective doctrinal framing.

A reassuring “wink” to Member States and a “stinky eye” to the Commission

Paragraph 52 deserves particular attention, as it is less about compatibility and more about constitutional boundaries. At first sight, it appears to be a side note on the Commission’s discretion under Article 107(3) TFEU. In reality, it performs a more structural function. The Court reassures that the Commission enjoys broad discretion when addressing compatibility under that Article. It then immediately warns that if every “general and abstract tax exemption” had to pass through that compatibility filter, the Commission’s assessment would risk systematically replacing that of the Member State, thereby encroaching upon their fiscal autonomy. As that is exactly the perception I had in my doctoral thesis when I raised the issue that Member States undergo through the notification procedure because the State aid control system is so difficult to predict, that this procedure became a safeguard of compliance to Article 107(1) TFEU prohibition—better to be safe than sorry.91 Paragraph 52 reads:

The Commission has a broad discretion to regard certain aid as compatible with the internal market under Article 107(3) TFEU (judgment of 31 January 2023, Commission v Braesch and Others, C-284/21 P, EU:C:2023:58, paragraph 94 and the case-law cited). If the exercise of that discretion were required to cover any general and abstract tax exemption, there would therefore be a risk of the Commission’s assessment systematically replacing the Member States’ assessment in the matter, thereby encroaching on their fiscal autonomy.

This is not an innocent remark. Formally, paragraph 52 concerns Article 107(3) TFEU. Substantively, however, it operates upstream, at the level of Article 107(1) TFEU. The Court is drawing a boundary: the Commission’s wide discretion in declaring aid compatible cannot become a mechanism through which general tax policy choices are indirectly harmonized. If general tax exemptions were too readily classified as aid, fiscal autonomy would be hollowed out not through prohibition, but through that procedural absorption.

Initially, this paragraph reads like a doctrinal interruption. It shifts attention from selectivity to institutional balance, making an already dense judgement even more layered (layers on top of layers). But closely reflecting on this paragraph, it resonates with the fear I raised in my doctoral thesis. Since the Adria-Wien Pipeline, lawmakers have increasingly been reluctant to pursue non-fiscal objectives through taxation, precisely because of the risk that such measures would be captured by State aid control as a non-notified aid.92

Paragraph 52 appears to be a calibration of the interplay between the EU (through the Commission) and Member States in the State aid control system. The Court appears to reassure Member States that pursuing pressing societal objectives (such as climate) through tax measures will not, by default, trigger Article 107(1) TFEU. At the same time, the message to the Commission is clear: compatibility control under Article 107(3) TFEU cannot become a substitute for Member States’ fiscal choice where no selective advantage exists.

We may be witnessing a certain maturation of the State aid control system, in which the CJEU appears increasingly willing to entrust substantive assessment—particularly the analysis of selectivity—to national Courts, rather than resolving every nuance at the EU level. This could be an encouragement to Member States to adopt more tax measures pursuing general-interest objectives and thereby strengthening their capacity to respond, through fiscal policy, to climate change and other sustainability issues.

At least, that is my hope.