State Aid Control in the Modernisation Era: Moving Towards a Differentiated Administrative Integration?

Despite contentions that state aid control is an instance of direct execution, the State Aid Modernisation (SAM) package has largely dispersed responsibility for giving effect to the common legal framework across national administrations. In this setup, the effectiveness of state aid law is confronted with asymmetric capacities, diffuse application of complex rules and difficulty in keeping dispersed powers under control. This article combines selected evidence on the operation of state aid control with a multidisciplinary analysis of the functioning of integration mechanisms in the European administrative space. The aim is to examine the potential effects of SAM. It contends that, while the reform has strengthened the Commission's dominance, the instruments of administrative integration might fail to secure adequate capacities and implementation performances at national level, due to constitutional constraints and opposing forces. This effect, it is argued, could be divisive, and risks undermining the integrity of the internal market.

addressing the structural problems arising from the decentralised architecture of state aid control?' This question is all the more pressing because, while the Commission has yet to provide an empirical evaluation of SAM, 11 the reform is nonetheless fully in place and proposals to extend its scope are on the table. 12 Given the wide ambitions and early phase of the reform, the article provides an initial overview of the operation of state aid control after SAM. This is achieved here by combining general knowledge on the effects of mechanisms of integration in EU law implementation derived from adjacent literatures -EU administrative law, EU governance and EU compliancewith a critical analysis of relevant legal sources, official documents and selected examples in the area of state aid.
After this introduction, the remainder of this article is organised as follows. Section two explains the main characteristics of the transformed organisational model of state aid control and the potential structural problems arising from it. Sections three and four evaluate the tools of administrative integration created with SAM and reflect on their expected effects in curbing the structural problems emerging from the reorganisation. The main argument is that, while the reform has extended the Commission's supranational dominance thanks to the close links it establishes with national state aid experts, its instruments of administrative integration might fail to secure adequate capacities, uniform implementation and coherent accountability mechanisms across the administrations involved, as constitutional constraints and conflicting forces might limit their effects. This raises concerns about the integrity of the internal market following SAM.

| SHARING RESPONSIBILITIES FOR STATE AID CONTROL AND THE EMERGING STRUCTURAL TENSIONS
Traditional scholarly narratives address state aid control as an instance of direct execution; 13 that is, a model whereby the Commission implements EU rules through its own structures. This classification stems from the distribution of competences laid down in the EU treaties. Pursuant to Article 108 of the Treaty on the Functioning of the European Union (TFEU), the Commission retains exclusive competence to decide whether national aid measures are compatible with the internal market. Member States are requested to block provisions until completion of a centralised assessment of their compatibility. 14 Due to the Commission's key position, much scholarly work has focused on the characteristics of EU-level enforcement, 15 on application of administrative rulemaking to limit the Commission's discretion, 16 and on the effects thereof on domestic obedience. 17 11 On 7 January 2019, the Commission launched, in line with its Better Regulation Guidelines, an evaluation of the rules adopted as part of SAM to assess whether they are fit for purpose and have delivered, in order to take a final decision on whether to further prolong them or possibly update them in the future. peculiarities that distinguish it from other sectors of EU administration. More specifically, unlike the adjacent field of competition law, the Commission retains overall responsibility for administrative functions. 28 This means that, like the Single Supervisory Mechanism (SSM) established by post-crisis EU-wide reform, state aid control is still operated in a model of direct administration. 29 Yet, by extending national institutions' implementation tasks and requiring increased national institutional capacity, SAM goes in an apparently opposite direction to the SSM and other reforms introduced after the crisis. 30 The resulting organisational arrangement is thus characterised by a co-existence of two opposing forcesone towards centralisation of control, and another towards dispersal of responsibilities. As these two forces may be driven by conflicting interests and pull in opposite directions, their co-existence might generate tensions around the common goal of giving effect to EU rules. Following Chiti's analysis of the structural problems affecting the EU administrative system, 31 we may single out three aspects of the administrative architecture of state aid control where such tensions could be particularly problematic.
First, SAM places a high premium on domestic implementation mechanisms to ensure effectiveness of state aid law, but this goal may be jeopardised by largely heterogeneous and asymmetric structures at national level. In many areas of EU law, scholarship has linked uneven implementation of supranational rules with domestic procedural limitations and asymmetries, as well as political unwillingness. 32 These risks appear particularly acute in the field of state aid. The high level of incompliance experienced under the previously centralised control has been attributed to Member States' lack of capacity, poor cooperation and the extraordinary variation of domestic structures. 33 In a model that brings further decentralisation, this problem is presumed to play an even greater part. This urges the Commission to develop effective mechanisms to shape adequate implementation capacity across domestic administrations. Yet, any pragmatic solutions in this regard must accommodate two opposing needs. First, under the principle of national institutional autonomy, domestic authorities can self-organise their own administrative practices and structures as they see fit. 34 Second, implementation activities are dispersed across a vast polycentric network that includes public and private bodies. 35  31 See E. Chiti, 'Is EU Administrative Law Failing in Some of its Crucial Tasks?', supra n 5. For Chiti, EU administrative law is failing to accomplish its main tasks necessary to ensure a stable functioning of the EU polity due to a number of structural problems. In the present paper, the argument is made that while these structural problems were also present in the model of state aid control prior to SAM, they become even more evident and urgent to resolve after the reform due to the redistribution of administrative activities occurring with the reform and the high level of interdependence across administrations that is needed to give effect to the supranational rules in the field. Therefore, Chiti's analysis is deployed here to emphasise the role of the instruments of integration created with SAM to address the structural problems mentioned. 32 On the general elements that explain correct or incorrect implementation of EU law by national administrations, see, among the many studies, E. Versluis, homogeneity in key organisational elements. 36 In this structure, promoting adequate implementation structures is not only constitutionally problematic, but also challenging in practice.
Second, application of state aid law raises complex issues of interpretation, 37 a feature that can provide breeding ground for instability and inconsistency of national behaviours. A typical instance is the notion of state aid set out in Article 107(1) TFEU. 38 As the notion represents, for the most part, an objective concept, national administrations and the Commission can apply it to specific measures, but the EU courts are responsible for its interpretation. 39 The problem is that, despite being an essential step in defining whether state aid law applies, such interpretation has become subject to endless variations, 40 due to the formulation of Article 107(1) TFEU as a rule of negative integration. Since the notion prohibits national measures that distort competition based on their expected effects on the internal market, its interpretation becomes dependent on the characteristics of the relevant domestic activity and the evolution of the relevant market conditions. This complexity, which is exacerbated by a lack of homogeneity in terms of subsidy measures across Member States, 41 suggests that, in the new decentralised architecture emerging from SAM, there is a need to ensure that national administrations can deal with this interpretative complexity effectively.
Third, accountability is crucial to keep the exercise of administrative powers under control, but requires reconsideration after SAM. 42 As traditionally understood, state aid control relies predominantly on a system of inter-administrative accountability through which the Commission checks any provision made at the Member State level. 43 Domestic courts, parliaments and state aid experts alike may act as tools for account-holding in the polycentric web of aid grantors, assisted by fire-alarm devices such as citizens and competitors. So far, however, these internal arrangements have displayed large variance in resources, powers and institutions, as well as in intensity and effectiveness. For instance, few countries have in place systems of mandatory control on domestic measures, while others use non-binding checks, voluntary self-assessment or ex post controls. 44 Yet, SAM increases the relevance of this broader accountability system. To compensate for its increasing focus on the most impactful cases, the Commission has expressed the intention to expand domestic tools of control and use state aid law as a mechanism to enhance the legitimacy of domestic expenses. The process started with SAM thus urges the design of a coherent accountability system that provides for effective and complementary structures of control to ensure the proper and legitimate exercise of powers across the administrative machinery. 36 See for instance on the characteristics of national administrations involved in the SSM and their independence, E. Chiti and F. Recine, 'The Single Supervisory Mechanism in Action', supra n 29, 112-114. 37 On the complex interpretation of state aid law, see A. Biondi  In sum, despite being an essential part of the European economic constitution, 45 state aid control appears vulnerable to structural problems that may endanger its effective functioning. Ultimately, each of these structural problems can be traced back to the unusually fragmented nature of the architecture, as administrative responsibilities are dispersed across a multiplicity of loosely coupled actors displaying dissimilar mechanisms and capacities. 46 While fragmentation has been always endemic in state aid, SAM's further decentralisation of tasks and limitation of ex post control make it more evident and urgent. This representation highlights the need for the Commission to develop effective mechanisms of administrative integration in order to avoid management deficits and, ultimately, incompliance. 47 Specific structures and processes are required that allow all relevant national and supranational administrations to cooperate closely and coordinate their activities to pursue the shared goal of giving effect to the common legal framework.
With SAM, the Commission has demonstrated that it takes these concerns seriously. While leaving existing control processes almost unaffected, it has established several new tools of administrative integration to manage and coordinate the activities of national institutions. In addition to clarifying key concepts of state aid law and fostering problemsolving through administrative rulemaking and networking, the Commission has sought to directly influence domestic practices and capacities through enhanced partnerships and conditionality agreements with the Member States.
To better appreciate the meaning and implications of SAM, a more fine-grained perspective is thus required in the study of state aid control. Beyond the traditional focus on direct execution, we need to expand our scope to examine the mechanisms of administrative integration created to connect the decentralised and fragmented infrastructure into a European administrative space (EAS). 48 The dilemma is whether these tools can effectively counter the structural problems that arise from SAM and ultimately ensure uniform application of rules across the whole architecture. This is the main challenge driving the following analysis.

| STRUCTURING ADMINISTRATIVE INTEGRATION FOR STATE AID CONTROL
This section discusses the functioning and potential effects of the mechanisms of administrative integration created with SAM. For this purpose, it adopts a twofold perspective. First, it clarifies the general characteristics and aims of each mechanism. In this sense, the Commission deploys with SAM a set of tools that makes use of different EU governance techniques to cope with potential pitfalls of decentralised implementation of state aid law. In addition to expanding the typical strategies of networking (through multilateral administrative networks) and convergence (through administrative rulemaking and strengthened partnerships), 49 the reform introduces the stronger novel approach of conditionality (through the ex ante conditionalities on ESI funds). Second, the analysis provides an early assessment of the potential effects of the tools established with SAM. It does so by combining theoretical expectations about the impact of each instrument on national implementation of EU policies with initial results experienced in practice in the field of state aid. While acknowledging some important improvements, the analysis elaborates also 45 F. De Cecco, State Aid and the European Economic Constitution (Hart Publishing, 2013); J. Piernas López, The Concept of State Aid Under EU Law, supra n 1. 46 In the present analysis, fragmentation is used to emphasise the plurality of actors, displaying a variety of organisational forms and practices, that, after SAM, exercise functions connected to state aid control. This understanding of the notion connects with the general observation of the large internal variation in the governance system giving effect to EU law, observed by different authors (see C. Harlow and R. Rawlings, Process and Procedure, supra n 34, 9-37). As such, fragmentation differs from the concept of decentralisation, and in the current analysis is used to clarify that SAM has led to more activities being exercised by Member States instead of by the Commission. on the limitations of each tool for delivering the Commission's ambitions with SAM to build national capacities from above and limit its centralised control in the area.
3.1 | The notice on the notion of state aid: the limits of steering domestic implementation through administrative rulemaking As noted, Member States typically struggle to identify state aid measures in practice. A main reason for this is their difficulty of applying the complex and evolving interpretation of the key legal concepts forming the notion of aid. 50 To address this concern, the Commission issued its Notice on the Notion of State Aid (the 'Notice') as part of SAM. 51 This document aims to clarify the Commission's understanding of Article 107(1) TFEU, with a view to contribute to an easier, more transparent and more consistent application of the notion of aid across the Union. 52 Since the correct qualification of a national provision as state aid is the first step to check whether the rest of the legal framework applies, the Notice represents one of SAM's cornerstones to build domestic compliance and limit the Commission's centralised control.
In legal terms, the Notice is an act of administrative rulemaking in accordance with Article 288(4) TFEU. This tool has often been deployed as a technique of regulation by information, to explain how openly worded and vaguely formulated rules are to be implemented. 53 Although this instrument lacks any direct binding force and legitimacy problems loom large, 54 its advantages are several: coping with complex political realities, 55 managing discretion and policy flexibility, 56 ensuring increased predictability and unity of administrative practices and sharing information. 57 In light of these benefits, the field of state aid has long been replete with administrative rulemaking. Traditionally, the Commission has issued unilateral rules to explain possible exceptions to the state aid prohibition laid down in Article 107(2) and (3) TFUE. 58 These provisions give the Commission the discretion to admit certain categories of aid based on their capacity to achieve objectives of 'common interest' in the internal market. In addition to shielding the Commission from political pressure and increasing time-effectiveness of its control, 59 administrative rulemaking in that area generates important legal and practical effects. First, administrative rules that structure how the Commission will use its discretionary powers in individual cases produce binding effects when combined with general principles of law. 60 As the Court of Justice clarified, the Commission may no longer derogate from them 'under pain of being found, where appropriate, to be in breach of the general principles of law, such as equal treatment or the 50 See the assessment of the European Court of Auditors, 'Do the Commission's procedures', supra n 7, para 28.  protection of legitimate expectations'. 61 Second, given that the Commission binds itself to enforce such rules in its day-to-day control, national administrations are brought into compliance to pre-empt potential investigations. 62 Through this proxy, the Commission has thus promoted positive integration 'from above' on national state aid policies. 63 The Notice marks however a departure from this practice. It represents the first example of administrative rulemaking's use to clarify Article 107(1) TFEU, a provision on which the Commission enjoys little margin of discretion. 64 This implies that the Notice represents for most an interpretative instrument pursuing mere informative purposes, 65 as it clarifies the Commission's understanding of an objective concept that the Court of Justice alone can interpret. 66 In practice, somewhat like the European Banking Authority (EBA) in the administrative mechanisms of the banking union, 67 the Commission has used the Notice to carve itself a 'meta-regulatory' role in order to build domestic compliance and limit its centralised control from above.
When seen against the background of misunderstandings and errors that has characterised national implementation in the field, post-legislative guidance that explains the concepts inherent to Article 107(1) TFEU represents a major development. It merits mention in this regard that actors have generally welcomed the document as a much-needed novelty. The Commission recently observed that the rules had become more familiar across national institutions. 68 This is partly attributable to the Notice which, thanks to its authoritative source, exerts powerful influence on the behaviour of national administrations. 69 In keeping with this, the document is discussed in a working group involving EU and national experts, 70 and is shared on many national coordination bodies' websites. 71 Yet, the Notice raises two specific issues that seem to constrain the Commission's ambitions to limit its centralised control and steer domestic compliance from above.
Firstly, by developing new or inaccurate interpretations of the treaties, the Notice risks triggering troubling consequences for legal certainty. Given the aim to contribute to an easier, more transparent and more consistent application of the notion of aid across the Union, one would expect the Notice to systematise complex issues already settled in case law, as interpretation is mostly a responsibility of the Court of Justice. Yet, on some points the fiscal aid granted through tax rulings falls under the scope of Article 107(1) TFEU. While this clarification is a welcome novelty in state aid investigations, it presents many uncertain points on which input from the Court of Justice is much needed. 72 Similarly, for the purpose of limiting its own investigation to bigger cases, the Commission introduces an understanding of the constitutive element of the 'effect on trade' of an aid measure that appears in conflict with case law on similar issues in other fields of EU law. 73 In addition to raising legitimacy concerns, 74 this use of administrative rulemaking looks questionable owing to the consequences it may generate. Outside the limited part where the Commission develops its own discretion, the Notice lacks legal effects to be challenged by the Court of Justice. 75 This means that the EU judges are precluded from direct scrutiny of the misleading understanding laid down in the Notice. Yet, the Court can always review the Commission's individual decisions that enforce the Notice on national measures. The problematic consequences of this are plain to see. A Member State, that treats the Notice as authoritative and complies with it, may be told by the Court that it has acted in violation of the treaties. At the same time, other national administrations would need to pinpoint themselves the impact of the judgment on the wording of the Notice and, since this activity may entail a more or less difficult interpretative effort, they would be tempted to start notifying cases to the Commission for reasons of legal certainty. Ultimately, private recipients relying on implementation of the Notice may be ordered to repay their grants. In sum, a mechanism created to introduce legal certainty and reduce the Commission's burden of investigation risks having the opposite outcomes. While these side effects have been somehow endemic to the Commission's control on individual measures, the use of the Notice to interpret the notion of aid have definitely made them more urgent. 76 At a more radical level, the question may be raised of whether a non-binding regulatory tool could be the most appropriate instrument to promote consistent understanding of a negative rule. As discussed earlier, 77 interpretation of Article 107(1) TFEU is in constant and fast-paced evolution, depending on the provisions that national administra- corner, unless the Notice is constantly updated as a 'living document'an onerous effort that the Commission has yet to promise.
Secondly, the Notice seems in danger of uneven application, depending on the actor that gives effect to it. 78 As noted, non-binding administrative rules that explain how legal provisions should be intended and applied may still exert important practical influence on national administrations. This seems to hold true for the Notice too, due to its origin from the primary agent of enforcement in the Union and to the wide discussion and general acceptance during its adoption process. 79 Yet, while significant practical effects can be reasonably expected, consistent application of the Notice at national level is not a foregone conclusion.
Previous research about how Member States' implementation of supranational guidance seems to confirm this threat. Domestic application of non-binding instruments has been found to depend largely on external incentives, such as supranational enforcement or peer pressure. 80 In the field of state aid, however, this force seems to be weak. Based on Expedia, 81 post-legislative guidance such as the Notice lacks coercive force on national administrations, and national courts are not mandated to enforce it in their judgments either. This means that the Commission's centralised control represents still the main strategy that could force Member States to ensure consistent application of the Notice. Yet, not only has the Commission's control already proved limited in ensuring consistent implementation of supranational rules by national administrations in the field 82 ; but the Commission itself has created SAM and the Notice to avoid precisely this consequence, namely to be consistently involved in checking domestic application of the rules in order to focus on the most important cases. The risk is thus that domestic authorities could be free to either cherry-pick elements of the Notice they will follow in daily practice, based on their interests to comply and the political costs connected to the provision of state aid, or opportunistically use the guidance to back up interpretations of the Treaties that would most favour their needs. 83 Additionally, compliance with non-binding guidance has been found to differ across countries, 84 as well as depending on the level of detail and complexity of the document. 85 Yet, as shown earlier, Member States still largely differ in terms of compliance performances and type of measures granted in the field of state aid. 86 Moreover, while the Notice has plenty of technical details, it principally aims to assist officials working for any granting authority, the majority being non-experts. 87 Implementation of the Notice is thus unlikely to be uniform across administrations, especially if internal coordination and control are missing. As such, while created 78 The author shares the same conclusion of A. Biondi  with the praiseworthy aim of contributing to a more consistent application of the notion of aid, the Notice's effects cannot be evaluated based on assumed efficiency gains. 88 Whereas its authoritative source is uncontested, it remains unclear whether this document could ensure uniform implementation of the notion of aid across the Union.
Given these limitations and risks, scope for legal uncertainty and uneven application may thus worryingly abound with the Notice. Both issues mentioned here, however, seem to undermine the capacity of the Commission to operate as executive centre to build compliance and limit its centralised control in the field of state aid. A second type is the network to strengthen transnational cooperation across Member States. This is found in the SAM working group organised by national experts to share best practices and debate common problems related to internal state aid control. Being an area protected by national procedural autonomy, discussions are held among national authorities, with the Commission acting as mere participant. 96 This multifaceted system illustrates how networking is deployed after SAM to hold the fragmented architecture of state aid control together. This objective is being pursued by establishing close interactions between the Commission and domestic state aid experts. These latter share information and provide advice on the application of state aid rules to other domestic administrations through multilateral networks, newsletters, databases and meetings. 97 Due to internal divisions of competences and specialised knowledge, they may also exercise accountability functions, to the extent that they could question and evaluate aid measures designed by other institutions, and at times even block certain provisions. In essence, they act as an essential linchpin between the supranational and national Yet, network governance also faces difficulties that cast doubt on its overall effectiveness. A first problem relates to the practical effects of multilateral networks on shaping domestic capacity, as these instruments seem prone to centrifugal forces generated by opposing national interests. One should not forget that, while cooperating closely with the Commission, state aid experts in most Member States are in ministerial departments and thus subordinate to their political principals. 102 In many sectors of EU governance, this dual loyalty or 'double-hattedness' has generated tension over common commitments across network participants, undermining network effectiveness. This consequence emerges particularly in times of confrontation and conflict over the application of common policies. In the field of state aid, such conflictual dynamics are likely to arise. 103 Given the high stakes often involved in the creation of aid measures, political representatives could be reluctant to build administrative capacity or comply with supranational rules. 104 This puts national state aid experts in a delicate position: since they primarily serve their national principals, they are unlikely to either disclose internal issues in supranational networks or to force domestic application of knowledge developed therein. 105 Unsurprisingly, effects on domestic capacity are still difficult to pin-point. As the Commission recently acknowledged, while well-equipped coordinating bodies were operative in some countries, others still lacked resources and expertise or had no influence on how state aid rules were executed. 106 To combat such centrifugal forces, scholarship on EU governance supports creation of reputational mechanismssuch as reporting obligations, review panels and scrutiny by peers. 107 Although unambiguous and conclusive evidence of their effects is lacking, 108 these network-level instruments may stimulate better coordination and achievement of common purposes across participants, in much the same fashion as 'new governance methods '. 109 Yet, the working groups introduced with SAM do not make systematic use of any such mechanisms needed to trigger inter-administrative accountability. As their description confirms, the existing networks intend to circulate interpretative guidance, present best practices and generate collective knowledge on shared problems. 110  Indeed, SAM networks rely on domestic experts' capacities to diffuse the knowledge developed to other national administrations and steer their administrative behaviours. The Commission's exclusive competences, however, may seriously hamper such capacity. This risk is particularly notable regarding domestic experts' support-giving and account-giving roles.
Firstly, since the Commission's exclusive competence means that only its centralised control may give legal certainty on how rules are to be applied, national state aid experts might find it difficult to support activities and solve , the Court of Justice considered that the measure at issue was the same as the one that gave rise to the Case C-262/12. By analogy, it concluded that Article 107(1) of TFEU must be interpreted as meaning that a mechanism for offsetting the additional costs imposed on undertakings because of an obligation to purchase electricity produced by power plants using solar radiative energy at a higher price than the market price that is financed by final consumers of electricity in the national territory, constitutes an intervention through state resources. 116

| Emergence of bilateral partnerships to strengthen domestic state aid control
As in other branches of EU law, execution of state aid law imposes great demands on internal coordination and capacity. With SAM, the Commission has concluded bilateral strengthened partnerships with national authorities to address this point. 118 This tool of EU governance is generally used to agree on solutions to country-specific needs linked to implementation of EU law. 119 By involving local stakeholders in a deliberative process that safeguards subsidiarity and national procedural autonomy, the instrument is presumed to garner a higher degree of acceptability within national authorities. At the same time, it enables the Commission to reinforce its position as executive centre and to target structural problems.
The Commission has signed formal non-binding agreements -'common understandings'with the central coor- The arrangement thus conveys the Commission's clear normative idea that an effective setup for state aid control 118 According to a presentation made by an official of the Commission in 2017, the Commission has concluded structured dialogue with 3 Member States, tailor-made cooperation with 9 of them and a case-driven cooperation with Czech Republic (see http://www.uohs.cz/download/Konference_a_seminare/ Konference-VP_kveten-2017/Da-Costa-Graca-SAM-after-3-years-final.pptx). While most of these partnerships are mainly focused on cases (oftentimes to co-define on which cases the Commission will give the priority in its investigation), the peculiarity of the partnership concluded with the Italian authorities is that the agreement was especially aimed at structuring internal administrative capacity. should consist of a polycentric network of national state aid experts that assist and monitor any potential focus of domestic incompliance. 128 Despite these positive effects, however, questions arise as to whether this tool is sufficient to build in-house capacity and control implementation at the national level. A first problem is that the goal to strengthen internal capacity might face national resistance and constitutional constraints. Being deployed to influence an area covered by national procedural autonomy, partnership agreements aimed at building domestic capacities depend on the goodwill and voluntary cooperation of national authorities. As empirical evidence corroborates, these factors vary widely across Member States. 129 Although the possibility is open to all countries, and compliance performance suggests that many Member States do face serious implementation issues, 130 partnership agreements have been concluded only with a few authorities prepared to accept the Commission's assistance. Moreover, as domestic constitutions usually protect the right to self-organise of national and subnational administrations, the arrangement may fall short in influencing each individual grantor within the signatory Member State. For instance, although capable of providing aid, the national parliament, private bodies and municipal authorities are not mentioned in the partnership agreement signed with the DPE 131 as much as the newly established Agency for the Territorial Cohesion, which is entrusted to coordinate implementation of EU cohesion policy. 132 Similarly, a recent study of the effects of the common understanding signed with the Italian authorities points to still-asymmetric resources, expertise and practices to implement state aid law across regional authorities and national ministries. 133 Because state aid rules have a wide scope that potentially cuts across any domestic provision of public money, these limitations could hamper the effects of strengthened partnership in structuring domestic capacity.
A second problem is that, due to the Commission's exclusive competence, the domestic polycentric network of national state aid experts might have limited practical influence in controlling rules application at the domestic level. In the Italian system, this weakness is particularly evident. The DPE recently observed that, while all regional and national administrations ultimately set up internal capacity to implement the rules, the functioning of the arrangement was watered-down by the reluctance of some granting authorities to submit measures to the control of their own bodies. 134 At the same time, inter-administrative checks at the national level might also be weak. For example, until recently, the Italian Ministry of Cultural Heritage and Activities and Tourism considered provisions to cultural activities as no-aid, even though they may entail the exercise of an economic activity and, as such, fall within state aid rules. 135 To combat the ministry's persistent refusal to notify, some regionswhich, according to the internal constitutional arrangement, are allowed to grant aid in a policy field pursuant to a national statute 136had no other option than to notify these measures to the Commission on the ministry's behalf. While this choice ultimately allowed the subsidy to be granted, it shows that duplication of control at the national level could fail to secure correct rule application and that the Commission could end up acting as an external mediator in domestic clashes. This is not necessarily a positive development in view of the Commission's limited resources and the goals of SAM. 137 Since the polycentric control system set out in the agreement with the Italian authorities is shared with other countries, its widespread promotion as best practice in domestic state aid control appears problematic at best.
3.4 | Ex ante conditionalities for ESI funds: Modelling domestic state aid control from the back door As in many post-crisis EU reforms, the Commission has tried to enhance domestic capacities in state aid through a more intrusive mechanism of administrative integration. To improve the effectiveness and efficiency of EU structural and investment funds (ESI funds), a type of subsidy that may entail application of state aid rules, 138 the 2013 budget reform imposed the conditionality principle on their use. Accordingly, payment of ESI funds is subject to a massive package of spending requirementsor ex ante conditionalitieswhich national authorities must fulfil. 139 Since state aid law has been the largest source of mistakes, 140 Member States are now requested to give evidence that national authorities have sufficient administrative capacity to execute state aid rules, as well as measures in place for staff training, information dissemination and effective application of the legal framework. 141 The logic of conditionality is not new in the EU legal system, 142 nor is it novel for state aid control. 143 During the enlargement process, conditionality served as a proxy to expand the acquis to Central and Eastern European countries and to induce them to adjust their administrative structures for EU membership, also in regard to state aid control. 144 The new ESI funding rules, however, relaunch spending conditionalities as a stand-alone governance tool to address structural deficiencies in implementation of EU policies connected to ESI funds. By requiring Member States to demonstrate administrative capacity under the threat of funding de-commitment, the Commission has deployed this tool as tin-opener to directly influence Member States' executive activities. 145 At the same time, the principle of partnership, entailing that all actors must cooperate to make the policy work, brings the benefit of negotiation and better-targeted solutions. Insofar as conditionalities combine the soft techniques of framework goals and deliberation with a stronger contractual logic, they seem to add a coercive edge to the mechanisms of integration discussed above.
Given these characteristics, ex ante conditionalities look set to enhance the effectiveness of state aid control in the area of ESI funding. Although the threat of funding de-commitment has yet to be tested, the partnership agreements do encourage transparency on the institutional capacities available for internal state aid control. In addition, Member States that lag in this regard have been pushed to elaborate institutional changes to catch up with others. In practice, the ex ante conditionalities appear to have extended to all institutions providing ESI funds a model of administrative governance entailing dispersed capacities and effective structures similar to those furthered in the strengthened partnerships seen above. 147 The challenge, however, is to verify to what extent an interfering tool like spending conditionality can influence the architecture of state aid control. In this regard, it may be noted that practical issues and constitutional constraints will likely impose limitations on their effectiveness.
First, empirical evidence shows that, like softer governance instruments, execution of conditionality depends on cooperation by Member States and adequate checks by the Commission. 148 Both elements, however, might be missing. Some national and subnational institutions have been remiss in duly fulfilling their commitments to the Commission, 149 also due to the substantial investments of time and resources required, which are not easy to secure in a context of austerity. 150 At the same time, the EU Court of Auditors recently noted that, as compliance with the ex ante conditionalities was based on self-assessment by Member States and the Commission had been ineffective in assessing the information provided, national administrations displaying severe problems in application of state aid law had not been duly addressed. 151 Second, questions arise regarding the scope of the conditionalities. A relatively small share of domestic aid grantors is required to exhibit their administrative capacity in order to grant ESI funds. In some countries, such as Italy, the Netherlands and Portugal, these operate only in ESI funding. 152 The problem is that constitutional constraints may also limit the effects of the conditionality instruments. While their constitutionality has yet to be tested, there is a tendency towards restrictive interpretation of conditions. When imposed by the EU institutions, a sufficiently direct link must be shown between the extended use of conditionality as a generalised governance tool to strengthen the condition itself and the measure to be funded. 153 In this regard, it may be noted that the use of conditionality for ESI funding is legally based on Article 317 TFEU. Pursuant to that provision, the Commission is directly responsible for implementation of the budget, and it executes this responsibility in accordance with rules adopted by the EU Parliament. As such, the Commission appears legitimised in imposing conditions linked to ESI of national administrative autonomy. 155 As often recognised by the EU institutions, the Commission has no competence in local execution of state aid law, since this falls within the exclusive responsibilities of the Member States.
Moreover, outside ESI funding, no same sufficient link is identifiable between the conditions imposed and measures financed, due to lack of EU co-funding. Therefore, while the Commission has tried to introduce intrusive administrative conditions impinging on state aid law, it has done so via the back door of ESI funding and has faced numerous limitations.

| TOWARDS A MULTI-SPEED STATE AID CONTROL: WILL THE CENTRE HOLD?
The thrust of the argument in the previous section is that the institutional configuration of state aid control changes pursuant to SAM. To allow the Commission to focus on bigger cases, the centralised architecture, based on hierarchical control and integration through law, has been supplemented by a complex web of mechanisms of administrative integration aiming to support rules application, develop capacity and structure coherent mechanisms of control. State aid thus seems akin to other fields of EU law where implementation is characterised by intense cooperation across EU and national actors. Yet, this analysis would also suggest that, in response to the structural problems emerging from the reorganisation process, SAM may be conducive to two deeper and apparently conflicting implications: increased supranational influence and emergence of a multi-speed administrative space.

| Divide et impera: Extended supranational influence across a decentralised architecture
Based on the above-discussed elements, the Commission appears with SAM to have extended its influence in the area of state aid. Its primary function as rule enforcer has been extended to include managerial tasks. These serve to integrate the dense network of domestic actors that now have broader responsibility for state aid control. In this new role, the Commission promotes policy discussion and information sharing to solve implementation problems. To create common understandings of key legal concepts and reduce centralised control, it uses administrative rulemaking beyond the traditional purpose of streamlining the exercise of its discretionary powers. Finally, by bargaining commitments of different intensity, it plays a decisive role in structuring the in-house capacity and control mechanisms of Member States, once areas considered outside its influence.
This leading role has been facilitated by the close cooperation created with national state aid experts. By establishing relatively tight connections and regular contacts, the Commission has de facto co-opted them as agents to ensure even implementation of EU rules throughout the decentralised and fragmented architecture. 156 As a consequence, civil servants specialised in state aid law can nowadays be identified in all Member States. 157 Albeit without the same legal powers and effects, they appear in principle capable of supplementing the Commission in each national system. In this role, in addition to implementing the supranational rules, they channel information, support and at times supervise behaviours of other authorities.
Overall, this has spurred emergence of a transnational epistemic community of state aid experts. 158 Through their constant interaction, the Commission and Member States are developing a common mentality on how state aid rules should be executed across the EU. 159 Furthermore, each is becoming more aware of how others operate. This effect Finally, this epistemic community appears to have penetrated beneath the surface of the central administration to reach local actors. Staff dealing with state aid policy in numerous Member States indeed maintain intense communication through a web of horizontal and vertical connections involving networks, newsletters, databases and informal contacts. 161 In practice, then, a long chain of integrated administrative structures seems to be emerging. These form an essential backbone to enable the new architecture to work. Yet, this structure is not based on a 'flat hierarchy' . Rather, it resembles an administrative pyramid with a clearly distinguishable top. Given the exclusive competence retained in this area, the Commission has a clear interest in operating as the infrastructure's main facilitator, in stimulating collective learning and in assisting implementation of EU rules, as well as in combining these activities with its traditional binding powers and the new conditionality approach.
To offset the structural problems that the reorganisation of state aid control may generate, a balance seems to have been found in a strengthened role for the Commission as executive centre in the area. This runs somewhat counter to the policy design pursued by SAM. As noted earlier, an initial goal of the package was to re-orient the Commission's control to the most relevant cases. Somewhat counterintuitively, the new architecture has produced more Commission interference in national activities in the field. Unsurprisingly, this evolution has increased the administrative burden and added internal complexities at the Commission level. 162 Yet, when considered in light of the field's rigid division of competences, it materialises as an extraordinary coup by the Commission, almost tantamount to the consequences produced by the modernisation of EU antitrust law. 163

| Administrative integration running amok and risks to the integrity of the internal market
The SAM package seems nonetheless to struggle in holding the whole administrative architecture together.
Conflictive forces and constitutional restraints risk limiting cooperation and coordination between national and EU administrations to address the abovementioned structural problems emerging from SAM. In such circumstances, a multi-speed administrative space seems to be emerging. That is, different levels of integration may be observed across the administrations involved in application of state aid law.
The idea of a multi-speedor flexible or variable geometry -EU was recently relaunched as a possible EU future scenario. 164 Roughly speaking, this means that the Union would embrace greater differentiation between Member States in degree of integration. In EU policymaking, this implies that Member States could opt out of common policies, select areas for enhanced cooperation and conclude international conventions with a restricted group of countries. 165 A slightly analogous process is observable in the after-crisis reforms of the EU administration. For example, in the SSM different administrative arrangements are emerging applicable to different groups of countries, beyond the cross-sectoral variation of administrative techniques typical of the EU administrative system. 166 In the field of state aid, a similar pattern of variable integration seems to be emerging, as the remits of some of the mechanisms described above allow links with only a subset of the actors involved in state aid law implementation. In addition, however, a multi-speed administrative space seems to arise from the uneven effects that the linking tools established with SAM could generate across the whole fragmented architecture. The potential consequence is twofold. Although beneficial for targeting specific problems and circumventing constitutional limitations, administrative integration at variable geometry might endanger the correct operation of state aid control by furthering development of inadequate implementation structures, different understandings of rules and malfunctioning accountability systems. Let us discuss each instance of differentiated integration and its effects on the post-SAM decentralised activities of state aid control.
As indicated, the reform design pursued by the Commission requires adequate implementation instruments be internally developed. This means that although the principles of proportionality and national procedural autonomy allow accommodation of structural heterogeneity across domestic administrations, a minimum degree of uniformity must be ensured in implementation mechanisms throughout the relevant administrations. Yet, while being created for this purpose, the instruments of integration established with SAM face limitations in achieving this goal.
First, though they are both helpful to reinforce internal structures, as discussed, administrative networks, strengthened partnerships and spending conditionalities rely on voluntary and correct implementation by the EU administrations involved. This could however give way to lack of cooperation or incompliance of domestic actors, as well as to lack of adequate control by the Commission. Those risks are substantial, given the long chain of actors whose activities are needed to give effect to the supranational provisions after SAM. Because the Commission normally interacts with only one or a few national representatives, and the latter lack competences or capacity to build up the implementation structures of other aid grantors, some actors risk failing to shape adequate internal capacity. Second, legal constraints could limit the scope of some of the tools created to tackle country-specific implementation problems. As noted, such invasive instruments as ex ante conditionalities can be applied only in the area of ESI funds. Of course, this reform may be beneficial to build up capacity in a sector where application of state aid law has traditionally been problematic. Yet, authorities involved in ESI funds represent a small share of the national executives tasked to implement state aid law, and the reform cannot be extended beyond this. Furthermore, strengthened partnerships to improve the domestic setup have been established with only a few Member States willing to cooperate more closely with the Commission. Yet, as discussed, figures on pending recovery decisions and Commission investigations suggest that other national authorities should also have been targeted by the Commission.
Ultimately, a cluttered administrative infrastructure seems to be emerging with gaps whereby uneven application is a likely consequence.
Similarly, although created to generate common understanding of how state aid rules must be applied in concrete cases, we cannot take for granted widespread effects of the administrative networks and the Notice throughout the fragmented system. Due to the intricate division of competences, knowledge produced through these instruments is non-binding and requires always an interpretative effort to be implemented. This leaves uniform application of state aid rules reliant on goodwill and effective coordination across domestic administrations. These conditions could be weakened, however, by conflicting interests or lack of involvement. This possibility is confirmed by preliminary evidence: while ultimately accomplishing the goal that at least 90% of aid measures be granted by Member States directly, SAM seems to have made little progress in raising the level of compliance in state aid law across national administrations. 167 An additional problem is generated by the ambiguous use of administrative rules made by the Commission to clarify the notion of aid. As the notion is in constant evolution and is interpreted in a rather intricate chain of control, the Commission's ambition to use the Notice to assist domestic authorities and ensure uniform 166 See in this sense E. Chiti, 'In the Aftermath of the Crisis', supra n 67, 318-322; C. Harlow and R. Rawlings, Process and Procedure, supra n 34, 292-294. 167 Speech by Director-General Laitenberger of DG Comp, 'State aid modernisation', supra n 68. application of rules could be weakened. Ultimately, different interpretations developed by the EU Courts or an imperfect application of national administrations might generate incompliance and negatively affect the legal certainty that private recipients need.
Finally, an analogous pattern may be emerging across the domestic actors established to supervise other actors in the area. Through partnership agreements, ex ante conditionalities and networks, the Commission has tried to establish a vast polycentric network of state aid experts that can supervise provisions granted by other domestic institutions. Yet, while definitely representing a step towards ensuring widespread checks on the application of rules, these instruments could fail to add up to a fully integrated accountability system. As demonstrated by the earliermentioned examples, national state aid experts might fail to apply supranational rules correctly themselves. Or they might fail to prevent other national administrations from misapplication, due to their lack of powers or limited capacities to address implementation failures. Either way, the integrated functioning of the EAS might be hampered, opening leeway for uneven application of rules.
The challenge here is that other systems of accountability can hardly fill these gaps. First, to refocus its centralised control on bigger cases pursuant to SAM's ambitions, the Commission offers non-binding guidance to assist national actors in their exercise of domestic tasks. 168 This has dubious legal value and effect, however, as it represents a mere prima facie assessment. 169 Since Commission monitoring of national measures is also limited and horizontal accountability across Member States largely underexploited, there is a risk that inter-administrative control could overlook domestic incompliance. Second, private enforcement of state aid rules is generally perceived as ineffective, because national courts often lack expertise in the area and private parties face constraints in demonstrating whether state aid law was violated. 170 Despite this, SAM has left this mechanism almost untouched. Third, due to a clear information imbalance, application of state aid law is hardly discussed in national and subnational parliaments. 171 In addition, although transparency has been a hallmark of SAM and the newly established register is certainly a positive step, opacity persists in how state aid law is applied at the national level in the vast majority of countries, as hardly anyone outside the administration is aware of how rules are intended and implemented in concrete measures. 172 The likely result is the emergence of uncoordinated structures to hold the fragmented administration to account.
On closer inspection, each instance of differentiated integration seems to relate to the ambiguous position enjoyed by the Commission in the structures created with SAM. Though acting as an equal partner with the Member States, the Commission retains exclusive competence to enforce Article 107 TFEU vis-à-vis national executives. This role duality might imply that both sidesthe Commission and national authoritiescan sometimes work with each other and other times work against each another. Clearly, when negligent or obstructive behaviours overwhelm national executives' cooperative attitudes, orderly EAS operation will be hampered. The Commission's centralised position and exclusive competence might prove double-edged, especially for the national state aid experts that are the essential linchpin making the whole infrastructure work. While close integration with the executive centre may afford these specialised officials a privileged position to influence their national peers, it may nevertheless reduce their effectiveness in addressing internal obstacles and incompliance, since it excludes the exercise of any binding 168 'Comfort letters' are soft law instruments issued upon a request from national institutions where the Commission, instead of providing a full-blown assessment, gives its opinion about whether a national measure complies with state aid law and is compatible with the internal market, based on the information issued. power at the national level. At the same time, while not being formally invested with the legal responsibility to interpret Article 107 TFEU, the Commission de facto exercises it with SAM. This however might limit the cooperation between the various administrations involved, especially when such interpretation is reversed or additional national level activities are requested to implement it. The ultimate consequence could be a breakdown of integration in the EAS, in which national authorities could 'take back control' on state aid measures and aid recipients could ultimately be treated unevenly. These risks were already present in the previous centralised system. Nevertheless, the new decentralised structure appears to have accentuated them, due to the combined effects of extended national responsibilities and limited supranational control.
Granted, multi-speed mechanisms of administrative integration may be advantageous and sometimes even a constitutional imperative in the area of state aid control. Not only might they afford the flexibility needed to tackle chronic issues of flawed decentralised activities, they could also provide a smart way for the Commission to make a virtue of its own constitutional competences, which may differ in scope across the policy areas in which state aid rules apply. In addition, one should not forget that subsidiarity and proportionality requirements may make heterogeneity necessary. 173 The essential question, however, is whether the system of integration is equipped to ensure that the fragmented architecture is capable of giving effect to the common legal framework. Looking at the resulting picture, some grounded fears might arise.
In sum, while this analysis constitutes a mere initial investigation of SAM's effects and, as such, faces inevitable limitations, it nevertheless suggests that the reform choice is not uncontroversial. SAM combines centralised control with distribution of administrative tasks across a fragmented and decentralised architecture. Constitutional constraints, problematic use of instruments of integration and doubts about commitments to a common purpose may generate tensions that hamper cooperation and coordination in the EAS. SAM might therefore struggle to integrate the components into a coherent whole. The problem is that the final outcome could be divisive and put the integrity of the internal market at risk. 174 Indeed, state aid law was created to check distortive aid provisions and ensure a level playing field throughout the EU internal market. 175 However, if rules are intended and implemented differently across the Member States or any of their territorial spaces, the ultimate goal of ensuring internal market integrity will be frustrated. This consequence would contravene the overall constitutional framework of the EU.

| CONCLUSION
This study represents a preliminary investigation of the potential effects of the complex transformations introduced with SAM. It provides an overall initial assessment of the reform based on scholarly insights, relevant legal sources and documents, and a number of examples. Once each actor gains sufficient experience in operating under the new infrastructure, the current analysis could be combined with investigations of selected case studies or longitudinal analyses on the post-reform evolution of the field. Yet, despite these limitations, the study makes some interesting observations about the potential of SAM to address the structural problems emerging from the reform process 173 Flexibility is intended here in its procedural dimension and pertains to the capacity of the instruments of integration to accommodate diversity in the administrative architecture, while subsidiarity focuses instead on the constitutional principle that allocates responsibilities at the most adequate governance level, thereby protecting national competences. As such, both concepts differ from the understanding of fragmentation explained supra n 46. See on the relationship between differentiation in EU and principles of law G. and affected by the tension between centralisation of control, on the one hand, and decentralisation and fragmentation of administrative tasks, on the other.
With SAM, the Commission has tried to effectively deal with these problems by establishing a number of instruments of administrative integration. 176 In addition to accommodating institutional diversity and constitutional autonomy, these instruments have been earmarked to influence domestic mechanisms, to ensure common understandings of rules and to structure coherent accountability regimes. Pursuant to this strategy, the Commission appears to have expanded its role as executive centre by closely integrating domestic state aid experts into its activities. Yet, the resulting administrative space appears to exhibit a pattern of integration at variable speeds, due to legal limitations and pressure exerted by opposing forces. Indeed, strengthened partnerships and conditionality arrangements have influenced domestic governance in only a few cooperative Member States and in a few selected areas, and sometimes even here have failed to do so fully. Moreover, the use of soft instruments to secure uniform execution of supranational rules may be hampered by national administrations' lack of cooperation and by their unambiguous use by the Commission. Finally, accountability structures have de facto multiplied under SAM, without however, providing definitive and coherent solutions outside the Commission's centralised control. As a side effect of all these struggles, integrity of the internal market and legal certainty are on the line, as the transformations triggered by SAM struggle to ensure uniform and consistent implementation of state aid law. While these problems are just starting to unfold, they will arguably emerge more strongly in the future. Responsibility for their resolution will ultimately fall to the Commission, as the executive centre, possibly with contributions by national administrations.
Therefore, whereas with the decentralisation of tasks introduced with SAM, it may appear inevitable and logical to focus Commission's attention on big cases, the effects of the reform warrant careful consideration, as they threaten to be divisive. Multi-speed integration across a fragmented administrative architecture risks unduly reinforcing national political interests and protectionism. As this could threaten the EU internal market's future, increased attention is urgently needed for the impacts of the reformed administrative governance of state aid control.