Institutional Design of Enforcement in the EU: The Case of Financial Markets

Enforcement of EU law has become increasingly ‘Europeanized’. But how is and can it be organized in the integrated legal order of the EU to promote effective enforcement? In light of the recent institutional and substantive changes in the area of EU financial markets regulation, this article identifies four models (S, M, L, and XL models) of enforcement of EU law. It discusses the possibilities and challenges to effective enforcement of each of such models and the major trade-offs which policy-makers face at the EU and national levels when designing enforcement frameworks, namely centralization vs. decentralization (an institutional perspective) and harmonization vs. differentiation (substantive and procedural perspectives). It argues that at least a minimum degree of institutional centralization is necessary to promote the uniform enforcement and implementation of EU policies in a Union with 28 legal systems. The more specific details, such as specific institutional shape of centralized bodies (should it be a network, an agency or an EU institution?) and of the distribution of functions between the national and EU level are better addressed on a case-by-case basis in light of the political, economic, and social characteristics of the sector at stake.

identified challenges in advance when making choices between different enforcement models. While this article derives its conclusions focusing primarily on the financial sector, the proposed models and respective analysis could be useful in searching for the most optimal models in various areas of shared regulation and enforcement in the EU.
This article proceeds as follows. Part II gives a brief historical overview of the latest developments in the financial sector, which shows how enforcement in the financial sector has evolved from a somewhat loose informal to a rather centralized and institutionalized structure. The institutional evolution has been prompted by the transformation of powers from the national to the EU level. Mapping out the institutional characteristics, advantages and challenges of each step in the mentioned development helps to identify four models of how the enforcement of EU law can be organized. Four models are categorized on a scale ranging from less to more powers given from the national to the EU level or S (small), M (medium), L (large), and XL (extra-large) enforcement models. Part III offers a comparative analysis of the identified models as to their institutional structures, degrees of sharing and division of tasks, possibilities and challenges to effective enforcement that individual models may face. The discussion of the major trade-offs which policy-makers face at the EU and national levels when designing enforcement frameworks, namely centralization vs. decentralization (an institutional perspective) and harmonization vs. differentiation (substantive and procedural perspectives), follows in Part IV. The final part contains the conclusion (V).
of national representatives were the place for exchanging best practices and the issuing of guidelines and standards for the national authorities to follow as a result of, for instance, peer reviews of administrative regulation and regulatory practices in individual countries. Such standards and guidelines were neither legally binding nor enforceable rules which had only indirect effect via the national application practice. 8 As a result, in contrast to the Lamfalussy Report's expectation that the outcome of the work of these committees 'would be non-binding although clearly it would carry considerable authority' , 9 'experience has rather shown that measures agreed at Level 3 have not been applied consistently enough in the dayto-day supervisory practice of the national supervisors. This is sometimes reinforced by the fact that some regulators issue at national level guidance diverging from guidance agreed in the Level 3 Committees. ' 10 The problem here was two-fold. On the one hand, standards and guidelines adopted by networks of national supervisors were non-binding, which caused an inconsistent interpretation and application of the agreed standards at the national level. On the other hand, it was the absence of any binding power of the committee 'to tell a national supervisor that its interpretation of European law was wrong ' . 11 Next to that, a troubling moment was that 'differences on measures and procedures to authorise and supervise investment firms' 12 or, in other words, differences in administrative and criminal sanctions at the national level remained the case. In its report on sanctioning regimes in the financial sector issued a few years later the European Commission concluded that 'the way sanctions are designed in some Member States raises the question whether sanctions are fully effective, proportionate and dissuasive. Furthermore, divergences exist as to the level of enforcement. In some Member States no sanctions were applied for more than two years. The Commission considers that this could be symptomatic of a weak enforcement of EU rules. ' 13 All in all, in the words of the follow-up De Larosière Report 'national supervisors did not cooperate sufficiently to converge either supervisory practices or interpretations -whether the reason is to protect a national champion, restrict competition, preserve a national practice viewed as a competitive supervisory or regulatory advantage or just sheer bureaucratic inertia' . 14

Replacing networks with non-binding legal powers by EU agencies with legally-binding powers
To address existing inconsistencies and to enhance enforcement greater convergence of national regimes was thought to be necessary. This became especially clear in the course of the financial crisis of 2008 when the networks of national authorities failed to achieve a coordinated response to the crisis. In this light the 2008 financial crisis seems to have played an important role in speeding up political decisions on rather complex and important issues in a remarkably short period of time. 15 The new European System of Financial Supervision, which includes the creation of the EU financial agencies replacing the previous network structure, has been constructed in some 20 months. 16 The ESFS comprises the following: (i) European Systemic Risk Board (ESRB), which is responsible for identifying and assessing potential threats to financial stability arising from macro-economic developments and from developments within the financial system as a whole ('macro-prudential supervision'); (ii) European System of Financial Supervisors (ESFS), comprising a network of national financial supervisors working with new European Supervisory Authorities (ESAs) to safeguard financial soundness at the level of individual financial institutions and to protect consumers of financial services ('micro-prudential supervision'). 17 The former networked organization of national supervisory authorities has been replaced by three financial agencies, the European Banking Authority (EBA), the European Insurance and Occupational Pensions Authority (EIOPA) and the European Securities and Markets Authority (ESMA). The institutional change was accompanied by the centralization of power to enhance enforcement through harmonization.  Council's determination of the emergence situation ESA's decision to national supervisors to take specific action If the national authority does not comply, in some cases ESA can direct its decision to financial market participant And there is more. Regulation 513/2011 has made ESMA exclusively responsible for the registration and supervision, including the right to impose sanctions, of credit rating agencies in the EU. 19 This implies transferring almost all supervisory powers from the national to the EU level in the respective field. ESMA's supervisory powers have become extensive. These include the powers to examine and to take copies of any relevant records and material, to ask for an oral explanation, to summon and to hear persons, to require telephone and data traffic records, and to interview persons (Article 23(b)), powers not available even to the European Parliament in its investigatory capacity. ESMA's sanctioning arsenal includes the withdrawal of registrations, temporary prohibitions on issuing ratings, suspending the use of ratings for regulatory purposes, and the issuance of a public notice (Article 24). With respect to the remedy in the enforcement procedures it cannot choose any other remedy than to impose a fine (Article 24), the minimum and maximum amounts of which are fixed. This rather punitive enforcement method seems to have been circumvented in practice by using a risk-based enforcement style and adopting a wide margin of discretion. ESMA is likely to impose sanctions only in exceptional circumstances. 20 It is important to note that ESMA does not replace relevant national authorities since its work depends on close cooperation with them. According to relevant cooperation agreements national supervisors may be required to provide information, assist ESMA and carry out investigations and on-site inspections on ESMA's behalf. ESMA may also transfer certain tasks to them, though such a transfer may not dilute ESMA's authority. All in all, the new regime aims to ensure a stricter governing of both compliance and disclosure by means of centralizing power at the EU level.

The new supervisory tasks of the ECB
There is even more than what has been discussed so far. The European Central Bank (ECB) has assumed the new banking supervisory responsibilities. 21 'Due to the financial crisis, governments invested trillions of Euros to rescue their national banks, as a result of which the national financial situation of many Member States became fragile. To break the link between the Member States and their banks, the European Stability mechanisms (ESM) is proposed to recapitalize banks directly at the European level. ' 22 The idea is to create a single European supervisory system for banks under the command of the ECB.
As of November 2014, the ECB directly supervises significant credit institutions, 'representing almost 85% of total banking assets in the euro area' . 23 The ECB has powers to authorize and to withdraw authorization from credit institutions, to assess qualifying holdings, and to conduct supervision, to name but a few. 24 Next to the possibility to withdraw a bank authorization, its enforcement arsenal includes administrative sanctions. With respect to the supervision of significant credit institutions, national supervisors will thus assist the ECB in the daily assessments of respective banks and the implementation of the ECB's guidelines and regulations. Interestingly, even in relation to less significant credit institutions, the ECB will be able to issue regulations and instructions to the national supervisory authorities. While national supervisors will remain responsible for other issues, such as consumer protection, the combating of money laundering and other tasks not conferred upon the ECB, the ECB will become a very strong prudential supervisor with the exclusive and most important supervisory and enforcement powers over all significant credit institutions in the EU. 25

III. Four models of enforcement
The evolution of financial supervision in the EU shows a process of the transformation of national supervisory powers to the EU level. This 'Europeanization' trend implies an institutional evolution and functional transformation. It is a development of going from somewhat loose networks of national supervisors with non-legally-binding powers to a rather centralized and institutionalized structure of EU agencies and EU institutions with legally-binding and even overruling powers in relation to respective national supervisors. Within the framework of this development four enforcement models stand out. In light of the so-called Europeanization trend, these models are best categorized on a scale ranging from less to more powers given from the national to the EU level or S (small), M (medium), L (large), and XL (extra-large) enforcement models (see Table 3).

Institutional characteristics of the S, M, L, and XL models
The differences between the models are first of all noticeable in the institutional characteristics of cooperation. Whereas the institutional setting of all models feature the representation of all Member States with the possible involvement of the EU institutions (the Commission and the European Parliament) and relevant stakeholders, the setting in which they cooperate can differ from an entity with (the M and L models) and without (the S model) legal personality. The XL model significantly differs in this respect since in this model the centralized supervisory and enforcement tasks are given to a treaty-recognized institution, i.e., the European Central Bank. The institutional choice between a network, an agency and an EU institution clearly has implications for such questions as democratic legitimacy and accountability. An EU institution is in this respect the most legitimate option since it is established by the Treaties, which are silent about the possibility of creating networks and agencies. This also concerns the question of networks' and agencies' powers and accountability, though the accountability of the ECB is not without reservation because of the latter's independence. While these issues remain debatable, even after the most recent ESMA shortselling case (C-270/12), 26 where the CJEU confirmed the possibility to empower agencies with powers to take legallybinding decisions of general application, networks and agencies have been created in nearly every EU policy area; from energy to food safety, transport, vocational training and police cooperation. While the Lisbon Treaty has become the first treaty to mention EU agencies in about 27 of its provisions, including Article 263 TFEU on the possibility to review decisions issued by 'bodies, offices or agencies' , yet excluding networks, it does not establish any framework on when an agency can be set up, with what powers and by whom. The debate on agencies' legitimacy and accountability is hence likely to persist. 27

Functional characteristics of the S, M, L, and XL models
The sharing and division of (non-)legally-binding powers in relation to different enforcement aspects are the most remarkable differences between the identified models. One talks about the sharing of powers when similar tasks are exercised at both national and EU levels whereas the division of powers implies preventing functional overlaps by drawing the line between the powers of national authorities on the one hand, and of the EU institutions and agencies on the other. The fewer the tasks that are shared between the national and EU levels, the less Europeanization there is (the S model). The degree of Europeanization increases with transferring more legally-binding and exclusive powers to the EU level institutions and agencies (the M, L, and XL models). The financial supervisory system under the Level 3 committees is the example of the S model in this respect. These committees could provide cooperation and coordination between national authorities' actions which was done with the help of 'newly' created tasks in the sense that no transfer of (legallybinding) powers from the national to the EU level took place. The tasks of issuing non-binding guidelines and recommendations, promoting supervisory convergence and coordination, establishing common databases and standards and conducting peer-reviews were not previously exercised at the national level. The creation of a network of competent national supervisory authorities even without legally-binding powers made the 'promotion of pan-European supervisory convergence' 28 possible. This is different from the M model where similar tasks can be exercised at both national and EU levels. While pursuant to the most recent changes it remains the task of national supervisory authorities to ensure the compliance of EU financial markets regulation, in case of a failure of a national authority to apply the law properly the EU supervisory agency comes into play. According to their founding acts, the three financial agencies have received powers which imply the possibility to issue a formal opinion and via the Commission to request a national supervisor or a financial market participant to stop any violation of the EU financial markets law (the earlier discussed Article 17 and Table 1). In this way, the EU agency may exercise similar tasks, such as conducting investigations and issuing compliance decisions, to those of national supervisors. The task of ensuring the compliance of EU financial markets laws is in a way shared.
The L and XL models differ from the M model because a division, rather than a sharing, of tasks between national and EU supervisory authorities takes place. The peculiarity of the L model is that EU agencies have received the power to overrule national authorities in emergency circumstances (the earlier discussed Article 18 and Table 2). In the XL model, the delegation goes further than emergency circumstances: the EU level bodies receive exclusive tasks to regulate certain things. This concerns, for instance, the powers of ESMA under Regulation 513/2011 which makes ESMA exclusively responsible for the registration and supervision of credit rating agencies in the EU. 29 ESMA does not share registration and supervision powers with its national counterparts, although it does rely on cooperation with them. Rather, the tasks are divided between the EU and national level authorities concerning who does what in relation to the credit rating agencies. ESMA registers and sanctions the credit rating agencies whereas national authorities provide assistance which ESMA asks for and they may exercise tasks which ESMA can delegate to them. All in all, the degree of Europeanization in the L model is higher than in the M model because the EU level bodies receive emergency powers to overrule national authorities. The same is true for the XL model where the EU level institution (the ECB) or agency (ESMA) has received exclusive powers, thus assuming the tasks which could have been previously exercised by national supervisors.
In light of the discussed example of ESMA which characterizes several models (M-XL), it is important to note that the distinction between different enforcement models may not be that clear-cut in practice because different models may exist in relation to the same institution but with respect to different tasks that the institution performs. This distinction is important, as it allows one to show how differently individual models can address similar challenges, which is discussed next.

Possibilities and limits of the S, M, L, and XL models
The discussed institutional and functional developments of supervisory tasks in the field of EU financial regulation have been taking place with the aim of promoting the uniform implementation and enforcement of EU financial markets' laws. Yet, how effectively can the identified models ensure the enforcement of EU laws and rules? Each model offers some possibilities and is subject to certain limits in this respect.
The possibilities that the S model brings can be divided into three categories: expertise, coordination, and peer review. 30 First, since networks comprise representatives from regulation and enforcement bodies of the Member States and relevant stakeholders, they are in a good position to provide expert advice to the Commission in the course of the drafting process. 'Examples of this include the involvement of the CEBS in drafting detailed implementing measures for directives via the Lamfalussy process. ' 31 Second, with the help of accumulation and exchange of information about enforcement laws, regimes, strategies, good and bad practices of individual Member States networks devise common standards in order to promote the uniform application of EU laws. They provide a venue to coordinate and mitigate the differences among and within Member States in the absence of centralized supervision. In fact, networks could be seen as minimum centralized supervision thanks to its third task, peer review. Enforcement laws, regimes, strategies, and results are being reviewed and commented upon. Furthermore, 'the Commission bases its compliance reports and benchmark reports regarding the implementation status of directives on input from the networks. ' 32 Thus, by accumulating information, establishing common standards, checking compliance with EU laws in individual Member States and exchanging good and bad practices and contributing to the EU legislative process, networks can promote the harmonization of enforcement laws and practices. Thanks to these powers they can be important tools to foster cooperation and coordination between relevant national supervisory authorities. They cannot, however, address all the problematic areas in the field of enforcement in the EU multi-layered legal order. When differences between Member States' enforcement laws and practices adversely affect the uniform implementation of EU policies, a more advanced form of coordination and cooperation may be necessary. This is where the other enforcement models come into play.
All other models (M, L, and XL) can address the limit that the S model faces, namely the impossibility to enforce the EU law and the network's non-legally binding recommendations and to sanction noncompliance with laws, recommendations, guidelines, requests to cooperate, and the outcomes of mediation and conciliation procedures. This is because EU agencies, directly or via the Commission and the ECB, have legally-binding powers to either bypass the national supervisory authority when the latter fails to enforce the EU law, or to enforce the EU law directly by using the national supervisors as assistants. Thanks to the 'power of the stick' , enforcement is thought to be more effective than would have been the case under the S model, also with respect to the soft law that the agency passes.
The difference between, on the one hand, the M and L and, on the other, the XL models is, however, how they go about another challenge that the S models cannot cope with, namely the existing differences in the implementation and enforcement of EU laws and the network's recommendations due to legal, political, and cultural differences in the Member States. The M and L models allow for differentiation among national enforcement laws and practices since the EU supervisory agency steps in only in exceptional circumstances, i.e., when the national authority fails to enforce EU law and in the case of an emergency established by the Council. There is thus institutional and to a certain extent functional centralization accompanied by a significant degree of differentiation in relation to the enforcement laws and practices of individual Member States. As a result of the differentiation, however, the risk of having differences in the implementation and enforcement of EU laws and the network's recommendations due to legal, political, and cultural differences in the Member States remains. Such a risk is certainly mitigated by the overruling power of the EU agency, even though it can only be used in exceptional circumstances, but the risk is still there. Since the system is new, it remains to be seen how the EU financial supervisory authorities, directly or via the Commission, will use their supervisory and emergency powers. But it is important to bear in mind that the sanction which is never used may be tantamount to having no sanction at all.
Conversely, the XL model centralizes all the supervisory power at the EU level. By making an EU agency or an EU institution responsible for the registration and supervision, including sanctioning, of all credit rating agencies or credit institutions in the EU the adverse effects from the different implementation and enforcement of EU laws due to legal, political and cultural differences in the Member States can be eliminated. Yet, solving one problem creates another concern. The new system of sanctioning is rigid in the sense of not being responsive to different workable enforcement strategies of individual national supervisory authorities and specific cases.
To this end, it is important to note that there is no genuine consensus on which enforcement strategy is the most effective. Moreover, different enforcement strategies may be more and less effective in relation to different countries, sectors, and the number and size of actors that are to be supervised. As an example, a so-called responsive regulation, a strategy that combines both compliance and deterrence enforcement approaches, must be found. 'When types of regulation can be identified, a specific and targeted, rather than a pyramidic, enforcement strategy may constitute a more effective use of resources. ' 33 By opting only for a rather punitive enforcement strategy in the case of ESMA's supervision powers over credit ranking agencies, for example, the Union Legislator runs the risk of affecting enforcement in a negative sense by ignoring the existing established and workable 'softer' enforcement strategies tailored to the specific characteristics of individual Member States and their markets' specifics. For example, in the Netherlands the Netherlands Authority for Consumers and Markets (Autoriteit Consument & Markt, ACM) has explicitly chosen for a problem-solving enforcement strategy, where it 'subsequently selects the instrument or a combination of instruments that offers the highest probability of producing a structural solution of the problem. In that regard it is crucial to look at the entire range of formal and informal instruments at ACM's disposal. In certain cases, a fine imposed on an undertaking for a violation could be perfectly complemented with an awareness campaign aimed at the market's demand side (consumer empowerment). ACM has a wide range of statutory powers, and in order to fulfill our tasks as efficient as possible, we aim for swift and pragmatic solutions wherever possible. ' 34 All in all, while the identified four models can promote the implementation and enforcement of EU laws at the national level in different ways, they all face challenges to effective enforcement. Some challenges of the S models are attempted to be addressed by the M, L, and XL models, although resolving one problem may lead to another concern. What binds all the models together is the fact that they try to mitigate negative effects from the existing great differences in the laws and practices of national supervisory authorities which have resulted in having inconsistencies and disparities in applying EU law. While enhancing coordination and cooperation can indeed promote the uniform interpretation and convergence of enforcement regimes, strategies, and institutional settings at the national level, what is interesting to note is that none of the existing enforcement models has addressed the differences in the capacity of national supervisors. However, it is unrealistic to expect the same level of enforcement from competent national authorities which differ greatly as to their human and financial resources. In the federal states, such as the US, the enforcement of federal laws can be (co)financed from the federal budget in relation to those states which may lack necessary resources to implement the federal policy. Perhaps this is something for the Union Legislator to consider when it regulates and considers different enforcement models.

IV. Trade-offs in designing enforcement frameworks in the EU multi-layered legal order
So far this article has discussed the institutional and functional evolution of enforcement powers in the field of EU financial markets on the basis of which four models of how enforcement can be organized in the EU multi-layered legal order have been distinguished. In the case of financial supervision these models can be seen as an evolutionary process where the networked S model was replaced by EU agencies or by conferring more supervisory and enforcement powers on the EU institutions (the M, L, and XL models). Creating networks of national supervisors, delegating EU agencies with overruling enforcement powers in exceptional circumstances and exclusive powers in relation to credit ranking agencies as well as giving exclusive supervisory powers to the EU institutions over significant credit institutions are all examples of the different degrees of the 'Europe factor' chosen to supervise the financial markets at the EU level.
The identified models in this article are merely examples of possible ways to promote the uniform interpretation of EU laws and enforcement. While similar developments can be identified in other policy fields, 35 the discussed models may clearly differ depending on the specifics of other sectors, the 33 R. Baldwin et al., Understanding regulation. Theory, Strategy, andPractice, 2012, p. 231. 34 ACM Strategy Document, August 2014, p. 8, to be found at: <https://www.acm.nl/en/publications/publication/11993/Strategy-Netherlands-Authority-for-Consumers-and-Markets/> (last visited 21 November 2014). 35 Consider the establishment of networks of national regulatory and supervisory authorities and agencies and the powers of the EU competences of the EU in other fields, the scope of differences in enforcement laws and practices and the risks from having those differences among and within the Member States. At the same time while the specifics of other sectors may differ, which may, therefore, require different responses from the legislator in relation to how enforcement responsibilities could and should be shared and divided, there are a few caveats which the policy-makers will face in designing enforcement frameworks for any sector in the EU. This section discusses the two central dilemmas in this respect: centralization vs. decentralization (an institutional perspective) and harmonization vs. differentiation (substantive and procedural perspectives); see Figure 1.

Centralization vs. decentralization (an institutional perspective)
The scope of regulation at the EU level is governed by the principles of conferral, proportionality, and subsidiarity (Article 5 TFEU and Protocol 2). The enforcement of EU law is left to the Member States under the principle of national procedural autonomy, which is based on other principles determined by the case-law and Article 4 TFEU as interpreted by the CJEU, such as principles of loyal cooperation, effectiveness, equivalence, proportionality, and dissuasiveness. The enforcement system in relation to EU laws has hence been institutionally decentralized, accompanied by a differentiation in laws and practices per definition (point A in Figure 1).
Since disparities among EU Member States' enforcement laws and practices may result (and have allegedly resulted) in the ineffective enforcement of EU laws and hence have put the uniform implementation of EU policies at risk, a certain level of centralization seems essential. At a minimum, a uniform implementation and enforcement require cooperation, exchange of information and good practices and the existence of common databases. Therefore, a complete decentralization of the enforcement of common policies in most areas seems unwise and in fact hardly exists. 36 The question of whether 'to centralize or not' is hence not as relevant as the question of how much centralization is functionally necessary and politically possible or desirable (in terms of creating new agencies or strengthening the existing institutions and bodies with more powers). This question will be addressed differently by the experts in specific sectors and politicians depending on political, economic and societal circumstances in particular policy fields. Here, as the example of the financial markets shows, certain events, like a crisis, may speed up the process of centralization. Among other influential factors could also be the degree of internationalization of the economy: the higher the degree of internationalization, the more centralization could be feasible and necessary to coordinate EU national policies to withstand international pressures and developments.

Harmonization vs. differentiation (substantive and procedural perspectives)
Whereas the necessity of institutional centralization is clear, at least in its embryonic stage for the sake of coordination and cooperation efforts, the other trade-off concerns the question of whether 'to harmonize laws and practices or not' and, if so, to what extent (in substantive and procedural terms). This concerns the choice between B and C options of Figure 1. The B option implies institutional centralization, for instance, establishing a network of national regulators or an EU agency, which provides cooperation and coordination between national players. National laws and practices are not harmonized. The integration of supervisory and enforcement laws and practices goes further in option C by setting certain (minimum) common standards for all to comply with and making a central body or institution responsible for ensuring compliance with the set standards.
The advantages of the B option is that it allows differentiating between national laws and practices. It thus respects the differences between the EU Member States in relation to their cultural, political and policy fields' specifics. Here, for example, if the Netherlands uses less repressive enforcement strategies than Bulgaria in enforcing specific EU laws, both countries continue to enforce EU laws in their own ways as long as they can both implement EU laws properly. At the same time, it is argued that the existing differentiation results in, at least to a certain extent, an inconsistent application of EU law in national law. Hence, differentiation may jeopardize the effectiveness of the implementation of EU policies. In this light, the C option seems more advantageous since it can address the problem of eliminating the differences between laws and practices among the EU Member States. However, the cost of bringing all Member States under one standard involves making enforcement laws stricter than may be necessary in individual Member States and adversely affecting the innovation spirit in search of the most optimal supervisory and enforcement strategies to adapt to national and sectoral diversities.

V. Conclusion
Having analyzed the change in the institutional landscape of financial supervision in the EU in the recent decade, this article has identified four models of how the enforcement of EU law can be exercised and at what level. The centralization from an institutional perspective can take different institutional shapes from networks (the S model) to EU agencies and EU institutions (the M-XL models). The models differ as to the degree of sharing and the division of enforcement tasks between the national and EU levels from centralizing cooperation and coordination (the S model) to harmonizing regulation and enforcement laws and practices (the M-XL models). The identified models show different degrees of the Europeanization of enforcement responsibilities, which raises the question of what is the most optimal model for enforcement in the EU. We believe that at least a minimum degree of centralization is necessary to promote the uniform enforcement and implementation of EU policies. In a Union with 28 legal systems, it is impossible to ensure consistent and uniform implementation without information exchange, coordination of efforts, identification of (common) problems, differences and goals, and an exchange of best practices. Hence, centralization from an institutional perspective, by, for instance, creating common databases and platforms for exchanging information by the Commission or a network of national enforcement authorities is a must for the great majority of policy areas.
The questions of the institutional shape of centralized bodies (should it be a network, an agency or an EU institution?) and of the distribution of functions between the national and EU level are better addressed on a case-by-case basis in light of the political, economic, and social characteristics of the sector at stake. In a dynamic, highly internationally-orientated market, such as the financial sector, a stronger EU regulator seems necessary in order to maintain a coordinated policy and to withstand international pressure. When markets are more nationally focused and the risks are limited or arise primarily at a national level, 37 a strong EU regulator may be less necessary. Clearly, next to functional necessities, the political realities and other circumstances, like a crisis, will influence the meaning of what is an optimal model of enforcement in a specific policy area and at a given period of time in the EU. ¶