Contesting the European Central Bank in Banking Supervision: Accountability in Practice at the European Parliament

The establishment of the Single Supervisory Mechanism (SSM) raised expectations regarding the ability of the European Parliament (EP) to hold the European Central Bank (ECB) accountable for its decisions. This article examines the accountability interactions between the two institutions in the first years of the functioning of the SSM (2013&#8211;18). The focus is on the extent to which the EP contests ECB supervisory decisions in practice through letters and public hearings. The analysis shows a frequently&#8208;used infrastructure of political accountability that is however limited in ensuring the contestation of ECB conduct in banking supervision. The study identifies problems with the performance of the EP as an accountability forum and with the tight confidentiality rules of the SSM, which allow the ECB to silence contestation on many politically salient issues. The findings are based on an innovative analytical framework on the study of accountability interactions.


Introduction
The accountability of the European Central Bank (ECB) is a perpetual topic of interest in academic research in political science, law and economics (Amtenbrink and van Duin, 2009;Collignon and Diessner, 2016;Curtin, 2017;De Haan, 1997;Elgie, 2002;Magnette, 2000). Initial debates revolved around the mandate of the ECB in monetary policy, grounded in two notions established by European Union (EU) Treaties: first, the specification of a primary objective to maintain price stability (Article 282(2) TFEU) and second, the requirement of independence from other EU institutions and national governments (Article 282(3) TFEU). The challenge of the ECB was how to reconcile two seemingly conflicting goalson the one hand independence from political interference, and on the other hand the need to be accountable to a democratically elected authority (Braun, 2017;Dawson et al., 2019;Issing, 1999).
In this context, the relationship with the European Parliament (EP) was placed front and centre. In the late 1990s, the two institutions established a 'Monetary Dialogue' through which the ECB President would appear four times a year before the EP's Economic and Monetary Affairs (ECON) Committee, a possibility now envisaged in Article The article is structured as follows. The first part describes the relationship between the EP and the ECB in banking supervision in relation to principal-agent theory and the public administration literature on accountability. The goal is to contextualize the accountability relationship under investigation and underline the need for an approach that examines day-to-day interactions between MEPs and the ECB. The second part introduces an analytical framework for the study of accountability interactions, focused on instances of contestation. The third part provides an empirical analysis of accountability interactions between the ECB and the EP in the first four years of SSM activity. The conclusion problematizes the accountability shortfalls of the SSM and highlights the contribution of the article.

The SSM and Political Accountability
The relationship between the ECB and the EP in banking supervision is an example of central bank accountability and, more broadly, of political accountability between a specialized executive agency and a parliamentary body. At first glance, there are similarities to the classic principal-agent model in which an authorized principal delegated powers to an agent with the expertise and policy credibility to carry out specific tasks (Strøm, 2000). The idea is that in a representative democracy 'those authorized to make political decisions [the principals] conditionally designate others [the agents] to make such decisions in their name and place' (Müller et al., 2006, p. 19). In the principal-agent logic, accountability is the counterpart of delegation, based on the understanding that 'A is obliged to act in some way on behalf of B' and, in turn, that 'B is empowered (…) to sanction or reward A for her activities or performance in this capacity' (Fearon, 1999). Delegation and accountability thus go hand in hand, allowing principals to exercise continuous control over their agents. In the literature on central bank accountability, the emphasis on control is common among advocates of 'contracting,' i.e. the view that political 'principals' can control their monetary policy 'agents' if they create the optimal incentive structure (Persson and Tabellini, 1993;Schaling and Nolan, 1998;Walsh, 1995). EU banking supervision deviates from this logic because the EP is not technically the principal of the ECB in the field. The SSM was established by a Council Regulation adopted through a special legislative procedure in which the EP was only consulted (Art 127(6) TFEU). In fact, the EP had the same role in the process as the ECB, which was also consulted (Amtenbrink and Markakis, 2019, p. 9). From this perspective, national governments in the Council remain the principal of the ECB in banking supervision.
As a result, there are formal limitations to the EP's powers to hold the ECB accountable for its supervisory decisions. If the EP depends on the Council for revising the ECB's mandate in the SSM, then its ability to influence the incentive structure in which the agent operates (through 'contracting') is automatically curtailed. Furthermore, the SSM Regulation specifies that 'the ECB should exercise the supervisory tasks conferred on it in full independence, in particular free from undue political influence' (Recital 75; see also Article 19). This means that any recommendations made by the EP in its 'Resolutions on the Banking Union-Annual Reports' might have an informal impact on supervisory conduct but do not constitute formal mechanisms to sanction the ECB because parliamentary resolutions are never legally binding. On the plus side, the EP can veto the appointment of the Chair and Vice-Chair of the Supervisory Boarda power it does not hold over the Executive Board of the ECB (Amtenbrink and Markakis, 2019, pp. 14-15). Moreover, in case of poor performance or serious misconduct, the EP must approve the dismissal of the Chair of the Supervisory Board following a proposal from the ECB on the matter, but the decision is ultimately settled in the Council by qualified majority (SSM Regulation, Article 26(4)). In addition, the ECB has the obligation to 'cooperate sincerely with any investigations by the European Parliament, subject to the TFEU' (SSM Regulation, Article 20(9)).
For these reasons, the principal-agent framework is not particularly helpful to understand political accountability in the SSM. A more fruitful approach can be found in the public administration literature and the widely-used definition of the term proposed by Mark Bovens (2007). He describes accountability as the relationship between two partiesan actor and a forumwhich imposes certain obligations of the former before the latter. From this perspective, accountability is conceptualized in stages: first, there are formal or informal obligations for the actor's regular disclosure of information about its activities (the information stage); second, there are ex-post mechanisms to scrutinize the actor through requests for information and justification of conduct (the discussion stage); third, there are additional ex-post mechanisms to pass judgement on the behaviour of the actor and impose sanctions if necessary (the consequences stage) (Bovens, 2007, p. 451). In respect to the latter, what matters is the 'possibility of sanctions' and not their actual application in practice.
Bovens's understanding of accountability is well-suited to describe the relationship between the ECB ('the actor') and the EP ('the forum') in banking supervision. Indeed, the Interinstitutional Agreement signed by the two institutions in the SSM framework creates 'a formal obligation' for the ECB 'to render account' to the EP 'on a regular basis' (Bovens, 2007, p. 451). Specifically, the ECB has to present to the ECON Committee an annual report of its activities and transmit confidential, annotated minutes of Supervisory Board meetings that allow MEPs to understand the substance of the discussions and decisions taken (Interinstitutional Agreement, Articles 1, 4). The first stage in Bovens's conceptualization (providing information) is thus fully covered, albeit within the limits of the ECB's secrecy regime (Interinstitutional Agreement, Article 5). Conversely, the second and the third stage of Bovens's definition are not clearly distinguishable. Since the EP's ability to sanction the ECB is limited, 'discussions' between the two institutions represent the platform through which the ECB can face informal consequences, for example when MEPs make statements condemning supervisory decisions or ECB conduct. The SSM framework requires the Chair of the Supervisory Board to participate in ordinary and ad hoc public hearings at the ECON Committee and, upon request, in confidential meetings with Members of the Committee; at the same time, the ECB is obliged to respond in writing and within five weeks to written questions asked by MEPs (SSM Regulation, Article 20). Hearings and letters therefore constitute the main form of accountability interactions between the EP and the ECB in banking supervision.
The centrality of accountability interactions in the SSM poses, however, significant methodological challenges. The analysis of hearings and letters is inherently open to qualitative interpretation, making it difficult to evaluate the extent of accountability in the relationship. To address this problem, it is necessary to adopt a systematic approach to the study of accountability interactions and identify clear criteria for assessing them. The broader objective is to complement analyses of [legal] accountability frameworks by examining how actors and forums interact on a routine basis. The next section introduces such an approach.

Studying Accountability Interactions -An Analytical Framework
The study of accountability interactions requires the application of an analytical apparatus that is ontologically and epistemologically consistent. The approach proposed here is coined 'interactionist' in reference to Erving Goffman's concept of 'interaction orders', which denotes situations in which individuals find themselves in each other's presence and influence one another's actions (Goffman, 1983, p. 2). As a sociologist, Goffman was interested in physical interactions between individuals; in contrast, accountability interactions are typically formalized in an institutionalized setting, hence the encounter between actors and forums is constituted by organized hearings, meetings, and, even more frequently, the exchange of documents between the two parties. Accountability interactions can thus take the form of both verbal and written communication.
The advantage of the interactionist approach is that it deconstructs accountability interactions into their constituent parts centred around different types of contestation. The underlying assumption is that when actors and forum interact with each other in an accountability relationship, what happens are instances of contestation through which forums challenge an actor's past decisions and conduct. Drawing on Bovens (2007, p. 453), it is expected that forums ask questions demanding something from the actorinformation, justification of conduct, change of decisions or sanctions for responsible parties. The interactionist approach proposes to keep an inventory of the type of questions posed by forums and the type of answers provided by actors. Accordingly, the basic idea is that accountability interactions have at least three steps, in which: (I) the forum contests the decisions of the actor, (II) the actor silences, rejects or engages with said contestation and (III) the forum follows up on the issue or not, thus continuing or ending contestation on the matter. Alternatively, the forum might use legal means (if available) to force the actor to engage with contestation or change its behaviour. Figure 1 offers an overview of the framework, which is explained in what follows.
Contestation begins (column I) when a forum puts forth a request before an actor making at least one of the following demands: A, provide information about [the context of] a decision; B, justify the decision taken or explain conduct in a given situation; C, amend the decision or change the conduct in a specific or more general way; and D, sanction individuals considered at fault for the negative implications of the decision or the conduct. Requests of type A include demands for information about existing policies or the process through which decisions were taken. Requests of type B cover demands concerning the justification of conduct, asking the rationale why one line of action was preferred over another. Finally, requests of type C and D comprise demands made of actors to change their decisions or conduct and sanction the people deemed responsible for specific errors. In addition, there is the possibility that a forum asks a question to the actor that does not contest anything about its decisions or conduct (requests of type E, marked as 'outside the scope' in Figure 1). For example, the forum could simply demand the actor's policy views on a specific topic, without challenging past decisions or conduct in any way. Requests of type E are the first indicator of an accountability shortcoming.
Next, in response to the demands of the forum (column II), the actor can (1) silence contestation, (2) reject responsibility on the matter or (3) engage with contestation. These three possible responses can be identified for each type of request made by the forum, creating multiple categories of reactions through which the actor can respond to contestation. When actors silence contestation, they can do so fully (by evading an answer or invoking secrecy requirements) or partially by invoking secrecy requirements and giving a general answer about their policies. Secrecy requirements are common in certain policy areas, such as national security, economic stability, or in reference to the need for a 'space to think' during sensitive negotiations (Hillebrandt and Novak, 2016). Likewise, when replying to requests for change of conduct, actors can silence contestation by evading an answerpretending that the question was not posed and keeping their decisions and staff as before. Second, when actors reject responsibility for a decision, they direct the forum to the actor considered competent to provide information, justify the decision or conduct, change policy, or sanction personnel on the particular issue. The rejection of responsibility might occur for a valid reason (actual lack of competence), which is why this category is distinct from silencing contestation. The validity of the reason must be investigated on a case-by-case basis. Finally, when replying to requests for information and justification of conduct, actors can engage with contestation with different degrees of transparency, depending on whether questions from the forum are answered fully or partially. Simultaneously, actors can engage with contestation by defending their conduct and maintaining their decisions, by changing their conduct immediately or by promising to change their conduct in the future. The same logic is present for the application of sanctions.
In a third step (column III), it is possible to have (1) no follow-up to the actor's response, or action by the forum, to (2) reject the validity of the actor's answer or (3) use legal means to force the actor to answer satisfactorily or intervene directly on the matter. The three envisaged reactions of the forum can occur regardless of whether the request was for information, justification of conduct, change of policy, or personnel sanctions.
A lack of follow-up can be caused by the forum's loss of interest in the topic, its acknowledgement of the futility to continue asking questions because the actor will not provide a full reply, or simply by the forum's acceptance of the answer, which effectively ends contestation on the issue. Conversely, when the forum rejects the answer of the actor, it can continue contestation or take direct action by using legal means to force the actor to provide information or justification, by amending the actor's decision directly, or by imposing sanctions. The forum's direct intervention can occur by changing the legal framework in which the actor operates, dismissing personnel, or through a court ruling.
Following this line of thought, the evaluation of accountability in the interactionist approach is made along a continuum ranging from 'less accountability' to 'more accountability' (see vertical arrow pointing downwards in Figure 1). The continuum is valid within each of the three steps: first, requests for change of conduct and sanctions logically involve a higher level of contestation than requests for information and justification of conduct, so they are taken as an indicator of 'more accountability'. Second, an actor who silences contestation automatically raises red flags and provides a sign of 'less accountability': while there might be legitimate reasons for secrecy, there is a trade-off with accountabilityas the forum does not receive an answer in response to its contestation. Moreover, the rejection of responsibility can also be problematic if it results in a 'blame game' in which nobody is held accountable for poor decisions. In fact, the rejection of responsibility needs to be judged on its own merit, depending on whether the lack of competence on the matter is factual or not. The only clear instance of 'more accountability' in column II occurs when actors engage with contestation and provide a full or even partial answer to the question raised. Finally, when it comes to the follow-up by the forum, the three scenarios also signal different degrees of accountability. A lack of follow-up is empirically ambiguous because it can have different reasons: the forum might have lost interest or belief in the merit of the topic; alternatively, the forum might have simply accepted the response of the actor. Conversely, if the forum rejects the actor's justification of conduct (or the lack thereof), then contestation on that particular issue continues and can later result in concrete legal measures such as changing the actor's legal framework, forcing the disclosure of information, amending the decision directly, or applying sanctions. For this reason, the continuation of contestation and the use of legal means stand at the opposite ends of 'more accountability'.
From a methodological standpoint, the interactionist approach proposes to adapt the method of political claims analysis developed in the social movements literature (Koopmans and Statham, 1999). Political claims analysis is a form of discourse analysis which combines actor-centred and content-centred approaches: in an accountability context, it allows the researcher to link actors and forums with the content of their interactions, including their respective positions and frames of justification used to support them. In line with the interactionist approach (Figure 1), contestation is operationalized following the 5 types of requests made by forums in column I, the three categories (and their subclasses) of replies envisaged for actors in column II, and the three possibilities (and their subcategories) for forums to follow up on the response of the actor listed in column III.
Having established the main tenets of the interactionist approach, the article now turns to the specific case of the SSM and the accountability relationship between the ECB and the EP in this context.

Accountability Interactions in the SSM
The formal adoption of the SSM Regulation in October 2013 was soon followed by the establishment of an institutionalized dialogue between the EP and the ECB on banking supervision. The analysis presented here covers the period October 2013-April 2018 and includes 283 written letters exchanged between the two institutions and 13 public hearings of the Chair of the Supervisory Board at the ECON Committee. The Online Appendix provides a breakdown of letters exchanged over time (Table A1) as well as an overview of the relevant hearings held at the ECON Committee every year (Table A2). Using the qualitative data analysis software Atlas.ti, all letters and transcripts of public hearings were coded to identify single-topic questions and their corresponding answers. Single-topic questions could entail several interrogative sentences clustered around one specific issue, which meant that MEPs could ask multiple questions at a time. The analysis additionally includes an inventory of the nationality and political affiliation of MEPs who ask questions, available in the Online Appendix ( Figures A1-A4). Following the interactionist approach, the purpose here is to illustrate (1) how MEPs contest ECB supervisory decisions and conduct and (2) how the ECB silences, rejects, or engages with their contestation.

Contesting the ECB in Banking Supervision
In the framework of SSM accountability, the ECB and the EP interact on a regular basis. In over four years since the creation of the SSM, the analysis identified a total of 706 single-topic questions and a corresponding number of answers in letters (337) and hearings (369). In line with the interactionist approach, the questions of MEPs were categorized along five types of requests for information, justification, change of decisions/conduct, sanctions and policy views. Moreover, it was possible to distinguish questions asked for the first time (initial contestation) from questions on which MEPs followed up because they were dissatisfied with the original answer (continued contestation). There were no instances in which MEPs used legal means to force the ECB to provide an answer/change its conduct/impose sanctions. The latter would have been possible if MEPs set up a Committee of Inquiry to conduct investigations involving the ECB (see Title II of the Interinstitutional Agreement), but such a committee has not been so far established. There is also no evidence of MEPs formally accepting ECB answers as valid because they do not follow up positively on the written or oral answers of the Chair of the Supervisory Board. Consequently, only negative follow-ups can be identified, when MEPs continue contestation on a given issue. Table 1 offers an overview of the classification of questions posed in letters, hearings, and cumulatedin terms of both absolute values and percentages of the total.
There are several observations coming out of the table. First, 15.7 per cent of the total number of questions are 'requests for policy views', meaning demands for the ECB's expert opinion on ongoing legislative files or domestic developments. According to the interactionist approach (Figure 1), these cannot be considered accountability interactions. While it makes sense for MEPs to ask the ECB's expert opinion on their legislative activity, this can be done separately and not filed under accountability. In fact, the ECB is formally consulted on proposals from legislation that concern the EMU, responding to official requests for opinions from the Council, the EP, but also from national authorities (European Central Bank, n.d.). A discussion between MEPs and the Chair of the Supervisory Board on future legislative proposals is a form of inter-institutional interaction but does not fall in the category of 'being held accountable' because there is no contestation involved. Even if such questions have the potential to be confrontational and reflect disagreements between the two institutions on the future SSM legal framework, this is not supported by the evidence analysed. On the contrary, there is a tendency for MEPs who are rapporteurs on legislative dossiers to use hearings with the Chair of the Supervisory Board to ask for her expertise on issues relevant to them. Requests for policy views are consequently more common in hearings (22 per cent of all questions asked) than in letters (8.9 per cent of all questions sent). From the perspective of the interactionist approach, this reflects poorly on the performance of the forum, as more than a fifth of all hearings is wasted on questions that are outside the scope of accountability. Second, in both letters and hearings, MEPs tend to focus over 85 per cent of their questions on requests for information (49.7 per cent overall) and justification of conduct (36.3 per cent overall). This is not surprising, given the institutional independence of the ECB and the lack of political mechanisms to directly change decisions or impose sanctions (see also Amtenbrink and Markakis, 2019). What is interesting, however, is that out of the 296 requests for information and 216 demands for justification, almost half of the questions concern the situation at specific banks under the direct or indirect supervision of the ECB (for a breakdown of questions per topic and per year, see Table A3 and Figure A5 in the Online Appendix). Banks that attract the most attention are those that perform poorly in stress tests and have a high level of non-performing loans (NPLs) (for example the Italian banks Monte dei Paschi di Siena, Banca Popolare de Vicenza, and Veneto Banca), were formally declared likely-to-fail (for example the Spanish Banco Popular), or were considered to receive preferential treatment in stress tests (for example the German Deutsche Bank). Other examples include the resolution of less significant institutions (LSIs) (for example the Portuguese bank Banif) or the recapitalization of state-owned significant institutions (SIs) with the approval of the Commission (for example the Portuguese bank Caixa Geral de Depósitos). Unsurprisingly, these are also the banks that are most often mentioned in press reports regarding the performance of the SSM. However, given the professional secrecy requirements laid down in the Interinstitutional Agreement between the EP and the ECB and in the Capital Requirements Directive (CRD) IV, the ECB 'cannot comment on the interactions with individual supervised institutions or on the supervisory measures taken with regard to them' (Nouy, 2016a). There is thus a mismatch between the issue that MEPs care most about and the likelihood that they will receive the information they publicly seek. Contestation is bound to be limited from the outset.
Third, follow-up questions are more likely to occur in hearings (65.3 per cent of questions) than in letters (41.0 per cent of questions raised). This occurs because some hearings have a central topic that dominates the Q&A session. One example is the ordinary hearing from November 2017, when MEPs contested the draft Addendum to the ECB Guidance on NPLs as overstepping the institution's mandate. The Addendum, published for public consultation on the ECB website on 4 October 2017, deserves close attentionnot least because the document sparked consensus among MEPs from different political groups and attracted the attention of EP President Antonio Tajani. This is also one of few, straightforward instances of MEPs demanding concrete changes to the ECB's conduct. The Addendum addressed a persistent problem in banking supervision, namely how banks should deal with high levels of NPLs on their balance sheets. The document was designed to supplement the earlier ECB Guidance on the matter by providing quantitative supervisory expectations which set minimum levels of prudential provisions for new NPLs starting 1 January 2018. The ECON Committee, after receiving the opinion of its Legal Service, contested some of these supervisory expectations as ultra vires because they effectively introduced additional obligations for banks beyond the current regulatory framework (see opening remarks of Roberto Gualtieri, ECON Chair, at the hearing on 9 November 2017). Moreover, MEPs considered that the ECB did not give legislators and the public sufficient time to provide feedback on the Addendum, as its date of entry into force was less than three months from the publication of the draft version. Ms. Nouy acknowledged during the hearing that the phrasing of several provisions could be improved, as their meaning seems to have been misunderstood from what the ECB had intended. The required changes were made in the revised version of the Addendum, whose date of entry into force was also moved to 1 April 2018 (European Central Bank, 2018a, p. 7). Overall, the episode demonstrates the effectiveness of the EP as an accountability forum when there is a clear agenda, coordinated at the level of political groups and supported by the leadership, about what to ask from the ECB.
Moving to the range of answers provided (Table 2), we can observe that the ECB engages with contestation in about two thirds of all its replies. The percentage of answers engaging with contestation is slightly higher in hearings (73.5 per cent) than in letters (65.5 per cent), showing the general openness of the Chair of the Supervisory Board to explain issues in person. In line with the questions received, the majority of answers refer to the provision of information (32.2 per cent in letters and 34.0 per cent in hearings) and the justification of decisions (23.8 per cent in letters and 27.4 per cent in hearings). Overall, the ECB's track record in answering questions is fairly positive.
When does the ECB silence contestation? Here there is a difference in relative terms between oral and written questions, as instances of silencing contestation in letters are primarily supported by confidentiality requirements of the SSM legal framework, while instances Such answers concern questions that require information or justification of decisions regarding a specific supervised bank. In the early years, the ECB would address the requests by invoking its confidentiality regime and offering no answer. The rationale for secrecy is well-established among economists, who understand the role of central banks in guiding market expectations (Cukierman, 2009). If the ECB were to reveal sensitive information about specific credit institutions, this could undermine the competitive position of the bank on the market (which is illegal in the EU) and could trigger bank runs and panic in the population, endangering financial stability more generally (Angeloni, 2015). There are hence substantive reasons why the ECB cannot answer questions about specific banks. At the same time, there are those who advocate the need to democratize supervisory decisions by allowing the public to know if the supervisor is acting in their interestas opposed to the narrower interest of individual banks (Gandrud and Hallerberg, 2015). In response to frequent requests from MEPs regarding specific supervisory decisions, the SSM Chair started to provide over time general considerations about the bank in question and what the ECB does to address similar circumstances for any bank. Such generic answers are still considered an instance of silencing contestation in the interactionist approach because the questions of MEPs are dealt with expediently and unsatisfactorily from the viewpoint of the accountability forum. If some MEPs ask multiple rounds of questions on the same bank, 2 they stop and move on to different issuesaware that there is nothing they can legally do to force the ECB to provide information or justification.
Finally, there are those questions for which the ECB rejects responsibility because they are outside its mandate in banking supervision or because they concern issues within the competence of national supervisors (16.6 per cent in letters and 11.5 per cent in hearings). Such questions include (1) the methodology of stress tests, especially the choice of adverse scenarios, for which the European Banking Authority (EBA) is responsible; (2) the resolution of specific banks, where the Single Resolution Board (SRB), the European Commission and/or the Council are the relevant authorities; (3) decisions to object to or cap the provision of Emergency Liquidity Assistance (ELA) to banks in trouble, for which the ECB is responsible in its monetary policy capacity (and according to the principle of separation from banking supervision); (4) issues of consumer protection, especially concerning unfair practices of banks, where national bodies have jurisdiction; and (5) cases of financial misconduct and money laundering in different member states, where national authorities are also responsible. The problem here is not only the poor preparation of MEPs when sending questions to the ECB on banking supervision, but also the complexity of the banking unionwhich makes it difficult to understand the differences between banking regulation/supervision/resolution or between national and EU-level competences (see Amtenbrink and Markakis, 2019, pp. 19-20).
However, when it comes to the overlapping competences between the ECB and national competent authorities (NCAs), the evidence suggests a tendency of the ECB to distance itself from national supervisors. One controversial case is the Portuguese bank Banif, an LSI put into resolution in December 2015. In response to multiple questions about the supervisory process prior to the failing or likely-to-fail (FOLTF) decision, the SSM Chair invoked lack of competence and referred to Banco de Portugal as the 'right addressee' for the questions (Nouy, 2016b). If the ECB is supposed to hold national supervisors accountable for non-complianceas suggested by some authors (Karagianni and Scholten, 2018)this is not visible in the relationship towards the EP.

Conclusions
Overall, the interactions between the EP and the ECB in banking supervision paint a layered picture of political accountability. First, following the interactionist approach, the extent to which MEPs contest the supervisory decisions of the ECB remains limited. Too many of their questions simply ask for policy views or are not addressed to the relevant authority, which means that the ECB can provide answers without actually facing contestation of its conduct as a bank supervisor. Second, the SSM confidentiality rules prevent MEPs from receiving full answers to questions that contest the situation at individual supervised banks. The ECB can silence contestation by invoking the professional secrecy requirements to which the EP agreed in the Interinstitutional Agreement and CRD IV. However, it would be inaccurate to conclude that the ECB refuses to engage with contestation of its supervisory conduct in general. In fact, the ECB is open to engage with contestation in response to requests for information about the internal organization of the SSM and the decision-making process in banking supervision. Moreover, in the few cases where MEPs demanded a change of conduct, the ECB demonstrated willingness to address their requests and subsequently made the required adjustments. So far, it seems that the EP can exercise more accountability when it has evidenceprovided by internal parliamentary servicesthat the ECB acted outside the limits of its mandate in the SSM, as for example in the case of the 2017 Addendum to the ECB Guidance on NPLs.
The findings fall in line with the arguments presented recently by Amtenbrink and Markakis (2019) but contradict those of Fraccaroli et al. (2018). In respect to the former, there are similar conclusions regarding the high number of questions outside the competence of the ECB in banking supervision, the use of Bovens's definition of accountability, and the observation regarding the lack of sanctioning mechanisms available to the EP (Amtenbrink and Markakis, 2019, pp. 22-23). The added value of the present article lies in its focus on contestation, the 'discussion stage' in Bovens's definition of accountability. The implication is that the lack of formal sanctions would not be a problem in itself if MEPs asked relevant questions and the ECB was simultaneously responsive to their contestation. In relation to the analysis presented by Eule and Fraccaroli in Fraccaroli et al. (2018), the difference is that they take the increasing frequency of interactions between the two institutions as evidence of improved accountability. The present article shows that frequent interactions do not necessarily translate into 'more accountability'. The latter depends on the relevance of questions asked by MEPs and the corresponding responsiveness of the ECB, both of which were shown to display problems in the analysis conducted here.
Under the circumstances, the question is what can be done to improve the record of accountability interactions in the SSM. In line with the interactionist approach, what is needed is for MEPs to contest relevant issues regarding ECB conduct in banking supervision and, in turn, for the ECB to engage with contestation and change its decisions under specific circumstances. These two conditions require minimising the asymmetry of information between the two institutions, which is not an easy task. One possible solution is for MEPs to develop in-house expertise on banking supervision in order to ensure that their questions are addressed to the relevant institution while substantively contesting [something about] the ECB conduct in the SSM. Another avenue of reform is to revise the SSM confidentiality rules by identifying specific conditions under which supervisory decisions can be disclosed, for example after a sufficient period of time has passed or after a bank was declared FOLTF. This has been done to a limited extent in 2017 with the publication of non-confidential versions of FOLTF decisions (European Central Bank, 2018b). The idea to reform the professional secrecy standards applicable in the SSM is not novel, having already gathered the support of other scholars (for a comprehensive proposal, see Smits and Badenhoop, 2019). Such a reform would require a review of the SSM legal framework (SSM Regulation, CRD IV), but more importantly a change of approach from the ECB leadership, which the new Chair of the Supervisory Board Andrea Enria seems to favour (Enria, 2019). The political feasibility of this reform will be decided in the years to come.