Accountability through mutual attunement: How can parliamentary hearings connect the elected and the unelected?

The increased authority delegated to independent agencies raises questions about the conditions of politically accountable governance, and specifically parliament’s role as a representative institution. Focusing on committee hearings as an accountability mechanism, we ask: How can a parliament employ hearings to ensure that the ends pursued by agencies have a democratic foundation? We propose a model of “mutual attunement” where accountability relations presuppose a process of working-out shared understandings of the ends, means and circumstances of policy needs. We test our argument through a case study assessing the interaction between the European Parliament’s Committee on Economic & Monetary Affairs and the European Securities and Markets Authority. Theoretically, we contribute to discussions on agency accountability and European governance, while providing a novel conceptual model and the first analysis of its kind.


Introduction
Independent agencies wield public authority at arm's length from elected representatives and partisan politics. The principles of democratic legitimacy, however, require that public authority is politically accountable to elected representatives. How can a parliament employ hearings to ensure that the ends pursued by agencies have a democratic foundation?
On the standard view, accountability presupposes a certain division of labour. Political bodies, like parliaments, choose the ends of policy; the role of independent agencies is to provide expertise regarding empirical consequences and to implement the adopted policy (Richardson, 2002;Vibert, 2007). Accountability can then be conceptualized in terms of a principal-agent relationship, where safeguards are institutionalized ex ante and performance control is exercised ex post.
On this account, independent agency expertise cannot be used to frame the political mandate itself, it is rather restricted to identifying empirical constraints: "expertise acts as a kind of external filter on the deliberations of other parts of the division of labour such as politicians and ordinary citizens" (Christiano, 2012: 42). In line with this, traditional principal-agent frameworks expect mechanisms such as written questions directed to agencies, agencies' annual parliamentary reports, and budgetary control to be used by the political actors as a source of technical information or reports on performance (Bach and Fleischer, 2012; van Rijsbergen and Foster, 2017). In a slogan, it is about the means of policy, not its ends.
In this article, we aim to contest the common conceptualization that ties accountability to a strict division of political labour. We will focus on hearings as a mechanism that can serve accountability interests through what we call "mutual attunement." In order for there to be a coherent mandate for independent agencies to comply with, there has to be a shared space of understanding. Ex post control measures cannot truly serve accountability unless the performance indicators are grounded in a sufficiently substantive justificatory relationship, which will be described in terms of an "authority of connection".
Insofar as hearings are governed by the aim of mutual attunement, we expect to observe three conditions. First, instead of a hierarchy where the principal sets ends and the agent reasons about the means, there will be reciprocal reasoning about ends. Second, there will an active interaction between actors where they constructively engage with questions and comments raised during the deliberation, rather than a passive statement of positions. Third, we expect to see a forward-looking outlook on policy that discusses potential future regulatory spaces, rather than a backward-looking account of the agency's actions.
We assess our conceptual argument through a case study. We conduct a content analysis of parliamentary hearings organized by the European Parliament's Committee on Economic & Financial Affairs (ECON) and its interaction with participants from the European Securities and Markets Authority (ESMA), illustrating a distinct mode of engaging with independent agencies. We conduct the first systematic study assessing the deliberation that takes place within committee hearings between the Members of the European Parliament (MEPs) and an agency. In doing so, we provide novel conceptual tools and material for further research into the relationship between agencies and elected representatives, and for normative assessments of its role in governance.
The EU context is an especially interesting test case, seeing as the literature has identified growing fears over weak accountability mechanisms and agency drift (e.g., Busuioc, 2013;Dawson et al., 2015;). Our argument is not that such fears are unwarranted, but rather that the standards of assessment should track a feasible and normatively attractive model of accountability relationships. Most scholars agree that EU agencies are not making purely technical decisions (Busuioc and Rimkut_ e, 2020;Egeberg and Trondal, 2017). Arguably, efficient agencies are impossible on a restrictive reading of the Meroni doctrine that constrains delegation of political discretion (Everson et al., 2014). Our argument brings out why the political nature of agency reasoning about political ends is not in and of itself a threat to accountability. What matters is that agencies pursue ends in ways that are appropriately attuned to the reason-giving processes of politically representative bodies, such as parliamentary committees.

Background
Before presenting the details of how accountability is served by mutual attunement, it is worth situating it in the theoretical accountability landscape. Specifically, to frame it against the backdrop of other alternatives to traditional control-oriented principal-agent approaches.
In developing a contrast to the control-oriented perspective on accountability, an important precursor is the "transaction-cost approach" that sees agencies as independent "fiduciaries". The approach's key claim is that "only an independent delegate, not subject to the power of direction of the delegating authority, can provide credibility to long-term policy commitment" (Majone, 2001: 69). Notably, in operationalizing independence, the idea of "not being subject to the power of direction" has in part been interpreted as not having "obligations to parliament" (Gilardi, 2002: 882-883). Presumably, having to participate in hearings is taken to imply reduced independence.
By contrast, our approach highlights that the appropriate form of political independence does not require deliberative separation. Talk of "insulating regulators from the political process" (Majone, 2001: 66) overplays the "credible commitment" aspect of delegation and underplays the expertise-providing function. Insofar as delegation is partly a matter of creating epistemic credibility, and not merely solving time-inconsistency problems, independence may support a complementary, knowledge-based form of inter-institutional reasoning with elected politicians.
If parliamentary hearings function as arenas for mutual counteraction of myopia and clarification of intentions, then hearings are conducive to cooperative independence. Agencies are still fiduciaries in one sense of the word; to trust someone is in part to stand in a relationship of mutual justification, as opposed to merely relying on someone or something to perform as expected (cf. the distinction between normative and predictive trust in Hollis, 1998: 10-11).
In recent years, the principal-agent framework has also been challenged by the "reputational" perspective on accountability (Busuioc and Lodge, 2016;Carpenter and Krause, 2012). A key point here is that accountability practices may serve the interests of account-givers as well as account-holders. It gives agencies a chance to track expectations and manage reputational risks (Busuioc and Lodge, 2016: 95-95). Regarding EU agencies, a key virtue of this approach is how it renounces a strict division of labor between "political" principals and "purely technical" agents: "regulatory legitimacy is much more of a moving target as agencies dynamically interact with, respond to, and attempt to shape audience expectations" (Busuioc and Rimkut_ e, 2020: 1261-1262. While our mutual attunement approach certainly agrees with the claims regarding the political nature of EU agencies and the importance of dynamic interaction, it does not go so far as to say that "accountability is not about reducing 'information asymmetry,' moral duties, containing agency losses, or ensuring that agents stay committed to the original terms of their mandate" (Busuioc and Lodge, 2016: 92). Accountability interaction may have certain reputational drivers that enable predictive claims about how organizational images will be managed. Nevertheless, concerns of commitment to mandate and reducing information asymmetries are not thereby analytically extraneous to accountability practices. As Jerry Mashaw puts it, accountability is about preventing "the (inappropriate) use of rules of behaviour that apply in one realm of human action in another" (2006: 119). Insofar as accountability is about managing warranted expectations, as opposed to mere strategic indulging of unreasonable requests, the interaction must somehow track fidelity to the entrusted mission. Hence, if we interpreted fidelity to mandate broadly as compliance with norms or principles that apply in the agency's domain of action, then that counts as a minimal constraint on what to count as accountability interaction (Eriksen, 2020).
In this regard, the concept of mutual attunement is a normative model of how mandates (broadly understood) can figure in dynamic interaction; in part, the mandate-constituting norms and principles are developed and elaborated through a process of inter-institutional reasoning that creates a shared space of meaning. In turn, this provides a sphere of answerability that can be tracked by more hierarchical or backward-looking modes of accountability (of the sort traditionally considered in principal-agent approaches).
Finally, the Calibrated Public Accountability model represents another relevant suggestion for how to "substitute or supplement" the principal-agent approach (Schillemans, 2016(Schillemans, : 1415. Drawing on experimental research, the model describes how key attributes of accountability practices affect decision-making. For our purposes, the most interesting claim is that "when the forum is 'owner' of a standard, it will guard the standard of evaluation more vigorously which, in turn, quite naturally stimulates the agent to take the norm more seriously as well" (Schillemans, 2016(Schillemans, : 1410. Here, "owning" means having devised the standard. A key conjecture on our part, however, is that in the case of expert agencies it is doubtful that a forum can "own" an effective standard that is also not justifiable to the agent. That is because expert agencies may prefer sanctions or irritating principals to the alternative of compromising their principles and professional judgment (Pollack, 2007: 7;Waterton and Wynne, 2004: 101-102). By contrast, co-ownership of the standard makes the agency answerable in a deeper and more bespoke sense; it must explain and justify how its behaviour is responsive to the kinds of reasons it has propounded and assented to. Hence, the idea of mutual attunement may be conceived as another tool for "calibrating" accountability practices to behavioural mechanisms.

Two kinds of authority
In this section, we want to clarify the accountability question by unpacking two distinct modes of engagement. Drawing on Anthony Simon Laden's differentiation between the authority of command and the authority of connection (2012, Ch. 2), we explain how the traditional principal-agent approaches misses a key feature of accountability.
The authority of command is the most familiar kind of authority. It is about having the unilateral standing to change the normative environment. Addressees of this kind of authority are subordinates liable to receive instructions or sanctions. By contrast, the authority of connection leads to a conception of accountability that does not presuppose that political intentions are settled. The authority of connection concerns an essentially mutual answerability, where both parties shape a shared normative environment. In terms of standards of political interaction, this has much in common with the ideas of reciprocity and reasonableness associated with deliberative democracy (Gutmann and Thompson, 1996;Rawls, 2005). What Laden brings out, however, is that there is a form of authority involved in relations of reciprocity. It is the authority to confront others with considerations that must be responded to in terms of reasons rather than mere volition or decisional fiat (Laden, 2012: 66-67). While one cannot command any specific action, one has the standing to demand that proposals be heard and given a reasoned response.
On the face of it, independent agencies do not have the political standing that the authority of connection requires. Formally speaking, they are executive or technical bodies, and as such they are considered end-takers rather end-shapers when it comes to political questions. However, this picture is misleadingly coarsegrained, and, in the end, it may obscure the conditions of a feasible and normatively attractive model of accountability. That is because the picture does not capture parliaments as potential addressees of political considerations articulated by independent agencies. In what sense are they addressees?
They are addressees in the sense that ignoring political reasons presented by independent agencies may detract from the legitimacy of the decision-making process. Independent agencies are institutionally committed to pursue public interest in a way that is guided by non-partisan considerations and ongoing consultation with relevant stakeholders (Mashaw, 2018;Pettit, 2004). Many administrative procedures (e.g. stakeholder meetings, consultations) are designed to promote impartial and inclusive reasoning as opposed to mere compliance with settled political intentions of the elected politicians. Regarding the authority of connection, the important point is that both representatives of independent agencies and MEPs are bound by a commitment to the common good rather than mere partisan strategy or non-public interests (cf. Lord, 2011: 916; on the public reason of parties more generally, see Bonotti, 2017).
Arguably, this joins the two bodies in a way that enables the authority of connection. Naturally, they have different areas of expertise and are bound by distinct standards of argument, but independent agencies may have a legitimate standing to reason with the parliament about what ought to be done (ends), not simply what can be done (means). We call this process mutual attunement.
The argument pursued in this article is not that the accountability of independent agencies should be conceptualized in terms of one kind of authority rather than the other. Instead, both the authority of command and the authority of connection are necessary features of the accountability relationship to parliaments. The point is that command without connection with independent agencies does not serve accountability. Insofar as accountability is supposed to be a virtue of institutions, mere authority of command may be morally reckless given the access independent agencies have to relevant public reasons.

Parliamentary hearings and measuring modes of authority
Hearings in most congresses and parliaments allow for the exchange of views between members of the parliament and various other actors over policy issues Katsaitis, 2015, 2019;Leyden, 1995). Hearings offer the grounds for interactions between elected representatives within specialized committees and agencies that fall under their political responsibility (i.e. where they have agency oversight). Significantly, because of their deliberative nature, hearings may provide a venue for mutual attunement.
Drawing on the work on social reasoning (Laden, 2012) and deliberative approaches to policy-making (Cohen and Sabel, 2006;Eriksen and Fossum, 2012), we identify three main criteria that can be used to assess whether we are observing authority of connection or authority of command. These criteria respond inversely under different modes of authority. The criteria are: (i.) relationship; (ii.) interaction; and (iii.) orientation. We discuss these measures below and present them concisely in Table 1.
Relationship: Hierarchy vs. Reciprocity. The principal-agent understanding of accountability assumes a distinct hierarchy under an authority of command, where the political principal dictates to the agent the limits and scope of its powers. In the context of deliberation, participants can appeal to their position (rank) to resolve disagreements, and/or force perspectives. Therefore, the participants are unequal in terms of formal authority, which spills over to their discussion in an observable manner.
By contrast, the authority of connection is grounded in reciprocal answerability. Each participant must appeal to reasons rather than mere expressions of will. As we will understand it here, the form of reciprocity required for connection can manifest itself against a background of institutional hierarchy. In our analysis, the relevant sign is how parties back their opinions and specifically their disagreements; are they appealing to mutually shared standards as opposed to mere decisional fiat? Do political representatives attempt to force their perspective on to the discussion? As Laden nicely puts it "particular instances of the authority of connection are not wielded like a sword, but jointly constructed like a bridge" (Laden, 2012: 72). Hence, based on this view, rejecting a suggestion is more like dissent than disobedience. In our analysis, we are concerned with how disagreements are dealt with; are they couched in terms of compliance or acknowledgment? If there is an authority of command structure in place, we expect to observe disagreements to be solved with final decisions made by the committee members.
Interaction: Active vs. Passive. Under a traditional system of agency oversight, due to the separation of ends and means, we expect principal and agent to exchange views on a broader theme without necessarily engaging in debate. Therefore, each institutional actor engages with a different aspect of the discussion's theme but not with the points raised by its counterpart. That is to say, under an authority of command, the communication between agency and the committee should resemble a series of monologues rather than a discussion.
Under authority of connection, there is a joint shaping of the normative environment. This requires that participants have the capacity to seek common ground by appeal to mutually acceptable reasons. It is a capacity to issue and respond to proposals, invitations, and questions rather than merely assertions, instructions and answers. This capacity must be exercised on both sides of the relationship; speech-acts like proposals, invitations, and question are unsuccessful without appropriate uptake and response.
Therefore, we consider how policy ends are shaped; is there genuine engagement or is one part merely subservient? We expect that under an authority of connection, elected representatives and agency representatives interact through discussion rather than passive speech reading that serves a theatrical management of expectations. They discuss and mutually attempt to shape means and ends together. Conversely, under an authority of command we expect a passive interaction between representatives and agency.
Orientation: Forward-looking vs. backward-looking: Having the standing (or de jure authority) to change the normative situation can be treated either as settled in the past or as depending on the ongoing interpretation of the relationship. The command perspective takes a backward-looking perspective; actors have been given prerogatives for unilateral use, and their authority is independent of the agreement of the addressee. In the connection perspective on authority, by contrast, the credentials are dependent on the interaction between participants. That is, the normative credentials of speech-acts depend on their ability to engage with the others in a way that is taken seriously and that enables mutual attunement. The authority of a proposal or invitation is to some extent acceptance-dependent.
Under an authority of connection, we expect that agency and representatives engage in discussions primarily over future policy actions in an attempt to reach common ground over forthcoming expectations, rather than assess actions in the past. Conversely, under an authority of command model we expect that agency addresses past actions, which it reports to its political principals.

Why expect authority of connection?
Having clarified the two modes of interaction and the associated measures, why should we expect to see one rather than the other? Some work on the EP's role in the accountability relationship with agencies has focused on the role of the budgetary committees, where a principal-agent relationship of ex post control has been identified (Bach and Fleischer, 2012: 161-162). Do we have reason to suppose things will be different in the specialized ECON Committee?
Based on a recent overview of the accountability practices ESMA is subjected to, one would suspect not. It suggests that political accountability involves the EP's and ECON's ability to "interrogate the actor and to question the adequacy of the information or the legitimacy of the conduct" (van Rijsbergen and Foster, 2017: 68). Practices of interrogation are much closer to authority of command than authority of connection.
Nevertheless, there are reasons for the EP and ESMA to seek mutual attunement under the authority of connection. In this section, we explain three reasons for expecting of authority of connection, where each consideration highlights general aspects of relationships between elected politicians and agencies.
The first is uncertainty. As they attempt to regulate in the face of unknown unknowns, neither the agency nor the elected representatives can be sure about the line between means and ends. Moreover, key regulatory terms like reasonable precaution and proportionality tie professional considerations up with political values in complex ways. Thus, to address the constantly evolving regulatory demands of any domain, the responsible institutional players need to work out a shared space of reasons.
The second is mutual dependence. Often, the public image of an institution is connected to another institution's performance. An agency's reputation is linked to the public acceptability of the ends it pursues, which means it has an interest in engaging in evaluative matters regarding legislation (Carpenter, 2010). Conversely, the parliament's reputation is linked to its capacity to enable efficient promotion of the public interest. For instance, limited support of an agency may turn it impotent, which in turn affects the parliament's public standing.
The third, which is particularly relevant in governance settings beyond the state, is the dynamic nature of institutional relationships. For example, it has been argued that the EU should be seen as a form of "deliberative polyarchy," where, at the limit, principal-agent accountability gives way to peer-review Sabel, 1997, 2004). Moreover, legal scholars doubt that a strong separation of powers between the legislative and executive branch along functional lines is either feasible or normatively attractive given the institutional realities (Carolan and Curtin, 2018).
We are not suggesting that these expectations are equally warranted in all specialized parliamentary oversight committees. As a counterweight, here are three scope conditions on the authority of connection. First, when interacting with an agency with a comparatively non-technical mandate politicians are less epistemically dependent and may have less incentive to attune themselves to agency judgment. Second, if an agency has a bad or controversial public reputation committees may have an interest in appearing firm and critical in the authority of command mode. Third, partisan interests of committee members may diverge from the agency's mandate, making the interaction more strategic or confrontational. We do not think these scope conditions are particularly salient in the case of ECON and ESMA. The agency is fairly technical and recent, and it was created in response to a broadly recognized problem.

Research design: Selecting a case study
To assess our expectations, we require information on the discussions held between a parliament and an agency during committee hearings. We focus on the EU context for two reasons. First, the explosion of EU agencies has led to a rich literature examining accountability relations (Wonka and Rittberger, 2010). Whereas the specialized committees of the EP have been identified as the "locus" of political accountability vis-a`-vis EU agencies (Busuioc, 2013: 116), researchers employ traditional approaches to assess its powers, such as MEPs written questions, agency parliamentary reports, and budgetary controls.
While there is no doubt that EU agencies are held accountable by the EP under an authority of command, we argue that it is only a mode of authority in place within a broader system, where different modes of authority co-exist. Therefore, the EU provides fertile ground to test complementary accountability frameworks.
Second, following the financial crisis the EP, and specifically ECON, gained substantive policy-making powers due to the Europeanization of financial regulation (Schoeller and H eritier, 2019). Part of this move included the creation of the European Supervisory Framework, and the creation of ESMA (along with the EBA, and the EIOPA they form the ESA). We chose to focus on ESMA because it is an important example of a recent move towards delegating additional authority to agencies in the EU such as, direct intervention and supervisory powers (Moloney, 2011).
Indicatively, the agency was the centre of attention in a much-debated case that the United Kingdom brought before the Court of Justice of the European Union, where precisely the mandated political discretion of the agency was a core matter of contention (C-270/12). ESMA is therefore already a salient agency when it comes to conceptualizing the political judgment exercised in supranational regulatory practice.
The agency mentions on its website that it is: "an independent EU Authority that contributes to safeguarding the stability of the European Union's financial system by enhancing the protection of investors and promoting stable and orderly financial markets". Nevertheless, ESMA continues: "Whilst ESMA is independent, there is full accountability towards the European Parliament where it appears before the Economic and Monetary Affairs Committee (ECON), at their request for formal hearings." 1 Therefore, from the EP's perspective, we decided to focus on ECON to which ESMA is directly accountable. The Committee is responsible for policy linked to the economic and monetary union. Moreover, it is responsible for the regulation of financial services, the free movement of capital and payments, taxation and competition policies, and policy linked to the international financial system. The EP's committee hearings are recorded and available to the broader public through the EP's online archive. To assess the type of authority in place during ECON's hearings, we used the available search engine and located all ECON hearings where ESMA was a participant. We found seven (7) hearings between 2011 and 2017 where ESMA was included; the relevant hearings were transcribed using f4transcript software, aided by a research assistant.
We conducted a content analysis taking into consideration our expectations and the outlined measures. Content analysis is a systematic examination and interpretation of a body of material in an effort to identify patterns and variation (see Berg, 2009) There are different types of content analysis which depend on the degree of inductive reasoning applied (see Hsieh and Shannon, 2005). In this article, we conducted a two-layered analysis. We first conducted a directed content analysis of the speakers' statements, which involves creating coding categories that have been derived from existing theories, in this case we developed measures drawing from deliberative theory and social reasoning (see Table 1).
To assess our measures, we require a unit of analysis. Considering the speech patterns observed we employed each speaker's statement as the unit of measurement. Using these counts as our context, we assessed to what degree our developed measures revealed an authority of command or an authority of connection. Against a theme, we coded whether it corresponded to our measures (0/1). For example, a theme could be forward looking (1) or not (0) or backward looking (1) or not (0); each theme was tested against each measure (orientation, relationship, interaction). We provide the breakdown in the appendix, and a summative figure and table in the analysis (see Appendix).
We would like to highlight that our analysis assessed first the manifest meaning of the statements, and following examined the potential latent meanings within each theme if we deemed there was one. We are particularly interested in latent meanings because we are assessing the themes within each speaker's statement, but also the potential reaction to the themes by the other side. As such, our analysis contains a second layer of summative content analysis that explores potential latent meanings in the discussion. Our assessment indicates that the manifest meaning in the transcribed speeches provides enough information to test our measures and deduct a reasoned conclusion on the mode of authority in place.
To improve the validity of our analysis, each author and a research assistant involved in the hearing's transcription conducted a content analysis of the hearings. Each conducted an independent assessment of the relationship observed between MEPs and agency representatives, taking into consideration the literature on agency oversight, and the proposed conceptual frameworks. Whereas there was some minor variation, all three identified a limited mode of authority of command, and identified a mode of authority of connection associated with the proposed measures.
Whereas automated text analysis provides an alternative methodology to our approach (see Benoit et al., 2009;Schonhardt-Bailey, 2013), we identified two key factors that led us to conduct a content analysis. To begin with, considering that even in automated text analysis ultimately the research must make some qualitative decisions, and bearing in mind the need to identify manifest meaning across themes rather than specific words in the text (making predictive text reading difficult); employing an automated text analysis would limit our analysis's scope.
Moreover, to the best of our knowledge this is the first study assessing the actual content of the discussion that takes place within an EP committee hearing between its MEPs and an agency. As such, there is no other point of reference in the literature from which to draw pre-assigned values or principles of behaviour, which in turn we can correlate with specific strings of words. The undertaking of such an enterprise is a research project in itself. Being the first to assess the content of the discussion between MEPs and agencies this article aims to act as a point of reference for future work; outlining their structure, some behavioural principles, and guidelines for future research in this area, whether through automated text analysis or content analysis. In this spirit, the hearings transcripts are available to researchers.

Analysis
We begin our analysis with an overview of the hearings' structure. Overall, we identified a protocol of interaction in committee hearings that can be broken down into seven steps, which we provide in Table 2 below.
Considering this pattern, we identified variation in the hearings' purpose, which influences their protocol of interaction, their participants, and to what degree they are forward or backward looking (see Figure 1 for a detailed breakdown). Specifically, we identified two central types of hearings: (i.) expertise-seeking hearing; (ii.) oversight-seeking hearing (see Table 2). Out of the seven hearings assessed, two can be placed under the oversight category where the European Supervisory Authorities were invited to discuss their activity so far and future perspectives (ESMA, EBA, EIOPA). While five hearings out of seven reflected a discussionpanel type of hearing with a variety of actors invited.
In the first case, hearings seeking expertise consist of a diverse panel of participants involving a mix of representatives from EU agencies, think tanks, civil society, national agencies and institutions, among others. The overall aim here is to conduct a debate/discussion where different perspectives over an issue are presented, for example hearing on FinTech or the MiFID II review. This hearing-type addresses either a space where a policy proposal has entered/soon will enter the legislative process (e.g. MiFID II); or discusses the potential needs for future regulation (e.g. FinTech). For the purpose of analysis, we call hearings that are primarily a deliberative forum Type I.
In this case, the hearing has all the steps mentioned above, but it is primed to be more forward-looking, and the discussion format is evidently more open. The MEPs ask questions concerning the agency's opinion on distinct political issues, which is not a form of control. Moreover, as the interaction between MEPs and agency is about reasoned engagement, the relationship shows reciprocity.  1. The committee chair makes a brief opening speech regarding the hearing's overall aim. 2. The rapporteur makes an opening speech linked to the specific objective of the hearing e.g. the policy proposal's focus and general questions. 3. The participants make a speech discussing the hearing's point of discussion from their perspective, this speech is often pre-distributed to the MEPs beforehand. 4. The chair opens the discussion, a set of MEPs' questions are directed to the panel. Most likely, the rapporteur and shadow-rapporteurs (or political group representatives) responsible for the issue will ask questions. Usually each political group will ask 1-2 questions per hearing, a question (or questions) can be directed to more than one panellist. 5. The panellists respond to the questions. 6. The Chair asks the rapporteur to draw conclusions. 7. The Chair closes the hearing with a brief speech.
In the second case, hearings that seek oversight involve the MEPs and an agency (or agencies) representative. In these hearings there is no rapporteur (i.e. step 2 is absent, while the chair conducts steps 6 & 7). These types of hearings are closer to an authority of command. Thus, they are relatively more backward-looking but not exclusively, the agencies discuss primarily their activity so far but also make future projections. Moreover, these hearings observe a greater overall engagement between agencies and MEPs, noted in the number of statements analysed. For the purpose of analysis, we call hearings that act as an oversight forum Type II.
Taking into consideration our analytic measures, it becomes apparent that there is a different mode of authority within the hearings. During steps 1-3, hearings take place under an authority of command where there is a distinct hierarchy, and the interaction is primarily passive as speakers essentially read written statements, which have been provided to the committee members before the hearing takes place. Moreover, speakers are more interested in making broader statements that contain a mix of forward-and backward-looking statements.
This result partially explains why committee hearings tend to be lumped with other static forms of accountability such as written questions: their format makes it plausible that one type of authority permeates the procedure. However, as we show below this ignores the actual discourse that takes place during the discussion/ debate (Zittoun, 2009). While the hearing's general frame somewhat affects its structure, and specifically its outward/backward-looking component. Assessing the speakers' themes, we did not find a clear mode authority of command in steps 4-7. Focusing on the discussion component across hearings, we noted that our measures corresponded to an authority of connection. Below we provide an overview of our analysis vis-a`-vis each measure, and some examples from the themes analysed to highlight our point.

Interaction
Assessing the themes content per speaker it became apparent that MEPs and agency representatives did not strictly divide their labour into a political jurisdiction of ends and an agency jurisdiction of means. Rather, both speaker categories employed a mix of means and ends in their speech, often under a theme. The general pattern observed was one where the speaker opens up with a broader comment that is linked to the political aspects of a policy/action in question, and follows up with a question linked to the agency's technical means. Significantly, this active engagement took place across hearings, including those that had an oversight objective. Moreover, we noted a deliberation in place where speakers engaged with the other side's points, i.e. the agency representative took up the comments and questions raised by the MEPs in a constructive fashion. Similarly, MEPs considered points raised by the agency and posed relevant questions. We highlight that this engagement took place under an amicable environment where both sides provided positive framing devices over the procedure, their invitation to attend the hearing, and the agency's presence at the hearing. For example: "Thanks for this response, but on. . .I raised a specific issue. . .which effort can ESMA make to ensure all over the common market that fees are limited and that they are fair and not burdening unfairly investors and their return?" Sven Giegold 30 September 2013 00:58:12

Relationship
Considering the relationship between MEPs and agency representatives during the discussion phase, we noted reciprocal reason-giving. To begin with, the extent of disagreement between committee members and agency was rather limited. Overall, MEPs or ESMA requested points of clarification or underscored key issues and/or objectives. Moreover, the response to these questions emphasized common reasoning and policy-making objectives. Furthermore, across all MEPs' statements we did not note an opinion pressed on to the discussion, or resolution of disagreement based on their authority forced onto the agency. The MEPs highlighted the agency's role in providing expertise necessary for the EP to progress with its policymaking responsibilities, while ESMA highlighted the EP's important role as a policy-maker. As such, the relationship presented does not reflect one of principal and agent, but rather policy-makers addressing different aspects of the policy domain's needs.
To the extent that a hierarchy was observed, this was noted in some specific instances where the MEPs and the agency highlighted the committee's role in shaping the agency. Nevertheless, these comments contained direct mentions by MEPs to expand the agencies powers. Thus, further highlighting an organizational fuzziness where national vs. European perspectives were underscored rather than legislature vs. agency. This lends support to our argument regarding the interconnectedness of institutional legitimacy, and the collective policy-making that takes place, while adding a Europeanization dimension to it.
". . .The EPP will support you all [ESAs] when it comes to the budget, we believe your agencies need additional resources. . .So our objective is your objective, we want your agencies to fulfil your remit to the full and we want you to have the resources that you need. And the last thing we want is for you to be scapegoats when things go wrong." Jean-Paul Gauzes 19  Orientation As we mentioned above, the forward-looking or backward-looking focus of the participants has a correlation with the hearing's purpose. Hearings seeking oversight tend to address more ex post issues. Nevertheless, even within these hearings a substantial component discussed future projections of the agency's activities and the necessary budgeting it would need to achieve said activities. In this case, the committee requested from the agency's representative an assessment of the budget it would require (political means), which the committee was supportive of.
Moreover, this forward-looking perspective is closely linked with an openended understanding of the hearings. Therefore, in a number of statements the agency opted to carefully assess a point raised by the MEPs and provide a response at a later time. As such, the hearings do not provide a closed set for the assignment of responsibilities but rather serve as learning enterprise that guides the policymaking process, and which can be re-visited as a point of reference by the EP and the agency in the future. ". . .On the longer-term funding of the ESAs, and that of ESMA specifically, I think the overall model where typically, the day-to-day supervision will be conducted at the national level. . .strong argument to do it at the EU level. . ." Steven Maijoor (ESMA Director) 30 September 2013 01:28:14.
"I think it's worth considering, but I don't have the answer today on that." Verena Ross (ESMA Executive Director) 13 June 2016 02:28:47 To visualize how the statements are categorized under different measures, reflecting either an authority of command or an authority of connection, we created Figure  1. The figure shows all the statements codified per measure; a central point raised in our analysis becomes clearer through this visualization. There is a mix of different modes of authority within hearings, however the type of hearing (i.e. if the hearing is expertise-seeking (Type I) or oversight-seeking (Type II)) impacts the interaction between MEPs and agency, and the mode of authority that is prevalent.
Thus, in Type I hearings that resemble a policy discussion the authority of connection is more present. However, Type II hearings that involve only the ESAs and the MEPs resemble a regulatory oversight mechanism where the authority of command is more present, raising the overall number of statements (see Figure 1).

Implications
In this article, we attempted to assess to what degree we observe an authority of connection or an authority of command in the discourse of committee hearings focusing on the exchanges between elected committee members and agency representatives. Based on our analytic measures, our content analysis provides a nuanced understanding of agency oversight in the context of committee hearings. While committee hearings' protocol of communication contains aspects reminiscent of an authority of command, the actual discourse that is part of the discussion section, holds characteristics closer to an authority of connection.
The EP through ECON supports the Europeanization of financial policy. In doing so, the analysis points to a dimension little addressed in the literature, which is the mutual support between elected representatives and independent agency in further empowering European regulatory authorities. Thus, one fails to capture the interaction with a perspective that sees the committee as geared exclusively to constraining, controlling, or steering regulatory agencies.
As indicated in the scope conditions of our study, this may to some extent be a "most likely" case both in terms of the chosen committee and the agency. A study of a more disorderly or controversial agency confronted by the EP budget committee would probably reveal more command-style interaction (cf. Bach and Fleischer, 2012). But given other reports of genuinely knowledge-oriented interaction between agencies and specialized EP committees (Scholten, 2014: 164), there is no salient reason to consider our findings idiosyncratic either.
Hence, it is worth considering the normative implications of mutual attunement as a standard practice in democratic systems. On the one hand, this mode of interaction may give agencies considerable influence. They are invited to discuss policy in ways that reach far beyond their ostensibly technical mandate. In this regard, some conceptions of "democratic autonomy" may see the parliament as too deferential regarding political values (Richardson, 2002). Others may see an agency's open engagement with political values as a corruption of its objectivity or facts-based legitimacy (Vibert, 2007).
On the other hand, it is not obvious that the political influence of agencies is problematic when it takes the structured and public form described above. Insofar as hearings are governed by the authority of connection, neither MEPs nor agencies are "subject to the power of direction" (Majone, 2001: 69) that distorts independence. That is, by themselves, hearings only serve to pressure each party to publicly articulate their evaluative stand. Moreover, it appears democratically preferable to have the political evaluations of agencies out in the open and subject to contestation rather than merely figuring implicitly in regulatory work (given that values cannot be eradicated).
As a matter of how to best serve political accountability, it is arguably unfruitful to consider authority of connection as a stand-alone phenomenon. It depends on whether it is embedded in a broader network of complementary accountability forums, some of which display authority of command to a greater extent. In this, we concur with the "ecological" conception of legitimacy: "It is often not possible to determine the legitimacy of an institution in isolation; instead, its legitimacy may be a function of how it fits into a network of institutions" (Buchanan, 2013: 198).
In this regard, our analytical framework may help unpack what it means for an institution to "fit into a network." In particular, by indicating how an accountability regime can be governed by distinct modes of authority, we have highlighted how hearings in specialized committees may complement the work of other institutions that are more command-oriented.

Conclusion
New governance confronts us with the issue of how to understand new modes of authority and accountability in dynamic settings. In this article, we have attempted to understand whether the political accountability of independent agencies can be understood outside a mode of authority of command with strict divisions of labour.
Drawing on work on deliberative theory and social reasoning, we have argued that there is an authority of connection, where agency and parliament engage in a mutual attunement of expectations. Agency and parliament deliberate and develop a shared space of expectations. This is likelier to be revealed in forums that allow discussion and deliberation in real time, such as parliamentary hearings.
Our results reveal a discourse with little hierarchy, neither actor is limited to either means or ends, actors engage in an attempt to create a common space of expectations based on reasoning not fiat, and they are forward-looking in terms of policy.
Significantly, the premise of division of labour where legislature defines normative ends that agencies aim to achieve, while agencies specify the terms of technical solutions, does not apply across the board. Rather, there are mechanisms where the labour is merged and both agency and legislature attempt to contribute under a collective policy-making logic. While this is generalizable vis-a`-vis ESMA, the proposed criteria should be tested on different hearings (other agencies) or on different accountability mechanisms.