Updating public accountability: a conceptual framework of voluntary accountability

ABSTRACT Voluntary accountability is widely documented. Yet, the concept does not sit comfortably within prevailing frameworks that conceptualize accountability as an obligation. This paper sets out to update our understanding of accountability by explicitly embedding this phenomenon within the broader conceptual structure of public accountability. It demonstrates how the proposed conceptual framework can help us distinguish between voluntary accountability and related, but distinct, concepts, such as transparency. It further illustrates how the framework can help us answer pressing questions about the varying rationales behind voluntary account-giving, the audiences targeted by accounts as well as the consequences of voluntary accountability for democracy.


Introduction
Voluntary accountability is a well-established phenomenon in modern governance. A host of empirical studies document the adoption of practices by public sector organizations or public officials through which accounts are rendered in the absence of any formal obligation (Busuioc 2013;Doering et al. 2021;Karsten 2015;Koop 2014;Laegreid and Rykkja 2021;Puppis et al. 2014;Schillemans 2011;Verschuere et al. 2006). Voluntary accountability reportedly relies on a variety of mechanisms such as self-initiated consultation procedures, social media, interactive websites, voluntary external reviews, requests for parliamentary hearings, consultative forums, codes of conduct, voluntary accountability programmes and press releases/conferences.
In policy documents and political texts, such practices are recognized as instruments for rendering account too. To illustrate, the Organisation for Economic Cooperation and Development (2021) writes on its website that public communication 'is a prerequisite for government transparency and integrity, and for the ability of citizens to hold them accountable and participate in public life'. In a similar vein, the World Bank strives for enhanced citizen engagement because '[c]itizens play a critical role in advocating and helping to make public institutions more transparent, accountable and effective' (World Bank 2021). Similar statements have been made regarding the adoption of digital technologies. For example, Obama (2009) wrote in a memorandum addressed to the heads of executive departments and agencies that '[t]ransparency promotes accountability and provides information for citizens about what their Government is doing. (. . .) Executive departments and agencies should harness new technologies to put information about their operations and decisions online and readily available to the public'. Nowadays, many public bodies -whether they are street-level bureaucracies, independent agencies or core government -are embracing social media and other digital technologies in the name of accountability. This suggests the existence of a whole new class of accountability practices that are largely voluntary in nature and whose resemblance to traditional accountability instruments (e.g. parliamentary hearings, formal reporting, legal procedures, financial auditing, etc.) is often difficult to grasp.
While voluntary accountability has been qualified as a 'persistent' (Reiss 2011, 647) phenomenon, its conceptual clarity needs further unpacking. The scope of the concept has expanded beyond the traditional understanding of accountability, often conflated with related, but conceptually distinct, practices that fall short of the accountability threshold. Moreover, the phenomenon of voluntary accountability does not sit comfortably within the broader understanding of accountability. While dominant and influential conceptualizations of accountability (Bovens 2007) note that an accountgiver ('actor' in accountability terminology) can in principle render account voluntarily to an account-holder ('forum' in accountability terminology), at the same time there seems to be an inherent tension between this phenomenon and the conceptualization of accountability put forward. A central component in seminal accountability frameworks (Bovens 2007;Mulgan 2000;Strøm 2000) that implicitly or explicitly draw on assumptions from Principal-Agent theory is the presence of an 'obligation' to render account: 'The most concise description of accountability would be: "the obligation to explain and justify conduct"' (Bovens 2007, 450, emphasis added). Yet, this sits somewhat at odds with the notion that voluntary accountability is self-imposed (Koop 2014, 565) and therefore, by definition, rendered in the absence of an obligation. It remains thus unclear how the concept fits in our broader understanding of public accountability.
Reconciling voluntary accountability with existing understandings of accountability is relevant, since a whole array of account-giving practices are undertaken voluntarily, in the absence of an obligation (even of the moral kind). Referring to non-majoritarian institutions, Busuioc and Lodge (2016, 253) note that '[s]eeking accountability (. . .) is an important strategy for such actors to manage reputational risks and demonstrate that account giving is being taken seriously'. Voluntary accountability has even been called 'an essential tool for active reputation management' (Doering et al. 2021, 376). Actors have been found to render voluntary accounts in order to build support for their policies (Moffitt 2014) and controversial decisions (Karsten 2015), to create mutual trust (Van Slyke 2006), to gain autonomy and reputation (Busuioc and Lodge 2017) and in the process may even offset principal control (Bertelli and Busuioc 2020). Such strategic motivations behind voluntary accountability are in line with a broader literature that speaks of entrepreneurial actors who are 'proactive in managing their own legitimacy' (Black 2008, 152). This literature stresses that bureaucratic actors gain power and autonomy through pro-actively engaging in and maintaining (accountability) relationships with relevant audiences (Carpenter 2010). Reputed organizations are able to (co-)shape the formal constraints on their authority and the terms of their formal mandate (Carpenter and Krause 2015). This could entail that rather than improving the general state of accountability in modern governance, practices of voluntary accountability could stand to subvert it. In order to advance our understanding of such mechanisms as well as the potential positive and adverse consequences of voluntary accountability, we need to first lay more analytical ground towards an integrated conceptual framework and consensus on the meaning of the concept.
This paper therefore sets out to update existing accountability frameworks by developing a conceptual framework of voluntary accountability. The proposed framework stays close to Bovens' influential accountability framework (Bovens 2007), while at the same time drawing attention to some elements that need recalibration in order to incorporate the phenomenon of voluntary accountability. In this sense, our framework -in light of recent empirical developments -'updates' existing accountability frameworks. Specifically, we identify four aspects that function differently when accountability is voluntary in nature: why the account is rendered, why actors are held accountable, how and to whom account is rendered. In structuring the account along these elements, we mirror Bovens' work on accountability to ensure adequate embedding and synergies between the two conceptualizations, while simultaneously adapting it to the specific demands of voluntary accountability. As will be argued in this paper, these adjustments are necessary because voluntary accountability, unlike mandated accountability, is not settled formally. In contrast to its mandatory counterpart, in the case of voluntary accountability, the relationship between the actor and the forum is not the product of a formal stipulation by an external authority (e.g. the legislator) imposing a duty to give account but instead emerges unforced through mutual acquiesce and engagement from both sides. It is argued here that recognition of this distinctive aspect of voluntary accountability is important for obtaining a shared conceptual understanding, for unpacking the underlying mechanisms behind the phenomenon, and for reflecting critically upon its normative implications.
A conceptual framework of voluntary accountability in the public sector is thus direly needed. The rationale for this is three-fold. First, seemingly unrelated phenomena, ranging from maintaining a website to voluntary accountability programmes with formal sanctions, have been classified as voluntary accountability or one of the many related concepts that mean approximately the same thing such as social accountability, horizontal accountability, direct accountability and stakeholder accountability. The paper hopes to help mitigate the existing conceptual confusion around the topic by establishing when a practice qualifies as voluntary accountability through a clear-cut definition and to contribute to greater consistency in how the concept is applied. It furthermore provides a systematic instrument for assessing political or bureaucratic claims, such as the ones mentioned above, that the adoption of a certain practice promotes accountability.
Secondly, the proposed framework completes the conceptual structure of accountability (see Sartori 1970) because every possible form of accountability is either mandatory or voluntary. While most studies examine mandatory accountability mechanisms, de facto account-giving does often not occur because of formal obligations (Romzek, LeRoux, and Blackmar 2012;Scott 2006). Thus, by situating voluntary accountability within our broader conceptual understanding of accountability, we can get a more complete picture of the state of accountability in modern governance. This can, in turn, help us to make better assessments of the potential advantages (e.g. remedying democratic deficits, enhancing organizational learning, ensuring balance of powers) and disadvantages (e.g. overloads, incompatible accountabilities, bias) of accountability in its full picture.
Third, we present a novel research agenda that highlights the implications -both theoretically and normatively -of the proposed framework. This agenda focusses on the varying rationales behind voluntary account-giving, the logic behind decisions regarding how and to whom account is rendered and the consequences of voluntary accountability for democracy. This opens up new pathways for empirical investigations that will help us better understand the drivers and the implications of voluntary accountability.

The challenge of accountability in modern governance
Traditionally, accountability has been conceptualized as a hierarchical chain that runs opposite to the chain of delegation. Citizens are the ultimate principals who 'have transferred their sovereignty to popular representatives, who, in turn, have transferred the drafting and enforcement of laws and policy to the government. The ministers subsequently entrust the execution of their tasks to the many thousands of public servants at the ministries, who proceed to delegate part of their tasks to more or less independent bodies and institutions' (Bovens 2007, 463; see also Strøm 2000). Implementing organizations are contractually and/or statutorily obliged to inform civil servants on their performance regarding the delegated task. Public servants are, in turn, accountable to the minister by means of hierarchy within the ministry and the minister is accountable to parliament whose members are accountable to the people during elections. Accountability can be said to function properly in this model when every actor is required to give account to its superior so that a single accountability chain runs from policy implementers all the way up to voters.
This model of accountability has come under stress as an indirect consequence of public sector reforms (Considine 2002). A wide variety of new accountability practices -both mandatory and voluntary -have been introduced to remedy accountability problems caused by the structural disaggregation of the public sector that has taken place over the last couple of decades (Michels and Meijer 2008). These new practices are often enacted outside the hierarchal relationship that is associated with the 'idealtypical' unitary model of delegation and accountability (Strøm 2000). As such, reforms in the public sector 'have stretched the elasticity of our received notions of accountability to the breaking point' (Considine 2002, 23).
In response to these changes, many new concepts -including voluntary accountability -have been proposed with the aim to align the accountability idiom with the new empirical reality (see Table 1 for an overview). Together, these concepts reflect the increasingly complex, multidirectional 'web of accountability' (Page 2006) in which public institutions operate these days. The growing role of the internet as a platform for societal accountability (Bovens 2007) has led to the development of concepts like webbased accountability (Saxton and Guo 2011) and dynamic accountability (Schillemans, Van Twist, and Vanhommerig 2013). Related concepts such as direct accountability (Meijer 2005), citizen accountability (Meijer and Schillemans 2009), participatory accountability (Damgaard and Lewis 2014) and citizen-initiated accountability (Schillemans, Van Twist, and Vanhommerig 2013) were coined to describe the growing influence of citizens in accountability processes. Other authors have proposed  O'Donnell (2003) concepts that describe the growing role of all kinds of stakeholders in horizontal, multi-actor and multi-level governance settings, such as media accountability (Maggetti 2012), informal accountability (Romzek, LeRoux, and Blackmar 2012), mutual accountability (Whitaker, Altman-Sauer, and Henderson 2004), extended accountability (Hall, Scott, and Hood 2000) and stakeholder accountability (Meijer 2007).
The vast majority of practices flagged by these new concepts are undertaken voluntarily by account-givers. As such, the proliferation of new accountability concepts should be seen in the light of the 'trend towards the introduction of forms of accountability which are not provided for in legislation' (Koop 2014, 565). Similar trends have been described in the non-profit sector (i.e. voluntary accountability clubs) and the private sector (i.e. corporate social responsibility). It is noteworthy that Bovens (2007) traces the emergence of horizontal accountability (which is primarily voluntary in nature) to the debate about corporate social responsibility: 'Influenced by the debate on corporate social responsibility and corporate governance in business, more attention is being paid to the role of non-governmental organisations, interest groups and customers or clients as relevant "stakeholders", not only in determining policy, but also in rendering account' (Bovens 2007, 457, emphasis in original).
Whereas scholars studying corporate social responsibility (Roberts 2003) and voluntary accountability clubs (Gugerty and Prakash 2010) have sought to construct all-encompassing frameworks for analysing these new phenomena, the public administration literature has been flooded with many new, somewhat disjoint concepts. While new concepts can be incredibly valuable, there is a need for a more integrated approach. First, it is often difficult to gauge how the different concepts relate to each other. In some cases, the similarities are even so substantial that concepts can be considered synonyms. Secondly, only relatively few concepts are clearly defined, demarcated and embedded within the broader conceptual structure of accountability. This is also reflected in the different uses of the same concept (see column 'compare with' in Table 1). Thirdly, most concepts are used sporadically, which hinders broader theory development.
In this paper, we propose a framework of voluntary accountability as a unifying concept. Our focus is on voluntary practices as existing accountability frameworks have proved to be valuable analytical tools for studying mandatory accountability. Yet, they are less well-equipped for studying voluntary accountability because they define accountability as an obligation. Hence, the gap in the conceptual structure of accountability relates to voluntary accountability specifically. The paper sets out to extend the analytical value of existing frameworks to the realm of voluntary accountability practices -which are increasingly relied upon in modern governance. We will concentrate exclusively on voluntary accountability within the public sector, as others have already developed conceptual frameworks for similar phenomena within the nonprofit sector (Gugerty and Prakash 2010) and the private sector (Roberts 2003). As our approach is firmly rooted in existing understandings of accountability, we believe the framework can contribute towards a more systematic and integrated analysis and evaluation of voluntary accountability in modern governance.

Descending the conceptual ladder of accountability: defining voluntary accountability
Voluntary accountability is a subtype of the broader concept of public accountability. Following the logic of Sartori's ladder of abstraction (Sartori 1970), a specification of another concept needs to have all properties of the higher-level concept. This implies that any property of accountability should also be incorporated in the conceptualization of voluntary accountability. To this end, it is important to clarify first what accountaility is and how voluntary accountability differs from other types of accountability before delving into the meaning of voluntary accountability itself. Accountability can be defined as 'a relationship between an actor and a forum, in which the actor has an obligation to explain and to justify his or her conduct, the forum can pose questions and pass judgment, and the actor may face consequences' (Bovens 2007, 450). This definition highlights two important properties of accountability that are commonly agreed upon in the literature.
First, accountability constitutes a social relationship between an account-giver and an account-holder (e.g. Black 2008;Bovens 2007;Busuioc and Lodge 2017). This relational element of accountability is regarded as crucial for discerning between accountability and related, but distinct, concepts such as transparency or openness. While the former represents a communicative interaction, the latter may be entirely one-directional. The interaction element is a critical element for a practice to qualify as accountability (irrespective of the form it might take). Secondly, accountability has three phases (Bovens 2007). The process begins with the information phase, in which the actor informs the forum on its conduct. Thereafter, the forum assesses this account in the discussion phase. The forum judges the adequacy of the provided information, may ask for more information and can pose follow-up questions. At the final consequences phase, the forum may apply sanctions or rewards in line with its judgment.
The fundamental difference between voluntary accountability and mandatory accountability lies in the nature of the obligation. Unlike mandatory accountability, voluntary accountability is rendered in the absence of a de jure obligation. Such obligations are usually set out in legislation, statutes, contracts, regulations and other legal documents that determine the rights and duties of an organization. However, this binary categorization -i.e. voluntary accountability versus mandatory accountabilityhides a whole spectrum of accountability. As noted, before, accountability processes, by definition, have three phases: information, discussion and consequences. In practice, obligations often pertain only to the first phase or the first and second phase, but not to all three phases. Approaching the concept in this manner reveals that two other accountability processes can be discerned that consist of a mix of both mandatory and voluntary phases (see Table 2).
First, mandated transparency occurs when an actor is required to release information without any further responsibilities for external parties to use this information for account-holding. Freedom of information acts, sunshine laws and the release of performance information by inspectorates are typical practices that fall under this category. These measures guarantee the disclosure of information, but they do not ensure that an assessment about the actor's conduct is made by a forum or that proper consequences are available and implemented (Meijer 2007). This is consistent with established accountability conceptualizations that have already noted that information provision/transparency alone is a pre-requisite of accountability but not a sufficient condition for it. Only when the information provided gives rise to scrutiny, such practices qualify as a genuine form of accountability (Bovens 2007, 453). Second, the category of weak mandatory accountability consists of practices whereby a forum is granted authority to demand information from an actor and formal rules warrant debate, but the forum lacks formal sanctioning power. Thus, the difference with traditional mandatory accountability pertains to the possibility of consequences, which is largely absent in the case of weak mandatory accountability. Examples of weak mandatory accountability are client panels and user councils. Such bodies are granted access to information and organizational leaders are required to appear a minimum number of times per year before the panel or council (Meijer and Schillemans 2009). Yet, there is no formally defined opportunity to impose sanctions or rewards. Such practices contain the critical element of interaction between actors and forums and as such qualify as accountability -albeit in its weaker form (due to a lack of consequences). Such bodies generally tend to resort to informal sanctions or rewards (e.g. public praise, expressing criticism and inducing feelings of shame) or rely on other forums to enact actual consequences.
Thus, what sets voluntary accountability apart from other accountability processes is the absence of an external obligation in all three accountability phases (i.e. information, discussion and consequences). Yet, all three phases need to be present for a fullfledged accountability process to occur. Coupled with the crucial notion that accountability is relational, voluntary accountability then refers to a relationship between an actor and an engaged forum, in which the actor voluntarily commits itself to offering information on, and explanation of, its conduct and may face consequences. While this definition does not contain any reference to an 'obligation', it is firmly rooted in existing conceptualizations of accountability. It stays close to these conceptualizations by stressing the relational core of accountability as well as the three phases.
Although the proposed definition stays close to Bovens' established conceptualization in many respects, there are important implications of this seemingly small adjustment. As we will demonstrate below, the voluntary element has important consequences for thinking about why account is rendered, why actors are held accountable, as well as how and to whom account is rendered. While these aspects are imposed -and, therefore, fixed -by an external actor in traditional, mandatory accountability processes, this is not the case when account-giving is voluntary. The whole relationship between the actor and the forum is not -in contrast to mandatory accountability processes -predefined by an external obligation. Consequently, the establishment of an accountability relationship is entirely based on discretionary decisions in the case of voluntary accountability.
It should be noted that this discretion exists on both sides of the relationship. The proposed definition underlines this by speaking of a relationship between an actor and an engaged forum. For it to be described as voluntary accountability (which requires the presence of a 'social relationship'), not only does the actor seek accountability, but there should also be a forum that wants to hold the account-giver accountable. This means in practice that voluntary accountability requires acceptance by both the actor and the forum. Actors commit to this relationship by providing information and explanations, while forums can accept this relationship by assessing the information provided and considering sanctions or rewards. Thus, there are two requirements that need to be fulfilled before a practice qualifies as voluntary accountability: voluntary commitment from the actor ('actor self-committal') and engagement from the forum ('forum engagement').
Importantly, scrutiny by a specific forum is always required to meet the accountability threshold; a process that stops after voluntary information disclosure does not qualify as accountability. To illustrate this point, a public institution may maintain a publicly available website with information about its own performance, but this fact alone does not mean that it is accountable. Whether maintaining a website can induce an expectation of accountability and help to keep the public institution in check ultimately depends on the extent to which another party 'accepts' the relationship. Concretely, this means the presence of an account-holder -for example, an individual citizen, a journalist, an interest group or a politician -that could, and is willing to, evaluate the website's information and pass judgment. It should be noted that not every release of information on the website needs to be followed up by a forum before it can be regarded as accountability. Yet, overall forum engagement should be substantial enough to induce a general expectation of accountability on the side of the actor. The rationale for this is that an actor will only adjust its behaviour in anticipation of being held accountable when it knows from past experience that scrutiny is a realistic expectation. Thus, maintaining a website does not qualify as accountability when forum engagement is largely a hypothetical scenario without much precedent in reality. This would for instance be the case when a website is not visible, attracts few visitors and is rarely used by stakeholders for interactions with the actor. Of course, the manifestation of forum engagement depends on the practice under examination. Forum engagement can be expressed in metrics such as the number of replies to a Tweet from a public institution, the extent to which press releases spur media coverage or the number of responses from stakeholders to a self-initiated consultation procedure.
With this definition of voluntary accountability in mind and along the lines of Bovens' framework, we identify four questions that need to be asked when accountability is voluntary (cf. Bovens 2007). The first question is why does the actor render voluntary account? This question relates to possibly the main difference with mandatory accountability, where account must be given regardless of the motivations of the account-giver. Whereas in the case of mandatory accountability the answer is straightforward ('because it is required by mandate'), in the case of voluntary accountability this becomes the central question to explicating the rise of voluntary accountability and its drivers. As we will see below, several competing rationales have been put forward in extant literature. The second question is why are actors held accountable? This question logically follows from our definition of voluntary accountability, which implies, as argued above, the presence of an engaged forum. The third question pertains to how is the account rendered? Unlike mandatory accountability, in the case of voluntary accountability actors can freely choose from a broad arsenal of accountability instruments which differ greatly in form and function. Recognition of this great diversity is crucially important, we argue below, for understanding the varying rationales behind accountability-seeking behaviour. The fourth and final question is to whom the account is rendered? This question is related strongly to the third question as certain instruments are typically geared towards certain forums (e.g. voluntary accounts on social media are often geared towards the general public). While answering these last two questions in the case of mandatory accountability results in a mapping of different typologies of accountability relationships provided for by formal design, doing so in the case of voluntary accountability allows us to understand which social relationships actors are actively pursuing and initiating of their own accord. Crucially, because in the case of voluntary accountability actors are not mandated but rather have a choice of which forum to give account to, and about what (i.e. the content of their accounts), tackling these questions simultaneously gives us a level of insight into actor motivations (the 'why' of voluntary account-giving). The first two questions will be discussed in the next section and the other two questions will be discussed thereafter.

Actor self-committal: why account is rendered
Whereas when it comes to mandatory accountability it is irrelevant to the existence of the accountability relationship whether the actor is actually motivated to be accountable -i.e. there is no choice -, this is in fact the focal point in the debate around voluntary accountability. This debate revolves around adjudicating between different driving rationales explaining voluntary accountability behaviours, specifically the logic of appropriateness and the logic of consequence. Koop (2014, 567) has argued in a key study on voluntary accountability that voluntary account-giving 'does not have clear advantages for the organizations that render account'. Consequently, accountabilityseeking behaviour should be understood as following from the logic of appropriateness. That is, it follows from 'internalized prescriptions of what is socially defined as normal, true, right, or good, without, or in spite of calculation of consequences and expected utility' (March and Olsen, 2011, 480). This explanation of voluntary accountability fits well within the democratic perspective of accountability which stresses the normative connotation of the concept and its role in sustaining democracy. Karsten (2015) questions the proposition that account-giving is costly for the account-giver and proposes the logic of consequences as an alternative explanation. This logic entails that an actor evaluates the consequences of possible actions and chooses to do what is expected to contribute most to the fulfilment of one's own goals (March & Olsen, 2011). In accordance with this logic, Karsten (2015, 697) contends that voluntary account-giving is motivated by political-strategic motivations. That is, it is the result of a cost-benefit analysis made by self-interested actors 'who estimate the strengths and weaknesses of alternative types of accountability behaviour under different circumstances'.
Other authors have proposed additional rationales that do not neatly fit within the logic of appropriateness or the logic of consequences (Busuioc andLodge 2016, 2017;Schillemans 2011). Various scholars have pointed to learning motivations. On the one hand, learning motivations can be seen as following from the logic of appropriateness. It has been argued that empowering stakeholders through accountability can help an organization to better cater to the needs of those who it is supposed to serve (Damgaard and Lewis 2014, 259;Schillemans 2011). On the other hand, there may also be an instrumental rationality at work. Many public sector organizations need to obtain information from stakeholders in their environment in order to draft high-quality policies and make sensible decisions. Hence, establishing accountability relationships with the possessors of valuable information can help to satisfy organizations' informational needs (Arras and Braun 2018).
Still other authors have pointed to reputational motivations. These scholars argue that account-giving affords the account-giver an opportunity to highlight favourable aspects of the organization (Busuioc andLodge 2016, 2017;Doering et al. 2021). Voluntary account-giving may therefore be part of a strategic effort to build broad public support, which is crucially important for sustaining (regulatory) power, gaining autonomy and overcoming opposition (Carpenter 2010). Such accounts put emphasis on voluntary accountability as a form of strategic reputational management. Yet, cultivating one's reputation can also be seen as normatively appropriate behaviour. As suggested by Maggetti and Papadopoulos (2018, 180), 'reputation may not only have an instrumental value but be intrinsically valued too'. Busuioc and Lodge (2016, 250, emphasis in original) wrote on this very point that the process of reputationbuilding through account-giving 'is not necessarily always strategic. Actors can come to internalize audience expectations, and to this extent, the relationship can come to be constitutive'.
This short overview of the extant literature shows that valuable efforts have been made to theorize and document possible rationales behind accountability-seeking behaviour. It further indicates that the jury is still out on what exactly motivates actors to seek accountability. What we will argue in more detail below is that the multiplicity of rationales should be seen in the light of the different manifestations of voluntary accountability as well as the social relationships in which it occurs. This implies that this question (i.e. why account is rendered) cannot be addressed separately from the other three questions that characterize voluntary accountability (i.e. why actors are held accountable, how and to whom account is rendered). Below, we elaborate on this argument, but we will first expand upon the notion of forum engagement.

Forum engagement: why actors are held accountable
As noted before, forum engagement is an integral aspect of voluntary accountability, and public accountability, more broadly and a key element to whether an accountability-seeking practice qualifies as accountability. Yet, it has often been overlooked that its presence cannot be taken-for-granted, even though there is no theoretical ground to assume the presence of forum engagement a priori in the case of voluntary accountability. As Meijer (2014, pp. 513-514, emphasis added) writes, 'there is no ex ante reason to assume that transparency and accountability are "friends": we need to evaluate their relation on the basis of empirical research'. Meijer (2014, 517) also notes that one lesson that can be drawn from the existing literature is that 'proactive transparency has little value for accountability'. In a similar vein, Busuioc and Lodge (2017, 98, emphasis added) write that what defines voluntary accountability is the 'onesided, nonrequited, voluntary character of the interaction, with a forthcoming account-giver and relative disinterested passivity on the side of the account-holder'.
The literature mentions at least five reasons why potential account-holders are hesitant to take up a role as engaged forum. First, they may fear that they will be blamed for unpopular decisions or insufficient oversight, as hinted by Damgaard and Lewis (2014) in regards to people's observed lack of motivation to participate in client councils. Second, actors may provide so much unstructured, decontextualized information that stakeholders need to invest a lot of resources to distinguish the relevant information from the noise -a strategy which Hood (2007, 204) illustratively labelled 'snowing'. Thus, voluntary accountability may not only contribute to 'accountability overloads' on the side of actors (Bovens et al. 2008), but also to 'information overloads' on the side of forums. Third, an actor can also decide to present highly polished, biased information that has little value for assessing the actor's conduct (Meijer 2014). As voluntary accountability happens basically on the actor's own terms, there are no formal rules that ensure that an actor's initiative is anything more than a presentational strategy. Fourth, it may be that potential account-holders lack necessary capacity, skills, resources or knowledge to assess complex information, such as in the case studied by Lindén (2015) where lay audiences were invited to review appeals of asylum seekers whose applications were rejected. Finally, Mizrahi and Minchuk (2019) have argued, and demonstrated empirically, that citizens face a collective action problem when trying to hold public sector organizations accountable; their survey results show that citizens are less inclined to monitor these organizations when the costs of monitoring are not shared equally. Such accounts underline that forum engagement cannot be taken-for-granted. Succinctly, forum engagement is critical because in its absence accountability becomes indistinguishable from transparency.
Apart from the substantive relevance of understanding the 'other end' of accountability relationships better, empirical assessments of forums' behaviour help to determine to what extent actor self-committal results in full-fledged accountability processes. Note that assessing the level of forum engagement is also relevant when studying mandatory information provision (i.e. mandated transparency), as forum engagement cannot be taken-for-granted in this case either. According to the Oxford Handbook of Public Accountability, transparency relates to 'the availability of information about an actor allowing other actors to monitor the workings or performance of this actor' (Meijer 2014, 511). Hence, forum engagement is essentially the marker of the difference between genuine voluntary accountability practices and mere transparency. Consequently, actor self-committal without any (possibility of) forum engagement can easily be misclassified as voluntary accountability when the forum engagement is left out of the analysis. As the literature has identified multiple reasons why potential account-holders do not become engaged, it may be that the most that can be claimed in many cases of actor self-committal is enhanced transparency or openness (rather than accountability). Conflating voluntary accountability with such concepts is problematic because 'transparency as such is not enough to qualify as a genuine form of accountability' for the reason that 'transparency does not necessarily involve scrutiny by a specific forum' (Bovens 2007, 453).
Hence, it is crucial to assess empirically the degree of forum engagement following from an actor's initiative. Not performing this empirical assessment will inevitably lead to a too 'rosy picture' about the state of accountability in modern governance by leading us to document phenomena as accountability that do not meet the accountability threshold.

How and to whom voluntary account is rendered
Besides discretion regarding the commitment to accountability, actors are also free to choose how and to whom account is rendered. Although voluntary accountability is more often than mandatory accountability geared towards society (Bovens 2007;Koop 2014;Verschuere et al. 2006), empirical studies on voluntary accountability reveal there is in fact a lot of variation in the practices adopted. This variation exists both in terms of the type of instrument relied upon to render account (i.e. how account is rendered) as well as in terms of the type of forum involved (i.e. to whom account is rendered). The variety of voluntary practices documented in extant literature can nevertheless be grouped in three broad categories, each of which represents a functionally different form based on how and to whom account is rendered. While some voluntary account-giving will -much like a 'message in a bottle' -be aimed at no specific forum in particular, other practices involve the provision of more targeted voluntary accounts, such as the use of consultative forums or presentations in which account is rendered to specifically selected societal stakeholders or to principals.
Answering these questions (i.e. how and to whom?) -beyond offering us different typologies of voluntary accountability -is important, we argue, in offering us clues as to why account is being given and the degree to which forums are generally engaging with the voluntary account (as summarized in Table 3). That is, each category of practices is related to a different set of driving motivations (see column 'logic of action' in Table 3) and different rationales for forums to engage with account-givers, leading to varying expectations about the presence of forum engagement (see column 'expected forum engagement' in Table 3). In other words, the four questions are interlinked and the answer depends on the type of practice under consideration.
The typology thus highlights that different types of voluntary accountability have distinct features that must be recognized in order to make conceptual and theoretical sense of the great diversity of practices that are classified as voluntary accountability. Most importantly, the motivations behind accountability-seeking behaviour will plausibly vary depending on how and relatedly, to whom voluntary accounts are rendered. Rendering account without a specified forum in mind ('message in a bottle') will presumably provide an organization with different benefits than those to be gained from engaging a specific constituency. While in the former case, voluntary accountgiving might serve to signal a general organizational commitment to openness, the latter will serve to gain support or resources that a specific constituency alone can provide. As such, discerning between types of voluntary accountability can potentially account for contrasting findings on motivations behind voluntary accountability towards the general public (Koop 2014) versus voluntary accountability towards political representatives and voters (Karsten 2015). In other words, moving beyond the current understanding of voluntary accountability as a unitary phenomenondriven by a common underlying logic -will help us to grasp the varied rationales behind it. We elaborate on each voluntary accountability type in turn.

A message in a bottle: anyone there?
Documented voluntary practices of this type are generally geared towards a wider, often unknown, public. Since the introduction of new communication technologies, particularly social media and the internet, this type of voluntary account-giving has • Political-strategic motivations • Medium become commonplace. Nowadays, many public sector organizations rely on these technologies to demonstrate accountability directly to citizens (Bertot, Jaeger, and Grimes 2012). Additionally, traditional media are increasingly recognized as a significant forum (Maggetti 2012). The media can spark new accountability processes by uncovering misconduct (e.g. the Watergate scandal), but the media are also the target of many voluntary accounts. Many organizations proactively render account to the media and, by extension, the general public through means of press releases/conferences and interviews (Puppis et al. 2014).
Practices aimed at informing such general audiences can be seen as a 'message in a bottle' because the actor disseminates information in the public domain about its behaviour without knowing in advance who -if anyone -will actually find and use this information to hold the actor accountable. While various authors have noted the potential utility of practices aimed at general audiences for enhancing accountability (Bertot, Jaeger, and Grimes 2012;Puppis et al. 2014;Schillemans, Van Twist, and Vanhommerig 2013) and some authors use the adoption of this type of practices as an indicator of accountability (Christensen and Lodge 2018;Koop 2014;Saxton and Guo 2011), forum engagement is arguably particularly difficult to achieve with practices of this nature. As shown by Mizrahi and Minchuk (2019), general audiences face a collective action problem because the costs and benefits of monitoring are distributed unevenly. Moreover, it is often unclear with this type of practices whether potential account-holders would be interested in the information at all. Consequently, forum engagement will likely be low when account is rendered to general audiences.
Given their non-targeted character, such 'message in a bottle' practices are likely driven by the logic of appropriateness -and democratic motivations in particularrather than consequential thinking. The most important function of accountability towards the general public is to democratize modern governance (Willems and Van Dooren 2012). It puts citizens in the position to hold organizations directly accountable (rather than indirectly via the ballot box), which explains why it is heralded as the remedy against alleged 'democratic deficits' (Bovens 2007). Given the strong connotation with democratic quality, public officials may see it as their democratic-civic duty to explain and justify the use of taxpayers' money to society (Olsen 2013), consistent with the logic of appropriateness. Moreover, 'message in a bottle' practices have the least tangible benefits for the account-giver because the general public typically cannot provide material rewards or expert information. Consequently, consequential thinking seems a less probable explanation for this type of account-seeking behaviour (Koop 2014). Nevertheless, beyond internalized expectations, proactively reaching out to the general public by explaining and justifying one's actions may perhaps be essential for securing more elusive resources, such as trust from citizens, a positive public image, support from one's environment and legitimacy (Busuioc and Lodge 2017).

A discerning voluntary account: targeting specific societal stakeholders
This type of voluntary account-giving involves practices in which account is rendered towards often specifically selected stakeholders who enjoy a privileged status that separates them from the general public. Members of these groups are often selected based on certain criteria, such as expertise, representation of interests or professional background. Consultative forums, stakeholder boards, visitations and external reviews (Arras and Braun 2018; Busuioc and Jevnaker 2020) are some examples. Another well documented example is the use of public advisory committees by agencies, in the absence of formal obligations in that respect (Moffitt 2014). Organizations can also engage with specific stakeholders by establishing public consultation procedures. Agencies in the United States (Magill 2009;Reiss 2011), several European countries (Thatcher 2002) and the European Union (Arras and Braun 2018) have been found to hold public consultation procedures in the absence of, or beyond, statutory responsibilities. Although such procedures are in principle open to everyone, they typically function as an instrument for advantaged stakeholders to provide technical feedback on specific issues (Yackee and Yackee 2006). Note that these practices are mandatory for some organizations (Schillemans 2011); it depends on an organization's formal obligations whether these practices are undertaken voluntarily.
One plausible dominant rationale behind these practices is learning. Public institutions may feel that such practices are necessary as to obtain relevant information and policy feedback from stakeholders in their environment. This motivation may be particularly strong among actors with high informational needs such as regulatory agencies (Arras and Braun 2018). Reputational considerations may also play an important role. Rendering account to allying stakeholders may be part of a coalitionbuilding strategy, whereby the actor seeks to build a coalition that can function as a counterweight against the democratic control exercised by principals (Carpenter and Krause 2015).
Forum engagement can be expected to be relatively high when voluntary accounts are targeted at specific societal stakeholders. Often, these stakeholders will have a strong interest in engaging with the account-giver, such as when regulatees try to exert influence on the rules of a new regulation through public consultation procedures (Joosen 2021). It has been pointed out that too much forum engagement, especially when certain interests are overrepresented in the forum, raises concerns about the potential undemocratic effects of accountability towards societal stakeholders. In such cases, account-giving may give way to bias or even capture (Arras and Braun 2018;Busuioc and Jevnaker 2020). Moreover, strong accountability ties with societal stakeholders can lead to offsetting the control of legitimate authorities (Bertelli and Busuioc 2021).

Paralleling the principal-agent relationship: re-engaging the principal
Voluntary accountability practices occur not only towards horizontal forums, as discussed above, but can also be vertical in nature -i.e. targeted towards principals. This type of voluntary accountability involves practices that are voluntarily initiated by actors upwards, towards new or existing principals. That is, they exist in parallel with mandatory accountability mechanisms such as appointment and dismissal procedures, parliamentary hearings and reporting requirements. The existence of voluntary accountability in vertical relationships is intriguing because it contrasts Bovens' claim (Bovens 2007, 460) that vertical accountability is always mandatory, whereas horizontal accountability happens basically on a voluntary basis. Requests for hearings before the European Parliament by European Union agencies (Busuioc 2013) and the European Central Bank (Jabko 2003) are instances of voluntary accountability running in parallel with the principal-agent relationship. Karsten (2015) has examined accountability-seeking behaviour by Dutch local political executives towards municipal councillors and voters. These relationships can be seen as principal-agent as well, since voters mandate municipal councillors through local elections, after which the elected representatives delegate part of their authority to political executives.
It has been suggested that account-giving towards principals is undertaken to reengage with principals (Schillemans and Busuioc 2015, 210). This strategy is particularly attractive when principals -which often hold powerful tools at their disposal (e.g. budgetary, mandate amendments, etc.) -show low interest in the actor. Under these conditions, the actor may run the risk of being forgotten, with potentially far-reaching consequences for the budget and the allocation of tasks. Hence, it can be expected that voluntary account-giving towards principals will occur in particular when the principal is not actively engaging with the account-giver. Yet, the principal being targeted with voluntary accounts will presumably not be completely unwilling to scrutinize the conduct of the account-giver since this would be detrimental for the relationship with the actor in question. Consequently, forum engagement can be expected to be medium when the account is geared towards principals.
Such voluntary practices are presumably driven by consequential thinking rather than the logic of appropriateness as it has clear political-strategic advantages for the account-giver. Schillemans and Busuioc (2015, 210) list several advantages, including 'gains in terms of reputation, relations, political salience, credibility and even autonomy'. Therefore, they suggest that 'rational public bodies and organizations will not necessarily choose to escape accountability, but to the contrary, it might be in their strategic interest to promote and expand on their accountability' (Schillemans and Busuioc, 2015, 210, emphasis added). Busuioc (2013) suggests that a resource reasoning is at play; agency leaders may view agency-initiated parliamentary hearings as an opportunity to profile their organization, to lobby for parliamentary support and secure financial resources. Similarly, Karsten (2015, 696, emphasis added) mentions four ways in which account-giving towards principals 'serves the political-strategic interests of the accountor'. Likewise, Jabko (2003, 721, emphasis added) has shown that the European Central Bank's decision to seek accountability towards the European Parliament was 'the result of a simple calculation -the expected gains in terms of consolidated independence are far higher than the expected costs of acknowledging the European Parliament's role as a privileged interlocutor.'

Implications and ways forward
To recap and bring together the various threads discussed above, the conceptual framework presented above (see Table 3) allows us to better understand the drivers, manifestations and implications of accountability-seeking behaviour by public sector organizations. The framework gives impetus to novel directions for research on voluntary accountability. Most importantly, the conceptual framework helps us to differentiate voluntary accountability from other phenomena with which the concept has increasingly been conflated. It helps to separate general communicative and bureaucratic practices from those specific practices that meet the accountability threshold. As such, the framework is a useful instrument for evaluating whether a voluntary practice meets the full accountability criteria. This is important since a whole range of voluntary practices have been described as accountability instruments without careful consideration of the mechanism through which these practices are supposed to bring about increased accountability. Hence, there is a serious risk that many practices have been routinely conflated with voluntary accountability. This can lead us to overdiagnose the incidence of voluntary accountability, lulling us into a somewhat false sense of security as to the state of accountability in modern governance. This demonstrates the importance of treating forum engagement as an integral aspect of voluntary accountability and the need for empirical researchers to assess the degree to which actors are actually held accountable through voluntary practices.
The framework could, for instance, be used to address questions about the promise of social media -a prototypical example of a 'message in a bottle'-type of practice -for accountability (Bertot Jaeger, and Grimes 2012). The framework reminds us that it is neither the intentions of public bodies nor the properties of social media that determine whether social media contributes to public accountability, but the presence of mutual interactions between account-givers and account-holders. Importantly, the framework shifts the attention from asking questions about whether social media can bring about more accountability (for which the answer is typically 'yes') to asking questions regarding the actual enactment of accountability on social media. The proposed framework can aid us in studying this empirically. For instance, little is known about the extent to which Twitter promotes actual accountability, although it is often discussed as an example of voluntary accountability. Our framework will lead the analyst to examine the extent to which voluntary information provision on Twitter by public bodies results in actual forum engagement (i.e. citizens engaging with the tweets of public bodies) and/or consequences (i.e. the extent to which public bodies face consequences, whether reputational or other). Alternatively, one could, for instance, survey citizens to measure their willingness to engage with public bodies on social media or conduct a content analysis of citizens' responses to messages from public bodies. Thus, the framework directs the analyst to relevant questions that need to be addressed to assess whether social media meets the accountability threshold and to make informed normative judgements about the promise of social media for promoting accountability in modern governance.
The conceptual clarification is also useful to move us forward in investigating the rationales behind voluntary accountability -a topic of considerable academic debate (Busuioc and Lodge 2017;Karsten 2015;Koop 2014;Schillemans 2011) and high theoretical relevance (Schillemans and Busuioc 2015, 209). A core concern in the literature has been to adjudicate between two competing logics -i.e. the logic of appropriateness and the logic of consequences -to identify the logic that serves as a general driver behind accountability behaviour. Instead, the discussion above points at the need to determine under which conditions a certain motivational logic is most likely to become a dominant driver of account-seeking behaviour. For instance, organizations facing insufficient resources might be driven towards one-on-one engagements with principals (e.g. voluntary accountability towards politicians who decide on muchneeded resources), whereas a reputational logic might drive organizations to embrace practices involving the participation of a diverse group of societal stakeholders (such as public committees), when faced with specific threats that require broader public support (Moffitt 2014).
Another important strand for future research refers to finding out which factors determine how and to whom actors seek to advance their accountability. It is uncontroversial to state that actors wish to be evaluated positively. However, this may shed light on the observed variation in voluntary accountability practices. As actors seek positive evaluation, they are likely to advance their accountability towards forums from whom they expect support and whose support they need. In the words of Busuioc and Lodge (2016, 255), actors 'prioritize their accountability-giving behavior toward those key audiences on whose endorsements they are dependent'. This argument is in line with findings by Potter (2019) who found that public agencies in the United States take more time to consult with societal stakeholders if there is greater ideological distance between the agency and Congress -one of the principals in the United States. Potter (2019) explains this as a strategic effort by agencies to gather support from influential groups in anticipation of opposition from principals. In other words, actors may seek to advance their accountability to forums with favourable opinions in order to build support for policies that are regarded as undesirable by another forum. Such dynamics could have important consequences for the democratic function of accountability (Bertelli and Busuioc 2021).
More generally, the consequences of voluntary accountability deserve more attention in order to provide a solid evidence base for any normative claim. Voluntary accountability is often considered a remedy for accountability gaps within the public sector and declining citizens' trust in public institutions (Bovens 2007;Schillemans 2011). However, these positive consequences do not sum up the full picture of possible effects when strategic motivations are taken into account. Karsten (2015, 695) writes in this regard, 'referring citizens and councillors to mechanisms through which they could be held to account and by stressing and facilitating their own accountability, executives tried to avoid citizen participation and councillors becoming involved in decisionmaking'. In addition, it has been suggested that actors may emphasize one kind of accountability to resist other accountability claims when facing multiple, conflicting accountabilities (Black 2008). For instance, hospital employees (Pellinen et al. 2018) and museum directors (Overman 2021) have been found to emphasize their professional accountability in order to deflect responsibilities in terms of aspects of their accountability towards democratic institutions. Furthermore, actors may seek to cultivate an accountability relationship with an allying stakeholder to resist more demanding accountability relationships. For instance, Day and Klein (1987) describe in their classic book on accountabilities in the public sector how schools produce 'selfaccounts' in order to create a dialogue with parents. These voluntary accounts 'can be seen as a defensive strategy for creating a constituency of support designed to weaken attempts to move towards a hierarchical system of managerial accountability (. . .). Downward accountability is invoked in order to resist upward accountability; giving an account is seen to be a way of avoiding being called to account' (Day and Klein 1987, pp. 170-171, emphasis added). These examples show that, paradoxically, more voluntary accountability does not necessarily mean more democratic control. What is more, if strictly undertaken as presentational strategy, rational actors can be expected to particularly engage in voluntary accountability towards non-specific, internally diffused forums. Under this condition, actors can expect maximal reputational benefits, as they appear accountable while facing low potential costs arising from close monitoring (Busuioc and Lodge 2016, 254). This highlights that studying voluntary accountability not only contributes to an improved accountability theory, but also can give societally relevant insights and a fuller picture of the state of accountability in modern governance.