The concept of accountability
The concept of accountability is rather elusive. The term accountability was widely used in public administration literature for decades, focusing on the organizational sphere and its political linkages. The concern over accountability in modern public administration was also associated with the debate over the administrative reforms put in place in OECD countries starting in the 1980s and the need to keep the administration responsible to those having political legitimacy (Hood 1991; Osborne and Gaebler 1992). Today, however, accountability is portrayed as a complex and multifaceted concept. The notion has expanded over the last 20 years as it has moved from the more straightforward notion of accountability within organizations in the public sphere to that of public accountability. Furthermore, discussions on different dimensions of accountability have resonated continuously in the literature (Bevir 2010; Considine and Afzal 2012).
Whereas broad conceptualizations distinguish between accountability as a mechanism and accountability as a virtue (Akpannuko and Asogwa 2011), the latter makes it very difficult to establish empirically whether an official or organization has accountability virtues (Bovens 2007; Bovens et al. 2008). In fact, this is related to the multiple ways in which professionals are accountable for their practices and activities, which raises the broad issue of their relations with society in general - certainly beyond the scope of our analysis. Therefore, a narrower definition of this concept based on the idea of accountability as a mechanism is required. Building on these insights, we take as a departure point Bovens's narrow definition of accountability as any '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 judgement, and the actor may face consequences' (Bovens 2007: p. 450).
Thus understood, accountability describes a relationship between power-holders and those affected by their actions. It comprises two key elements: 'answerability' - making power-holders explain and justify their actions - and 'enforceability', allowing the participants in the forum to judge and punish poor or criminal performance (Schedler 1999). Translated to democratic practices, this means that the whole citizenship is the ultimate forum, that each public officer and public institution remains in dialogue with their particular forum, and they should expect to face some specific consequences if they fail to produce particular results. Concerning the regulatory state in a democratic society, accountability does not mean that the publicity principle is to be introduced in all cases, but rather that it requires a legitimate association to the forum (Luban 2002).
The scope and meaning of accountability for public organizations has been extended in a number of directions beyond its core sense of 'being called to account for one's actions' through a social exchange by external bodies or groups having some rights over one. This 'extension' of accountability refers also to the creation of new accountability instruments, which include tools to report not only to parliament or the corresponding ministry, but also to direct stakeholders and the public. Authors have emphasized particular accountability mechanisms over others, namely, citizens' oversight (Graham 1997) and output-based mechanisms (Majone 1999). Thus, the literature often introduces several distinctions to classify such diverse mechanisms according to the characteristics of accountability obligations, or to the features of the relations of those who are to be accountable towards particular actors within the forum.
As to accountability obligations, there are different categorizations. Mandatory vs voluntary accountability identifies whether there is or not a legal basis for accountability mechanisms (Koop 2013). Differentiation between formal and informal accountability mechanisms represents a similar division, depending on how accurately the notion of informal mechanism could be formulated. Distinguishing between ex ante systems of control and ex post, results-oriented systems of control refers to the prospective or retrospective nature of the activity the agent is to account for. This distinction, however, has been problematized on the basis that accountability should only focus on the actor's obligation to answer for past actions (Mulgan 2011). In fact, three steps have been identified to constitute a complete accountability relation, namely, information, discussion and consequences (for example, Mulgan 2003). The first one involves explaining conduct, aims or rationales, while the second one entails establishing interactions between the specific forum actors and the subject, and the last one refers to the judgements actors in the forum make and communicate - and which can eventually involve sanctions (Brandsma and Schillemans 2013).
In respect to the characteristics of the accountability relation, traditional discussions on accountability have remained largely focused on institutional mechanisms, especially on the importance of input and parliamentary channels (Lodge 2004). However, the literature has further distinguished between upward and downward relations (for example, Verschuere et al. 2006). While upward mechanisms concern those involving hierarchical relations based on the principal-agent logic, downward accountability implies non-hierarchical mechanisms, namely rendered to relations with citizens, users and stakeholders in particular. Nevertheless, the division between upward and downward accountability cannot be related to horizontal accountability, as far as we understand horizontal relations as those occurring between agents or institutions of similar status and where only under particular circumstances may a hierarchical relation apply. To solve this inconsistency, and building on Scott (2000: p. 42), we adopt a more precise distinction that implies three directions: upwards, horizontal and downwards. More generally, this is what Mulgan characterized as a '360-degree' view of accountability (2011: p. 4).
To better clarify these distinctions, upward accountability can be identified with relations between citizens and representatives, or elected politicians vis-a-vis bureaucrats; whereas downward accountability is related to delivery organizations vis-a-vis consumers, or regulatory agencies vis-a-vis interests groups. Finally, horizontal accountability in the public domain can be understood basically as inter-institutional accountability. As Pasquino details, 'it occurs when the various institutions in different ways are in a position to control the activities of another specific institution' (2011: p. 17). Typical cases include relations between the executive and the legislative or between the judiciary and other independent authorities. In each case, mechanisms of accountability can be either two-sided or one-sided, involving strong sanctions or not. Still, all cases are subject to a specific set of rules defining the conditions under which these mechanisms will be activated.
Beyond these classifications, and in an attempt to categorize theoretical approaches on public accountability, Bovens et al. (2008) identified three perspectives on accountability mechanisms, based on different assessment criteria: democratic, constitutional and learning. Democratic accountability refers to the goal of providing mechanisms to make the executive and its agencies explain and debate their actions in a democratic regime. Clearly, upward accountability mechanisms predominate here. According to this, the main objective is to keep active and vital the 'democratic chain of delegation' from citizens - the audiences they serve - to the executive. The second perspective - constitutional accountability - focuses on the prevention of governmental abuses, and its corresponding assessment criteria aim to control government activities and prevent the exercise of potential privileges. In this case, horizontal mechanisms prevail given that inter-institutional accountability is the main procedure to assure constitutional stability. Finally, the third perspective expects accountability to trigger the learning capability of governments, whereas the criteria are whether this process induces the executive to achieve the most desired societal goals. Downward mechanisms are dominant within this perspective given that they involve often continuous interactions and dialogue with multiple stakeholders.
During recent decades various literatures have increasingly used these notions of accountability, both from national and international perspectives. From the domestic perspective, the concept of accountability has come a long way, starting with the introduction of the New Public Management (NPM) approaches. The NPM doctrines that confronted the traditional public administration paradigm (see mainly Hood 1991; Pollitt 1993; Christensen and L^greid 2006) challenged hierarchical controls by representative institutions and, in consequence, made it necessary to promote a more diversified understanding of accountability for new public managers (Dunleavy and Hood 1994; Peters and Wright 1996; Olsen 2003). In fact, some of the major prescriptions for administrative reform, such as the autonomy of public managers, empowerment of front line workers and market-like competition to create incentives, were perceived as a challenge to traditional public administration theories (Deleon 1998). However, the NPM concentrated on expanding the control perspective in hierarchical contexts, without considering the other theoretical perspectives identified by Bovens et al. (2008); namely, the learning potential of accountability or the design of mechanisms to enhance democratic accountability in autonomous institutions.
The international relations literature, namely debates on 'governing without government', has focused on the accountability deficit of institutions of global governance, either from the perspective of democratic deficit, the lack of control, or the need to increase dialogue and learning (Rosenau 2000; Keohane and Nye 2003; Grant and Keohane 2005; Philp 2009). In a similar vein, a growing literature has analysed the democratic accountability deficit at the European Union (EU) level and the EU regulatory state (Majone 1994; Majone 1996; Thatcher 2002; Geradin et al. 2005; Christensen and L^greid 2006).
The increasing fragmentation and autonomy of the public realm within and across levels has been accompanied by the rise of rather sophisticated forms of accountability, eventually also including a transnational dimension as well. Consequently, concerns over democratic quality have also fostered more interest in a variety of innovative accountability mechanisms (Verschuere et al. 2006).