Accountability in the Shadow of Hierarchy: The Horizontal Accountability of Agencies

In a narrow sense, many authors understand accountability to be a communicative interaction between an accountor (person or organization) and an accountee (see Pollitt 2003: 89), in which the former’s behaviour (in the broadest sense of the word) is evaluated and judged by the latter, in light of possible consequences (c.f. Day and Klein 1987: 5; Romzek and Dubnick, 1998: 6; Scott 2000: 40; Mulgan 2003: 9). Accountability then requires an actor with a duty to render an account and a second actor with the authorization to judge and, usually, impose sanctions. In this sense, accountability refers to answerability to someone for appropriate conduct and expected performance. Accountability as it is understood here, refers to the processes by which actors provide reasons for their actions against the backdrop of possible negative (or positive) consequences (Dunn 1999: 335). Accountability implies that the accountor explains and justifies his behaviour towards a significant other (see Orbuch 1997: 455; Roberts 2001: 1551; Keohane 2002: 4; Dubnick 2005: 1).

The communicative interaction between accountor and accountee presupposes a relationship. This relationship often exhibits formal as well as informal elements. The accountee is usually in an authorized position to demand answers from the accountor. Accountability is first of all based on a formalized relationship that stipulates the rights, authorities and available sanctions of the accountee (see Broadbent et al 1996: 269; Aucoin and Heintzman 2000: 54). The accountee has the right to demand information, the duty to pass judgment and the opportunity to sanction dissatisfactory conduct. This formalized relationship distinguishes accountability from the many other communicative relations of public agents with other parties. In addition, however, there is usually also a subjective element to accountability. As different authors note, the accountor feels obliged to render an account of his conduct, behaviour and performance to the accountee (Lerner and Tetlock 1999: 255; McCandless 2001: 22). The accountor acknowledges the position of the accountee as his accountability-holder. In order to trace this subjective aspect of accountability, scholars have often asked people in public administration to indicate to whom they felt to be accountable (see Day and Klein 1987; Sinclair 1995; Romzek and Dubnick 1998; D. Dunn 1999; Verschuere et al. 2006).

From an analytical perspective, processes of accountability normally involve three phases (Mulgan 2003; Bovens 2005). In the first phase, the accountor renders an account on his conduct and performance to a significant other. This may be coined the information phase. In the second phase, accountor and accountee engage in a debate on this account. The accountee may ask for additional information and pass judgment on the behaviour of the accountor. The accountor will answer to questions and if necessary justify and defend his course of action. This is the debating phase. Finally, the accountee comes to a concluding judgment and decides whether and how to make use of available sanctions. This is the sanctions or consequences phase.

Sanctions may vary from formal disapproval to tightened regulations, fines, discharge of management or even the termination of the organization. Many authors notice that there is a hierarchy of sanctions. As Hood et al (1999: 47) indicate: ‘It emanates with the ‘ability to shame’, escalates to lighter weapons such as certificates or formal (dis)approvals and culminates in the ‘nuclear weapon’ of liquidation’. From a legal and institutional design perspective it is imperative that the accountee has sufficient investigative and sanctioning powers. The literature on accountability often focuses on the ‘heavier weapons’ amongst the formal sanctions and the adoption of performance standards. This is enhanced by the fact that many authors favour a principal-agent approach to accountability, in which controlling the agent (=accountor) is a crucial concern (see Przeworski et al. 1999; Strøm 2000; Broadbent et al 1996; Besley and Ghatak 2003).

In addition to formal sanctions, informal consequences may also follow from accountability. This was already implicit in the above quote from Hood et al (1999): their hierarchy of sanctions started with ‘the ability to shame’. Negative publicity may also be seen as a form of sanctioning, even though no formal retribution is used (see also Harlow and Rawlings 2007: 545). Accountability arrangements can be depicted on a scale, with strong mechanisms on the one extreme, where the accountee is armed with the ‘nuclear weapon’ of liquidation’, and relatively weak mechanisms on the other extreme, where the accountee may only ‘shame’ the accountor in public, in front of his peers or before the eyes of a political principal.

It is not necessarily the case that stronger forms of accountability are more influential than weaker forms, even though this would be a logical assumption. As Braithwaite (1997) indicates: big sticks rebound. The application of heavy sanctions may have negative implications for the accountee, invoke defensive reactions and perverse effects, or, at the very least, put a strain on the relationship between accountor and accountee. Conversely, authors have noted that relatively weak mechanisms can sometimes have huge impacts on public organizations. One of the reasons is that the voice of a reputable accountee may damage the reputation of the accountor if it appears in public (see Smulovitz and Peruzzotti 2003: 6). The impact of negative voice depends on reputations and the saliency of the issue involved (see Lerner and Tetlock 1999: 258). Managers of public agencies therefore often understand public evaluations as accountability processes (see Mulgan 2003: 63). In addition, the voice of a reputable accountee with limited sanctioning powers may alert and activate hierarchical superiors. For ‘weak’ accountability arrangements to be effective in this sense, it is necessary that the outcomes are published or presented to superiors or in important policy networks. Or at least it is important that the accountee anticipates that this might happen. In those instances, relatively powerless forms of accountability can become influential because they operate ‘in the shadow of hierarchy’ (Scharpf 1997). As, as we will see, many horizontal accountability arrangements have limited formal sanctioning powers—they are found at the weaker end of the scale—they are quite dependent on published voice and voice in policy networks.

Public administrations feature a wide variety of accountability arrangements. Different authors have made different categorisations of accountability. Romzek and Dubnick (1998) for instance distinguish between political, hierarchical, professional and legal accountability; and Behn (2001) between accountability for results, process and finances. In this paper, different accountability arrangements are categorized on the basis of the type of relationship between the accountor and the accountee. In traditional forms of accountability, a superior demands accountability from a subordinate. This is based on the principle of ownership (Mulgan 2003: 12) and is implicit in principal-agent approaches. Hence they can be understood as vertical forms of accountability. In order to categorize as a case of horizontal accountability, the accountee needs to be independent from both the agent and the principal. There is not a full principal-agent relationship between accountor and accountee. For semi-autonomous agencies, independent evaluators, boards of stakeholders or commissioners, journalists, interest groups and clients can all act as horizontal accountees. This is often based on the principle of affected rights and interests (Mulgan 2003: 13).

Numerous scholars have recently referred to horizontal accountability (Day and Klein 1987: 28; Sinclair 1995: 223; Scott 2000: 42; Goetz and Jenkins 2001: 363; Keohane 2002: 22; Mulgan 2003: 26; Bovens 2005). The concept is most strongly associated with the research of O’Donnell (1998; 2003. See also Kenney 2003). His conception of horizontal accountability is in the core similar to the one developed here, but differs in its implications. In the core he also asserts that horizontal accountability applies to non-hierarchical accountability. In its consequences, O’Donnell’s research (see 1998: 117–9) deals with forms of accountability between the different constitutional powers of the state. He is particularly concerned with the question whether the legislative and the courts are in a position to effectively hold the executive to account. This paper also looks at non-hierarchical accountability, but looks at different accountors: O’Donnell focuses on the accountability of the executive, this paper at agencies. From the perspective of agencies, accountability to parliament would be a form of vertical accountability whereas it is horizontal in O’Donnell’s perspective. Because of the similarity in the core of the argument this paper has chosen to make use of ‘horizontal accountability’ above other available (but narrower) concepts such as downward accountability (Verschuere et al 2006), citizen accountability (Paul 1992) or societal accountability (Smulovitz and Peruzzotti 2003).

If we summarize the argument so far, we come to the following criteria for horizontal accountability (Table 1). It is a purely analytical definition, that does not address the rich normative debate on accountability (see for normative discussions Romzek and Dubnick 1998; Behn 2001; Bovens et al 2008). The following elements define horizontal accountability arrangements.

The sum of all accountability arrangements of one agent may be called its accountability regime (Scott 2000: 55; Bovens et al 2008). The introduction of horizontal accountability produces an accountability regime in which different accountees demand accountability, even though the ministry is still the most important. The prime evaluative question for additional arrangements is then how their role is defined in contrast or addition to ministerial responsibility. The question that needs to be answered is whether horizontal accountability is to be considered as an extension of ministerial responsibility or whether it is complementary and adds something new.

The literature tends to favour the second position. More specifically, authors have claimed that accountability to all sorts of stakeholders can be beneficial particularly from a learning perspective (Aucoin and Heintzman 2000; Behn 2001; Bovens et al 2008). Accountability creates feedback information with which public organisations may learn how to improve their conduct. Accountability forces power-holders to reflect upon their behaviour and this stimulates their learning capacities. ‘Learning’ in this respect means the ‘revision of cognitions and beliefs as a result of the transformation or recoding of information based on the observation and interpretation of experience’ (Hemerijck and Visser 2003: 4–5). Horizontal accountability is expected to produce feedback information on operations that is particularly supportive of first order, or single loop, learning processes (Argyris and Schön 1978). Additional accountability arrangements contribute to learning if they:

A minority of authors holds the opposite position that horizontal accountability should be judged from the question whether it enhances democratic control on the execution of public duties. This position implies that horizontal accountability is regarded as an extension of the democratic chain of controls on bureaucracies (Strøm 2000). From this perspective, the major evaluative questions are:

The contributions of horizontal accountability arrangements in the Netherlands will now be assessed in order to investigate which of the above positions is most appropriate. Before proceeding, however, the research design and methods will be described in short.

The information phase of horizontal accountability processes generally focuses on major strategic policy issues, such as budgets, tariffs, quality standards and political developments in the field. The different accountees are mostly informed about the conduct of the agencies through the same documents that are also presented to the principal. These documents generally cover the most important strategic aspects of the activities of agencies. Documents such as annual reports, mid-term strategic plans and all documents about services are discussed with the ministry, board of commissioners, board of stakeholders, and they are embodied in the visitation procedure. The minutes of meetings show that the different boards comment and give advise on documents such as mid term strategic plans, plans to reorganize the organization, the annual report, new tariffs, and relevant legal and political developments.

The (re-)use of the same information in different accountability processes is generally considered as a positive feature, or at least as something very practical. As one respondent notes: ‘It is true that we discuss the same documents with the ministry, the board of commissioners and the board of stakeholders. Luckily, I should add. Otherwise it would be an enormous task providing all the information’.

Most respondents indicate that the topics that are covered in the information phase can be loosely divided into two categories. One respondent indicates that the ‘inner world’ of the general strategy of the agency is the first important issue. This also regards issues such as communication with customers and quality standards. The second major issue is, according to the same respondent, the ‘outer world’ of relationships with political principals and societal stakeholders. This last finding is a little surprising, as they formally have to contribute to the quality of services and stability of the agencies (the inner world). In practice however, all of these horizontal accountees display a keen interest in the politico-strategic issues of the outer world with which the agencies have to deal. Boards therefore demand to be informed about all major external and political developments.

It is important to note that there is a more or less supply-driven information provision in the information phase that distinguishes horizontal accountability from traditional forms of vertical accountability. In vertical accountability, agencies have to provide government, inspection services, Ombudsmen, Auditor-General and judges with all the information they require. The exchange of information is usually driven by the demands of the accountees. In horizontal arrangements, the accountees take more passive stance. They concentrate on the “drill of the annual planning and control cycle”, as one respondent noted. The accountees are more dependent on the information that is provided to them because they usually have less time at their disposal and have limited knowledge of what information is available. There is an information-asymmetry between accountor and accountee that stems from the fact that the organisation of course knows a lot more about its own operations than the accountees do. When asked, none of the members of the different boards found this problematic. They were convinced that they received all the information they needed, because they fostered friendly relationships with the managers of the agencies.

In the debating phase of horizontal accountability, the dialogue between accountors and accountees generally covers the most important aspects of the ‘inner world’ and ‘outer world’ of the agencies. Most respondents describe the tone of their dialogue as ‘cordial’ and ‘friendly’. All accountees have the dual task of assisting and advising the organisation on future action and evaluating past performance. In practice, the former issue often prevails. Accountees do not concentrate their scrutiny on breaches of performance standards. Rather, their focus is reflective and aimed at future improvements. As different respondents indicate, accountees tend to ask: ‘what can we learn from this?’ And another respondent adds: ‘We prefer to operate in the realm where actions are still to be taken’.

When asked, most respondents describe the role of the different boards and visitations in terms that clearly suggest’reflection’. ‘Their role in practice is to look ahead’, one observer notes. Others add that they operate as ‘sparring partners’ to management over major strategic issues, that they abound in posing ‘difficult questions’ on new policy decisions and that their role is to ‘enhance critical reflection’.

This cooperative and reflective mode of operation is enhanced by the interests of both parties. Members of boards and visitation committees have to deal with information-asymmetry. The agencies are large bureaucracies with hundreds or thousands of employees. They perform complex tasks in often turbulent political contexts. Members of boards and visitation committees simply do not have the time to penetrate into all of the issues involved. Self-estimations of invested time run from 50 to a 100 available annual hours. In addition, they are supported by a small staff of employees of the agency. In this context, a reflective and cooperative mode of operation is rational. Some members of boards reflect upon this openly. They indicate that the quality of the relationship also has strategic benefits, because: “you simply receive more information” and: “we attain more this way”.

Fostering a more or less cooperative accountability relationship is also beneficial to the other party: the managers of the agencies. They indicate that the different members of the boards are all connected to relevant outside constituencies. Members of boards are sometimes seen as interlocutors to the minister, they have networks in important bureaucratic circles or they are connected to the strongest societal organisations and pressure groups. It is in part against this background that managers explicitly indicate that is important to develop a positive relationship in horizontal accountability. They indicate that this could ‘enhance their reputation’ and that the members of boards may ‘serve as ambassadors to our cause’.

However, irrespective of the motives of accountors and accountees for developing a cooperative and reflective relationship, this reflection-enhancing role strengthens a bond of trust between accountor and accountee. This creates a form of accountability wherein the social distance between accountor and accountee decreases. This is what Roberts calls a form of face-to-face accountability that can have socializing effects (Roberts 2001: 1567). The repeated dialogue between the two parties fosters trust and cooperation and this may challenge explicit or implicit policy assumptions. Over time, their repeated dialogue stimulates learning processes. In accordance with this view it is clear that the different accountees all aim to formulate lessons for future improvements.

In comparison to traditional forms of accountability, it is noteworthy that the horizontal accountability processes evolve in relatively safe settings. As one director of an agency notes: ‘It is not similar to meetings with the ministry, where I always go with a slightly increased level of adrenaline. It’s not like that when I meet with the board’. However, the dialogue is not devoid of tension either. Both parties acknowledge—and notify this explicitly and usually without being asked directly in the interviews—that the principal is a silent listener to their discussions. In some instances, minutes of their meetings are sent to the minister and screened by his civil servants. As one departmental respondent puts it: ‘I always listen carefully to what the board has to say, because you know that the opinions [of board members] are likely to recur later on’. In other instances, members of boards sometimes meet formally or informally with the minister or senior ministerial officials. In these circumstances, boards can serve as important ‘ambassadors’ to the cause of the agency, or, in a less positive fashion, they can function as a ‘snitch’, as another manager feared. And in still other instances, the outcomes of horizontal accountability can and will be used by departmental personnel as monitoring information. The ambiguity of the possible consequences adds tension to the dialogue between accountor and accountee that circumvents the perils of a collusion of interest.

Processes of horizontal accountability can have different sorts of consequences for agencies. It may directly influence policy decisions and may also work indirectly, when the judgments and voice of the accountees activates other parties, such as the ministry, a regulator or the media. And accountees may use formal sanctions, when available. The different sorts of consequences were found in the investigated agencies. In the observed cases there was some direct influence of horizontal accountability and there was some additional influence through the interaction with the ‘shadow of hierarchy’.

The only mechanisms for horizontal accountability with strong formal sanctions are (most of the) boards of commissioners. Important powers that normally reside with the minister have been partially transferred to these boards. These include budgetary powers, the appointment and discharge of management, tariffs and the possibility to lay down binding directives for operations. Boards address all major operations and organizational strategies, just as boards of commissioners do in the ‘two tier system’ of the private sector.

The fact that there are little formal sanctions and that the available sanctions are not used might suggest that horizontal accountability had no influence on the policies of the agencies. This suggestion is not correct. The research showed many policy decisions that were influenced through horizontal accountability. Processes of horizontal accountability generally have a clear, but somewhat modest, influence on the decisions of agencies. For most of the agencies that were examined, it was possible to find a number of specific decisions that were influenced by horizontal accountability. Influence was acknowledged to exist when two independent respondents together with ‘evidence’ from formal documents, such as minutes of meetings, pointed in the same direction. For boards, most instances of influence concerned improvements to systems of management or the relationship with the ministry. For boards of stakeholders, instances of influence were mostly concerned with specific aspects of services, and particularly communications with clients. The influence of visitations is primarily directed towards transparency and responsiveness. Table 3 presents an overview of decisions and issues that were found to be influenced by horizontal accountability.

The length of this table perhaps falsely suggests that horizontal accountability was found to have a very strong direct influence on the decisions of agencies. But that is not the case. Most listed items are decisions that can be taken without further consequences for the overall policies of the agency. They are important to the system or aspects of quality care, but can not be considered as major strategic changes. Horizontal accountability was then found to have a clear influence on the agencies—but also clearly less influence than hierarchical powers with stronger authorities. The fact, however, that all mechanisms at least did have some impact on agency-policies, even without strong formal sanctions, can be understood through the ‘strength’ of published voice and the policy networks of the different members.

Horizontal accountability was found to have indirect consequences by activating the ministry and the minister—or by the anticipation that this might happen. All respondents indicated that they were aware of the fact that hierarchical actors might and often did ‘listen’ to the dialogue between agencies and the different boards. Departmental respondents stated that they scanned agenda’s, minutes, reports and other evolving documents in order to trace ‘new developments’, ‘risks’ or ‘unwanted policy interpretations’. Particularly the visitations showed several of these informal consequences. This is the case because the resulting report is often used as input for vertical accountability. Visitations are used by ministerial departments as forms of fire alarm monitoring (McCubbins and Schwartz 1986). The visitation procedure is highly public. The reports are published and the press write about the outcomes. This is possible, because the process creates new and often highly informative information on the agencies. This is due to the outspoken style of the evaluation committee, the level of new information that is brought out through the self-evaluations and the clear-cut evaluative perspective that is used. Through this all, the process is always informative to outsiders. The visitation procedure therefore has started to play a role that extends the confines of the relationship between agency and visitation committee. Some reports have been discussed in the ministries. Some ministers have drafted formal reactions and often it has lead to discussions between parent-department and agency about specific outcomes. In one of the cases, the permanent secretary of the parent-department did an unsuccessful attempt to prevent the publication of the report. In another case it was agreed that the ministry was to receive the draft-report before publication, which left some room for intervention. The visitation thus adds to the monitoring capacity of the principal. Also, the voice of the boards in particular was sometimes taken over by the ministry. One example is where a board of stakeholders issues a public advice on the tariffs of an agency that is subsequently adopted by the minister.

More important however than the question whether or not the ministry acts upon the signals from horizontal accountability is the insight that the agencies were found to anticipate this. Given that the boards are potential ambassadors of the organisation, negative voice by boards in public or policy networks is clearly unwanted. Ministerial respondents point out that this is exactly what they expect horizontal accountability to accomplish. And furthermore, should for some reason a negative voice evolve, they would certainly act upon it. As one respondent puts it: “I assume that we would consider a negative voice from the board as a rather serious signal. Given the fact that the organisation will try to prevent this to occur (…) so if it does occur, then it needs to be taken really seriously”.

The ministerial respondents thus correctly expect that agencies try to prevent negative conclusions in horizontal accountability. Loss of reputation and organizational legitimacy are important motivational factors in this respect. As Verhoest et al (2007: 487) note, the possibility of sanctions creates a political pressure to adapt policies and this can lead to innovative behaviour. Managers foster the legitimacy of their organisation. They anticipate that negative news may lead to negative future consequences.

Horizontal accountability is important in respect to the legitimacy and reputation of agencies. Through horizontal accountability, agencies are able to build a coalition of supporters. Carpenter (2001) showed how agencies actively look for support for their policies and try to build networks in different important arenas. In the long run, it is important that the agency has a political multiplicity, which means that it draws support from a variety of proponents. This is a safer situation for the agency than the situation where an agency is fully dependent on a single department, political party, dominant interest group or segment of society. Horizontal accountability can be used to build long-term constituencies. It can contribute to reputational uniqueness, which means that the agency can demonstrate that it possesses skills and expertise that no other agency or political actor can provide. Positive voice by a reputable third party, boards or a visitation committee, is favourable from this perspective. Their support strengthens the reputation of the agency and increases trust by political principals.