Tobias Bach, Birgitta Niklasson, Martin Painter, The role of agencies in policy-making, Policy and Society, Volume 31, Issue 3, September 2012, Pages 183–193, https://doi.org/10.1016/j.polsoc.2012.07.001
Navbar Search Filter Mobile Enter search term Search Navbar Search Filter Enter search term SearchThere is an extensive literature on the proliferation of agencies and the delegation of authority to such bodies across different countries. Much of this research asks whether New Public Management (NPM)-style agencification reforms have been implemented according to the original objectives, and what consequences – intended or unintended – the reforms have produced. Yet much of this research lacks an explicit link to the literature on the policymaking functions of public bureaucracies and their interactions with elected politicians. What are the consequences of agencification for the policy process at large? What policy relevant tasks do agencies perform, to what extent are they involved in policy-making, and what factors influence the quality and quantity of their participation? This introductory article gives an overview of key concepts such as “public agencies” and “policy autonomy” and the research literature. Moreover, it critically discusses relevant theoretical perspectives, outlines the articles included in this themed issue and argues for a more systematic and theoretically guided analysis of agencies’ role in policy-making.
This themed issue brings together several articles on the role that government agencies play in policy-making processes. The policy activities of government agencies (a more precise definition follows below) constitute an important aspect of their autonomy vis-à-vis their political principals and we claim that there is a need for a more systematic, theoretically guided analysis on the role of agencies in policy-making. The articles in this collection take a step in this direction.
There is an extensive literature on the proliferation of agencies (or “agencification” as it is often somewhat clumsily labeled) and the delegation of authority to such bodies across different countries ( Christensen & Lægreid, 2006; Lægreid & Verhoest, 2010; Pollitt & Talbot, 2004; Verhoest, Van Thiel, Bouckaert, & Lægreid, 2012). This literature has expanded in recent years in response to the spread of agencification within the context of New Public Management (NPM) reforms. A lot of this research addresses whether these administrative reforms have been implemented according to the original objectives, and what consequences – intended or unintended – the reforms have produced. The main focus of this literature is on different aspects of the managerial autonomy of public agencies, for example agencies’ accountability toward politicians and stakeholders; the use and effects of performance management techniques; and problems of coordination across organizational borders. Yet much of this research lacks an explicit link to the literature on the policy-making functions of public bureaucracies and their interactions with elected politicians.
There is an implicit assumption in the NPM agencification model that, in return for enhanced managerial autonomy, agencies will refrain from becoming involved in policy-making. If this proposition has been fulfilled, agency policy autonomy should be more restricted after NPM reforms. But has this indeed been so? What are the consequences of agencification for the policy process at large? What policy-relevant tasks do agencies perform? To what extent are they involved in policy-making? What factors influence the quality and quantity of their participation? These are the general questions addressed by the introductory article and the individual articles in this themed issue.
All these questions touch upon the perennial question of democratic control of the bureaucracy ( Aberbach & Rockman, 2006; Painter & Yee, 2011; Svara, 2006). This question is concerned with what the division of labor between democratically supported political leaders and the civil service looks like and whether it conforms to some kind of ideal-type. Researchers on these fields are far from reaching an agreement on what the ideal should be, whether different ideals can in fact be reached (and in that case how), and how close to the ideals we are today. Our hope is primarily that this issue will shed some more light on the last of these questions. However, before taking a closer look at the ideals that have been suggested, we clarify what we mean by “public agencies” and by “policy autonomy”.
In many countries “agency” is used as a generic term for all kinds of public organizations and will include “central agencies” headed by ministers as well as “executive agencies” headed by public officials. The articles in this issue explicitly focus on the latter group. The organizations studied in this themed issue are public sector organizations structurally separated from the government offices but “close enough to permit ministers/secretaries of state to alter the budgets and main operational goals of the organization” ( Pollitt, Talbot, Caulfield, & Smullen, 2004, p. 10). Thus, they belong to the wide range of semi-autonomous bodies that Chester (1953) places in his third category of public sector organizations (the first being departments; the second, local authorities; and the third, “the rest”). According to Thynne (2004) agencies are “executive bodies, as well as those statutory bodies which are not incorporated and do not have responsibilities that rightly distance them from ministerial oversight and direction (…). They are all public law, non-ministerial organisations which relate to ministers or the government as agents to a principal.” (p. 96). This category also covers organizations sometimes discussed in terms of quasi-non-governmental organizations (quangos), fringe bodies, and non-departmental public bodies ( Wettenhall, 2003). We will refer to these organizations as government agencies, or just agencies. Whereas some countries have a long tradition of delegation to public agencies (such as Sweden, Germany, or the US) others (such as the UK, the Netherlands, or Belgium) have only more recently, as participants in the agencification trend, witnessed an increase in the number of such bodies.
Formally, the main task of these agencies is usually some form of policy implementation, such as service delivery, regulation or exercising different kinds of public authority ( Pollitt et al., 2004; Van Thiel, 2012; Thynne, 2004). Normally policy formulation is formally the task of the ministerial departments (which we will also refer to as the parent ministries) and possibly the ministerial cabinets. The chief executives of agencies are usually civil servants, who are appointed by the government or the minister in charge. In most countries, the political responsibility for the activities of the agencies also remains with the ministers or – like in Sweden – with the cabinet.
These agencies are thus not strictly comparable to many US agencies, which are “caught in the middle” ( Weingast, 2005) between Presidential and Congressional influence. In parliamentary contexts, the control of agencies by the legislature, by and large, takes place via the parent ministry and the politically responsible minister. All empirical articles in this collection focus on agencies in parliamentary systems of government, with the partial exception of the contribution on Hong Kong agencies. However, Hong Kong has a machinery of government inherited from British rule that strongly resembles the “ministerial” governments typically found in parliamentary systems.
To sum up, the agencies covered in this themed issue operate at arm’s length from their political principals and enjoy some degree of autonomy. However, it is generally assumed that they possess little, if any, policy autonomy. In the following section we clarify what we mean by “policy autonomy” and outline the empirical literature on public agencies’ policy autonomy.
The concept of autonomy refers to a capacity to act independently from the control of other actors. In our context, it relates in the first instance to a dependency relationship between two or more actors, one or more of whom are formally configured as superior or supervising actors (the “principals”). In parliamentary systems of government, in addition to the competencies delegated to agencies by statute, agency autonomy is a function of the delegation of roles and competencies from the parent ministry. We focus on one particular aspect of agency autonomy, namely policy autonomy, which can then be understood as the degree of policy-making competency enjoyed by an agency in relation to its parent ministry ( Verhoest, Peters, Bouckaert, & Verschuere, 2004). However, policy-making competencies come in many forms because policy-making is a complex process in which there are multiple roles. Thus, the degree of policy autonomy can vary across different roles in the policy process.
A fruitful distinction to make at the outset is that between formal and de facto autonomy. The actual (de facto) policy autonomy that agencies enjoy may diverge from their formal policy autonomy (what the agency’s statute and other legal documents specify as the appropriate agency role in policy-making) ( Yesilkagit, 2004). All the papers in this issue aim at describing and explaining the de facto political autonomy of public agencies.
The concept of policy autonomy is closely linked to Carpenter’s notion of “bureaucratic autonomy” which he defines as a situation in which “politically differentiated agencies take sustained patterns of action consistent with their own wishes, patterns that will be not checked or reversed by elected authorities, organized interests, or courts.” ( Carpenter, 2001, p. 14). In other words, bureaucratically autonomous agencies are able to sustain preferred policies through their administrative actions (perhaps against demands for change); they are capable of designing new policies according to their own preferences; they may fight for those policies in the public realm; and their actions may change the policy preferences of their “principals”, potentially with the result of new authoritative policy enactments. The bureaucratic autonomy perspective is useful because of its explicit focus on power and influence over policy outcomes through multiple agency interventions in different stages of the policy process. In addition it highlights the possible importance of a wider stage of action and of actors than simply the dyadic relationship of minister and agency. Agencies’ relations of dependency with these actors (organized interests, clients and ‘stakeholders’ more generally) may also significantly affect their policy autonomy vis-à-vis ministers and parent ministries.
A bureaucratic autonomy perspective in one sense turns the policy-operations split on its head and argues that granting autonomy (of any kind) may result in an enhanced policy role for an agency. The empirical task becomes one of observing the impacts of an agency on policy-making and policy outcomes across all phases of the policy cycle. We argue that a distinction should be made between (1) agency involvement in and influence over authoritative policy decisions and (2) agency involvement in and influence over a wider range of decisions and actions that significantly affect policy outcomes, which we propose as a measure of policy autonomy more generally. These dimensions essentially reflect different stages of the policy process, which will be elaborated in more detail below.
Scattered findings on agencies’ policy autonomy can be gleaned from a diverse existing literature. First, it is a common finding that agencies do have policy roles and functions. A quantitative, comparative study analyzing the perceived policy autonomy of agencies in Flanders, Ireland, and Norway ( Verhoest, Roness, Verschuere, Rubecksen, & MacCarthaigh, 2010) found that perceived policy autonomy is fairly high in all three countries, particularly among older agencies. In an in-depth case study, Egeberg (1995) compares the role of two central agencies in transportation policy in Norway. He finds that a lot of policy formulation takes place at the agency level and attributes variation across agencies to organizational factors at the ministry and agency level. In a cross-national design based on survey data, Maggetti (2009) analyzes the “centrality” of regulatory agencies in specific policy processes and finds that regulatory agencies generally had central positions in those processes. Substantial policy-making activities among agencies are also detected in the Netherlands ( Yesilkagit & Van Thiel, 2008) and Germany ( Bach, 2010).
Policy involvement has been found to vary according to formal roles and functions. Agencies do not all conform to a single blueprint. In the UK, different mixtures of formal roles can be found, with some agencies having only executive tasks, others having both policy and executive tasks, and a third type sharing both functions with the parent department ( Talbot, 2004). This variety is reflected both in formal mandates such as executive agencies’ framework documents and also theses agencies’ de facto involvement in policy formulation ( Gains, 2003). Yesilkagit and Van Thiel (2008) found significant differences between legal types, suggesting that politicians may be more willing to delegate policy roles where they can exercise close oversight. An in-depth case study of the participation of a large service delivery agency in Flanders in two policy processes explains variation in the agency’s influence by the agency’s capacity for policy development, the agency’s policy preferences, and the political principal’s willingness to include the agency in the policy process ( Verschuere, 2009; Verschuere & Bach, 2012).
Cross-national differences have also been observed, pointing to the importance of national traditions and contexts. Elder and Page (1998) compare the policy roles of agencies in Sweden and Germany. A key finding is that Swedish agencies seem to have more extensive policy-making roles compared to their German counterparts. This difference is attributed to the consensus-orientation of the Swedish politico-administrative system (i.e. many issues which in other contexts would be considered “policy” are understood as “operational”) and the consultation procedure during which “draft reports of commissions of inquiry intended to lead to legislation are circulated to interested administrative agencies inter alia for their comments” ( Elder & Page, 1998, p. 37). They report several episodes of policies that originated in public agencies, which does not seem to represent a consistent pattern, though. According to Pierre (2004) several Swedish agencies have become “sources of policy” in the 1990s and have “either put forward extensive policy proposals or – by virtue of their institutional autonomy – formulated programs to guide their actions” (pp. 210–212).
In sum, the involvement of agencies in policy formulation is quite common, even in contexts where norms of separating policy and operations prevail. This is the case both for NPM-type agencies and for other types. Moreover, agency involvement in policy formulation comes in many guises, such as developing policy proposals, drafting laws and regulations, commenting on draft laws in order to assess the practicability of policy proposals, evaluating existing policies, or formulating programs or rules as part of the agency’s formal sphere of authority. Several studies suggest that agencies’ contributions to policy formulation by and large are based on their experiential knowledge from policy implementation. Also, parent ministries seem to have a gatekeeping function with regard to agencies’ policy participation. Finally, the literature review underlines the relevance of politico-administrative contexts, something that is also followed up by some of the studies in this issue ( Niklasson & Pierre, 2012; Painter & Yee, 2012; Verschuere & Vancoppenolle, 2012). In the next section we link the literature on agencies’ policy autonomy to a “stages” or “policy cycle” understanding of the policy process.
Mintrom (2012) defines public policies as “the choices that governments make on behalf of citizens living in their jurisdictions. These choices are codified in the rules, plans, principles, and strategies that guide government action” (pp. 1–2). This definition presents a somewhat static picture of public policy, but it is important from our perspective to recognize that it also a dynamic process. Policies are complex series of decisions that stretch over long periods of time. “Policy in action” involves a series of stages and events, sometimes depicted in simplified form as a policy cycle. Standard models describe a series of stages such as problem-definition, agenda-setting and advocacy, setting of goals and targets, formulation and analysis of options, selection of the preferred option, policy implementation, evaluation, and feedback (for critical reviews of these models see Colebatch, 2006; Jann & Wegrich, 2007).
One common way of defining the ideal roles of politicians and civil servants in the policy process is to tie these actors to a specific part of the policy cycle. An ideal that was advocated many years ago by prominent scholars such as Weber (1978) and Wilson (1887) is that policies should be formulated by politicians and implemented by civil servants. Thus, all stages except for the implementation, evaluation, and feedback stages should be the domain of the politicians and the later stages should be that of the civil servants ( Hill, 2009). The ideas of NPM, which started to spread in the 1980s, pushed for administrative reforms that were in line with this ideal ( Hood, 1995). The principle of a policy-administration dichotomy was given fresh impetus in NPM by drawing on principal–agent theories of delegation, in which a dichotomous division of roles and responsibilities was a central feature ( Verschuere, 2009). This resulted in the agencification trend; in many countries, agencies were created or reformed in ways inspired by NPM-thinking ( Van Thiel, 2012). New public agencies, operating “at arm’s length” from their political principals, were founded in order to secure the managerial autonomy of the civil service in the implementation process. One idea behind these reforms was that agencies should become more like private companies, supposedly leading to efficiency gains. Overarching policies, on the other hand, were to be created by politicians with the support of ministerial departments and policy staffs.
There are both normative and also empirical objections to the ideal division of labor within the policy cycle as presented above, however. From a democratic point of view, one could claim that it is legitimate for politicians to exercise control over the outputs of the civil service, including the implementation process. Reserving the latter for the experts suggests technocracy, not democracy. Politicians, not civil servants, will be held accountable for government output, including the acts of the administrators. Leaving the normative arguments aside, from an empirical viewpoint it is hard to keep the policy stages apart in practice, since different parts of the policy cycle may be occurring simultaneously. For example, elaboration and measurement of the pros and cons of different options (which is mostly a technical process often done by professional civil servants) can help to re-formulate the problem or to envisage different goals and targets (which is what political leaders should be doing). The stages may even occur in a different order; the analysis of options may be retrospective, so as to demonstrate that the preferred option is the right one. Similarly, there is no sharp distinction between when a policy is created and when it is implemented. Policy decisions can just as well be made during the implementation stages of the policy process as during the earlier stages. As part of their implementation role, agencies are often delegated the power to make decisions on what tools and methods should be used in order to achieve a goal that has been set up by the politicians. These kinds of decisions contribute to the shaping of the policy in significant ways and they certainly affect the outcome and the future development of the policy. Decisions made by agencies during the implementation stages can therefore be viewed as policy-making decisions ( Howlett, Ramesh, & Perl, 2009; Jann & Wegrich, 2007; Verschuere, 2009). Hence, several researchers refer to agencies’ authority to make these kinds of operational decisions as policy autonomy ( Yesilkagit & Van Thiel, 2008) or implementation autonomy ( Bach, 2010).
Policies are in fact going through constant change and re-formulation throughout the whole policy cycle. Consequently, there is also not a simple or neat fit of roles for different actors in the policy cycle that can be captured in formal statements. In reality, the policy roles occupied by political and administrative leaders are diverse and multiple. Moreover, roles and contributions are not static. Societal pressures and events continually place agency chiefs and politicians in new situations to which they must respond (indeed, politicians often respond by diving headlong into the details of agency administration) ( Pollitt et al., 2004). In the process, agency chiefs are called upon both to use their administrative discretion and also to perform roles in on-going deliberative (or policy) processes. While they may recognize limits and constraints on their roles due to the nature of the formal delegations that constitute their office, their de facto discretion (the space they enjoy to define their own roles) may be extensive. For example, policy-making includes giving and receiving advice based on various forms of knowledge, which comes from many quarters including administrators. In a technical sphere (so long as it remains defined as such), agency chiefs may well be trusted for their policy knowledge and advice, primarily because they are considered better at understanding a particular field and it has become the norm that they take the lead ( Gains, 2003).
There are clearly conflicting considerations and criteria on how to divide roles and functions in the policy process between political and administrative actors, or ‘principal and agent’. But in spite of the problems related to attempts to distinguish neatly between policy-making activities and roles in different stages of the policy cycle, this framework allows us to analyze and compare the involvement of agencies in the policy process systematically through observing their different contributions. Several papers in this issue mainly look at the earlier stages of the policy cycle, that is, those before the actual implementation ( Bach, 2012; Handke, 2012; Niklasson & Pierre, 2012), but some articles cover the whole policy cycle ( Painter & Yee, 2012; Verschuere & Vancoppenolle, 2012).
This section highlights how the approach to the study of policy autonomy in this themed issue relates to some of the main theoretical perspectives in contemporary research on government agencies. Much of the discussion of agency roles in the traditional public administration literature is embedded in theories of ‘formal institutionalism’. That is, the starting proposition is that the legal-formal definitions of institutional rules and structures primarily determine behaviour and outcomes ( Moe, 1997). However, as our conceptual discussion has already argued, formal and structural explanations must be supplemented by theoretical understandings of the informal dimensions of agency roles and contributions within a wider social and political context. But staying with institutional approaches for the moment, a powerful strand of theory in the literature is drawn from rational choice institutionalism. The delegation perspective on agencies assumes that political decision-makers delegate authority to (executive) agencies which may behave as rational agents pursuing their own preferences, which differ from those of their political principals. There is now a vast literature on how politicians supposedly engage in deliberately designing administrative structures and procedures in order to curb agency discretion and to prevent bureaucratic drift ( Braun & Gilardi, 2006; Huber & Shipan, 2002; Pollack, 2002).
Our conceptual analysis has already covered ground that suggests this approach is of limited application for understanding the policy roles of agencies. The dichotomous division of labor between politicians and civil servants implicit in delegation theories and principal–agent accounts of politico-administrative relations is notoriously slippery in principle and it is even more difficult to sustain in practice. The assumption that policy is equated with goals or strategy is not always helpful in analyzing how policy and administrative decisions are actually made. One meaning of policy to a politician is whatever, for the moment, is politically controversial ( Pollitt et al., 2004). This does not always include policy in the sense conveyed in principal–agent theory but it often includes operations, particularly when they go wrong. Most scandals are generally revealed in the detail, not the big picture, at which point politicians cannot but escape being directly engaged. This confusion of roles may work the other way round as well. Sometimes, policy may be delegated where governments are facing a hot potato and see no way out, other than to shift the blame or put it off by ‘passing the parcel’. Other times, they are pressured to intervene reluctantly by public demand and media criticism. What politicians accept responsibility for is always politically contingent and does not conform to neat, logical prescriptions. More generally, the strong focus of the delegation literature on formal structures as means of preventing “bureaucratic drift” of agencies tends to underestimate the dynamics of politico-administrative relations in the post-delegation stage.
A further confounding factor for an unambiguous principal–agent view of the policy process is that there are layers of principals and agents, not just one relationship. In the case of Hong Kong, for example, the principal official (a position similar to a minister) is clearly the principal in relations with the permanent head of the policy bureau (equivalent to a ministry). For most purposes, however, the head of the bureau is viewed as the principal by the department or agency head, although the latter is supposedly conveying the principal officials’ preferences. Exactly who might be pulling the wool over whose eyes is not always clear. In this sense, studying the policy autonomy of agencies in many instances will necessarily have to focus on ministry–agency relations, rather than minister–agency relations, or perhaps triadic constellations involving all three parties ( Gains, 2003).
Most theories of agency delegation do not in fact discuss policy autonomy ( Bendor, Glazer, & Hammond, 2001; Moe, 1997; Pollack, 2002). They talk about agency autonomy in a broader sense. They do not even always focus on delegation between ministries or departments and agencies; they talk more generally about delegation between voters and elected representatives, legislators to the executive branch, the government to different ministries, and ministries to agencies. They have used game theory and rational choice models in order to decide when a principal is likely to delegate power to an agent. Pollack (2002) concludes that it is (1) when there is only imperfect information available and there is a demand for policy-relevant expertise, (2) when the politicians need to be tied down to making credible commitments and there is a demand for independent, credible agents, and (3) when there are conflicting preferences among principals and between principals and agents. However, this body of literature tends to be almost exclusively focused on the US context and more particularly on the legislative policy-making. Moreover, these studies generally pay little attention to the mechanisms underlying how organizational structure and agency tasks shape the de facto autonomy of public agencies ( Painter & Yee, 2011).
The shortcomings of principal–agent analysis for our purposes suggest also that a starting point in NPM views of the world will not be particularly fruitful. Many researchers take their starting point in NPM, at least for formulating working hypotheses ( Gains, 2003; Verhoest et al., 2010; Verschuere, 2009; Yesilkagit, 2004). However, because NPM does not represent a coherent theory, analyses that explicitly tests propositions concerning the implementation of NPM on the organizational level, or the fit between empirical findings and NPM models and prescriptions often leads to suggestions for other analytical perspectives. For example, a fundamental problem of rationalistic approaches is highlighted by Yesilkagit (2004), who aptly summarizes the implications of agencification for the policy process:
What does happen when the agency is formally established and left alone to perform its tasks? How does the political world become a different place? What is the nature of the interactions between the agency and the political and administrative actors in this new environment? ( Yesilkagit, 2004, p. 534).
He proceeds by arguing that rational choice accounts of delegation assume that principals and agents instantly adapt their behavior to the new circumstances, which seems to be an all too optimistic assumption. He claims that “game theories of delegation neglect the time period newly established political actors need to understand and interpret the rules that are in play after the delegation decision” ( Yesilkagit, 2004, p. 535). Drawing on Berger and Luckmann’s (1966) ideas of the social construction of reality, Yesilkagit addresses these weaknesses by using the concept of habituation, which describes the process of institutionalization: “It refers to the initial steps human actors take when they encounter a world of a social state of affairs that was unknown to them before.” ( Yesilkagit, 2004, p. 536).
Variables such as time and the degree of ‘habituation’ by actors to new roles and relationships bring into play historical, social and political dimensions that economistic theories of delegation and principal–agent interaction exclude. For example, Gains (2003) takes her starting point in policy network and power dependency theory. She claims that this kind of approach is fruitful when analyzing the impact of agencification on policy outcomes:
The differential financial, legal, managerial and structural autonomies granted to agencies clearly led to differing relationships between agencies and the parent department and its minister. Agencification in each case led to the necessity for agency and department to exchange resources to achieve their goals. This process created power dependent networks operating under formal and informal rules of the game ( Gains, 2003, p. 75).
This research shows that even though “[a]gencies were deliberately set up to implement and not make policy (…) these ‘bureaucratic’ networks did impact on policy outcomes” in various ways ( Gains, 2003, p. 74). For instance, the author demonstrates that agencies may have a strong influence on the definition of operational goals and may even determine the general policy agenda, depending on the type of resource exchange involved.
In this collection, the authors do not adopt a single theoretical approach or test a common set of propositions drawn from a single perspective. Historical institutionalism is perhaps the most common, but not the only, starting point for the separate authors. That aside, they have in common the view that policy autonomy is a feature of social and political relationships within a complex, on-going set of processes that constitute policy-making. Just as they adopt varying theoretical perspectives they also adopt different methodologies, in common with the approaches already adopted in the literature: single country or cross-country comparative analysis using survey data, in which many variables are controlled for one another ( Verhoest et al., 2010); qualitative comparative approaches ( Maggetti, 2009; Pollitt et al., 2004); and single-country case study analysis based on rigorous data collection ( Carpenter, 2001; Egeberg, 1995; Verschuere, 2009).
The article by Birgitta Niklasson and Jon Pierre takes a historical-institutionalist perspective on agencies’ policy autonomy in Sweden. The main theoretical claim is that the norms and values regarding agencies’ role in policy-making which dominated at the time of agency creation have a long-lasting effect on those agencies. Such norms and values can grow to permeate the entire organizational structure and it may take a substantial policy shift or radical administrative reform – often connected to a crisis situation or severe performance failure – to change them. The introduction of Program Budgeting in 1963 is depicted as such a paradigm shift, whereas administrative reforms following NPM ideas starting from 1990 are something like “variations on a theme” which have not altered the administrative system in a similar way as the earlier reforms. The authors test their hypothesis based on a comprehensive survey of Swedish agencies, complemented with other sources. Moreover, they control for variables related to the agencies’ external relationships, their tasks and their organizational structure. The empirical analysis confirms that indeed agencies founded before 1963 even today report higher levels of policy autonomy than agencies founded after 1963, whereas agencies created after the NPM reforms do not report higher levels of policy autonomy. Also, the analysis reveals that in addition to the variable measuring the period of creation, the level of public attention has the strongest effect on agencies’ policy autonomy. This corresponds to the survey findings of Bach (2012), but contradicts the case study findings of Handke (2012) who finds that high salience reduces policy autonomy because the ministry takes over activities previously performed by the agency.
The contribution of Tobias Bach covers the involvement of agencies in policy formulation in Germany, analyzing a comprehensive survey among federal agencies. The politico-administrative context is characterized by strong norms of separating policy and operations among ministries and agencies. Against the background of – mostly anecdotal – evidence of agency involvement in policy formulation and a supposed decline in the importance of ministerial bureaucracies in policy formulation, the article aims at explaining variation in policy autonomy by referring to agency features related to task, structure and culture. The empirical findings confirm a high degree of variation in policy autonomy among federal agencies. Moreover, it shows that task characteristics seem to be most relevant for explaining variation in policy autonomy, although the research also finds significant effects of structural and cultural characteristics. Among other things, the political salience of an agency is identified as an important task characteristic for explaining policy autonomy, a finding which is also made by Niklasson and Pierre (2012). Also, contrary to many recent writings about the supposedly increasing power of regulatory agencies over policy decisions, the analysis does not support those claims.
The article by Martin Painter and Wai-Hang Yee addresses the empirical puzzle raised by an earlier survey that senior executives of Hong Kong agencies report high levels of policy autonomy, regardless of organizational form. To this end, the authors conducted case studies of several Hong Kong agencies and their relationships with their political principals. The case studies exemplify different types of agencies’ policy roles and rationales of politicians to delegate policy autonomy. The first type refers to a segmentation of policy fields, which may either take place through deliberate delegation of policy autonomy for specific parts of a policy field (while still keeping an eye on the agency’s activities) or through permanent delegation on part of the government in less salient policy fields. The second type of dividing policy roles among political executives and agency executives consists of a segmentation of the policy process, which may for instance take place through agencies having a strong role in agenda-setting or high levels of discretion in policy implementation. In theoretical terms, the case studies illustrate the poor performance of well-known accounts of separating policy and operations and conflicting interests among political principals and administrative agents in explaining the policy roles of agency executives in a system of government characterized by a tightly integrated political and administrative elite. In such a context, decision-making may better be explained from a cultural perspective which explains decision-making as matching of behaviour with role expectations (a “logic of appropriateness”). The segmentation of policy fields and processes can be understood as a mechanism to avoid competition over control of policy, which would inevitably create conflict within a political and administrative elite that highly values consensus and cooperation.
In his article Stefan Handke studies the roles and inter-organizational relations of the German Ministry of Finance and the Financial Supervisory Authority in financial market policy. Similar to the contributions of Verschuere/Vancoppenolle and Painter/Yee this is a qualitative study mainly based on expert interviews with senior officials and politicians. Handke argues that the relationship between the ministry and the agency is characterized by a substantial imbalance in terms of expertise and manpower to the detriment of the ministry’s capacity in policy formulation and control of the agencies’ operational activities. This imbalance became obvious during the financial crisis in 2008. A key finding is that the ministry heavily depends on the regulatory agency in daily business (including the development of regulatory standards and the like) and lacks qualified personnel to effectively control the agency. The control problems are particularly obvious in the context of EU or transnational agency networks in which the agency is heavily involved. Also, the article shows that the agency is better characterized as a “trustee” acting on behalf of consumers and market participants rather than an “agent” acting on behalf of the ministry. Although the agency has been formally delegated wide-ranging operational as well as standard-setting tasks by law, yet the ministry may hierarchically instruct the agency on any matter. Thus in formal terms the agency is not legally independent, yet in practice it has a lot of autonomy both in terms of implementation as well as policy formulation. Also, the article addresses the agency’s preferences, i.e. what do agencies actually want? Three orientations are described, namely organizational survival and growth (“budget maximization”), bureaucratic discretion, efficient policies and organizational reputation (“policy seeking”), and a convenient job (“bureau shaping”). Handke shows that the policy autonomy of the agency may also be explained by deliberate delegation of policy functions to the agency by the ministry, which was not considered problematic under conditions of low political salience (see also the article by Painter and Yee on how political bureaucracies deliberately vacate policy space which is filled by agencies).
The article by Bram Verschuere and Diederik Vancoppenolle investigates the roles of ministers and their political advisors, ministerial departments and semi-autonomous agencies in two policy processes in Flanders (Belgium). The case of Flanders allows for a straightforward empirical test of how the separation of policy and operations works in practice, which was at the core of a comprehensive administrative reform heavily inspired by NPM thinking. The authors study policy decisions in the fields of social policy (care for disabled people) and transport policy (public transport), in which the agencies responsible for operations were involved in various stages. Several key findings emerge from their study: First, they show that a dyadic model of department–agency relations fails to account for the high influence of ministers and their “cabinets” of political advisers. In the policy processes they study, policy related interactions take place between the cabinets and senior agency officials, literally by-passing the ministerial departments which according to the reform objectives should be the primary policy advisers of the ministers. This suggests that management reforms might be “blind” to the reality of political decision making, which in the case of Flanders (and presumably other countries with cabinets of political appointees) relies heavily on ministerial cabinets. Second, their work indicates a rather clear division of tasks between the political leadership (ministers and cabinets) and agencies: whereas ministers and cabinets decide on strategic lines and supervise the further steps in the decision-making process, the influence of the agencies is higher in the stages where strategic decisions need to be made operational. Thus, there seems to be some division of labor along the policy-operations dimensions, yet agencies are granted substantial autonomy in putting general policy lines into practice. Finally, they emphasize that executive agencies are not homogeneous actors when it comes to policy formulation, as representatives of stakeholder groups in the agency’s governing board may be involved at various stages of the decision process.
To summarize, besides variation in terms of context and methodology, the articles in this collection also vary in terms of their empirical and theoretical approaches. Several studies are rich in quite detailed description, looking into what role specific government agencies play in the policy-making process ( Handke, 2012; Painter & Yee, 2012; Verschuere & Vancoppenolle, 2012). These studies use different theoretically-grounded scenarios of role divisions among oversight authorities and agencies ( Verschuere & Vancoppenolle, 2012); political explanations for ‘voluntary vacancy’ by political leaders and segmentation of the policy process between different actors ( Painter & Yee, 2012); and formal delegation approaches such as principal–agent and trustee theory ( Handke, 2012), to structure the case analysis. Other studies in this collection aim at explaining the level of policy autonomy of government agencies by drawing on comprehensive survey material which allows for multivariate testing of hypotheses ( Bach, 2012; Niklasson & Pierre, 2012). Yet another group of studies improves our understanding of the effect of specific explanatory variables by analyzing causal mechanisms over time. This particularly applies to the study by Niklasson and Pierre, who provide an in-depth account of how “critical junctures” in history affect policy autonomy.
This introductory article had several objectives: first, we reviewed the empirical literature on the role of agencies in policy-making and set out the research agenda of the themed issue. Then we proceeded with some conceptual clarifications of “public agencies” and “policy autonomy”, using the policy cycle framework for studying agencies’ policy-making roles in the different stages of the policy process. Such an approach allows for a systematic empirical scrutiny and comparison of agencies’ bureaucratic autonomy while acknowledging the complexity and contingency of the policy process.
There are many different ways through which public agencies potentially contribute to “who gets what, when, and how”. This variety is reflected in the contributions included in this themed issue, which address agency contributions to policy in different stages of the policy process. In terms of theoretical approaches, we argue that rational choice institutionalism and principal–agent accounts of delegation offer only limited insights into de facto bureaucratic autonomy. Partly, this can be related to a bias in this literature toward legislative control over bureaucracy, which tends to reduce the bureaucracy to an instrument in achieving political ends ( Moe, 1997). For instance, as indicated by the Hong Kong paper, agencies’ policy autonomy is not necessarily the result of a power struggle between politicians and the agencies. Instead, the level of policy autonomy appears to be a consequence of political unwillingness (disinterest). We briefly outlined more promising theoretical angles which are also used in the contributions in this themed issue. These include approaches rooted in historical institutionalism, resource dependency theory, and contingency theory of organizational structure. In the remainder of the conclusion, we highlight some “loose ends” which will require further attention in the future.
The contributions in this themed issue cover the role of agencies in policy-making in different parts of the world. However, none of the contributions is explicitly comparative in character. Nevertheless, the articles contain (mostly implicit) assumptions about how politico-administrative contexts affect the phenomenon under scrutiny, which may guide future research. For instance, the article on Hong Kong ( Painter & Yee, 2012) suggests that similar role perceptions and career patterns of politicians and senior bureaucrats open up for potentially high degrees of agency policy autonomy. The other articles pay less attention to the question whether politicians and agency officials share similar values and career patterns, though. More generally, studies on politico–administrative relations focus on senior ministry bureaucrats, rather than agency executives ( Aberbach & Rockman, 2006). Taking into consideration the growth in the number of public agencies, we believe that senior agency bureaucrats deserve more attention in this kind of research.
Another dimension of comparison is more explicitly related to the formal institutional context and the effect of administrative reforms. As the Swedish article ( Niklasson & Pierre, 2012) shows, administrative reforms may have a profound effect on the role of agencies in policy-making. The contribution on agencies in Flanders ( Verschuere & Vancoppenolle, 2012) illustrates very well how structural changes create confusion about appropriate roles in policy-making. The articles show that creating or reforming of agencies is not a mere technical or managerial issue, even though it may sometimes be labeled as such by decision-makers. A more general point is that reform history needs to be taken into account when studying and comparing agency roles in policy-making. For instance, the creation of agencies will almost inevitably lead to more complex relationships which more often than not will involve multiple principals and stakeholders ( Verschuere & Vancoppenolle, 2012). Thus, in order to understand the policy role of agencies, we possibly need to look at their interactions with other actors than just the parent ministry, e.g. NGOs, other agencies, and private businesses.
A relevant topic which has been raised in comparative research on ministry–agency relations (although explicitly excluding the issue of policy autonomy) has to do with the relative explanatory power of context versus task characteristics. There is reason to believe that agencies providing similar services (e.g. labor market policies) share many similarities, irrespective of their politico-administrative context. Such a perspective suggests that a structured, comparative analysis along the lines of Pollitt et al. (2004) could be a way forward to draw more robust inferences on the relative importance of context for agency policy autonomy. Such an approach could possibly also shed more light on the question whether regulating agencies play a more pivotal (or perhaps different) part in policy-making than other kinds of agencies.