‘And the rest is implementation.’ Comparing approaches to what happens in policy processes beyond Great Expectations

Abstract

Implementation occurs as a ‘late’ part in the stages model of the policy process. As such, it is seen as following upon and subordinate to the preceding stages of agenda-setting and policy formation. Hence, implementation is often addressed as ‘the rest’. This view on implementation as a presupposed residual in goal achievement implies little attention to ‘political’ dimensions, like ambiguity and conflict. Therefore, the view can only partially explain the – sometimes disappointing – results of policy processes. Alternatively, approaches to the policy/implementation nexus with an explicit focus on what happens at the street level have a greater explanatory potential. They are not taking implementation for granted as a seemingly technical matter, simply prescribed by policy objectives.

Keywords

Implementation policy implementation public policy public policy process policy formation policy formulation

Introduction

‘I got a lot of PhD-types and smart people around me who come into the Oval Office and say, “Mr President, here’s what’s on my mind.” And I listen carefully to their advice. But having gathered the advice, you know, I say, “This is what we’re going to do.” And it’s, “Yes, sir, Mr President.” And then we get after it, implement policy’.¹

Here made within the White House, this statement expresses a broadly accepted view on the relationship between policy formation and policy implementation. Once policy goals have been specified and decided upon, implementation is seen as an a-political administrative activity. Since it is normatively attractive, this ‘top-down’ view is still widespread in practice but also in policy implementation research (Hupe, 2014; Johansson, 2010). At the same time, its explanatory capacity is limited. In fact, the view as such has a specific character. First, it implies a chronological order in which expressed intentions precede action. Second, a linear causal logic is assumed: goals determine instruments, and instruments determine results. Third, a hierarchy is inherent: policy formation is more important than policy implementation. Implementation follows policy, in every sense of the word.

Pressman and Wildavsky (1973) expressed this view on implementation, more or less explicitly, in the first edition of Implementation. The lengthy subtitle of their book (see list of references) shows disappointment about the inadequate results in Oakland of the pursuit of ‘Great Expectations’ formulated in Washington. Such disappointment would follow every time anew when policy implementation is being looked at from a straightforwardly top-down view. In the literature, the normatively appealing and widely recognized but actually specific nature of this view has been unveiled as the thesis of incongruent implementation (Hupe, 2011).

The objective in this article is to explore the meaning of implementation. The focus is on the variety of ways the treatment of implementation can take in studies of policy processes. More specifically, the aim is to identify and categorize research approaches to the relationship between policy formation and policy implementation. What is the variation in the way the relationship is perceived between, on one hand, policy as formulated-and-decided-upon and, on the other, implementation as the ultimate realization of the policy goals? Our assumption is that different specific views on implementation (policy-in-practice) correlate with different ways policy-as-to-be-implemented is being seen (policy-on-paper). The way in which the policy/implementation relationship is perceived has consequences for the way empirical variation in policy objectives is explained. The analysis of the variety of these approaches enables an assessment of their capability to deal with the multi-dimensional character of policy processes. In this context, we recognise the phenomenon of symbolic policy processes (Edelman, 1971). We do not explore it, however, inasmuch it is related to situations where there is no serious concern to implement.

The article has five sections. After the introduction, the relationship between policy and implementation is explored theoretically in the ‘Public policy and implementation’ section. It arrives at a distinction between relevant dimensions on which the various types of research approaches can be investigated. In the ‘Approaching the policy/implementation nexus’ section, the approaches are identified, before they are compared on the identified dimensions in the ‘Comparing approaches of policy-as-to-be-implemented’ section. The latter enables a discussion of the explanatory potential of each type of approaches for studying what happens after policy goals have been formulated and decided upon. The article ends with some conclusions.

Public policy and implementation

Beyond implementation as a stage

Wilson (1887) introduced an influential distinction between politics and administration. Since then, this distinction normatively functions as a justification for the view that the two embody entirely different realms, implying completely different activities, performed by separate actors – or at least, that they should. Wilson’s (1887) distinction reappears in the stages model of the policy process. On the other hand, in their debate with those looking from a ‘top-down’ view, ‘bottom-up’ advocates argued that the distinction between policy formation (politics) and policy implementation (administration) is misleading (a.o. Barrett and Fudge (1981); for an overview and review of the current state of the field, see Hill and Hupe (2014)).

These quite different stances – implementation as a separate, identifiable, stage versus policy-making as an ongoing process – concern an issue, which is deemed important. In his discussion of the respective merits of top-down and bottom-up approaches to implementation, Sabatier (1986) addresses this issue, while making a case for the distinction. The main thrust of his argument is based on what he sees to be empirical evidence that in most cases the distinction can be easily made. In addition, Sabatier (1986) offers two more cogent arguments:

First, it (disregard of the distinction – the authors) makes it very difficult to distinguish the relative influence of elected officials and civil servants – thus precluding an analysis of democratic accountability and bureaucratic discretion, hardly trivial topics. Second, the view of the policy process as a seamless web of flows without decision points (…) precludes policy evaluation (…) and the analysis of policy change (Sabatier, 1986: 31)

Sabatier’s (1986) second argument specifically sets out the concern of this article to raise questions about the identification of decision points. If one considers the latter as useless, this implies a sweeping dismissal of a great deal of policy analysis that takes as its central concern questions about what happens after decisions that aim to structure action have been made.

The contemporary study of government reflects both the opposite stances mentioned above and more pragmatic, or ‘realistic’, views, expressing a concern about how to identify decision points in policy processes. In the preface to the first edition of their Implementation, Pressman and Wildavsky (1973) stated: ‘A verb like “implement” must have an object like policy’. Making an assumption that policy is readily and simply identifiable, they thus ran the ‘risk of catching themselves in a linguistic trap of their own making’ (Hill, 2013: 209). In that sense, they identified themselves with the Wilsonian tradition of making a sharp distinction between politics and administration.

Pressman and Wildavsky (1973) were then criticised for a very top-down perspective and of course later work involved asserting a distinctive ‘policy/action relationship’ (Barrett and Fudge, 1981) or seeing the identification of ‘stages’ in the policy process as problematical (Hill and Hupe, 2014; John, 1998). Other influential authors too have been unwilling to draw a sharp line between an initial policy formation part and a subsequent implementation part of policy processes. Lindblom, for instance, is particularly identified with a view of the policy process as essentially incremental. His book with Woodhouse, The Policy-Making Process, only contains a small discussion of implementation (Lindblom and Woodhouse, 1993). Their starting point is to say that it is not ‘accurate to suggest that there is a certain step at which policy must be “decided” ’ (Lindblom and Woodhouse, 1993: 11). They go on to argue:

An attempt to implement one policy almost always brings new problems on the agenda, meaning that the step called implementation and the step called agenda building collapse into each other (Lindblom and Woodhouse, 1993).

Simon (1957), whose more rationalistic approach is often contrasted with Lindblom’s incrementalism, took a similar view on ‘continual decision making’, arguing:

(…) it has not been commonly recognized that a theory of administration should be concerned with the processes of decision as well as with the processes of action. This neglect perhaps stems from the notion that decision making is confined to the formulation of overall policy. On the contrary, the process of decision does not come to an end when the general purpose of an organization has been determined. The task of ‘deciding’ pervades the entire administrative organization quite as much as the task of ‘doing’ – indeed, it is integrally tied up with the latter (Simon, 1957: 1).

Simon (1957) attacks the view of decision making as reserved for politics and thus external to administration. Albeit organized in a normative hierarchy – minor decisions being consequent upon major ones – the issue who takes which decisions in a policy process is an empirical question. The part of a particular policy process leading to stated policy objectives is characterized by interaction between policy design and decision making. The former concerns the ‘rational’ formulation of means/ends relationships in policy texts, while the latter refers to bureaucratic politics and other dimensions of action. The result of the interaction between the two, in a relation between Reason and Power, can be called policy formation. It often produces ambiguous results. For those expected to implement the policy goals involved, the latter do not function as a technical instruction. Various values and stakes may be involved. The public policy as stated is both result and object of sometimes conflicting world views and belief systems, while the process functions as an arena for competing interests (Hupe et al., 2014).

Dimensions of comparison

From the previous subsection, the conclusion can be drawn that normatively there is a sustained relevance to the view of implementation as a separate stage, while scholars reflecting on the empirical reality of decision making have an eye for the fact that ‘decisions’ can be seen throughout a policy process – even at the street level. The relationship between policy goals as expressed and policy as implemented is not an obvious one. This relationship cannot be taken for granted, neither in practice nor in studying it.

This situation makes it worthwhile to explore how ‘implementation’ and ‘policy’, both separately and in their relationship, are being viewed when these terms are being used in studies of policy processes. As indicated above, our assumption is that the view of implementation correlates with the way policy-as-to-be-implemented is being seen. The subsequent approach to the policy/implementation relationship has consequences for the potential to explain empirical variation in what happens with policy objectives.

In this context, it is clear that the number of studies of policy processes in which the relationship between policy and implementation, one way or another, is addressed, is endless. The stream grows daily, making bibliometric studies like the ones of Sætren (2005, 2014) relevant snapshots, while state-of-the-art overviews of the field need to be updated regularly (cf. the 3rd edition of Hill and Hupe, 2014). Instead, for this article, a selection has been made of publications with a character that may be deemed to present a particular type of approach. Without any claim on quantitative scope, we have aimed to choose ‘exemplary’ publications, often referred to in earlier work of the present authors.

In the next section, we sequentially identify studies in which policy is approached as prescriptive input, as broadly desired outcome, as circumscribing permitted divergence, as institutional mandate and as actual output. In each type of approach, implementation is treated differently, respectively as applying instructions, as realizing an ideal, as adopting the legislator’s intentions, as using institutional preconditions, and as something to be measured comparatively. The goal in this article is to give an assessment of the potential of each type of research approach for understanding and explaining what happens after policy goals have been formulated and decided upon. In order to compare this potential, relevant parameters need to be outlined.

Two dimensions for such a comparison can be highlighted here. First, the degree to which the policy/implementation nexus in the publication at hand is problematized at all may make a difference. To what extent and how has the relationship between implementation and policy been specified? Second, it matters if and how the intrinsic multi-dimensional character of policy processes gets attention in the studies involved. Public policy processes have an intrinsically political character, in the sense that policy makers are trying to deal with real problems and do so all the way. Dimensions of ‘implementation politics’ have been recognised by various authors. A well-known article is the one in which Matland (1995) stresses ambiguity and conflict as identifiable sources of varied implementation. A more recent discussion is embodied in May and Jochim’s (2013) argument for the use of a regime perspective in implementation analysis. Such a perspective is concerned with ‘how policymakers and others advance the ideas that are central to a given policy approach, how institutional arrangements reinforce policy cohesion, and whether the approach engenders support or opposition among concerned interests’ (May and Jochim, 2013: 427). These aspects can be used as indicators for attention to the multi-faceted character of policy processes: To what extent ‘political’ dimensions, like ambiguity and conflict, are acknowledged as inherent characteristics of policy processes?

Approaching the policy/implementation nexus

Policy as clearly defined prescriptive input/Implementation as applying instructions

Explaining ‘the transformation of collective decision in the policy performance(s) of implementation agencies’, Torenvlied and Akkerman (2004: 31) call the central concern of the ‘state of the art’ or ‘standard model’ of implementation (Torenvlied and Akkerman, 2004: 33). This ‘standard model’ implies a conventional approach in line with Sabatier’s empirical argument that ‘in most cases’ it is possible to make the distinction between policy formation and policy implementation.

It is pertinent to note how exponents of stages theory identify activities that lie between policy initiation and implementation. In, for instance, Jenkins’ (1978: 17) detailed list, ‘initiation’ is followed by ‘information’, ‘consideration’ and ‘decision’. Hogwood and Gunn (1984: 4) follow ‘deciding to decide’ with ‘deciding how to decide’, ‘issue definition’, ‘forecasting’, ‘setting objectives and priorities’ and ‘options analysis’. Only after these items, do these authors mention implementation.

Similar arguments about the need to see policy-making as involving a sequence of subordinate decisions are found in the work of Knoepfel and Weidner (1982) writing of ‘policy programming’ (see also Knoepfel et al., 2007: chapter 8). They are using a model that sees the details of a policy as forming a series of layers around a policy core. These will include:

More precise definitions of policy objectives.

Operational elements, which include the ‘instruments’ to be used to make the policy effective.

‘Political-administrative arrangements’ that involve the specification of the authorities whose duty will be to implement the policy, the notion that such authorities need money and other resources to do that follows self-evidently from that point.

Procedural elements, namely the rules to be used in the implementation of the policy.

Specifications of what the examination of implementation should involve have seldom been unaware of these problems. For example, Van Meter and Van Horn identify in a 1975 article ‘standards and objectives’, ‘resources’ and ‘inter-organizational and enforcement activities’ in their model of the implementation process (Figure 3.1: 463).

The problem lies in the difficulty of integrating these considerations into the empirical study of implementation. The strength of the conventional approach to the study of implementation is that it accepts the notion of a cut-off point at which formation is seen to be completed. It offers a model that looks just right for large-n studies in which the activities of different agencies implementing the same policy can be contrasted. Cut-off points therefore need to be carefully specified in terms of both space and time.

Interestingly, Sabatier’s argument in his 1986 article about the frequency of straightforward ways of specifying the policy formation/policy implementation distinction seems to be contradicted by much of his later work on advocacy coalitions. In that work, his concern to use a long time frame leads him to acknowledge ways in which there is interaction between ‘dominant’ and ‘minority’ coalitions, taking us a long-way from straightforward implementation processes (Sabatier and Weible, 2007).

All in all, the relationship between policy formation and policy implementation in this type of approaches is assumed rather than problematized. Political dimensions like ambiguity and conflict are not addressed as playing a role within policy processes.

Policy as broadly desired outcome/Implementation as realizing an ideal

A policy may have been formulated in a white paper or other public document, as is the case in the previous type of approach. As such, what needs to be implemented – the implemendum – may be the result of negotiation and struggle, turning the policy-on-paper into a compromise, which needs interpretation in order to be implemented.

There are studies in which the line between policy intentions and their follow up is situated further ‘upstream’ in the policy history. An example of these can be found in Marsh and Rhodes’ edited book Implementing Thatcherite Policies (1992). It is clear that this is a book of case studies in which authors differed in exactly how they interpreted their brief. At the same time, a discussion of implementation theory in the editors’ introduction indicates that the volume aims to take the first word in the book title seriously. The book is identified as about policy change and about ‘to what extent policy change resulted from a distinct policy agenda and legislative programme’ (Marsh and Rhodes, 1992: 4). In effect, the success and failure of the ‘Thatcherite’ policy change programme is explored right from processes occurring as the specifics of the agenda were worked through in Parliament, the Cabinet and even the Conservative Party through to policy outputs.

Implementation is a word used in both daily speech and academic discourse. In our view, academics should generally be reluctant to appropriate words from everyday use in order to apply them in very specific ways. They should accept variation in modes of usage, but should point out their implications. Therefore, we accept the validity of Marsh and Rhodes’ use of the concept of implementation. However, it must be recognised that this usage embraces a very large part of the policy process as a whole.

We admit that there are situations in which it is quite impossible to draw the policy formation/policy implementation distinction. Examples of this occur in the field of macro-economic policy. Given their substantial ramifications as soon as they are identified by economic actors, there policy decisions are more or less self-implementing. It is interesting to note here some of the difficulties that have been faced in practice by policies that try to establish a separate implementation mechanism (delegating powers to a central bank, for example).

In selecting Marsh and Rhodes’ book (1992) for discussion, we have taken the argument to an extreme example of including within an implementation study activities that are very far ‘upstream’ in the policy process. A less extreme example is in fact provided by Pressman and Wildavsky’s (1973) seminal work. The analysis of the interactions involving organizations between Washington and Oakland makes much of the implications of a lengthy decision chain (even using a mathematical model to explain how slight changes at any one stage multiply).

What has been given too little attention is that this seminal approach to implementation was explicitly developed in the context of American federalism. Paradoxically, it diverts attention from variation that might be considered as legitimate policy changes in a federal system of inter-governmental relations (see Ferman, 1990). Later, authors on implementation in the context of federalism recognize this issue, but do so nevertheless still with a need to refine analysis of the implementation process. Thus, Goggin et al. (1990) write of ‘federal messages’ (Chapter 3) and identify the need to formulate hypotheses about these. Stoker (1991) uses the term ‘governance’ to delineate situations in which reluctant partners to an implementation process are induced to collaborate. A recent study of implementation in the United States, May's (2014) exploration of the Affordable Care Act, includes consideration of the processes necessary to put the Act into operation. These processes include negotiated arrangements with States (mandated by the original legislation) that might be described as the continuation of policy formation.

The federalism theme finds echoes in discussions of implementation in the European Union. The term implementation is often used about the responses of member states to EU policies. In this case, an exploration is needed of the extent to which such policies are mandatory and about the scope for derogations. There is a potential for confusion about embracing situations in which there is a clear quasi-hierarchical expectation of obedience with those where policy evolution is expected. It is perhaps more appropriate to speak of weak forms of policy transfer where there are forms of ‘mutual adjustment’ or the copying of a model developed by others in a context of inter-governmental relations (cf. Holzinger and Knill, 2005; Knill and Lenschow, 1998).

Attention to the links between layers of government is relevant, because it enables researchers to identify where ‘implementation’ and where legitimate ‘policy co-formation’ are situated (see Hill and Hupe, 2003). In the type of approach analysed in this section, however, the relationship between policy formation and policy implementation is assumed rather than problematized. Similar to the first type, political dimensions like ambiguity and conflict are not addressed as playing a role within policy processes. In both types of approaches, explanation of perceived ‘implementation gaps’ or ‘policy failures’ is sought in what has gone wrong underway from well-articulated policy intentions (in Washington) to their flawed implementation (in Oakland).

Policy as circumscribing permitted divergence/implementation as adopting the legislator’s intentions

Torenvlied and Akkerman (2004) describe ‘multilevel policy-making systems as empirically characterized by the existence of relatively autonomous layers of decision making’. Such systems often produce ‘non-binding “soft policies”, such as recommendations, information campaigns, and action plans rather than collectively binding decisions’ (2004: 32). Their concern is then to adapt implementation studies that follow what they call ‘the state of the art model’, mentioned above. This adaption consists of a reformulation of:

(…) the dependent variable of the implementation model…in terms of cross-level policy coherence [and of] the causal mechanisms behind implementation in order to incorporate the relative autonomy of multiple layers of collective decision-making (Torenvlied and Akkerman, 2004).

Torenvlied and Akkerman’s (2004) article goes however no further than to explore ways in which ‘cross-level policy coherence’ may be achieved and argues that this is possible with the use of ‘soft’ policy. This is in many respects simply a variation on May’s ‘mandate’ analysis; see the next subsection.

What needs more attention is the notion of ‘soft policy’. It is a concept that can cover many possibilities. Torenvlied and Akkerman (2004) do no more than define it in the terms quoted above and significantly describe these elsewhere as ‘instruments’. It is used by Kennedy et al. (2011) in a discussion of education curriculum reform, where it is seen as applying most obviously to the extent of discretion granted to implementers. An alternative formulation making much the same point is found in a study of the implementation of change in the delivery of health policy where actors are seen as engaging in a ‘sense making’ activity influenced by past events, and policy is described as ‘loosely specified’(Coleman et al., 2010).

Oosterwaal and Torenvlied (2012) examine the implications of political conflict at the formation stage for what they call ‘policy divergence’ at the implementation stage. They treat ‘policy divergence’ as a dependent variable ‘measured by computing the absolute difference between the value of the policy decision on the policy scale with the value assigned to the actions and behaviours of the relevant implementation agency on the same scale’ (Oosterwaal and Torenvlied, 2012: 207). They then explore the impact of policy conflict both in the formation process and in the implementing agencies.

In the work described here, two issues, soft policy and policy divergence, are explored. These may be seen as separate issues, sometimes connected. In other words, soft policy is sometimes a product of policy divergence, but there may be other reasons why policy is ‘soft’ such as deliberate commitment to delegated discretion. Conversely apparently quite ‘hard’ policy may emerge from a policy formation marked by conflict.

Torenvlied and his collaborators (2012) show explicit attention to policy in their explorations of the implementation part of a policy process. They do so, by trying to identify a clear intention as laying at the basis of the public policy involved, enabling them to measure ‘divergence’ from that intention. Hence in this work, a commitment to what we have called the prescriptive input approach (A) can be observed. In fact, in the circumscription of the dependent variable, an ‘implementation gap’ is presupposed.

At the same time, with their characterization of policies as ‘soft’ or otherwise and their eye for political conflict, Torenvlied and his collaborators (2012) also show attention to dimensions of policy processes other than design ones. In fact, they explore empirically what authors like Matland (1995) observe at a general level, stressing the implications of policy dissensus for implementation.

Policy as institutional mandate/implementation as using institutional preconditions

One of the concerns of those who wrote in prescriptive terms about implementation from a top-down perspective is giving advice on how to design policies to ensure that they get implemented the way actors higher-up in the hierarchy of vertical administration want them to be implemented (see, e.g. Gunn (1978)). Mandating studies go back to these issues. The authors who have concerned themselves with the relationships embodied in federalism have contributed to taking this out of an explicitly hierarchical context, seeing compliance as something that cannot simply be ordered. May (1993) provides a particularly useful development of that approach in writing of mandates that ‘(…) do more than simply announce a set of goals or standards. They facilitate implementation through various inducements, system changes, and capacity building features’ (May, 1993: 636). This work can be seen as a long way from top-down approaches that identify sources of non-compliance. May (1993) suggests that implementation failures ‘stem in part from mandates that neglect to specify desired actions or to include features aimed at facilitating implementation’.

May’s (1993) work then takes this a step further to suggest that mandates may imply flexible principal/agent relationships. There may be not so much a single set of demands from a superordinate organisation to all subordinate ones but of forms of ‘negotiated compliance’ (1993) in which varied relationships are negotiated. In later work, May (1995), drawing upon his researches on regulatory environment policies in Australia and New Zealand, makes a distinction between ‘coercive’ and ‘cooperative’ mandating styles. It is the latter which are of interest here. May says:

Cooperative intergovernmental policy designs are aimed at enhancing local governments’ interest in and ability to achieve policy goals. Local governments act as regulatory trustees in seeking appropriate means for reaching performance standards. Cooperative policies prescribe planning or process elements to be followed (a form of policy mandate), but they do not prescribe the particular means by which local governments achieve desired regulatory outcomes. Cooperative mandates stress capacity building through financial and technical assistance. By offering various inducements for compliance, they also build commitment to policy goals. Sanctions may be applied if local governments do not undertake the mandated planning process, but wide latitude is allowed in local policy development (1995: 91).

Clearly, this work takes as its focus the nexus at the policy formation/policy implementation boundary. Less explicit attention seems to be given to ambiguity, conflict and comparable, intrinsically ‘political’ dimensions of policy processes.

Policy as actual output/implementation as to be comparatively measured

The plea for studying implementation from the policy delivery end is associated with the bottom-up perspective in the study of implementation (Elmore, 1979). At the same time, a bottom-up approach can be justified combined with the acceptance of the fact that public policy is sourced from the top. Elmore (1979) writes of the ‘noble lie (…) that policy-makers (= actors in policy formation – the authors) control the organizational, political and technological processes that affect implementation’ (Elmore, 1979: 603). Nevertheless, ‘backward mapping’ as he describes it ‘takes the policy-makers perspective on the implementation process’ (Elmore, 1979: 604).

In the context of this article, backward mapping represents a particular approach because it focuses on policy outputs and asks questions about what determines it. It may in this respect be seen as more sympathetic to the perspective of those at the bottom than, for example, an approach that stresses implementation deficits. Elmore, like Lipsky (1980), sees the world of the street-level bureaucrat as one in which action is determined by multiple pressures. The question likely to be asked in a backward mapping study is about the extent to which a new policy makes a difference to everyday life at the bottom.

Apart from this normative, or ‘political’, aspect, there is an analytical argument, well-articulated by Winter (2006). Whilst Elmore’s (1979: 607) article refers to the need to focus on output, he does not go as far as explicitly endorsing the case for starting an implementation study at the policy output end. In his review of implementation research, however, Winter (2006: 159–161) draws out the research consequences of Elmore’s stance, by stressing the need to explain the empirical variation in the outputs of a particular policy. Instead of examining the relationship between policy goals and ‘goal achievement’, policy input may be seen as an independent variable with a varying impact.

Hence two very different views on ‘policy’ can be recognized:

Policy as defined by an input, as involved in studies that focus on policy gaps or policy deficits.

Policy as defined by output, without an immediate and singular reference back to an input. Questions here are about why outputs differ from time to time or place to place.

In the latter type of approaches, there is an eye for the policy/implementation relationship, but turned ‘upside down’. Factors that have never been foreseen or given attention during policy formation, or exert influence from outside policy design activities – cf. ambiguity and conflict – are acknowledged as having an impact in the actual policy-making taking place while the policy is being implemented at the street level.

Comparing approaches of policy-as-to-be-implemented

In the previous section, we distinguished five ways of addressing the relationship between policy formation and policy implementation. For each, we answered two questions. First, to what extent and how has the relationship between implementation and policy been specified? And second, to what extent ‘political’ dimensions, like ambiguity and conflict, are acknowledged as inherent characteristics of policy processes? The result was an identification of the following types of approaches of the policy/implementation nexus, which now can be given characterizing labels.

A technical approach. Policy as clearly defined prescriptive input/implementation as applying instructions.

A normative approach. Policy as broadly desired outcome/implementation as realizing an ideal.

A control approach. Policy as circumscribing permitted divergence/implementation as adopting the legislator’s intentions.

An institutional approach. Policy as institutional mandate/implementation as using institutional preconditions.

A comparative approach. Policy as actual output/implementation as to be comparatively measured.

It now becomes possible in a comparative formulation to assess the potential of each type of approaches for understanding and explaining what happens in policy process once policy goals have been formulated and decided upon. We use the two questions mentioned above as dimensions of comparison by looking at the degree of specification of the relationship between policy formation and policy implementation, and at the level of attention to ‘political’ dimensions, like ambiguity and conflict, as inherent to policy processes. The comparison results are presented in Table 1.

Table 1.

Comparing types of research approaches to the policy/implementation nexus

Types of approaches Indicators	Technical	Normative	Control	Institutional	Comparative
Specification of the relationship between policy formation and policy implementation	Low	Low	High	High	High
Attention to ‘implementation politics’	Little	High	Moderate	Moderate	High

While accepting that policy-making goes on towards and at the street level, for both research and normative reasons, we deem it worthwhile trying to identify a separate implementation part of a policy process for which policy is an input. The first variant we called the technical type of approaches to the policy/implementation nexus. Policy is seen as a clearly defined input prescribing the further actions to be taken in the policy process at hand. Implementation then is a matter of applying the instructions implied by the policy-on-paper. In the normative type of approaches, policy is seen as broadly desired outcome, expressing an ideal to be realized via implementation. In research approaches to the relationship between policy/implementation labelled as the control type, the policy-on paper is looked at as circumscribing the degree of divergence permitted in implementation. Implementers are supposed to adopt the legislator’s intentions. In the institutional type of approaches, the nature of the institutional mandate is central, as enabling implementing actors to use the given institutional preconditions in function of the policy involved. Finally, in the comparative approach, the focus is ‘on the other end’: the actual outputs of a policy process, as to be measured comparatively from a ‘horizontal’ perspective.

Each of the types of approaches identified has merits. A great deal depends upon the research objective, the central question and the specific aspects the researcher wishes to explore. In particular, we recognize that in the study of policy processes, there is an inevitable division of labour.

The first two of the five types of approaches characterized above, the technical and normative types, have a twin relationship. Their different logics (narrow versus broad focus) justify a distinction. At the same time, both types of approaches share a lack of concern about the difficulties in specifying the policy/implementation relationship. They also have a neglect of the multiple dimensions of policy processes in common. Implementation is seen as a technical matter, merely a matter of ‘administration’, distant from and subordinate to politics.

The other three types of approaches have characteristics, which, more or less, enhance their potential explanatory value. The essential implementation question is: What determines the output of a policy process? Then what is important here – taking the cue from backward mapping – is going back into factors in addition to the behaviour of the implementing actors. Of the three types labelled as, respectively, control, institutional and comparative, approaches of the comparative type offer the most pertinent way to identify relevant causal factors and mechanisms. They do so, treating policy-on-paper as an important but perhaps not the most important factor determining what happens in the implementation of a particular policy.

Elmore’s (1979) contribution on backward mapping has long been seen as helpful for the debate about methodology. It seems worthwhile to revisit his view here. Elmore’s (1979) message hints at an approach to the specification of independent variables independent of policy intentions. Such an approach can eliminate the notion of gaps or deficits and therefore be used without clarity about intentions (or indeed in the absence of any very explicit intentions at all). A measure without any origin is problematical; when there is only one implementing agent, there is nothing against which to measure output. However, this is not a problem where there is more than one agent, which is a frequent situation in the practice of public policy implementation. Then, questions are about explaining differences between outputs by different agents.

The examination of output in ways which treat it as independent of input has two implications. First, there is an issue about the choice of output over outcome as a dependent variable. Winter (2006: 161–163) stresses the undesirability of working with the latter, inasmuch as it involves a need to deal with influences on behaviour that are independent of the implementation of other parts of a policy process. However, he recognises that there may be exceptions when it is necessary; for instance, with ‘soft’ policies, outcome goals are specified leaving implementers to determine the means to achieve those goals. In such cases, attention to outputs may be beside the point.

Second, the ways ambiguity and conflict are to be incorporated in a research design need attention. Both the approaches of the control type and the institutional type have relevant points to raise on this issue. Even if we cannot identify it very precisely, it may be argued that if policy input is uniform across two or more implementing agencies whose performance nevertheless varies, then surely the place to look for an explanation is within the agencies. Oosterwaal and Torenvlied (2012), however, draw attention to the possibility that dissensus within the latter may be a reflection of dissensus within the former. May’s work goes even further in indicating that how a superordinate organisation mandates subordinate ones may vary. Some examples help to make that point explicit:

It may be that a central authority negotiates rather differently with large local authorities than with small ones.

There may be cases where the formal independence of the lower tier authorities vary (as in countries where there are quasi-federal arrangements, like Spain).

A central authority may deal differently with local authorities dominated by compatible politicians than with those dominated by opposition parties.

Where multiple dimensions figure in the independent variables in implementation studies, there is a need to explore their characteristics.

Conclusion

Authors like Simon (1957) and Lindblom and Woodhouse (1993) underline the view that policy processes entail ongoing action and interaction, marked by a range of ‘decision points’. In the beginning of this article, we adopted this view. Then the identification of implementation beyond its characterization as a separate and subordinate ‘stage’ becomes relevant and, particularly, the way policy and implementation are addressed in their relationship.

In this article, we have explored how this relationship is looked at in various types of approaches and what the differences mean for understanding and explaining what happens after policy objectives have been formulated and decided upon – the latter in their combination addressed as policy formation. By looking into the meaning of implementation, we have demonstrated how behind the same term, very different research approaches to the relationship between policy formation and policy implementation are hidden. The policy/implementation relationship appears to be specified to a varying degree, while also various levels of attention to ‘implementation politics’ can be observed. In what has been characterized as the technical and normative types of approaches implementation is assumed rather than researched. In the mainstream studies following up on Pressman and Wildavsky’s classic, ‘congruent implementation’ is being presupposed, while actually an evaluation of goal achievement is being made.

Treating implementation as a presupposed residual in goal achievement, as occurs in the technical and control type of approaches of the policy/implementation nexus, has consequences for the explanatory potential of those approaches. First, the relevance of discretionary actors, their capacities and often legitimately exercised influence are ignored. Because the black box remains unopened, a lack of understanding of what happens ‘underway’, not in the least at the street level, may induce disappointment about good intentions laid down in policy statutes. Second, while the political dimensions of policy-making are neglected, chances may be missed to advise policy makers to mobilize political support towards the realization of the very objectives of the policy involved.

Third, the fact that behind the same term varying approaches to the policy/implementation nexus are hidden may function as a potential source of confusion, contributing to a literature in which scholars talk past each other. As a consequence and despite efforts by scholars to specify issues for research on implementation, this confusion is feeding the eagerness of those at the top of political-administrative hierarchies to blame implementers when a policy is perceived as going wrong. Strengthening control then is the standard reaction.

The degree to which the view on implementation as a presupposed residual in goal achievement is still widespread in contemporary practice of public administration deserves empirical research. It may be that not all top-level policy makers have the same perception of implementation as President Bush quoted in the beginning. However, in this context, it is worthwhile to observe recent attempts aimed at bringing about an ‘implementation science’ (Nilsen et al., 2013). This development, apparent at a global scale, indicates a revival of the rational ideal of technocracy. As such, it seems an expression of the enduring quest for control inherent to the modern era (Van Gunsteren, 1976). The argument here accepts the importance of that concern. At the same time, it must not be at the expense of a realistic assessment of the relationship between implementation and its environment. It is possible that ‘implementation science’ and policy implementation research can learn from each other. In any case, for analysts of public policy processes, it seems wise to acknowledge that ongoing policy-making and therefore politics are at stake, when it is argued that ‘the rest is implementation’.

Footnotes

Acknowledgements

The authors thank the three anonymous reviewers for their constructive comments on an earlier version of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Note

References

Barrett

Fudge

(1981) Examining the policy-action relationship. In: Barrett

Fudge

(eds) Policy and Action: Essays on the Implementation of Public Policy, London: Methuen, pp. 3–34.

Coleman

Checkland

Harrison

Hiroeh

(2010) Local histories and local sensemaking: A case of policy implementation in the English National Health Service. Policy & Politics 38(2): 289–306.

Edelman M (1971) Politics as Symbolic Action. Chicago: Markham.

Elmore

(1979) Backward mapping: Implementation research and policy decisions. Political Science Quarterly 94(4): 601–616.

Ferman

(1990) When failure is success: Implementation and Madisonian government. In: Palumbo

Calista

(eds) Implementation and the Policy Process: Opening Up the Black Box, New York: Greenwood Press, pp. 39–50.

Goggin

Bowman

AO’M

Lester

(1990) Implementation Theory and Practice: Toward a Third Generation, Glenview, IL: Scott Foresman/Little, Brown and Company.

Gunn

(1978) Why is implementation so difficult? Management Services in Government 33(4): 169–176.

Hill

(2013) The Public Policy Process, (6th ed). Harlow: Pearson Education.

Hill

Hupe

(2014) Implementing Public Policy: An Introduction to the Study of Operational Governance, (3rd ed). London: Sage.

10.

Hill M and Hupe P (2003) The multi-layer problem in implementation research. Public Management Review 5(4): 471–490.

11.

Hogwood

Gunn

(1984) Policy Analysis for the Real World, Oxford: Oxford University Press.

12.

Holzinger

Knill

(2005) Causes and conditions of cross-national policy convergence. Journal of European Public Policy 12(5): 775–796.

13.

Hupe

(2011) The thesis of incongruent implementation: Revisiting Pressman and Wildavsky. Public Policy and Administration 26(1): 63–80.

14.

Hupe

(2014) What happens on the ground: Persistent issues in implementation research. Public Policy and Administration 29(2): 164–182.

15.

Hupe

Hill

Nangia

(2014) Studying implementation beyond deficit analysis: The top-down view reconsidered. Public Policy and Administration 29(2): 145–163.

16.

Jenkins

(1978) Policy Analysis: A Political and Organizational Perspective, London: Martin Robertson.

17.

John P (1998) Analysing Public Policy. London: Pinter.

18.

Johansson

(2010) Implementing evidence-based practices and programmes in the human services: Lessons from research in public administration. European Journal of Social Work 13(1): 109–125.

19.

Kennedy

Chan

JK-S

Fok

(2011) Holding policy-makers to account: Exploring “soft” and “hard” policy and the implications for curriculum reform’. London Review of Education 9(1): 41–54.

20.

Knill

Lenschow

(1998) Coping with Europe: The impact of British and German administrations on the implementation of EU environmental policy. Journal of European Public Policy 5(4): 595–614.

21.

Knoepfel

Weidner

(1982) Formulation and implementation of air quality control programmes: Patterns of interest consideration. Policy & Politics 10(1): 85–109.

22.

Knoepfel

Larrue

Varone

(2007) Public Policy Analysis, Bristol: Policy Press.

23.

Lindblom

Woodhouse

(1993) The Policy-Making Process, (3rd ed). Englewood Cliffs, NJ: Prentice Hall.

24.

Lipsky

(1980) Street-Level Bureaucracy: Dilemmas of the Individual in Public Services, New York: Russell Sage Foundation.

25.

Marsh

Rhodes

RAW

(1992) Implementing Thatcherite Policies: Audit of an Era, Buckingham: Open University Press.

26.

Matland

(1995) Synthesizing the implementation literature: The ambiguity-conflict model of policy implementation. Journal of Public Administration Research and Theory 5(2): 145–174.

27.

May

(1993) Mandate design and implementation: Enhancing implementation efforts and shaping regulatory styles. Journal of Policy Analysis and Management 12(4): 634–663.

28.

May

(1995) Can cooperation be mandated? Implementing inter-governmental environmental management in New South Wales and New Zealand. Publius 25(1): 89–113.

29.

May

(2014) Implementation failures revisited: Policy regime perspectives’. Public Policy and Administration. Epub ahead of print 12 September 2014. DOI: 10.1177/0952076714561505.

30.

May

Jochim

(2013) Policy regime perspectives: Policies, politics and governing. Policy Studies Journal 41(3): 426–452.

31.

Nilsen

Stähl

Roback

(2013) Never the twain shall meet? A comparison of implementation science and policy implementation research. Implementation Science 8(63): 1–12.

32.

Oosterwaal

Torenvlied

(2012) Policy divergence in implementation: How conflict among decisive legislators reinforces the effect of agency preferences. Journal of Public Administration Research and Theory 22(2): 171–195.

33.

Pressman

Wildavsky

(1973) (3rd ed. 1984) Implementation: How Great Expectations in Washington are Dashed in Oakland; or, Why It’s Amazing that Federal Programs Work at all, This Being a Saga of the Economic Development Administration as Told by Two Sympathetic Observers Who Seek to Build Morals on a Foundation of Ruined Hopes, Berkeley: University of California Press.

34.

Sabatier

(1986) Top-down and bottom-up approaches to implementation research: A critical analysis and suggested synthesis. Journal of Public Policy 6(1): 21–48.

35.

Sabatier

Weible

(2007) The Advocacy Coalition Framework: Innovations and clarifications. In: Sabatier

(ed.) Theories of the Policy Process, Boulder, CO: Westview Press, pp. 3–17.

36.

Sætren

(2005) Facts and myths about research on public policy implementation: Out-of-fashion, allegedly dead, but still very much alive and relevant. Policy Studies Journal 33(4): 559–582.

37.

Sætren

(2014) Implementing the third generation research paradigm in policy implementation research: An empirical assessment. Public Policy and Administration 29(2): 84–105.

38.

Simon

(1957) Administrative Behaviour, (2nd ed). New York: Macmillan.

39.

Stoker

(1991) Reluctant Partners: Implementing Federal Policy, Pittsburgh: University of Pittsburgh Press.

40.

Stone

(2012) Policy Paradox: The Art of Political Decision Making (3rd ed, revised), New York: W.W. Norton & Company.

41.

Torenvlied

Akkerman

(2004) Theory of ‘soft’ policy implementation in multilevel systems with an application to social partnership in the Netherlands. Acta Politica 39(1): 31–58.

42.

Van Gunsteren

(1976) The Quest for Control: A Critique of the Rational-Central Rule Approach in Public Affairs, London: John Wiley & Sons.

43.

Van Meter

Van Horn

(1975) The policy implementation process: A conceptual framework. Administration and Society 6(4): 445–488.

44.

Wilson

(1887) The study of administration. Political Science Quarterly 2(2): 197–222. .

45.

Winter

(2006) Implementation. In: Peters

Pierre

(eds) Handbook of Public Policy, London: Sage, pp. 151–166.