The idea of policy design: Intention,process,outcome,meaning and validity

Policy, design and the theorising of governing

The idea of design can be traced back to the earliest academic writings on policy, going back at least to Plato, focusing on the actions of the wise ruler, taking thought, acting appropriately, and ruling well. It drew on the development of the social sciences in the 19th and early 20th centuries, leading to Lasswell’s (1951) call for a ‘policy science’ which would be problem-focused, multi-disciplinary and explicitly normative (Lasswell, 1951). There followed the growth and institutionalisation (notably in North America) of ‘policy analysis’ as a field of professional practice and academic study (Radin, 2000), from which emerged a focus on ‘policy design’. However, its leading advocates claim that academic work on policy design has ‘declined precipitously’ because of shifts in analytical focus, and call for a ‘renewal’ of academic attention to the subject (Howlett, 2014; Howlett and Lejano, 2012: 364). This paper examines the argument for seeing design as a central element in the practice or the analysis of policy, as compared with the alternative approaches. It points to the assumptions underlying the design approach, the ways in which its advocates have responded to critiques of its empirical accuracy, and draws on newer, interpretive modes of analysis to offer an explanation of the continuing claims made for the design approach. It is widely argued that in the last half-century, there have been major changes in governing practice and in the analysis of this practice. If there is now less interest in addressing the question of policy design, is it still the right question?

It is worth going back to the analytical assumptions within which ‘policy design’ is located. It is part of a perspective which sees the process of governing as an exercise in what could be called ‘authoritative instrumentalism’: it is something accomplished by authorised leaders (‘the government’), who exercise their authority to accomplish their objectives. These leaders are seen as coming to power with agendas to pursue, and preferences which guide them, though they are also confronted with ‘problems’ not of their choosing, which call for the exercise of the power of government. Policy is a tool which they use in this process, something that they ‘make’ as part of the exercise of control. ‘Public policy’, says Dye, ‘is whatever government decides to do or not to do’ (1976: 3). The authoritative leaders make decisions; these are ‘policy’.

Policy, then, is seen as an artefact, the creation of a small group at the top of the hierarchy of officialdom – ‘the government’, or ‘the policy-makers’ – but this raises the question of what all the others are doing. ‘The government’ is a vast conglomerate of organisations, in which a large number of people perform diverse and often quite specialised tasks. How do their activities relate to those of the authoritative leaders? The answer, in this perspective, is that the leaders make policy, and officials either advise them (before the decision), or implement the decision after it has been taken. The activity of governing is explained in terms of these critical decision points: giving advice, making decisions, and giving effect to them. The experts design the policy and the authoritative leaders approve it (making it a little unclear just who ‘the policymakers’ are). ‘Policy design’ is part of the diversity of activities which are absorbed into a narrative of authoritative decision.

Policy activity as skilled practice

The perception of policy as a field of skilled practice (including policy design) has emerged relatively recently. Before then (at least in Anglophone countries), policy was seen as emerging from constitutional processes: leaders (and parties) identified areas of concern and desirable responses to them, and mobilised support by the enunciation of these concerns and planned responses. At election time, these became the basis of party platforms, and election success was therefore viewed as popular endorsement for the platform and justification for following it (‘the mandate’). UK Cabinet minister Richard Crossman (1975) noted in his memoirs that ministers needed the platform ‘to protect us against the Dame’ (referring to the civil servant head of his department) – in other words, having an electoral justification for action as well as, perhaps as a counter to, an expert bureaucratic one. In the US, political scientists tended to see government as essentially about distribution: e.g. Lasswell’s (1936) Politics: Who Gets What, When and How, the Founding Fathers having put state structure there to regulate the competition – ‘a state of courts and parties’, as Skrowneck (1982) put it – rather than to be an independent actor.

The post-war years saw changes in these perceptions. In the US, the much-expanded role of the federal government in the war carried over into peacetime; the air force moved to set up the RAND Corporation, building on wartime work on operations research and systems analysis (McCloskey, 1987a), and Lasswell (1951) called for a ‘policy science’ that was interdisciplinary, problem-oriented and explicitly normative. In this context, ‘policy analysis’ – variously seen as a technique for the comparison of options, a field of professional practice, and a body of academic knowledge – emerged in the US.

As a technique, ‘policy analysis’ (initially ‘systems analysis’) as it emerged in the US was seen as a form of ‘decision support, grounded in operations research and microeconomics’. It was argued that a policymaker must specify the objectives, identify alternative ways of achieving them and the costs of each, evaluate the consequences of each option, and choose the alternative that delivers the greatest net benefit – Jeremy Bentham’s ‘felicific calculus’ (Majone, 1989: 12–13; see also Torgerson, 1985). This should be done before the policy decision (ex ante) and (preferably) also after (ex post), in which case it would probably be called ‘evaluation’. The elaboration of these calculations stimulated an interest in the application of cost–benefit analysis in relation to outcomes to which it was difficult to attach a market price, but the assumption tended to be that such questions were incidental, technical difficulties, and that well-planned quantitative analysis would show the optimum outcome, which should settle the debate about the action that should be taken.

This approach to policy had wide currency in the US in the 1960s, and ‘policy analysts’ and organisational units containing them became common features of American government agencies, legislatures and organised interests (Meltsner, 1975; Radin, 2000). While the federal requirement for agencies to use PPBS (the Planning, Programming and Budgeting System) as the basis for their budget requests was soon abandoned (Schick, 1973), the rhetoric of quantitatively based systematic choice retained great force, and it provided the analytic format for the evaluations which came to be demanded of federally funded programmes.

This emerging occupational specialisation was further institutionalised by the foundation of the Association for Public Policy Analysis and Management, with a journal, conferences and awards, lending support to a belief that policy analysis was a profession that had ‘come of age’ (Radin, 2000), though this was qualified by findings that many policy analysts had not done the university courses which were seen as professional training (Howlett and Wellstead, 2012).

This innovation in government was paralleled by the development in universities of graduate programmes to train people for this emerging function, mimicking the Masters of Business Administration (MBA) of the business schools. There might have been a debate about whether these programmes should be based in political science or economics, but in the US, economics was deeply embedded in the business schools, and graduate programmes in policy tended to have a strong political science component, though they usually included a core unit on policy analysis which was grounded in microeconomics. ‘Policy design’ was not in general use as a course label.

This was a distinctly North American development; though elements of it appeared in other jurisdictions. In the UK, while the Beveridge Report of 1944 had mapped out a much more extensive role for state action, particularly in health and welfare, this did not lead to attempts to use the techniques of planning and programming that had been learned in the war (McCloskey, 1987b) to facilitate the accomplishment of the more ambitious policy agenda, and even the term ‘policy analysis’ was not widely used until the closing years of the century. The Blair government released a White Paper on ‘modernising government’ which included a reference to ‘designing policy around outcomes’, but the subsequent handbook on ‘policy making for the 21st century’ (Cabinet Office, 1999) tended to refer to ‘good practice’ in ‘policy making’, and British policy staff were less likely to have been exposed to the systematic comparison of options taught in the North American graduate courses. In the UK and other Westminster jurisdictions, a perhaps more significant change was the growth in the numbers, and significance in the policy process, of political staff, directly attached to ministerial offices and acting as intermediaries between the bureaucracy and political authority (Anderson, 2006; Connaughton, 2010; Eichbaum and Shaw, 2011; Maley, 2000; Smith, 1977; Paun, 2013). This perhaps suggests that leaders were more concerned about their authority over the governmental process than the ‘design’ of policy programmes.

Similarly, ‘policy design’ did not attract much attention on the European mainland, even though there was a great deal of policy development going on in the closing decades of the 20th century as the EU moved to establish a single European market, and the regimes of order which had been built up over the years in the member states were exposed, declared obsolete, and replaced by new regimes which were negotiated, in a relatively evident way, between officials, interest group representatives and political leaders. Attention was focused on how the conflicting interests could be brought together into a mutually acceptable order, and the process was more likely to be referred to ‘the new (or ‘European’) mode of governance’ rather than ‘policy design’. The craft of policy was seen to be more concerned with inducing collaboration among a diversity of participants than with skilful drafting in advance.

The relationship between narrative and practice

But, even if the North American thinking and practice was not faithfully copied in comparable polities, ‘policy analysis’ had strong rhetorical appeal. A form of work had been identified, positions created for people to perform it, educational qualifications devised, and many of the elements of professional identity had emerged. But in what way did this modelling of policy activity as bureaucratic decision support – officials doing analysis and giving advice on the appropriate choice, based on mobilising knowledge for the systematic identification and comparison of options – relate to the experience of the governmental process?

The first question might be where the story starts. Many accounts of the policy process start with an actor called ‘the government’, or more vaguely ‘the policy-makers’, addressing a ‘problem’ which demands attention. This raises questions about what was happening before this point: how was this aspect of social practice being governed, in what way was ‘the government’ (or ‘the policy-makers’) involved, and how did the situation become a problem? It may be convenient for ‘policy analysis’ to posit a starting-point in order to focus attention on how actors responded to this identified problem, but practitioners tend to find that they are engaged in the managing of continuing patterns of governing, in which particular issues attract attention from time to time, but clear points at which these issues were raised or resolved are hard to see.

Nor is it self-evident what sort of knowledge will guide these systematic comparisons. Government is a conglomerate of specialised agencies, which tend to be organised around particular sorts of knowledge, and this can vary not just by functional area (Roads, Education, etc.) but also by what could be called the institutional orientation: to purpose (e.g. Health), to process (e.g. Auditing), to the people served (e.g. to old people), or to place (local knowledge) (Gulick, 1937). Each of these is likely to view the policy problem (and what would be useful to know about it) in a distinct way. North American-style policy analysis therefore was a new knowledge-based skill, an expertise of systematic choice, which was an implicit challenge to the adequacy of these ‘technical’ forms of knowledge as a guide to decision. Another, more explicit, challenge came from ‘overarching’ claims, such as the ‘whole of government’ ambitions of central agencies, or the ‘meta-value’ claims that the whole workings of government should be scrutinised for their impact on (for instance) gender relations or sustainability. In many areas, official expertise had to contend with the expert knowledge of researchers and organised interests. And despite the rhetorical power of claims to expert knowledge, there are also strong claims for participation by the public, and for the recognition of the value of their ‘ordinary knowledge’.

So there are many forms of knowledge in play, and which form should be mobilised depends on the context. What is the question being asked, and what are acceptable ways of finding an answer? There are likely to be ‘competing rationalities’ (Lin, 2003). Tenbensel (2006), going back to Aristotle, identifies the contenders as knowledge from study (episteme), knowledge from practice (techne) and judgment (phronesis), and argues that policy workers have to be able to ‘read’ the situation and deploy evidence from the type of knowledge which will be persuasive in that context. Weiss (1991) argues that it is the nature of the question being asked that determines what is policy-relevant knowledge: whether it is to clarify the situation (‘data’), to develop a response (‘ideas’), or to justify a preferred course of action (‘arguments’).

This makes the field of policy-relevant knowledge crowded and rivalrous, and the claim by mainstream policy analysis to hold the ‘trump suit’ by virtue of its ability to predict the outcomes of alternative policies and to have an objective metric for comparing their costs and benefits is not likely to go unchallenged. Nor is it clear that expert knowledge is being used to frame the action to be taken. March and colleagues point out that in many cases, detailed studies are made, but the recommendations are ignored; consultants are hired to advise, but after the decision has been taken; reports are demanded, but not read. Expert knowledge is invoked to demonstrate that action has been taken in an appropriate manner: it is part of the performance, not the determinant of the action (Feldman and March, 1981; March and Olsen, 1989).

As we saw, the sort of knowledge that might be mobilised in policy reflects the position that participants held, so a critical question for policy analysis has been who would be recognised as participants in the process. If ‘policy’ is defined in terms of what government decides to do or not to do, it can be assumed that the relevant actors are all officials, and the only question is whether to limit ‘the policy-makers’ to a small group of political leaders, to expand it to include senior bureaucrats, or to take in the wide array of officials involved in the process – ‘policy with a cast of thousands’, as Page and Jenkins (2005) put it. But a theme running through political science for at least the last half-century has been the clustering of participants – government and non-government – around the fields of practice needing to be governed – from Griffith’s ‘“whirlpools” of special social interest and problems’ (1939: 182) through Davies’ ‘gatekeepers… permanently camped around each source of problems’ (1964: 3) to Richardson and Jordan’s ‘policy community’ (1979), and culminating in the widespread use of the term ‘governance’ (Colebatch, 2014).

Rhodes’ (1996) original claim, in introducing ‘governance’ as an analytical term, was that the mode of governing the UK had so changed that it was no longer possible to speak of government by authoritative direction, and instead there was ‘governance’ by negotiation among self-organising networks. If so, this would raise questions about whether ‘policy design’ was possible, and how would it be done, and by whom. Rhodes’s assertion was, however, contested empirically (Bache, 2000; Johansson and Borell, 1999) and conceptually (Marinetto, 2003), and Rhodes tended to retreat from this usage. But the term came to be widely used, though with widely varying meanings, such that Offe (2008, 2009) asked if it had become an ‘empty signifier’. In particular, he pointed to the way it had changed from being a label for a particular way of governing (Gegenbegriff) to an umbrella term embracing all forms of governing (Oberbegriff), resulting in comparisons of ‘market governance’ with ‘hierarchical governance’ and ‘network governance’ (Bell and Hindmoor, 2009). But if ‘governance’ is limited to the existence of relatively autonomous participants and the need for negotiation between them, it is hard to see much scope for design as prior and expert activity. The waters were muddied further by claims for the existence of ‘metagovernance’, through which ‘the state’ controlled and limited the scope for negotiation among the allegedly autonomous participants (e.g. Daugbjerg and Fawcett, 2017). Would these ‘metagovernors’ count as policy designers, or would they be simply boundary-riders keeping the designers in check?

So while the use of the term ‘governance’ demonstrated the now widely shared realisation that the activity of governing involved more participants and forms of authority than ‘the government’, it does not help much in analysing the significance of this for policy design; we can address this question at both the collective level and the level of individual behaviour. At the collective level, it means that non-government organisations are increasingly recognised – in a stream going back at least as far as Schmitter’s argument about ‘corporatism’ (1974) – as part of the structure of public authority. This recognition was overtaken by the drive for market-like forms of governing, which has in turn been overtaken by the enthusiasm for ‘governance’. At the level of individual behaviour, it has long been recognised that most exercises in governing involve continuing forms of collaboration between participants from different institutional locations drawn together by shared understandings and mutual dependence. In making sense of their situation, participants tend to ask themselves questions like ‘who can I talk to about this?’, ‘in my experience of governing, what is this situation most like?’, and ‘what did we do that time?’. Familiar situations and familiar counsel tend to stabilise practice, and some observers label this tendency to use existing relationships to address policy questions as ‘networking’. Others go further and identify ‘networks’ as entities, sometimes as actors in the policy process, even though for the most part they can only be observed through the actions of the individual participants. Richardson and Jordan (1979) saw this as leading to identifiable ‘policy communities’. In any case, recognising the importance of this practice is of great significance in the way that governing is theorised, and we will return to this.

Going further, the key design assumption – that expert prior thought will deliver improved policy outcomes – is challenged by scholars applying ‘complexity theory’ to the policy process (e.g. Cairney, 2012; Geyer, 2012; Room, 2013; Sanderson, 2000, 2009; Teisman and Klijn, 2008). They argue that there is a ‘growing challenge of social complexity’, that increasing uncertainty makes traditional cause-effect assumptions unreliable, and that increased knowledge leads to more uncertainty, not less (Sanderson, 2009). Traditional ‘command and control’ modes of governing have to give way to ‘soft management’ (Cairney, 2012) and the involvement of non-state actors (Sanderson, 2009). In this perspective, policy work is about interaction, diversity and learning, rather than the execution of already-formed designs.

Instrumental and process-based accounts

We can see here two quite distinct accounts of the policy process emerging: an instrumental account focused on interventions to secure outcomes, which describes it in terms of expert advice followed by specific choice leading to exact implementation, and a process-based account which sees it in terms of ‘collective puzzling’, stakeholder involvement, interaction, ‘satisficing’ and ambiguity. Certainly, participants’ accounts of their place in the policy process do not suggest that their work is mainly concerned with ‘advising the Prince’ on the optimal choice. Rather than systematically comparing options for choice, practitioners find themselves enmeshed in the detailed work of ‘making things happen’. ‘Half the job is about implementation’, said one Australian policy worker (Adams et al., 2015: 103). Studies of policy practice found that many policy analysts were engaged more in managing matters of currents concern (‘firefighting’ – e.g. Howlett and Newman, 2010: 125) than in systematic, strategic planning. Radin found that as policy units became an established part of the governmental structure in the US, policy workers were more likely to be negotiating with their counterparts in other agencies to secure a mutually acceptable outcome (Radin, 2000: 35).

So policy practitioners have to deal with both accounts of their practice. For political leaders, it is important that the work of governing is seen in terms of (their) authoritative interventions, and academic observers often share this perspective. ‘Public policy is how politicians make a difference’, as one text puts it (Althaus et al., 2013: 5). And specialist officials want to bring their functional expertise to bear to shape these interventions to ensure their beneficial effect. But at the same time, they are aware that the policy process is a flow of continuing attention and activity and interaction, involving the maintenance (as well as the modification) of existing practice and response to the concerns of other participants, and what can be seen as ‘interventions’ are likely to involve dealing with these others, so that policy analysis becomes a matter of ‘making sense together’ (Hoppe, 1999).

So, practitioners have to recognise multiple accounts of the policy process, and deal with the discrepancy between them. Radin found that those who had done the graduate courses found a significant contradiction between what they had been taught and what they did as policy analysts:

‘There seems to be a disconnect between the analyst’s perception of self-worth (often drawn from the rational-actor model) and the real contribution that the individual makes in the nooks and crannies of the policy process. Many analysts… seem to need a language to describe what they do and to convince themselves – as well as others – that they contribute to the process’. (Radin, 2000: 183)

Managing multiple accounts in policy work

This tension between a prior, analysis-based conception of the task and the practice-based perception of the need for interaction and negotiation to achieve an acceptable outcome presents a problem for those employed as policy analysts – and even more for attempts to depict their activity as ‘policy design’. There are, in a sense, two problems for the analyst here: how to do the job, and how to talk about it. If ‘relationships matter’ in getting policies put into effect, should the policy analyst take note of them in doing the analysis? If the analyst decides that the recommendation coming from an analysis will not be adopted simply on its own merits, should she/he act as an advocate? Should the analyst bear this in mind when doing the analysis, and interact strategically with the other participants so as to maximise the chance that the recommendation flowing from the analysis will be followed? Or should the analyst simply identify the optimal course of action and leave it to others to bring it about?

The analyst’s second problem is how to talk about what they do. The difficulty is that analysts have been trained to identify the best course of action, but find themselves needing to negotiate around a range of acceptable courses. Howard (2005) finds that policy practitioners distinguish between the ‘textbook model’ (Nakamura, 1987) and operational logic. One of them said:

‘So long as they understand that [the model] is a theoretical construct… [it is] really about theory, not practice… These words are so neutral. It’s not about consultation. It’s really about stakeholder engagement.’ (Howard, 2005: 10)

Prior thought and theorising policy

We can see, then, that the idea of a prior, conceptual and authoritative stage in the process of governing has a long history. As Howlett says, the roots of policy design theory ‘lie in the very origins of the policy sciences’ (2014: 190). In the initial conceptual formulations about policy (e.g. Lasswell, 1951), the focus was on ‘policy-making’: ‘design’ was not identified as a specialist function in policy practice. There soon emerged a relatively clear model of the nature of policy practice and the underlying conceptual framework, which saw ‘policy analysis’ as both a prior stage in policy formulation and a specialised form of work. Analysis was concerned with the identification and comparison of options: it produced the raw materials from which authoritative figures (‘the policy-makers’) ‘made’ policy. This was, of course, consistent with the constitutional democratic model: bureaucratic experts might prepare options, but political leaders ‘made policy’. In this sense, it was a ‘good account’ of policy, not because it was an accurate depiction of the activity, but because it validated the outcome – a ‘normative’ rather than a ‘pragmatic’ account, in Bailey’s (1969) terms.

So the idea of prior, expert judgment was already well-established in theorising about policy. This was part of a presentation of the policy process as what Deborah Stone (1988: 4) tagged ‘the rationality project’ – policy is (or should be) instrumentally rational:

‘Public policies are the result of efforts made by governments… to carry out some end or purpose… These efforts can be more or less systematically designed.… Some policies emerge from processes such as log-rolling, patronage or bargaining and cannot be thought of as having been ‘designed’ in any meaningful sense of the term.… In some circumstances, policy decisions will be more highly contingent and ‘irrational’ than others: driven by purely situational logics and opportunism rather than careful deliberation and assessment.’ (Howlett, 2014: 188)

What was not clear was whether this new term referred to the activity previously known as policy analysis, or to a new field of practice. Dryzek, in one of the earliest pieces to use the term in the title, sought to shift the focus ‘away from methods emphasising the assessment of pre-ordained and well-defined alternatives, and towards policy design’, which he saw as ‘the process of inventing, developing and fine-tuning a course of action with the amelioration of some problem or the achievement of some target in mind’ (1983: 345–346). This suggested an empirical orientation, taking note of the experience of policy practitioners. On the other hand, Linder and Peters (1988, 1990), who dismissed Dryzek’s work as a ‘fashionable’ denigration of conscious approaches to policy choice (1988: 738), were at pains to stress the importance of ‘freeing the concept of design’ from any ‘entanglement’ with a consideration of the policy process, which ‘militates against systematic design’ by paying attention to how the policy process operates. Once the concept of design is freed from any awareness of the policy process, it ‘ceases to be a real-time phenomenon’ and ‘becomes an activity directed towards the policy process but separated from it in time and even space’ (1988: 742). This appeared to be a field for academic construction, unfettered by any link to actually existing policy activity.

This gave rise to ‘a specific self-referential literature on policy design in the mid-1980s (Howlett, 2014: 191), in which scholars sought to classify the ‘instruments’ of policy, the ‘generic tools of government action’ (Salamon, 1981: 256), the circumstances of their use and the outcomes. The argument was that a better understanding of these factors would lead to better-crafted, and more efficacious, policies: this field of knowledge was ‘policy design’. This enabled those scholars so inclined to construct models of policy instruments without reference to any specific policy practice. Whether anyone was doing the ‘policy design’ thus identified was another question.

The challenge to the design orientation

But as Howlett notes, by the early 1990s, this ‘design orientation’ was running out of steam, and ‘was largely replaced by a new emphasis on the study of decentralized governance arrangements’ (2014: 189). Howlett’s response to this finding has not been to scrutinise the reasons for this apparent loss of faith in the design orientation, but to call for a revival of interest, basing this call on two quite divergent defences.

The first, following Linder and Peters, is that empirical evidence of the policy process is not relevant to the study of design: ‘the actual public decision-making process can be divorced from the abstract conception of policy design, in the same way that an architecture [sic] can be divorced conceptually, if not in practice, from its engineering’ (Howlett and Lejano, 2012: 360). Consequently ‘the study of design is properly “meta-oriented” and, therefore, one step removed from the study of policy and policy-making’ (Howlett and Lejano, 2012: 361, quoting Linder and Peters, 1990). So while ‘[T]he design process is complex,… and usually much less accessible to public scrutiny than many other kinds of policy deliberation’, students of design can consult the literature of the 1980s and 1990s and ‘focus discussion on the key design parameters already identified within the literature’ (Howlett and Lejano, 2012: 370). The characteristics of design can be found in the classical texts: empirical evidence is evidently unnecessary.

The second line of defence is to redefine the subject matter, distinguishing an ‘old design orientation’, which saw policy design as the stage preceding decision and was concerned with the impact of particular sorts of policy instrument, from a ‘new design orientation’, in which design is seen as a continuing process, not starting with a blank page, but with new initiatives being added to existing settings, perhaps making for overlap and contradiction and a continuing need for policy maintenance (‘patching’). While it still talked about ‘designers’, it identified not only ‘“traditional” advisors in government’ but also ‘non-government actors in NGOs, think tanks and other similar organizations, and less formal or professional forms of advice from colleagues, friends and relatives and members of the public and political parties’ as ‘influential actors involved in design decisions’ (Howlett, 2014: 195). But the discussion is still about ‘designers’ facing ‘challenges’ about whether to ‘patch’ or ‘redesign’ existing regime elements which may be ‘multiple policy tool portfolios’ perhaps lacking ‘any degree of coherence and consistency’ (Howlett, 2014: 198). And we are told that ‘policy design… extends… to the consideration of the practices… of policy formulators or “designers”‘ (Howlett and Lejano, 2012: 359); we have design, but apparently the authors are not sure whether we have designers. The response to the inconsistency between the concept of ‘design’ as prior, expert activity and the observable evidence of policy practice has been to redefine design: everything that happens in the policy process can be accounted ‘design’.

Conclusion: Theorising policy design and policy analysis

This means that the discourse about ‘policy design’ has followed a similar trajectory as that on ‘policy analysis’, and for much the same reasons. Both were seen initially as prior to action, instrumental, calculative, expert, and giving rise to a coherent and efficacious course of action. But while it is convenient to think of policy in terms of discrete, single-purpose ‘interventions’, the process of governing (of which policy is a part) is a continuing flow of action in a wide range of fields, involving a diversity of actors, agendas and relationships, and new policy initiatives have to find their place in this continuing drama. Early policy theorising tended to ignore this dimension of the action; later, it was announced as a discovery – the ‘problem of implementation’ – and dealt with as a problem for ‘policy-makers’, until it was overtaken by a range of metaphors of collectivity (‘issue networks’, ‘policy communities’, etc.) and finally subsumed in the conceptual Sargasso Sea of ‘governance’. ‘Policy design’ was also seen as quite distinct from the messy world of policy activity; rather it was seen as ‘the Platonic form to which the real-world arrangement of elements must seek to conform’ (Linder and Peters, 1988: 743). But faced with a declining interest in this perception of the task, some design scholars came to accept that policy was not created by a single mind on a blank slate:

‘many existing policy mixes or portfolios were not “designed” in the classical sense of conscious, intentional and deliberate planning according to well established or often-used principles, but rather emerged from a gradual historical process… [or from] several different change processes – layering, drift, replacement, conversion and exhaustion – which helps explain why many existing policy mixes suffer from inconsistent, incoherent or incongruent elements’. (Howlett, 2014: 198)

But while speaking of ‘old design’ and ‘new design’ reflects an attempt to reformulate policy design thinking, the basic assumptions have not changed. It is assumed that policy is made by actors called governments which have objectives and choose designs to accomplish these objectives, that these choices determine the actions of officials, and that evaluation confirms that the action had the desired effect:

‘Governments… continue to design systemic modes of governance and establish the nature of the policy instruments adopted for the pursuit of these goals… The design orientation hence remains a key one in policy studies, and research in this vein is necessary to advance our understanding of both designs themselves and the processes which lead to their adoption, implementation, evaluation and reform.’ (Howlett, 2014: 200)

But does this matter? Is this anything more than a re-statement of a ‘traditional’ distinction between ‘rational’ and ‘political’ modes of policy-making? I argue that the claims of ‘policy design’ call for renewed attention because:

there is a very vigorous advocacy of the concept by high-profile figures in the policy analysis literature like Michael Howlett;

As we noted, Michael Howlett complained that the previous interest in policy design had run out of steam in the 1990s, and is actively striving to sustain enthusiasm for the concept, including sponsoring an online ‘Annual Review of Policy Design’, reprinting articles from various sources with a policy design focus. His paper on ‘the rise and fall (and rebirth?) of policy design studies’, presented to the 2011 ECPR conference had 19 pages of text and 225 references, and links the policy design focus to other analytic themes in good standing, including globalisation, multi-level government, and policy advisory systems. Howlett claims that the articles in a special issue of Policy Sciences which he edited ‘show that the design orientation is still very much alive and well as both a subject of policy practice and an object of scholarship’ (Howlett, 2014: 190). This is not a concept whose time has gone.

Other research on the policy process has thrown light on why the design perspective holds its appeal. The ‘interpretive turn’ in policy analysis has drawn attention to the way that the achievement of governing depends on the generation of shared understandings about what is of concern and why, who should act, and what sort of action is appropriate (see, for instance, Hoppe, 2010; Stone, 1988; Yanow, 1996). This means that it is important for the action taken to be seen as properly determined: resting on decisions by the proper authorities after consultation with appropriate persons and likely to lead to an improved outcome. Talk of policy as being ‘designed’ is entirely congruent with this, although in Howlett’s ‘new design orientation’, the policy designers appear to include ‘a complex web of policy advisors’ or a ‘policy advisory system’ (Halligan, 1995), which includes both ‘traditional’ advisors in government as well as non-governmental actors in NGOs, think tanks and other similar organizations, and less formal or professional forms of advice from colleagues, friends and relatives and members of the public and political parties, among others’ (Howlett, 2014: 195) – in other words, anyone seeking to have a voice in the process, which is likely to include participants with diverse perspectives, histories and reasons for acting. How this wide array;

‘should ensure that any new design elements are coherent in the sense that they are logically related to overall policy aims and objectives; that they be consistent in that they work together to support a policy goal; and that both policy goals and means should be congruent, rather than working at cross-purposes’. (Howlett, 2014: 195, n.4),

What is missing here is the recognition that framing the activity as ‘policy design’ is itself part of the action. Presenting the activity of negotiating an acceptable outcome as ‘policy design’ is a rhetoric for facilitating acceptance, mobilized by actors in the accomplishment of governing. Government is an arena rather than an actor, and many of the players in the arena are not part of government. Objectives may be hard to discern and of little consequence in shaping action. There may be quite diverse perspectives on the nature of the problem and the appropriate response, with policy practice seeking to mask these differences rather than clarify them. Much of the activity is aimed at generating a tacit acceptance of the action to be taken among relative ‘outsiders’ (see, e.g. Colebatch, 2009; Colebatch et al., 2010; March and Olsen, 1989). So giving an account of policy activity is itself part of the action. Wide-ranging and hard-fought negotiations among a diversity of participants are ‘enacted’ (Weick, 1979) by being announced as a government decision. The negotiations are then labelled ‘policy advising’ and the participants as part of the ‘policy advisory system’ (Halligan, 1995). In this way, the contest, overlap and ambiguity of the process can be validated in terms of the ‘sacred’ authoritative instrumental paradigm. (This is not the only way in which outcomes need to be validated, but it is the most conspicuous way.) Participants know that to achieve an outcome, a wide range of participants must be included in the negotiations, but that outcome must be represented (re-presented) as an authoritative choice. In this context, academic observers (like Howlett) can re-present these negotiations as ‘policy design’ without identifying who are the ‘designers’ or in what way their activities can be seen as ‘design’. There is an outcome, so it must be a ‘policy design’ and the people who put it together must be ‘policy designers’. This is understandable as part of official practice, but not as social science. It simply re-labels an existing ambiguity about ‘policy-making’: is it only the official endorsement that makes the policy, or does it include the activity before this point, and afterwards? And if policy is being made by the decisions of authoritative leaders, are they ‘designing’ policy? If not, how is ‘policy design’ differentiated from ‘policy making’? If it does not address these questions about the nature of practice and the representation of practice, the ‘new’ design-talk will be as self-referential, and as ready for redundancy, as the old.

This shows that the labelling is not simply validating the outcome, but also validating the action of participants in a particular way; they are not simply (for instance) business lobbyists trying to reduce their customers’ tax liability, but participants in ‘policy design’. Boxelaar et al. (2006) describe a similar example of this labelling, where agricultural extension agents worked with farmers to develop an agreed ‘best harvesting practice’ which would reduce the risk of fire, and reported this as encouraging peer-group governing of farming practice; departmental managers, however, insisted that it be reported as the department ‘delivering services’ to its ‘customers’ – thus validating its own activity (and expenditure) to the managers’ ‘relevant others’ – ministers and the Treasury.

As we have noted, policy practitioners are well aware of the strategic significance of labelling, and become adept at adapting their language to the context (see, e.g., Tenbensel, 2006). But this poses a problem for academic observers of the policy process: do they adopt the official (‘sacred’) language in their analysis of the action, or do they apply a more detached, analytical language? And if they do, will they be seen as being less ‘helpful’, perhaps even hostile? This becomes particularly significant when funding bodies demand evidence that the research is being ‘used’ by the subjects of the study. Does this create an incentive for researchers to adopt the official language of the participants in order to be seen as ‘useful’, and therefore fundable? This calls for researchers to be aware of the significance of interpretation in policy practice, and in the way that the language use of researchers, practitioners, and relevant others like funding bodies (in government and in academia) relate to this practice, but there is little of this awareness in the policy design literature to date. Good analysis has to recognise that analysis itself is part of the action.