Saylor.org's Comparative Politics/Implementation and Choice of Means

1 Models of implementation

by The Open University

1.1 Policy delivery

The question of policy delivery seems to be growing in importance. So, for example, the Blair governments in the UK were, from the outset, preoccupied with ‘delivery, delivery, delivery’ as ministers and prime minister grew increasingly frustrated with what was often viewed as the intransigence of public service professionals. The constant cycle of change, in which new policies and initiatives were introduced in rapid succession, producing what critics described as ‘policy overload’ or ‘initiativitis’, can be understood in part as a result of prime ministerial and ministerial frustration. This also produced an explosion of new regulatory mechanisms – targets, standards, audit requirements, inspections and new performance regimes – in an attempt to ensure that organisations conformed to the policy requirements of government. But, as you will see, this was only the latest in a long series of attempts by central governments in the UK and elsewhere to enforce their policies. This unit focuses on the relationship between ‘structure’ and ‘agency’. The questions pursued here can be summarised as follows:

•How do people (public service professionals, managers, front-line staff, as well as those working in voluntary and non-profit organisations) act (or exert their agency) in the context of the funding and policy constraints (the structures) that govern their work?

•How do they negotiate the dilemmas that they are faced with as different policies interact, or as new policies conflict with professional, organisational or community-based values?

•How do they work in partnership with others to deliver complex policy outcomes at the same time as trying to meet the performance requirements placed on their ‘own’ organisations?

•How do they manage the interface between different conceptions of ‘what works’ – especially where their own professional view of good practice conflicts with a policy based on a different evidential base?

This section introduces four different models of change and assesses their relevance to understanding the policy process. Exploring these models involves addressing their relevance to particular policy trends – the focus on ‘joined-up government’ and partnership, the emphasis on ‘what counts is what works’ in policy making and delivery, the shift towards involving the public itself in the policy process, and so on. Note that such models are both explanatory (they help illuminate reality by highlighting particular features of the policy process and suggesting their benefits or flaws) and normative (they carry implicit assumptions and prescriptions about how the process should work). Disentangling these different ways of working with models is not always easy, and you will return to this discussion at the end of the unit. Finally this section also explores the importance of new governance forms that, it is argued, create a need for new ways to conceptualise the policy process.

1.2 The machinery of government: policy as rational planning

Much of the policy literature is imbued with a rather mechanical conception of change: ideas about ‘pulling levers’ to make things happen, or about applying different ‘tools’ or ‘instruments’, all conceive the policy system as something like a machine itself. Component parts – the government departments, regulatory bodies, delivery organisations, and even the people who staff them – are viewed as connected though static and predictable mechanisms. The system is seen as non-adaptive and non-learning. That is, to make change happen it is always necessary to apply a new mechanism from above in order to pull the system ‘into line’ – or, in the case of previous failed attempts to do this, to redesign all or part of the system from scratch, based on a new blueprint. In this model, only policy makers are viewed as capable of learning; everyone else has to carry out the policy mechanically. The conception of structure and agency holds that structure is all: people, it is assumed, will act according to new incentives built into the design. All the government (the mechanic) needs to do is restructure the institutions (a merger here, a decentralisation there) or tinker with the incentive structure (altering the funding streams and/or introducing more competition between organisations) and, it is assumed, change will follow.

Susan Barrett (2006) discusses the policy equivalent of this image in her critique of the rational planning, top-down view of the implementation process. But it should also be considered how far such critiques apply to a rather more ‘modern’ version of the policy system: that based on contract. The new public management (NPM) reforms of the 1980s and 1990s in many western countries produced a ‘fragmented’ public sphere. The old departmental hierarchies of the state were slimmed down and the policy and delivery roles of the state itself were increasingly separated. In the UK this led to the creation of a number of different civil service agencies (those covering child support, highways, passports, defence, etc.), while in the USA it reinforced already existing divisions, to the extent that some influential authors called for a transformation of the public sector, with an emphasis on the need to mobilise (or empower) social or community entrepreneurs to take on increased responsibility (Osborne and Gaebler, 1992). The introduction of quasi-markets and competition extended the process of fragmentation, producing what Rod Rhodes (1997) terms a ‘plural polity’ based on a dispersal of state power. Some authors dispute the term ‘fragmentation’ and suggest that these reforms resulted in a modest disaggregation or loosening up of the arrangements for providing public services.

How, then, were the rods, levers and pulleys of the machine to function, given the greater distance between government and those delivering public services? One solution draws on ‘principal-agent theory’. This is based on a clear separation between ‘agents’ (those performing a particular function) and ‘principals’ (the clients who specify the functions to be performed). The relationship between them is the contract. This model underpins the ‘framework agreements’ between ministers and civil service agencies in the UK, as well as the contracting out of services to the commercial sector associated with NPM. Principal-agent theory states that contracting will be successful where there are tight specifications for the outputs to be delivered, where outputs can be easily measured and where inadequate suppliers can be quickly replaced (Donahue, 1989). The rationale here is the competition of the market-place rather than rational planning within hierarchies, but the relationship between principal and agent is nevertheless often viewed in somewhat mechanical terms. Indeed public service organisations acting as clients initially tended to manage their contracts in highly bureaucratic ways, producing overly tight specifications that squeezed the capacity of contractors to innovate, and monitoring outputs with an exaggerated zeal to the extent that damaged the possibility of trust between client and contractor.

The same charges might be laid at the door of governments acting as principals in relationship to other parts of government acting as their agents, delivering public services. Rather than a formal contract, the relationship here is governed by the specification of outputs in the form of performance indicators, achievement against which is then tightly monitored in a climate of low trust between government and public service organisations. Many benefits are claimed for the clear separation between principals and agents, and for the shift to contractual relationships. These include better performance, greater value for money, the tightening of accountability, and the generation of new providers – commercial, voluntary and non-profit organisations – whose involvement opens out the possibility of innovation. But there are also disadvantages, notably the problem of writing precise specifications or targets for complex services where the output depends in part on professional judgement, the tendency for organisations such as schools or hospitals to cream off clients or customers who are likely to enable them to meet their targets more easily, and the unwillingness of managers working to tight targets to undertake work other than that which is actually specified – a problem in areas where needs or conditions are fast-changing. A useful example here is provided by private finance initiatives (PFIs) and instances where contracts have been too rigidly put together. In order to enable the policy system to respond to complexity, uncertainty and rapid change, it becomes necessary to draw on alternative models that transcend some of the dysfunctions of the machine model.

Mechanical instruments based on a clear separation of principals and agents, coupled with tight specifications as to what is to be delivered, might be suited to simple tasks, such as refuse collection or catering. But when attention turns to policy problems such as the fear of crime, social deprivation, run-down communities, social exclusion, or a population without the so-called modern skills deemed necessary for survival in a global economy (all issues high on the policy agenda in many European countries and in the EU itself, in the form of the Lisbon Agenda), the principal-agent metaphor falls short for at least three reasons.

First, the definition and analysis of each of these problems are contested. That is, governments are confronted with a variety of explanations as to why the problem exists and a range of solutions for tackling it are proposed.

Second, such problems have a very weak fit with the classic departmental structures of government. The machinery of government, in other words, is not able to tackle these issues without completely restructuring the machine itself. Attempts have frequently been made to tinker with the machine – so, for example, in the 1960s an Office for Economic Opportunity was set up in the USA to lead the ‘war on poverty’; Australia in the early 1970s saw a short-lived Department of Urban and Regional Development with a strong emphasis on coordination between departments and levels of government; while in the 1990s in the UK a Social Exclusion Unit was established early in the first Blair administration to address what was deemed to be a ‘cross-cutting’ problem that did not fit the remit of any specific department of government. But creating new structures can only ever be a partial solution; even if this locates responsibility for such problems within a new unit, it is unlikely to be successful in providing that unit with the necessary authority and funding.

Third, and most important, governments require the collaboration of networks of actors to address such problems – and, increasingly, the public itself is invited to be part of such networks, since policies are more and more directed towards bringing about change in public attitudes and behaviour.

1.3 The perils of partnership: policy as an adaptive system

Here the focus is on an organic way of understanding the relationship between policy and action. From this perspective, government, public service organisations, contractors, staff and, more recently, the public themselves are viewed not as cogs in a machine but as mutually interacting elements of an adaptive policy system. As in other organic entities – populations, species, even the human body itself – change takes place around an equilibrium point at which the entity is in balance with its environment. This equilibrium is sustained by feedback loops: as the environment shifts, so the organism (or in this case the organisation) must adapt or perish. And few organisms can survive alone: collaboration is needed in order to better secure the survival of the whole species or population.

This ecological analogy helps in focusing on ways in which the policy system might better adapt to the complexity of many of the tasks or problems which the modern state has to address. Agency, then, is dispersed; and rather than being constrained by structure, it is shaped by mutual relationships and reciprocal dependencies. As Barrett (2006) argues, this requires a different approach to understanding policy implementation. Here the study of implementation is viewed, first, as an integral part of the policy process rather than as a final stage subject to formal administrative processes. Second, the approach acknowledges the ambiguity of many areas of public policy: objectives may not be precise, and different objectives may be in conflict. Third, it focuses on policy as a multidimensional, multi-organisational field of interaction – what Susan Barrett and Colin Fudge (1981) term a ‘policy-action continuum’.

Many policy innovations have emerged from bottom-up processes of agency. And sometimes an ‘enabling’ approach to policy on the part of government can elicit innovation in a positive fashion, by encouraging and rewarding experimentation and/or by judging performance on the basis of longer-term outcomes rather than short-term outputs. In the US context, David Osborne and Ted Gaebler (1992) have constructed a vision for the new public sector and its management (endorsed by the former Clinton administration) that is based around a simple mantra – steering, not rowing – which questions what the authors see as the traditional models of public administration. Their emphasis is on finding ways of working through a de-centred and fragmented policy field, held together by the exchange of best practice between social (and public) entrepreneurs, rather than by any regulatory framework or over-arching national policy. Similarly the rise of multi-level governance in the EU was a de-centred way of managing policy across national and sub-national jurisdictions, and a means of building direct links between EU members despite domestic constraints.

In the more centralist UK context, the balance between nationally driven policy and local partnership-based initiative was more uneasy than such a vision might suggest. But examples of such initiatives in the first and second terms of the Blair government were legion. Projects such as the education action zones, health action zones, SureStart and Neighbourhood Renewal encouraged local actors across the public, private and voluntary sectors to form partnerships. Each was required to develop plans by involving ‘the community’ in setting local priorities and finding the best ways of working to achieve positive outcomes. Differences of approach between projects in different areas were initially encouraged, and the results were closely evaluated to determine what lessons might be learned for mainstream policy making. Such an approach also enabled the focus to shift from organisational outputs (based on targets set by individual government departments) to policy outcomes (based on joined-up working).

The significance of these developments is reflected in the focus on joined-up government and the increasing emphasis on partnership as a means of both shaping and delivering policy. But, as Janet Newman (2006) argues in relation to partnership working, enabling and learning approaches to policy delivery tend to be undermined by the strong traditions of rational planning and the continued centralisation of power associated with top-down, mechanical models of the policy-action dynamic.

The organic model, as a metaphor, helps in looking again at the policy/implementation divide that characterises many of the classic policy texts. In this approach, rather than policy being made in one place (by government) and implemented in another (by public services and other contractors), policy and delivery are mutually implicated in an iterative process of change. That is, organisations in a dispersed field of power confront new problems and develop new ways of tackling older problems on an almost daily basis. Some of these new approaches are transmitted to other organisations in the system through a process of imitation or learning; some are left fallow, going nowhere; and some are taken into the next cycle of formal policy making.

The next two models can help explain why systems may fail to adapt successfully. Section 2 focuses on why some ideas fail to be translated into action, and suggests ways in which power and interests can act as barriers to change.

1.4 The contribution of culture: policy as meaning making

The third model of the relationship between policy and action, between structure and agency, is based on the idea that human agency cannot be understood by simply regarding people either as cogs in a machine or as elements in an interactive system. Rather, human beings are meaning makers and act on the basis of their understandings and interpretations of events. In other words, they construct their own reality. Such constructions are not unique to them as individuals, but draw on a stock of socially circulating repertoires of meaning to which new ideas are sometimes added while old ideas fall away. So, for example, interventions that seek to respond to criminal behaviour among young people might be framed by ideas of anti-social behaviour, parental responsibility and/or restorative justice. Each is a relatively recent addition to the anglophone policy vocabulary, but each has now become an established discourse through which policy makers, police services and public and voluntary agencies interpret behaviour and prescribe solutions. Each gives prominence to certain meanings (‘responsibility’ and ‘concern for the victim’) and marginalises others (the search for social causes of crime and the perpetrator's need for therapies). In a similar way the concept of ‘youth offending’ has, in part, now displaced that of ‘juvenile delinquency’, thus focusing attention on the criminal act rather than on the character of the young person concerned.

Repertoires are recurrently used systems of terms (e.g. metaphors, figures of speech) employed to characterise and evaluate actions and events. How does this help in understanding how change happens? These repertoires of socially circulating meanings – which, for simplicity, might be termed discourses – form, if you like, a kind of already existing structure of meanings that shape how individuals interpret situations or events, and thus guide the way they respond. But in order to better understand the relationship between discourse and social agency it helps to turn to a range of what Frank Fischer terms ‘post-empiricist’ understandings of policy and the policy process. He argues that:

ideas and discourses can have a force of their own independently of particular actors … Discourse, in this view, does more than reflect a social or political ‘reality’; it actually constitutes much of the reality that has to be explained … Instead of understanding power only in negative terms – such as the power to control or manipulate others – the approach … also emphasises that discursive power can determine the very field of action, including the tracks on which political action travels.

(Fischer, 2003, pp. vii-viii)

In other words, the language of politicians and policy makers, whether transmitted through policy texts or other communicative processes, does not just reflect a pre-given reality – it actually helps to constitute that reality. So shifts in the language within which policy is framed, or new ways of linking older ideas in new configurations, provide new resources of meaning making. This in turn, it is argued, can influence – but not determine – social action (there are few if any echoes of the more mechanical structure-agency dynamic here). Such discourses cannot be deterministic, since no one can force another to interpret events in a particular way or to adopt a new identity or self-image within the structural parameters of the policy system. However, the power of discourse can be discerned in the ways in which public service managers tend to adopt ‘new’ policy language in order to legitimate change to their staff, or in order to win credibility and/or funds from government.

Once again, this model disrupts the idea that policy and implementation are separate stages of the policy cycle. Governments may draw on new discourses emerging from the professions, from the business world or from other stakeholders, and adapt them to their own purposes. But rarely do these discourses emerge from government fully formed; it is up to managers, the professions and public service staff to interpret them, bringing their own constructions into the process of interpretation.

It is these actors, then, that are deciding what the policy actually means and enacting it accordingly. Policy is made, and enacted, in a myriad spaces: in schools, for example (as teachers construct for themselves what notions of parental involvement and parental responsibility actually mean); in hospitals (as doctors and managers decide how to work with emerging ideas of choice and consumer power); or in local government (as staff try to reconcile notions of community leadership with ideas of devolution and working in partnership).

The idea of language and other symbols as shaping reality is familiar territory in accounts of culture change programmes in public service organisations. However, the experience of such programmes also suggests some flaws in the notion that new symbols, linguistic or otherwise, can produce new meanings and thereby new practices and behaviours.

However strong the vision and leadership, however convincing the new rhetoric of change, however inspiring the mission, organisations tend to find themselves stuck in repeating the behaviours of their pasts. The same is true of the policy system. New forms of rhetoric on the part of politicians – joined-up government; what counts is what works; delivery, delivery, delivery; responsibility and choice – all provide points around which the agency of those charged with delivering policy can be mobilised. They enable those who are already committed to such developments to emerge from the sidelines and take centre stage. They enable new forms of experimentation and action on the part of those willing to lead organisational change. And, by providing a new rationale and purpose, they legitimise actions which organisations may have been wishing to develop.

Yet despite all this, change may still not occur. This can be explained in part by the continued dominance of mechanical models of implementation that tend to produce compliance rather than commitment. But any discussion of why change may not happen – implementation failure, in other words – would be incomplete without addressing the issues of power and resistance.

1.5 The problem of power: policy as political

The plural polity that characterises contemporary policy making means that many stakeholders are involved in the policy-action relationship dynamic, from commercial firms, public and non-profit organisations, the professions, central and local government, service delivery organisations, trade unions and the media, to organised groups of the public itself. Viewing policy as political, then, does not mean simply focusing on politicians. Rather, it signifies adopting a stakeholder perspective in which multiple groups, each with their own interests or preferences, seek to influence the outcomes of policy making and delivery. The focus on politics means analysing the kinds of power each stakeholder may seek to exercise and assessing the balance of power between them. It means rather more than looking at how different players may seek to obtain advantage through political manoeuvrings or game playing. Questions relating to power inequalities, and to the interplay of different power and resource dependencies belonging to stakeholders, are equally significant.

Issues of power have now been incorporated into analyses of policy formulation to a quite significant extent, with notions of advocacy coalitions, policy communities and so on coming to the fore. Rather less attention, however, has been paid to issues of power in the process of delivery; and such attention, where it does exist, has tended to focus on relatively simplistic assumptions about resistance. It is assumed that where policy is not delivered to the extent that, or in the way in which, government intended, this is because public service organisations or professions resist its implementation in order to defend their own interests. This issue is returned to in Section 2, which discusses policy failure and its causes. Here, however, it should be noted that there are two other perspectives on power that may be helpful in analysing the policy-action dynamic.

The first is the idea of discretion. Michael Lipsky's (1980) famous study of the ‘street-level bureaucrat’ highlights the significance of front-line staff as agents in the policy-action dynamic. They act within the structures (the rules and guidelines) of the bureaucracies where they work, but these can never anticipate every situation that staff face in the course of their everyday work. Thus, because public work is complex, front-line workers inevitably exercise some discretion. Lipsky argues that ‘the decisions of street-level bureaucrats, the routines they establish, and the devices they invent to cope with uncertainties and work pressures, effectively become the public policies they carry out’ (1980, p. xii). Other work, such as Janet Newman's (2005), highlights ways in which senior managers, newly cast as ‘leaders’, exercise agency in a dispersed field of power. Lipsky's and Newman's studies lend support to the organic metaphor, since both suggest that policy happens bottom-up as well as top-down. But the argument here is rather different. In neither study did those concerned exercise their power in terms of a narrow definition of their interests. In the case of front-line staff, the way in which they exercised discretion was, in Lipsky's view, based on coping mechanisms designed to deal with the pressures placed on them, or to simplify the uncertainties and ambiguities of their work. In the case of senior managers, Newman argues that while some used their discretion to reinforce or extend managerial power, others used it either to try to reconcile top-down governmental targets with bottom-up priorities derived from consultations with local populations or with service users, or to innovate in ways which went beyond, rather than conflicting with, current government policies. The second perspective on power to be noted here is the increasing emphasis on empowering citizens or service users in the policy-action dynamic and the extent to which power is actually devolved to the public. While there has recently been an explosion of participation and empowerment initiatives, vary rarely, it seems, is power actually ceded by the organisations concerned in favour of citizens or service users.

2 The models in action: what counts is what works?

As noted at the start of Session 1, the models of change can be both explanatory and normative. As explanations, each corresponds to a different theoretical tradition. So do you just pick the one that seems most compelling? Or do different theories help explain different kinds of phenomena? The answer suggested here closely follows the work on metaphors by Gareth Morgan (1986), who sets out a number of different models of organisations (some of which map on to those outlined here). Morgan argues that managers need to be able to use multiple models in order to highlight different features of the reality they want to describe or explain. You could think of the models as different lenses in a pair of methodological spectacles: put on the ‘rational planning’ model and some things will be highlighted and others obscured; but if you then switch to an ‘organic’ or ‘cultural’ model other things will come to your attention. Each one offers different possible explanations of why things work or, quite often, don't work. Another way of saying this is that each offers different kinds of explanation for implementation failure, and that using multiple models in this way encourages a shift from ‘either/or’ thinking.

The mechanical model. For those who turn to the mechanical model the assumption regarding policy failure is likely to be that the plans were defective, the targets flawed or the contract badly specified. The danger here is that implementation failure is liable to lead those working with an implicit mechanical model to draw more power back to the centre – in other words, to strengthen the levers of control – rather than to view the model itself as potentially flawed.

The organic model. Looking at policy failure from this perspective leads to a focus on failures of communication, on barriers to relationships working effectively, or on other factors that prevent the system from responding or adapting to new policy needs or imperatives. So, for example, policy failure might be attributed to government having failed to listen to those with knowledge of how things might work best on the ground, or having failed to draw on the lessons provided by earlier policy experiments. Here policy failure might lead to calls for a shift from top-down to more bottom-up approaches, or for more enabling policies that allow for innovation and experimentation.

The cultural model. This draws attention to problems of vision and leadership, either on the part of government or on the part of senior managers charged with the task of implementation. That is, there may not be a clear enough policy steer from government or managers, the policy itself may be so amorphous that it gives confusing signals about its purpose and goals, or senior managers may not be providing sufficient leadership in ‘selling’ the value of change to their staff. This, in turn, means that policy is likely to be implemented in a relatively shallow or tokenistic manner, with little depth of commitment on the part of those involved. Those drawing on this model might call for a more coherent policy process (in contrast to policy overload, in which different policies might be in conflict with each other) and a clear policy direction that focuses on the long-term outcomes to be delivered, rather than a rapid succession of new policy initiatives that are tightly monitored in terms of short-term outputs.

The political model. Finally, policy failure might be explained in terms of outright resistance, whether passive or active, on the part of public service professions or organisations seeking to defend their interests. This is a very common explanation for politicians, and senior managers, to draw on, and it invokes attempts to weaken the power base of those viewed as resisting change. This kind of explanation has underpinned a succession of government reforms, such as the introduction of competition, the restructuring of services that are deemed to be clinging to old ways of working, and the shift to enhancing the power of consumers in order to break open entrenched professional and/or organisational power. However, the political model might also be used to offer other explanations of why policy has failed to be implemented. First, it may be that there was not a strong enough coalition of interests in the policy formulation process. Government departments may not have been in agreement; civil servants may have been less committed to this policy than to others; key professional or commercial stakeholders may not have been sufficiently consulted, or their concerns may not have been addressed in the policy development stage. Remember, too, that it is not enough for a policy and its programme of action to be broadly supported across government. Often it will also be necessary for agreement to be reached on what the government will stop doing. In other words, success or failure may depend as much on the opportunities for gradual budgetary reallocations at appropriate levels as on resistance to the policy as such. Second, resistance may not be simply a matter of recalcitrant groups defending their interests. Those involved may, rather, be defending values that they hold dear, or trying to protect the communities or users they serve from what they view as damaging consequences of the policy concerned.

As you read this section you probably saw the value of each of the models in turn. And it is important that, as managers, you try to use each of them, rather than just sticking to one that seems most compelling or that fits best with your existing (probably implicit) framework of analysis. In the case of failures in partnership working, for example, the organic model takes us only so far in attributing problems to failures of communication or relationships within the partnership. As Newman (2006) argues, partnerships may also be undermined by, on the one hand, tensions between the commitment of partners to joint working in order to deliver long-term outcomes, and, on the other, by the imperative for each of the partners to meet organisation-specific targets. In other words, prescriptions based on the mechanical model may counter those derived from the organic. Partnerships may also fail because they are not sufficiently driven by shared values, perhaps because of a lack of leadership (the cultural model), or because of a failure to pay sufficient attention to power imbalances (the political model).

3 Governance, policy and action

It was noted earlier in this unit that the models you would meet are both descriptive/explanatory and normative. In Section 2 they were used as explanatory tools to illuminate different possible causes as to why change might not happen in the ways that policy makers intended. This might be viewed as failure, or it might signify the system adapting to circumstances that were not covered by the original policy. In other words, not all implementation failure is necessarily a policy failure: policy makers cannot anticipate all the ways in which change might take place within a policy cycle. This may mean that a policy is outdated before it is fully realised in action. This leads into the difficult territory of viewing the different models as normative, offering prescriptions as to how policy should be implemented as well as explanations of what actually happens.

Shifts in governance have important implications for normative ideas about the policy-action dynamic. Governance shifts associated with the neo-liberal reforms of the 1980s and 1990s introduced new frameworks. The policy-action dynamic shifted, in part, from one of hierarchical control to one based on principal-agent relations and the importance of contracts. While this offered significant change, you saw in Section 1.3 that this shift may have served to reinforce, rather than undermine, mechanical and top-down processes of implementation. We are now witnessing another shift of governance, one that places much more emphasis on the importance of networks and partnerships. What are the implications of this shift – described by Erik-Hans Klijn and Joop F.M. Koppenjan (2006) – for the policy-action dynamic? It would seem, at first sight, to imply the need for a more organic, evolutionary or even bottom-up approach of the kind described in Section 1. And indeed you can trace the emergence of a new language of policy making that does stress the need for more devolution, more involvement by policy stakeholders, and more emphasis on evaluation and learning. Yet moves in this direction – such as the policy experiments cited in Section 1.3 – seem to be undercut by a further tightening of central government control, together with a greater proliferation of targets and other mechanical measures that allow little scope for experimentation and learning.

Similarly, new ideas can have a normative impact on the policy-action dynamic. The idea of evidence-based policy – based on the rubric ‘what counts is what works’ – is one of the most significant to have influenced the policy process in recent years. It offers a de-politicised image of the policy process, one in which scientific evidence, rather than political preference, informs the selection of policy solutions. This is a seductive idea: who would want policies that don't work, or to argue against the need for evidence? However, critiques of the evidence-based approach have focused on the narrow model of science on which it is based (one founded on positivism and the assumption that there are scientific facts that speak for themselves), on the framing of policy in terms of instrumental rationality rather than values, and on the consequent preoccupation with means rather than ends. The evidence-based policy approach, then, readily lends itself to the forms of rationality and the top-down approaches that characterise the mechanical model discussed in Section 1.2. Evidence-based policy is not necessarily always top-down, however, and some groups (for example community, carer or interest groups) may be the ones to draw attention to problems that need to be solved before public-sector organisations have even thought about them.

In the early 1970s Peter Marris and Martin Rein (1972) were already highlighting some of the challenges faced by policy makers and professionals in trying to learn the lessons of research. They note that it is often impossible to know what would have happened if a particular intervention had not occurred, because social processes are fluid and some change would have taken place even without that intervention. More important, perhaps, the intervention being evaluated is itself likely to change over time, as those engaged in ‘action’ modify what they are doing in response to feedback. Since those concerned do not wait for the outcome of the evaluation before undertaking those modifications, and may even modify their objectives over time, there is never any completed process to analyse (Marris and Rein, 1972, pp. 191–207).

Sandra Nutley and Jeff Webb (2000) highlight the problems of linking an evidence-based approach to a centralised, rational model of policy in which evidence is used to legitimate a single set of solutions to what may be a complex and differentiated set of problems. They also set out an alternative model that corresponds more closely with the organic, incremental approach discussed in Section 1.3 above, in which adjustments to policy are made in the light of learning emerging during the implementation process, so that the learning process is seen as a political one based around a continued process of communication between those involved.

The normative power of this alternative model, however, is very weak compared with that of the value of rationality as an idea and with top-down, central government control as a practice. And the practice of evidence-based policy – however seductive as an idea – is often undermined by the exercise of political power. For example, governments often choose to ignore evidence where it does not fit with their political preferences; policies may be abandoned while evaluations of their effectiveness are still being conducted; or they may be launched while evidence of what works is still being gathered. This is a reminder that the policy process is inherently political, and that the policy-action dynamic cannot be reduced to a series of debates about the relative merits of different policy instruments or implementation measures. In other words, it is not possible to escape from the messiness of political decision making through the promise of greater technical proficiency and the accumulation of ‘evidence’. Even if it is clearly preferable to have some evidence that can inform the process of adaptive action, the nature of that evidence will always itself be uncertain and subject to challenge and reinterpretation.