Implementation and the Governance Problem

This article has two aims: to qualify the UK government’s ‘problem’ of governance in a comparison with Scotland and Wales, and to use implementation studies (the ancestors of the new governance literature) to explore policy developments since devolution in Britain. It presents a puzzling finding from extensive interview research: that while we may expect UK government policy to suffer a bigger ‘implementation gap’ based on distinctive governance problems (such as greater service delivery fragmentation and the unintended consequences of top-down policy styles), pressure participants in Scotland and Wales are more likely to report implementation failures. Using a ‘top-down’ framework, it explores three main explanations for this finding: that the size of the implementation gap in England is exaggerated by a focus on particular governance problems; that pressure participant dissatisfaction follows unrealistic expectations in the devolved territories; and that the UK government undermines devolved policy implementation, by retaining control of key policy instruments and setting the agenda on measures of implementation success.


Introduction
The study of implementation enjoys periodic revivals, with suspicion among its proponents that it lives on in the new alias of governance (Barrett, 2004, p. 254;Exworthy and Powell, 2004, p. 263;Hill and Hupe, 2002;O'Toole, 2000, p. 276;Saetren, 2005, p. 572). Yet, while the term 'governance' has risen in popularity, the intractable top-down versus bottom-up debate contributed to the demise of implementation as a widely used concept. In this light, is there a valid reason to make the links between old implementation and new governance? The value of these links from the 'bottom-up' is demonstrated well by Hupe and Hill's (2007) discussion of the accountability of 'street level' bureaucrats (Lipsky, 1980) in an era of multilevel governance. The aim of this article is to highlight links from the 'top-down'. It focuses on strands of the implementation and governance literatures which characterise diminishing central control as a problem. This literature describes the gap between expectations at the top and perceived policy outcomes at the bottom (the 'implementation gap') and explores the normative problem of ensuring that policies made by elected governments are carried out by subordinate authorities such as local governments and delivery agencies (Hill and Hupe, 2002). 1 To outline these issues, the article compares the UK policy process with the experience of the devolved governments in Scotland and Wales. 2 It then draws on extensive interview data to explore the perceived 'success' of policies in each country.
The primary advantage to this comparison is that it enables us to explore policy developments in political systems that appear to suffer fewer problems of governance. In Scotland and Wales there is less evidence of a fragmentation of service delivery organisations or the same unintended consequences associated with the pursuit of a top-down policy style. If such factors are used to explain the implementation gap in England, then we would expect to find fewer problems in Scotland and Wales. In this light, the findings from a series of interviews with 'pressure participants' (or policy influencing organizations such as interest groups - Jordan et al., 2004) and decision makers are surprising. This interview data did not confirm a smoother and more successful delivery process in Scotland and Wales. Rather, respondents were more likely to suggest that Scottish and Welsh governments suffer greater implementation problems. Therefore, a key aim of this article is to account for these puzzling findings, exploring the following: • the potential for the new governance literature to exaggerate problems of implementation in England; • the links between a perception of failure and the unrealistic expectations of pressure participants; • the attitudes of each government to the implementation process; and • the influence that the UK government has on the implementation of devolved policies.
A further benefit of this comparison within Britain is that the identification of a significant 'implementation gap' in Scotland and Wales allows us to gauge the level of policy divergence since devolution in a way not yet found in the literature. Although much of the literature challenges the idea of radical policy change, there is still a focus on a series of headline-grabbing policies that set the devolved territories on a different path (see Keating, 2005a;Keating, Stevenson, Cairney and Taylor, 2003;McGarvey and Cairney, 2008;Schmueker and Adams, 2005; 1. There is an understanding of clear and consistent policy objectives. 2. The policy will work as intended when implemented. 3. Tasks are fully specified and communicated to skilful and compliant officials. 4. The required resources (including political will) are committed to the programme. 5. Dependency relationships are minimal and support from interest groups is maintained. 6. External, or socioeconomic, conditions do not significantly undermine the process (see Marsh and Rhodes, 1992).
In contrast, the portrayal of governance as 'self-organizing networks' (Rhodes, 1997: 50) resembles a bottom-up approach that charts the implementation of policy through self-selecting clusters of organizations in which a variety of public and private organizations cooperate. In this vein, Barrett and Fudge (1981) criticize a focus on success and failure and point to the inevitable problems that governments face when they pursue a top-down mode of decision-making. While central government policy may be the main influence, it competes with a variety of demands (including contradictory government policies) and competing pressures within local implementing organizations. The notion of policy failure is criticized because: (1) departments are made up of more than one programme, and hence intra-departmental conflict may occur; and (2) few policies are fully implemented by one organization (Hjern, 1981;Hjern and Porter, 1982). It is difficult to force decisions on actors within the implementation structure who are employed by other organizations, so it is unrealistic to think that a sole central actor could secure its own aims and objectives irrespective of the actions of the others involved. Inattention to the complexity of these implementation clusters causes difficulties in the administration of policy. Hence, the exaggeration of policy 'failure', as difficulties lead to feelings of powerlessness since no one seems to be in charge. This circular link between top-down policy making and feelings of policy failure is central to Bevir and Rhodes' (2003, p. 6) argument that: 'centralisation will be confounded by fragmentation and interdependence that, in turn, will prompt further bouts of centralisation'. Therefore, the governance literature has ties to both bottom-up and topdown approaches. This extends to the top-down versus bottom-up debate on the capacity of the centre, its ability to control policy and its dependence on others to deliver policy. Yet, an interesting development is that the positions of Marsh and Rhodes have diverged on this issue. When studying the implementation of Thatcherite policies in the UK, Marsh and Rhodes (1992) were critical of accounts of implementation that exaggerated the lack of power at the top and the degree to which implementation structures were 'self-selecting' at the bottom. Yet, many discussions of Rhodes's hollowing-out thesis now effectively characterize it as a bottom-up position.
Many have questioned the 'hollowing out' thesis by pointing to the growing range of powers that the core executive enjoys, and exploring the extent to which changes at the centre have reinforced its power (Hogwood, 1997;Holliday, 2000;Marinetto;. Problems of 'overload' in the 1970s suggested that the government was never effective at controlling peripheral functions of the state. Governance changes such as privatization and civil service reforms mark a return to core competencies, with the centre making strategic decisions and creating accountability mechanisms to ensure that these are carried out by others. While this may involve negotiation and the trade of resources, the centre is still the most powerful actor and the loss of control described by the hollowing-out thesis is exaggerated. Similarly, Marsh et al.'s (2003) 'asymmetric power model' (APM) is critical of the (alleged) assumption of pluralism in Rhodes' 'differentiated polity model' (DPM), which highlights, 'not one but many centres linking many levels of government' (Rhodes, 1997: 3). Marsh (2008, p. 255) argues that 'strong government, although increasingly challenged' is a more realistic description than the 'hollowed-out state'.
In contrast, Bevir and Rhodes (2003, p. 6) question the ability of the government to impose policies. While 'the British executive can act decisively' and 'the centre coordinates and implements policies as intended at least some of the time', on the whole, 'to adopt a command operating code builds failure into the design of the policy'. This builds on Lipsky's (1980) discussion of 'street level' implementation, which suggests that the focus on the powers of the centre misses the point. Since public sector professionals are subject to an immense range of requirements laid down by regulations at the top, they are powerless to implement them all successfully. Instead, organizations create standard operating procedures as a way of satisfying a proportion of central government objectives while preserving a sense of professional autonomy necessary to maintain morale. The irony is that this cumulative pressure associated with central government power effectively provides them with a degree of freedom to manage their budgets and day-to-day activities.
There are two paths to choose on the basis of these parallels between governance and implementation. The first is a normative exploration of how much discretion implementers should have. This acted as a polarizing force in the top-down versus bottom-up debate. While some stressed the threat to accountability through representative government and the loss of 'certainty and consistency', others focused on the adaptability to local circumstances, the competition for legitimacy from local elected governments (Linder and Peters, 2006, p. 31) and the willingness of governments to include compromises in the policy design (Hill and Hupe, 2002, p. 71). These arguments are replicated in the governance literature, ranging from Rhodes' (1997, p. 54) argument that, 'hollowing out erodes accountability' to Hooghe and Marks' (2003, p. 233) suggestion that governance should be 'dispersed across multiple centers of authority' because, 'centralized authority -command and control -has few advocates' (see also Bevir and Rhodes, 2006;Richards, 2007).
The second path is descriptive, with debates regarding how best to explain policy 'failure' and where to study it. In the UK governance literature, competing models (APM and DPM) provide a similar broad narrative of change since the Thatcher period. Rhodes (1997) highlights the irony of hollowing out arising (with the exception of the Europeanization of policy) as an unintended consequence of Thatcherite attempts to present an image of governing competence. In addition to privatization and the reduction of the public sector, the UK government created contracting-out arrangements and a range of quangos to remove delivery functions from local authorities, introduced quasi-markets in the NHS and separated policy and management functions in the civil service. The unintended consequences were service delivery fragmentation, reduced communication between senior and junior levels of government and obscured accountability (Rhodes, 1994).
A further irony is that while the Labour government recognized these problems it also contributed to them -not only by devolving power to UK territories and English regions, 3 but also by granting independence to the Bank of England, extending the influence of the EU through the social chapter, furthering quasimarkets in health, relying more on political advisers than civil servants and groups involved in implementation, and further fragmenting service delivery with an emphasis on voluntary sector provision of public services (Richards and Smith, 2004). Labour's first response to the problem of governance was the modernization agenda on cross-cutting issues, seeking solutions based on trust and networks. This was replaced by a more straightforward top-down style in Labour's second term of office following frustration with a lack of progress on joined-up government at the centre. Cross-cutting targets coordinated from No. 10 were transferred to the Treasury and more strongly linked with the control of expenditure (Richards and Smith, 2004, p. 106). 4 The difference regards how we characterise the effects of these developments. The different approaches of APM and DPM mirror the two approaches in implementation research -describing the process either as a pyramid with central government at the top (following policy along the delivery chain from top to bottom) or as a sphere, with the street level organizations in the middle (suggesting the need to study how they interpret their decision-making environments) (Bevir and Rhodes, 2006;Hupe and Hill, 2007, p. 285;Richards and Smith, 2006).
The aim of this article is not to seek a resolution to these longstanding debates, particularly since each approach is self-fulfilling: a focus on the bottom highlights a multiplicity of influences and distance from central government, but misses systematic patterns of adherence to targets set at the top; a focus at the top highlights central control and meeting targets that relate to a small part of government business, ignoring the bulk of government responsibilities which are delivered out of the public spotlight. Rather, it adopts a pragmatic approach that treats models as complementary rather than contradictory (Linder and Peters, 2006,p. 28) and draws on a range of ideas on governance and implementation when appropriate. The top-down focus is useful because it allows us to situate the new governance problem within a broader implementation framework. It also qualifies the literature on devolution that gauges policy divergence according to policy choices rather than their effects. The bottom-up approach is useful to explain pressure participant dissatisfaction about the limited involvement of devolved governments in the implementation process. The descriptive and normative paths highlight the models of decision-making that governments and pressure participants choose to gauge the success of policy (since the evaluation of policy is as much political as empirical).

Governance, Implementation and Devolution
All three countries share 'hollowing' elements such as the Europeanization of policy and a reduced public sector following privatization. They also share implementation constraints, such as the need for clear and consistent objectives and to devote significant resources to policy. However, a focus on new governance problems suggests that England will face relatively unsuccessful implementation: 1. The fragmentation of service delivery organisations in England makes it difficult to control the direction of implementation (Rhodes, 1997). In Scotland and Wales the culture of contracting out was never as strong. In Wales it was often subverted (interview, former Chief Executive Swansea Council, 2005). In Scotland, the requirement to contract out 80 per cent of residential care for older people was never introduced (interview, UK Department of Health, 2006). Overall, there is more reliance on local authorities, while the devolved governments have taken greater control over significant public bodies (such as Communities Scotland and the Welsh Development Agency). In Wales there is a particularly strong commitment to supporting and integrating the public sector and its workforce (Welsh Assembly Government, 2006a,b,c). 2. The UK government has adopted a new top-down style to 'regain control over policy outcomes' (Richards and Smith, 2006, p. 343). Organizations must adhere to strict targets, which cause unintended consequences (Hood, 2007). The size of the English state is large and the monitoring arrangements are remote. In Scotland and Wales there are targets but the regime is less punitive and based more on the types of personal relationships that can be developed in smaller countries (Laffin, 2004;interviews, Convention of Scottish Local Authorities, 2004, Estyn, 2005Healthcare Inspectorate Wales, 2005, HMIE, 2006, Improvement Service, 2006, Welsh Local Government Association, 2005. 3. This top-down approach extends to consultation styles. The policy process in Wales and Scotland is arguably more 'consensual' and the systematic inclusion of pressure participants during the formulation process contributes to better policy and aids implementation (Entwhistle, 2006).
Yet, the interview evidence did not confirm a smoother and more successful delivery process in Scotland and Wales. Approximately 250 interviews were conducted from 1999-2007 with a wide range of interest groups (professional, voluntary, business, trade union, religious), civil servants and elected officials in England, Scotland andWales (1999-2007) in the 'most devolved' policy areas (health, education, local government). 5 These interviews probed: the nature and frequency of links between pressure participants, civil servants and ministers; the distinctiveness of policy decisions in the devolved territories; and the extent to which participants felt that these decisions were implemented successfully. These pressure participants were more likely to highlight problems of implementation in Scotland and Wales. So how do we explain this puzzling finding?
The primary hypothesis, explored using the top-down framework, is that the governance literature exaggerates the likelihood of UK government policy failure. This argument may refer to the inappropriate caricature of English governance, which focuses on strict targets rather than 'new localism' and 'double devolution' (see Walker, 2007), or the high profile disputes between ministers and interest groups rather than more consensual relationships within policy communities (Cairney, 2008a(Cairney, , 2009. It may also suggest that we focus more on the less visible but still significant 'regulatory state' in Scotland and Wales (McGarvey and Cairney, 2008, p. 150). However, in broader terms it suggests that a focus on new governance problems diverts attention from the old problems of implementation such as the lack of clarity in objectives or the lack of resources behind policy. Further, although each government may use different delivery mechanisms, they all face similar problems of interdependence with, and compliance from, implementing bodies.

Skilful and Compliant Officials
We may assume that the UK government suffers most from top-down implementation factors 3 (compliant officials) and 5 (dependency relationships and support from interest groups), based on the problems of fragmentation and interdependence exacerbated by top-down policy styles. Yet the differences should be examined rather than assumed. In Scotland and Wales the relative closeness between senior decision makers and implementing officials enables the greater ability to 'micro-manage' policy. Yet, this exaggerates the ability of devolved executives to monitor policy networks and quangos that operate at arms length. Neither Scotland nor Wales have developed a department that can match the Treasury's capacity to control funding and monitor targets (Cairney, 2006a;McGarvey and Cairney, 2008). Similarly, we should not overstate the value of a traditional local government relationship. Thatcherite attempts to bypass local authorities were based on their obstruction of central government objectives (Marsh and Rhodes, 1992). The devolved governments both face constraints to local authority compliance. In Scotland the best example regards the implementation of free personal care for older people (for the policy background see the Bell and Bowes, 2006;Shaw, 2003;Simeon, 2003;Sutherland, 1999). The Scottish Government does not implement this policy directly. Rather, it effectively reimburses local authorities for the care they provide or commission. It is not ring-fenced and service provision becomes linked to negotiations on the (in)adequacy of the local government settlement. The consequence for residential care is incomplete implementation (see Cairney, 2006b, p. 73;McGarvey and Cairney, 2008). First, many councils have lengthy waiting lists for people who qualify for care. Second, the funding shortfall leads to disputes over the coverage of the payments (e.g. does it cover meal preparation?). Third, local authorities pass on insufficient funds to private providers who then make up the shortfall by overcharging on accommodation costs to fund personal care. These issues resurfaced in 2008 when, first, the SNP Government (from May 2007) signalled no desire to challenge local authority behaviour and, second, the new Sutherland report highlighted a range of problems (see Cairney, 2008bCairney, , pp. 13-14, 2008c). While the main success of FPC has been a significant rise in care at home, the evidence suggests that, until recently, British local authorities already subsidised home care (although local authority policy varied, many charged an amount equivalent to the £40-60 per week Attendance Allowance benefit paid by the UK Government -interview, CERETAS Wales, 2005, Director of Social Work, 2004). Therefore, for a large part of the Scottish population the new policy replicates arrangements already in place. 'Free' care becomes a lower than expected increase in funding (since the new payments of £145 per week replaced rather than supplemented Attendance Allowance), with issues of poor quality personal and residential care still similar across Britain (interview, Help the Aged, 2006).
In other areas, compromises are often built into implementation to allow local authorities to adapt their policies in line with other priorities. For example, the Labour-led Scottish Executive's (1999Executive's ( -2007 policy on maximum class sizes in schools referred to an overall average (in theory allowing classes of 10 and 40 to meet a target of 25), while the SNP Government merely provided the funding without 'ring-fencing' it (as part of an overall strategy to reduce central government control of local authority spending -see Cairney, 2008d). The Welsh Assembly Government (hereafter 'Welsh Government') has also rejected 'hypothecated' funding. It provides local authorities with aims, but rarely tries to impose a ranking of those aims. Rather, it influences local authority behaviour through its consultation style, seeking partnerships and creating a sense of policy ownership (Laffin, 2004: 217;Rawlings, 2003;Trench and Jarman, 2007: 119;Welsh Assembly Government, 2002interview, WLGA, 2005).

Support from Interest Groups
The latter point suggests diverging policy styles, with governments in England pursuing compliance through targetry in contrast with devolved government reliance on more flexible measures backed by consultation and negotiation (Greer and Jarman, 2008). Consultation minimizes problems of compliance by creating a sense of involvement and greater commitment to policy success. It also allows the government to benefit from the practical experience of those consulted (Jordan and Richardson, 1987, p. 242;Marsh and Rhodes, 1992). Yet, Whitehall departments also enjoy these benefits. While even the rejection of consultation during the Thatcher era is debatable (Cairney, 2002), Marsh et al. (2001) argue that consultation rose significantly under New Labour (i.e. the comparison for devolved governments). This regularity of consultation is often missed twice when we focus at the top. First, most policy is produced by low-ranking civil servants seeking information from pressure participants (Cairney, 2008a;Jordan and Maloney, 1997;Page, 2006, p. 4). Second, the policy styles literature describes consultation during both formulation and implementation. Groups have at least one of two 'bites at the cherry': if consultation appears to be rejected (by ministers, in a small number of cases) during formulation, the process will return to 'normal' during implementation (Jordan and Richardson, 1982, p. 3). The significance of regular consultation with the UK government is confirmed by interviews with participants in health, education, local government, children's and older people's policies. The British Medical Association and Royal College of Nursing enjoy close relationships with the Department of Health at various levels (interviews, 2006); most teaching unions enjoy privileged access within the Department for Education and Skills (now the Department for Children, Schools and Families) in exchange for a formal commitment to the 'social partnership' (interviews, Association of Teachers and Lecturers, Professional Association of Teachers; Association of School and College Leaders, 2006); the Local Government Association has a good working relationship with the Department of Communities and Local Government (interview, LGA, 2006); and voluntary groups report regular, valuable contact with government (interviews, Help the Aged, National Council for Voluntary Organisations, NCH, 2006). Therefore, while many groups express dissatisfaction with inflexible targets and excessive centralization, their response is to accept the principle of targetry and attempt to negotiate around the details (interviews, LGA, RCN, and NHS Confederation, 2006; the exception is teaching unions on pupil testing - Cairney, 2008a).
This UK process of consultation is not necessarily 'inferior', particularly since the benefits of the devolved consultation processes may be exaggerated. For exam-ple, the devolved governments do not always consult widely (the British Dental Association Scotland was periodically excluded from policy on NHS dentistry -interview, 2006; the decision to subsume public bodies within WAG was done with little effective consultation - McAllister and Stirbu, 2007, p. 298;interviews, CBI Wales, IOD Wales, FSB Wales, Prospect Wales, 2005). Further, their greater propensity to maintain close relationships with pressure participants may be based as much on a lack of policy capacity and the need to consult as the wish to create a sense of ownership (particularly during the early years of devolution) (Cairney, 2008a;Keating, 2005a). Devolved policy styles also have their own unintended consequences. First, the 'everyone round the table' approach gives equal weight to each pressure participant. This contradicts the 'logic of consultation' (Jordan and Maloney, 1997) with the most affected and most directly involved in implementation. For example, in Wales, Shelter Cymru (interview, 2005) points to the early influence of homelessness groups, which exceeded their expectations, based on a low civil service capacity and crowded out local authority representation. Second, we may find consultation only when the principles rather than the details of policy are discussed. A classic example in Wales is policy on free prescriptions. While ministers engaged in partnership meetings to discuss the principles of policy, the same process was not followed by civil servants examining the details. This led to serious errors in the regulations to stop 'prescription tourism' (interviews, Community Pharmacy Wales; Royal Pharmaceutical Society in Wales, 2005). This example signals the possibility that Welsh and English groups have different 'bites at the cherry': while the former may be consulted more during formulation (since its links are often with ministers rather than civil servants), the latter may enjoy more fruitful consultation during implementation.
Further, the UK consultation process does not necessarily cause the biggest implementation gap, because: (a) success relies on a broader range of implementation factors; and (b) the UK consultation process shows a remarkable ability to compartmentalise issues even when they involve the same participants. Both factors are demonstrated by mental health policy, which arguably highlights the most significant breakdown of group-government relations in contemporary UK politics. This involved a 10-year stand-off between UK ministers and the vast majority of pressure participants which united under the Mental Health Alliance to oppose government legislation (Cairney, 2009). The end-result was a protracted legislative process and the withdrawal of many policy proposals. In contrast, in Scotland, ministers fostered high levels of policy 'ownership' among stakeholders (interviews, Penumbra, Mental Health Foundation Scotland, Scottish Association for Mental Health, National Schizophrenia Fellowship Scotland, 2006). This was harnessed during the implementation process, with the requirement in the Mental Health Act for health and local authorities to cooperate supplemented by delivery structures (community health partnerships) less subject to top-down controls and blessed with more stability than in England. Problems of compliance were minimised, dependency relationships were managed through stable partnerships and the political will was significant (Cairney, 2007c, pp. 77-80;interviews, Mental Welfare Commission, Head of Scottish Executive Mental Health Division, 2006).
Yet, the experience in Wales highlights a poor service delivery record despite consensual policy styles and less fragmented service delivery mechanisms. Interviewees (Hafal, July 2006, Mind Cymru, July 2006) point to close links with health ministers and a partnership approach to develop Wales' National Service Framework (NSF), but also a lack of resources and political drive. Further, problems in Wales resulted from measures to integrate services. Service delivery now flows through 22 small local health boards that were introduced to share coterminous boundaries with local authorities. As a result, the expertise required to commission specialist services is spread too thin and the reduced scope for economies of scale in direct provision often means buying services from England. In contrast, the UK government's relatively top-down and prescriptive NSF has faced fewer implementation problems (at least before the NHS funding crisis of 2006), in part because consultation was more participative and less confrontational than the parallel process on legislative reform (interviews, Mental Health Alliance, 2006).

Unrealistic Expectations and Bottom-up Implementation
Interest groups are not impartial observers of the policy process and their evaluations of policy success are linked to their expectations. Indeed, groups may exaggerate their expectations and demands when bargaining with government. Therefore, their apparent dissatisfaction with policy outcomes may exaggerate their lack of influence (Dür and de Bièvre, 2007). This point applies to groups UK-wide. However, there is good reason to expect further group disenchantment in Scotland and Wales, which is not explained by bargaining strategies. A potential irony of the devolved solution to compliance and interest group support is that it exacerbates pressure participant dissatisfaction. If we define the implementation gap as the gap between expectations at the top and perceived policy outcomes at the bottom (Hill and Hupe, 2002), then in Scotland and Wales there is a bigger gap because there were greater expectations. These are associated with the term 'new politics' (see McAllister, 2000;McGarvey and Cairney, 2008) and were fuelled by three factors: 1. Many policies were sold as a significant break from UK policy and associated with terms (e.g. 'free' rather than 'subsidised' personal care; the 'abolition' of student fees) that exaggerated their scope. 2. Devolved administrations fostered their own policy networks, giving the impression that their governments were the key decision-makers rather than part of a multi-level governance process. 3. Scottish and Welsh consultation fosters high levels of 'ownership' and participants count on policy to be implemented according to agreements reached at the top.
Therefore, Scottish and Welsh participants are more likely to express dissatisfaction if compromises with implementing bodies are built into the policy design. Yet, ironically, the devolved governments are the most likely to embrace bottom-up methods: minimizing ring-fencing, deferring to locally defined policy priorities, fostering service-led improvements, relying on outcome measures (e.g. improvements in quality of life) and promising to intervene only if insufficient progress is made (Cairney, 2007c;Laffin, 2007;interviews, COSLA, 2004, Improvement and Development Agency, 2006, Improvement Service, 2006, LGA, 2006, WLGA, 2005. The gap can therefore be explained by different attitudes (between pressure participants and devolved governments) to implementation styles, and therefore different perceptions of what constitutes implementation success. Devolved participants do not object to a less prescriptive implementation style per se. Rather, they lament their inability to influence policy when the centre relinquishes control (interview, One Parent Families Scotland, 2003). This frustration relates to their lack of resources to lobby outside of the central government arena. Much depends on the status of groups before devolution, with devolved arms of UK organizations (approximately half of all Scottish groups - Keating, 2005a, p. 65) the most likely to report insufficient resources (arguably because their parent organisations have not responded sufficiently to devolution -see Cairney, 2008a). Some may have one full-time member of staff, and in Wales some may have less than one. This is a problem felt particularly by groups in the children's and older people's sectors where policy issues extend across departments and many levels of government. In such cases, groups become involved in new policy communities, make a meaningful contribution to the formulation process, but then see their influence decline as implementation authorities set their own agendas.
These issues are demonstrated well by housing policy. Both devolved governments introduced higher minimum housing standards than England, but this merely produced the perception that Scotland and Wales make strategies while England delivers (interview, Chartered Institute of Housing, May 2006). In Wales, many local authorities did not contain the housing standard in their business plan; in Scotland there were serious delays on local authority submissions and civil service assessments of plans to fulfil the standards. In contrast, the UK Government was more focussed on enforcing a lower standard and monitoring local authority performance. Homelessness presents a similar picture (interviews, Scottish Council for Single Homeless, January 2004, Shelter Cymru, July 2005, Shelter Scotland, November 2003. Scotland and Wales introduced wider definitions of need but did not provide additional ring-fenced funding to match the expansion of policy (suggesting more people chasing the same housing). In Scotland, groups describe the 'best homelessness legislation' but worst social housing conditions in Europe (based on the lack of political will to direct local authorities to implement policy). In Wales, low civil service capacity and low levels of local authority interest in homelessness limited discussions between the Welsh Government and councils at senior levels. This contributed to implementation problems, including the use by local authorities of bed and breakfast accommodation for families (against Welsh Government policy).

External and Socioeconomic Conditions
'External' factors undermining implementation are felt most in devolved territories for several reasons: 1. Socio-economic conditions, such as a relatively ill or more dispersed population, are more likely to constrain policy success (Greer, 2004). 2. 'Europeanization' increases the role of the UK (as the member state) in monitoring the implementation of policy by devolved governments (Cairney, 2006c;Jeffery and Palmer, 2007;Keating, 2005a). 3. Devolved governments may be pressured by the UK government to emulate decisions made by the UK. In practice, this depends on whether or not there is a shared party of government (with most pressure coming from UK ministers to their devolved counterparts) and the attitudes of first ministers. For example, Jack McConnell appeared more willing to cooperate than Rhodri Morgan (BBC News, 2002;Keating, 2005a). 4. Devolved policies may be undermined by the reserved aspects of policy and the effects of English decisions. These constraints are most significant in Wales, which has responsibility for fewer policy areas and lacks the ability to pass primary legislation (although the Government of Wales Act 2006 has increased its powers to some extent - Paun, 2006). Wales is historically tied to policy in, and enjoys a fluid cross-border flow of goods and services with, England. They are also felt in Scotland, particularly when Scottish decisions relate to a UK-controlled tax and benefits regime (McGarvey and Cairney, 2008, p. 160). Examples include housing, child or fuel poverty, while the unintended consequence of free personal care was terminated entitlement to UK Attendance Allowance (Scottish entitlement to council tax benefit would also cease if the Scottish Government introduced a local income tax).
Although the UK-effect may be felt most in Wales, the attitudes of devolved governments are also significant. For example, although both faced a combination of strict Treasury controls on public borrowing and financial incentives to transfer local authority housing stocks, the effect was different in each country. In Scotland there was a history of co-operative, tenant and community ownership models of housing and so stock transfer did not represent a wholesale break from local authority control. The Scottish Executive (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007) embraced stock transfer as a means of reducing debt and increasing investment. In Wales, there was a stronger history of local authority control and more rhetorical links were made between housing associations and privatization (although tenant votes against transfer have also passed in Scotland). As a result, the Welsh Government has been less enthusiastic, seeking in vain a new policy that satisfies Treasury rules. This may further perceptions of policy failure in Wales (interviews, Scottish Federation of Housing Associations, 2003, Welsh Housing, 2005. Similarly, Treasury rules on the capital funding of schools encouraged Scotland (until the election of the SNP in 2007) but not Wales to follow wholeheartedly the English reliance on public-private partnerships (Cairney, 2006a). This exaggerates the appearance of policy failure in Wales. The inability of the Welsh Government to pursue alternative funding mechanisms for schools led to unfavourable comparisons of working conditions for teachers (interview, ATL, 2006). The devolved governments also face significant reserved issues in higher education (Keating, 2005b, p. 431) and this affected early policies (2001) on tuition fees. In Scotland, policy divergence was signalled by the promise of an 'abolition' of tuition fees (in fact, up-front fees were replaced by a smaller 'endowment' fee payable after graduation; this endowment was abolished by the SNP Government in 2008) but implementation problems were built into the policy design. While a key recommendation of the Cubie Report was the repayment of 2 per cent of income at an income level of £25,000, this was constrained by existing arrangements managed by the Inland Revenue. The Scottish Executive could have set up a separate system for fee collection, but it accepted the UK threshold because it could not opt-out of the arrangements for student loan payments (now 9% of income over £15,000 per year). Therefore, students still took out loans to cover living expenses and the endowment and were expected to pay this back after graduation at the same rate as English students (Cairney, 2006c). In Wales, the lack of legislative power was addressed in 2001 with an imaginative use of finance. A means-tested National Bursary Scheme for Welsh residents (including further education and part-time students) effectively paid for the up-front element of tuition fees (Rees, 2005).
Perhaps more significantly, both governments have been forced to react to the UK's introduction of top-up fees to supplement the income of English universities. For example, the Scottish Executive planned to charge English medical students studying in Scotland, to ward against Scottish students being crowded out (Cairney, 2006d). In Wales, although the tuition fee regime was formally passed to the Welsh Assembly in 2004, its new power was undermined by crossborder pressures. Indeed, the 'Rees review' (Welsh Assembly Government, 2005) concluded that any significant departure from English policy would have profound consequences on student and staff movement across the border. Lower fees would mean too many English students studying in Welsh universities, while lower investment could cause staff flows in the opposite direction. As a result, the Welsh Government recommended following the English lead. Party politics initially caused the National Assembly for Wales to reject this in favour of targeted additional grants to Welsh students studying in Wales (interviews, Higher Education andFunding Council Wales, 2005, Professor Theresa Rees, 2005), but in 2009 the grants were abolished following the Review of Higher Education in Wales . In both cases, decisions made by the UK Government undermined the implementation of policies made previously by devolved governments.

Setting the Agenda on Success
Such examples suggest different types of policy transfer Marsh, 1996, 2000) when devolved policy is driven by the UK government. The direct use of reserved powers suggests 'coercive transfer', while devolved policy based on the disproportionate effects of UK decisions in Scotland and Wales suggests 'indirect coercive transfer': policy decisions are taken voluntarily but 'driven by perceived necessity' (Dolowitz and Marsh, 2000, p. 13). A further type of indirect coercive transfer relates to changes in devolved policies driven by the appearance of UK success. Since this is unlikely to be based on a long-term analysis of policy outcomes (at least while devolution is in its infancy) the appearance of success in England is fostered by the use of outputs masquerading as outcomes. A perception of English achievement is based on the UK Government's ability to set the agenda and influence the measurements used to gauge policy success. Although the issue of measurement extends to most areas (see Andrews and Martin, 2007, which attempts to give a statistical overview of Welsh and English public service performance), we can see this effect most clearly in healthcare policy.
There are many factors which should qualify levels of health policy success: the effects of government policies on overall health improvement are difficult to track (see Alvarez-Rosete et al., 2005); the surgical operations associated with waiting lists account for a very small proportion of NHS spend (Andrews and Martin, 2007, p. 155); and the perfect attainment of targets does not necessarily produce successful outcomes (Hood, 2007). Yet, the UK Government often succeeds in equating targetry success with real success in healthcare efficiency. By extension, attempts by devolved governments to set their own agenda with different proxy indicators are often undermined by a public and media focus on targets, often fuelled by UK ministers bemoaning the lack of 'modernization' outside England. The dominance of the UK agenda means that devolved government success is not measured on its preferred terms (such as patient satisfaction -interview, Welsh Government special adviser, 2005). This is a crucial constraint to policy implementation when each government has already taken different policy decisions based on their respective strengths: England furthering the managed market approach with a pool of experienced NHS managers; Scotland moving towards the pre-internalmarket system of professional consensus, given its wealth of senior clinicians and Royal Colleges; and Wales drawing on close links to local government, focusing on public health and closer integration between local authorities and local health boards by introducing coterminous boundaries (Greer, 2004). The effect of the UK agenda on waiting times and waiting lists is that the devolved governments generally feel the need to respond and adopt policies that no longer play to their strengths. The UK has more resources available for implementing policies based on healthcare targets: • More 'new' money to reduce waiting times following a period of generous Treasury allocations to the Department of Health (although the NHS deficit crisis in 2006 and the current economic crisis may now undermine this advantage). While Scotland and Wales benefited from this spending increase (via the Barnett formula -see McGarvey and Cairney, 2008), the devolved agenda has been more concerned with 'streamlining' services before committing further investment. • A more developed 'command and control' function, particularly compared to Wales, which devolved decision-making to NHS Trusts and removed central capacity functions under the former Conservative government. The size of commissioning in Wales (22 LHBs) represents a sacrifice of efficiency for coterminosity, with expertise spread too thin and a limited ability to challenge the Trusts. • A bigger pool of senior NHS managers. In England the punitive regime for missed targets was made possible by the ability to threaten chief executives with unemployment. In Scotland and Wales there is less competition for posts and less pressure on managers to meet targets. • Less inertia based on medical commitment to previous policies, including the rejection of 'marketisation' and management-led reform in Scotland. This inertia may undermine a sudden insistence by Scottish ministers to devote more resources to meet targets that distort clinical priorities.
Yet, we can qualify this centralizing effect in three ways. First, the effect of the English agenda is not felt equally. In Scotland, there is more control over the presentation of success. For example, while (in September 2005) Tony Blair was publicly critical of the rejection of English NHS reforms in Scotland, the Scottish Executive's Permanent Secretary presented evidence suggesting that, 'we are now doing substantially better than England' (see Cairney, 2006a, p. 118). The effect in Wales was best demonstrated when Jane Hutt was replaced by Brian Gibbons as Health Minister -a sign that the Welsh Government was willing to change tack and play 'catch-up' with England (interview, Welsh NHS Federation, 2005). This reflected stark cross-border comparisons of statistics, made more significant by the flow of staff and patients between England and Wales. Yet, Gibbons' difference in approach has largely been presentational. For example, tackling waiting times by giving the 2nd offer scheme was devised before the current minister was in office. While it appears to follow the English agenda on Patient Choice, 'it is an equity, not choice, based system' (interview, Welsh Government special adviser, 2005). In England the individual drives the scheme on the assumption that s/he is an informed, motivated, articulate consumer. In Wales the NHS identifies patients and negotiates an alternative time and location, with the cost and ability to travel built into the calculation. Second, tobacco policy shows that agenda-setting is not one-way. The decision to ban smoking in public places in Scotland, combined with commitments made in Wales and Northern Ireland, acted as a catalyst for change in England (Cairney, 2007a,b). Third, this type of indirect coercive transfer related to success may be restricted to areas such as health, which display very similar starting points, easily comparable results and a significant flow of staff and (relatively interested and knowledgeable) consumers. This may contrast with, for example, compulsory education, which has always been organized differently, particularly in Scotland (this includes the development of its own indicators) but also in Wales, which has traditionally demonstrated a higher commitment to comprehensive schools and a stronger role for local education authorities (interviews, GTC Scotland 2006, General Teaching Council Wales, 2005. Further, the UK agenda on performance is less persuasive and subject to greater professional opposition than in health, while the flow of staff and students between England and Wales is not as significant as health or higher education. Therefore, Scotland and Wales have been able to distance themselves from performance league tables and there is less emphasis on testing.

Conclusion
This article explores the idea that the implementation gap will be bigger in England than in Scotland and Wales because the UK government faces unique problems: a top-down policy style alienates organizations charged with implementation; strict targets have unintended consequences; and service delivery fragmentation undermines central control (Rhodes, 1997). These factors are less apparent in the devolved arenas characterized by partnership working, a more flexible approach to targets and a closer relationship to delivery organizations. Yet, the interview evidence does not confirm that devolved governments suffer fewer problems of implementation. Indeed, it is possible to present an opposing narrative: in healthcare, the agenda is dominated by proxy measures (waiting lists and times) that England is better placed to deliver on; in social care, Scotland's flagship policy has not produced a better quality of care for older people; in housing, England has delivered more with less; and in education the results are mixed, with Scotland and Wales pursuing different agendas in compulsory education but struggling to maintain differences in higher education.
A key explanation for this finding is that there is significant potential for the new governance literature to exaggerate problems in England and underestimate the implementation gap in Scotland and Wales. 6 A more extensive discussion of top-down implementation factors suggests that each government faces similar problems related to interdependence, the compliance of implementing bodies and support from interest groups. For example, England's confrontational style is often exaggerated and there are unintended consequences to devolved policy styles, while dependency relationships with local authorities are not necessarily less problematic than with other service delivery organisations (particularly when the powers of devolved governments are limited). Devolved governments also face additional 'external' problems associated with reserved powers, the effects of English policies and the failure to challenge UK government measures of policy success. In other words, UK government influence on devolved policy does not end with policy formulation. Indeed, its (often intentional) contribution to the devolved implementation gap has undermined policy divergence significantly.
A further explanation for the gap between expectations and perceived policy outcomes can be found in the different attitudes of pressure participants and their respective governments to the implementation process. In England, if policy is imposed following inadequate consultation then problems with implementation will not be criticized by pressure participants. Rather, they link implementation 'failures' to inadequate policies, the inflexibility of government targets and excessive levels of centralisation. In contrast, devolved pressure participants count on the policies that they influence and approve to be implemented fully. Yet, ironically, their governments are the least likely to impose policies from the top down. This presents a problem for participants who develop a strong ownership of policies formulated at the top, but do not have the resources to influence policy implementation at the bottom.
In most discussions of implementation and governance, this problem leads to normative debates stressing the inappropriate ability of organizations to subvert democratically produced policies from the top-down. However, in this case the waters are muddied by government attitudes. While pressure participants may bemoan a lack of central control, the devolved governments themselves appear to foster bottom-up development as long as broad commitments are met. Compromises are often built into the policy, particularly when the same actors (such as local government) are involved in negotiations at the formulation and implementation stages. Although this leads to the intermittent appearance of policy failure, it may be preferable to the unintended consequences (most visible in England's focus on targets) of more successful levels of compliance. Notes 1. For the purposes of this article the Scottish and Welsh governments are treated as comparisons rather than 'subordinate authorities'. Each government has its respective subordinate authorities. 2. The article uses 'UK' to avoid confusion with governing practices in English regions. The term 'England' refers to UK Government policies for England. 3. Although note that the main focus of this article is not the contribution of devolution to the UK governance problem. 4. It is too soon to tell if Building Britain's Future (Cm 7654, 2009), promising the replacement of targets with 'entitlements', represents a distinct, third approach. 5. The broader project draws on over 300 interviews conducted between 1999-2008.
Two rounds of interviews in Scotland were supported by the ESRC's Devolution and Constitutional Change Programme (Keating, 2005a). Interviews in Wales in 2005 and England in 2006 were supported by the University of Aberdeen. Further