Policy failure and the policy-implementation gap: can policy support programs help?

Abstract There is an increasing awareness that policies do not succeed or fail on their own merits. Within complex messy systems, it is unclear how best to ensure effective policy design and implementation. However, rather than just let policies drift into full or even partial failure, governments are now beginning to take an interest in ways in which the policy process – especially the implementation phase – can be strengthened and supported. This article contributes to the debate in three ways: by unpicking the key factors behind policy failure; by exploring different approaches to policy support; and by identifying key messages for policy practitioners.


Introduction
There is an increasing awareness that policies do not succeed or fail on their own merits; rather their progress is dependent upon the process of implementation. The normatively attractive top-down view of policy and its implementation is predicated on three questionable assumptions: a chronological order in which expressed intentions precede action; a linear causal logic whereby goals determine instruments and instruments determine results; and a hierarchy within which policy formation is more important than policy implementation (Hill and Hupe 2015). Despite several decades of criticism, it is a model that still retains some popularity with policy-making authorities.
The policy context is, however, now understood to be much more complex than had been previously recognized. The earlier literature on the "policy-implementation gap" (Gunn 1978) has been supplemented in recent years by complex systems thinking informed by notions of unpredictability, nonlinearity, and adaptability (Braithwaite et al. 2018). Here the factors that shape and influence implementation are seen to be complex, multifaceted and multileveled with public policies invariably resembling "wicked problems" that are resistant to change, have multiple possible causes, and with potential solutions that vary in place and time according to local context (Rittel and Webber 1973).
As this level of understanding has increased, so governments have accordingly come to recognize that more needs to be done to try to ensure intentions are turned into resultsin short, that policy failure is avoided. Rather than just let policies drift into full or even partial failure, governments are now beginning to take an interest in ways in which the policy processand especially the implementation phasecan be strengthened and supported. This article contributes to the debate in three ways: by unpicking the key factors behind policy failure; by exploring different approaches to policy support; and by exploring the potential role of policy support programs.

The persistence of policy failure
In order to better understand how to improve policy support it is first of all instructive to appreciate the nature of policy failurelogically the reasons why things go wrong should help to guide the search for potential solutions. There is now growing interest in the notion of "policy failure" (Volker 2014) but as McConnell (2015) has noted, "failure" resides at the extreme end of a success-failure spectrum where it is characterized by absolute nonachievement. Such a situation will be unusual. As he observes, "failure is rarely unequivocal and absolute … even policies that have become known as classic policy failures also produced small and modest successes" (p231).
Four broad contributors to policy failure can be identified: overly optimistic expectations; implementation in dispersed governance; inadequate collaborative policymaking; and the vagaries of the political cycle.

Overly optimistic expectations
It might be especially thought that the bigger and more expensive policies -"major projects"would be those most carefully assessed for risk, yet "overoptimism" was the title given to an influential review of failure in major government projects in the UK by the National Audit Office (2013). It noted this to be: "A long-standing problem widely recognised that too frequently results in the underestimation of the time, costs and risks to delivery and the overestimation of the benefits. It undermines value for money at best and, in the worst cases, leads to unviable projects." (3) The study identified five interacting factors contributing to such overoptimism: complexity (underestimation of the delivery challenges); evidence base (insufficient objective, accurate and timely information on costs, timescales, benefits and risks); misunderstanding of stakeholders (optimism about the ability to align different views); behavior and Incentives (interested parties boosting their own prospects); and challenge and accountability (decision-makers seeking short-term recognition).
This problem is not confined to the UKa comparative study from the OECD (2015), for example, also notes successful delivery performance to be an ongoing challenge for centers of government. This is especially the case where policies require a long-term focus. A study by the Institute for Government (Ilott et al. 2016) of four such policy areas -Irish anti-poverty strategies, UK climate change, UK international development and rough sleepers in Englandidentified three common features that complicate delivery: costs and benefits are distributed unevenly over timethere is a long time lag between implementation and positive outcomes; they tend to be intellectually contested, politically contentious and hard to deliver; and the causes and effects span government siloes. In short, inadequate policy design is still far too prevalent.

Implementation in dispersed governance
Policies formulated at national level may face the challenge of ensuring some degree of consistency in delivery at subnational level, a process that is especially fraught where the subnational level has some separate degree of political authority (Norris et al. 2014). Sausman et al. (2016) draw on the concept of "local universality" to describe the process whereby general rules, products, or guidelines are shaped and tailored to fit into local contexts and enacted within practices. What is less clear is how central authorities can respond to this reality, especially where it occurs hidden from the view of the policy-making authorities.
Even where governance is concentrated rather than dispersed, implementation will still be highly dependent on local contextit is known from the literature on complex systems that an intervention that is successful in one location does not necessarily (or evenly routinely) deliver the same results elsewhere (Braithwaite et al. 2018;Allcock et al. 2015). All of this ties in with the long-standing literature on "receptive" and "nonreceptive" contexts for change pioneered by Pettigrew et al. (1992) and emphasizes the need for policy-makers to confront the "messy engagement of multiple players with diverse sources of knowledge" (Davies et al. 2008).
There is then the further complication that those operating at higher levels cannot succeed without having some grasp of what actually happens on, or close to, the frontline. This is the premise of the "bottom-up" school of thought on policy implementation and echoes Lipsky's notion of the "street-level bureaucrat" whose discretionary power can prove instrumental in determining the success or failure of a policy (Lipsky 1980). One of the salient features of many policiesespecially those requiring face-to-face contact with the publicis that "lower level" staff have considerable contact with outside bodies and often enjoy discretionary powers which accord them de facto autonomy from their managers. Although many of the decisions of these actors may seem small individually, in aggregate they may radically reshape strategic policy intention (Hudson 1993).

Inadequate collaborative policymaking
Policy-making has tended to be developed in distinct administrative siloes even though most interventions will almost certainly have wider implications that affect external parties. Moreover, despite growing academic interest in developing ideas and tools for promoting interorganizational partnering, improvement has been at best patchy and limited (Gazley 2017). The weakness of collaborative policy-making and the failure to establish a common ground for public problem-solving through a constructive management of difference remains one of the key reasons for subsequent implementation difficulties.
Other than for the most simple of tasks, policy design requires continuous collaboration with a range of stakeholders at multiple political, policy-making, managerial and administrative levels as well as the engagement of local "downstream" implementation actors such as end users, frontline staff and a range of local service agencies. Ansell et al. (2017) emphasize the need for policies to be designed in a way that "connects actors vertically and horizontally in a process of collaboration and joint deliberation." This, the authors say, should not be equated with a long and cumbersome search for unanimous consent; rather it is a search for sufficient common ground to proceed, without which there will be ongoing conflicts over policy legitimacy and organizational mission. The imperative therefore is for policy design and implementation to become an integrated process rather than simply a series of discrete and distinct stages. Whether policy-makers are equipped with the requisite skills, competencies, capacities, and capabilities to address such systemic flaws and succeed in such an endeavor is another matter (Williams 2012).

Vagaries of the political cycle
Politicians tend not to be held accountable for the outcomes of their policy initiatives in the event of failure the likelihood is that that they will have moved on or moved out. One consequence of this is that they are too easily attracted to the prospect of short-term results. This can lead to the pushing through of policies as quickly as possible, rather than getting involved in the messy, protracted and frustrating details of how things might work out in practice. In general, there is evidence to suggest that the political will necessary to drive long-term policy-making tends to dissipate over time (Norris and McCrae 2013). The concern here is that policy-makers are more likely to get credit for legislation that is passed than for implementation problems that have been avoided. Indeed the latter will probably tend to be seen as "someone else's problem" (Weaver 2010). Ilott et al. (2016) usefully identify three phases of political behavior in the policyformulation and policy implementation process: Phase 1: Rising Salience: In this phase an issue becomes politicized, gaining the attention of ministers. It is the point at which the problem to be tackled is defined and articulated, and some indication of what success would look like is identified. Phase 2: Building Blocks: Here politicians and officials put in place the policies, institutions, and targets aimed at resolving the problem. These actions should serve as a rallying point for the coalitions of support needed to sustain longterm focus. Phase 3: Embedding: This constitutes the period at risk of diminishing political interest but it is the time when the "building blocks" need to deliver some evidence of success.

Developing a policy support program
The four threats to successful policy implementation outlined in the preceding section are so widespread, that simply hoping that normal procedures and channels will be sufficient to resolve them is no longer realistic. At a minimum, a better understanding is needed of the processes through which policy moves and how, at each of these points, policy can best be supported. Four sequential points can be identified: preparation; tracking; support; and review.

Policy preparation
The aim at this point would be to ensure policy-makers are more alert to the practicalities of implementation by scrutinizing the feasibility of policy proposals more carefully at the outsetin effect, better "policy design" (May 2015). Faulty policy design can stem from many causes: a poor understanding of the problem; insufficient knowledge of the implementation context; unclear and even contradictory goals; poor quality evidence; and an absence of political backing.
The failure to draw upon, or be transparent about, the use of evidence has already been noted and is highlighted further in two recent examinations of the robustness of policy design by Sense about Science (2015Science ( , 2017. In these reports the question was posed: "could someone outside government see what you're proposing to do and why?" A framework was developed covering diagnosis (the issue to be addressed), proposal (the chosen intervention), implementation (how the intervention will be introduced and run) and evaluation (how will we know if the policy has worked?). A scoring system with four levels was then applied to 593 discrete policy proposals by thirteen domestic policy departments in England. Although a handful of examples of good practice were identified there were some recurring shortcomings around sharing work done, poor referencing, unclear chains of reasoning, and a failure to consider other policy options. Gold (2014) has noted that few countries have mechanisms in place to ensure a more robust policy design. In the UK, the Civil Service Reform Plan (HM Government 2012) required senior officials to warn before a political decision was taken where there were likely to be implementation concerns, but in practice, the central machinery only tends to be activated once an established policy is seriously off track. An exception in this regard can be found in Australia, where the Department of the Prime Minister and Cabinet has issued guidelines (Australian Government 2014) for those policy proposals considered likely to exhibit "significant implementation risks or challenges." These policies are defined as fitting one or more of several criteria: addresses a top government priority; has significant budget implications; makes major or complex changes to existing policies; involves significant cross-agency issues; is particularly sensitive; requires urgent implementation; involves new or complex delivery systems; and has been developed over a very short period.
In such cases a full implementation plan has to be developed during the drafting process, covering seven domains: planning, governance, stakeholder engagement, risks, monitoring, review and evaluation, resource management and management strategy. Each of these is further broken down and made available in the form of implementation "toolkits." There does not yet appear to be any evaluation of the effectiveness of these ambitious arrangements. In the UK a more modest suggestion to create an independent agency to scrutinize the assumptions underpinning key government decisions about public spending (Institute for Government 2017) has yet to receive a positive response.

Policy tracking
The focus here is on the establishment of some form of central "delivery unit" to track progress on the progress of policy implementation. Gold (2017) sees the proliferation of such units as a global trendthey are now reckoned to exist in 25 countriesthat aim to fulfill several functions: performance monitoring: tracking progress against key policy priorities through the analysis of a constant stream of departmental performance data problem-solving: undertaking field visits to identify obstacles to delivery and flagging up where additional resources may be needed to address specific problems progress assessing: supplying heads of government with routine progress reports While most such units have been located at the center of government machinery, this does not necessarily have to be the case; others can be established in key ministries or simply created for specific priority programs. It remains far from clear how effective these different bodies have been, and the barriers to compliance with their edicts are formidable. Weaver (2009), for example, identifies: monitoring problems, where target compliance may be difficult or costly; resource problems, where targets lack the resources to comply even if they want to; autonomy problems, where targets do not have the power to make decisions that comply with policy; information problems, where targets lack information that would make compliance more likely; and attitudinal problems where there are hostility and mistrust of targets.
In a global review of the effectiveness of delivery units in one policy areaeducation servicesa number of key lessons for effective practice are identified (Aid Effectiveness 2014). These include: focusing on a limited number of key priorities; being able to assume that budgets for each priority are adequate; developing good quality data and metrics to measure what matters; producing mutually agreed targets that are realistic and achievable; ensuring a clear understanding of delivery systems and active stakeholder engagement; and constructing an effective communications strategy.
More generally the Institute for Government's global review of delivery units (Gold 2017) identifies that although when well constructed they can make a useful contribution to supporting implementation, weak units can simply lead to additional significant problems. Their report sets out six key recommendations for successful practice: make sure there is strong, highly visible political backing; commit to a tightly defined remit; select a physical location close to a political sponsor; adopt the right hiring strategy, organizational structure and leadership model; ensure cross-government ownership of the delivery unit's results agenda; and put routines in place to review effectiveness and refresh operations.
The relevant theory of knowledge here is the positivist tradition with its assumption that social phenomena can be divorced from their context, and that objective knowledge about them can be achieved through empirical observation and quantitative expression. This constitutes a rational-linear model of decision-making in which unambiguous objectives are established, action upon them flows in predictable and unidirectional ways through established implementation structures, and outcomes are monitored against them. However, as noted earlier, it is the growing realization that implementation is complex, contextual and as much a bottom-up as a top-down imperative, that has led to an interest in an alternative or supplementary approach to policy implementation support.

Implementation support
Tracking performance delivery alone is unlikely to be sufficient to ensure effective implementation, especially where the policy is complex and long-term in nature. The question then arises as to whether some type of implementation support mechanism might be needed and, if so, what approach is appropriate. All such approaches would require close liaison with, and an understanding of, the position of the implementing agencies.
In a review of the components of service improvement for the Health Foundation, Allcock et al. (2015) point out that those who work on the front line, whether managerially or professionally, know more about the challenges of delivery than national policy-makers. A crucial task for implementation support is, therefore, to tap into the perceptions and experiences of those whose behavior will shape the implementation process. This support is not so much about explaining legal obligations or the requirements of statutory guidancethough this is importantthan about promoting the art and craft of policy implementation. It involves assessing existing capacity to deliver, knowing what is being done well, what needs improving and how best to build new capacity.
All of this is likely to require finding some way of bridging the understanding between national and local narratives via some intermediary body. One approach is the formation of what has been termed "implementation support centers" (Pew Charitable Trust 2017)entities of various types that work alongside and often at the direction of government to support effective implementation. Franks and Bory (2015) develop a similar concept in their exploration of "intermediary organizations" which, they conclude, "appear to play a critical role not only in implementing model programs, but also in developing the necessary capacity for systems change" (54).
The skills and qualities of those who work in such intermediary agencies will be crucial in determining acceptance and receptivity, but the availability of those with these attributes may be limited. Roberts and King (1996) even go so far as to suggest that such actors appear to have a unique identity, indeed certain innate personality characteristics. These are said to include being: highly intuitive; critical analytical thinkers; instigators of constructive social action; well-integrated personalities; highly developed egos; high level of leadership; and above average creative potential. The skills required of these "implementation entrepreneurs" are likely to be both in short supply and in high demand elsewhere.
All of this suggests that the way in which offers of implementation support are couched and perceived is vital in understanding their likely effectiveness. The danger is that such interventions try to straddle several strands of activity, some of which are in tension with each other or even simply contradictory. Three purposes have been identified: managing and regulating; problem-solving; and capacity building (Gold 2014(Gold , 2017.

Managing and regulating
Here the focus is on the identification of procedures for the measurement and scrutiny of performance and ensuring required standards are met. The detailed tracking of multiple implementation agencies is a formidable and expensive task. The temptation will be to allow agencies to self-assess their performance which in turn creates the risk that performance rating will be vulnerable to "grade inflation." It is an approach better suited to implementation tracking than to implementation support.

Problem-solving
The assumption here is that a problem has been sufficiently well defined to permit a close focus on how to address it. This could be pursued in a range of ways such as technical support, troubleshooting, the brokering of areas of dispute and encouraging the utilization of research and evidence. It is likely to require a flexible staffing model able to respond to the different needs and demands of local contexts.

Capacity building
Whereas problem-solving focuses on "what" questions, capacity building concentrates on the "how." It involves investing in skills and competencies that will be sustainable in meeting future implementation challenges. Training, peer learning, information, guidance, project management skills and other such interventions could all have a part to play. This sort of approach has informed the World Health Organisation's Regional Office for Europe's health systems transformation initiative, which is focused on how to make change happen and put in place the necessary infrastructure for this to succeed (WHO 2016(WHO , 2018Hunter et al. 2018).
Combining all three of these functionsespecially a fusion of compliance and supportwithin one agency is problematic. There seems to have been relatively little attempt to explore these tensions in contemporary institutions but there is still much to be learned from Henkel's seminal comparative study of the Audit Commission, the Social Services Inspectorate and the Health Advisory Service in England (Henkel 1991). In all three cases, attempts were made to straddle compliance and support, only rarely with any success. The key to managing the tension was the perceived independence of the agency from government, alongside credibility within the relevant policy communitiesthe assurance that those offering support understood the world from the perspective of the implementation agencies.
More recently Furnival et al. (2018) have also noted the ways in which, in UK healthcare, regulatory agencies are increasingly concerned not just with assessing current organizational performance but also with assessing their "improvement capability"a proxy of sorts for acting as implementation support agencies. However, they note that of eight measures of "capability," it is those that are relatively easy to measure that are dominant, namely process improvement and learning, and strategy and governance. More complex dimensions like organizational culture and a serviceuser focus tend to be marginalized. All told, although there are ways of offering or providing implementation support, the tensions and complexities involved in doing so will need careful consideration.

Implementation review
Post-implementation review is the last step in a policy support cycle, providing the opportunity to assess the extent to whichby some agreed pointpolicy objectives have been met. The literature on policy review and evaluation is long established (OECD 2013;Cairney 2016). Where done well it will seek answers to several key questions: Was the problem correctly identified? Were any important aspects overlooked? Were any important data left out of the analysis? Were policy expectations fulfilled? Have lessons been learned for future programs? In a rational evidence-based context, the answers to these questions can help to modify implementation trajectories and support decisions on whether or not to renew, expand or terminate an initiative.
What is less understood is how to conceptualize and evaluate policy support programs themselves. A generic framework open to adaptation is that arising from the work of McConnell (2015) in his discussion of degrees of policy failure, where he Low ambiguity low ambiguity and low conflict the pre-requisite conditions for a rational decision process are in place an activity associated with a generally shared and straightforward objective suitable for the application of a top-down approach key organizing concept: resources low ambiguity but high conflict a straightforward but strongly contested activity amenable to interaction and feedback implementation outcomes determined by the location of authority key organizing concept: power distinguishes between three interlinked activitiesprocess, programs, and politics. It is the first of these that is most applicable to the role of implementation support. Criteria for assessing the contribution of a policy support program might cover the extent to which such an intervention has helped to secure policy legitimacy; developed stakeholder support; exhibited clarity of purpose; demonstrated a comprehension of complexity; sustained political support; and contributed to the wider attainment of policy objectives.

Discussion: policy types and policy support tactics
It is likely that policy support programs will more readily flourish in some contexts, and in some situations, than others. Policies that seem to be relatively simple to address and around which there is a high degree of consensus may require little or even no support, whereas those that are contentious and complex may be doomed to fail without it. A useful framework for understanding the significance of policy type in relation to support programs is Matland's classic work on the impact of conflict and ambiguity on policy implementation (Matland 1995).
Matland's premise is that the varying characteristics of different policies have implications for the way these policies are implementedand, by extension, for the ways in which policy support programs might best be constructed. He uses his distinction between policy ambiguities on the one hand, and issues about conflict on the other, to develop the framework shown in Table 1 below.
There are important implications arising from this framework for ensuring the right model of policy support is associated with each domain. Broadly it can be hypothesized that: Administrative Implementation is amenable to a model associated with guidance, regulation and top-down performance management Political Implementation is amenable to a model associated with guidance, regulation and performance management but will also require flexibility and collaborative working Experimental Implementation is amenable to a model associated with a bottom-up approach, sensitivity to the implementation context and support for problem-solving Symbolic Implementation is amenable to a model associated with the same features as experimental implementation but may also require support for capacity building.
Alongside these broad messages for selecting a type of support program arising from Matland's analysis, there is the further question of developing a repertoire of tacticspractical measures to aid policy practitioners in their activities and decisionmaking. The starting point for this analysis was the notion of "policy failure." It was suggested that through a better understanding of how policy can go wrong, it is possible to develop a better appreciation of the potential role of policy support programs. Such an understanding could determine the point at which support might be most helpful (if at all) as well as the most appropriate mode of support and its potential duration.
Four broad causes of policy failure were identifiedoverly optimistic expectations, dispersed governance, inadequate collaborative policymaking and the vagaries of the political cycle. Each of these can be roughly equated with the four chronological phases of preparation, prioritization and tracking, support and review as shown in Table 2.
Although aspects of each of these elements can be routinely found it is not usual for all of them to be combined in a comprehensive manner to address all four stages. The most common mode of intervention tends to be around performance tracking, with policy-making bodies keen to unearth some evidence that their efforts are bearing fruit. Much less attention is being given to the other three elements identified in Table 2.
Given the relative novelty of such comprehensive policy support programs, there is correspondingly little empirical evidence to draw upon at this stage. However, the issues, literature, and evidence covered in this article do offer an opportunity to tease out some provisional messages for policy practitioners on how best to approach the task. These are outlined below.

Policy design preparation
Exploration of policy options and their feasibility with key implementation agencies Creation of forums for collaborative policy design: the more consensual the design process the less the likelihood of disagreements at the implementation stages Development of policy design assurance frameworks: identification of significant implementation risks and challenges along with risk management strategies Production of robust implementation statements: clear expectations of what should reasonably be expected to be delivered and under what circumstances Use of the best available evidence base to inform policy design Agreement on what would constitute an adequate funding stream for anchoring the policy and achieving the program objectives Ensure the agencies tasked with implementation can reasonably be expected to succeed in the task 3.5.2. Policy tracking Two-way communication processes: progress reports from implementation agencies to the policy-making center; responses back from the center to implementing agencies Use or creation of intermediary bodies between the policy-making and policy implementing levels Development of proportionate primary and secondary targets along with agreed timelines Separation of monitoring, regulating and inspecting roles from support mechanisms: use of policy support programs to better understand the stories behind the statistics Realistic expectations of what constitutes "success": policy objectives might never be fully delivered in the case of "wicked issues" 3.5.3. Policy implementation support Ensure the common ground developed with key stakeholders at the preparation stage is also applied to those putting policies into effect in managerial and professional roles: understanding bottom-up discretion and dilemmas Recruitment and development of a cadre of experienced and trusted "implementation brokers" to offer support tailored to local contexts Offer implementation support where it is needed or requested: ongoing assistance with problem-solving and capacity-building to develop sustainable implementation skills and knowledge 3.5.4. Policy implementation review Short, medium and longer-term review landmarks: clarity on what should have been achieved by when Routine use of action research for formative and summative evaluations Political acknowledgment that complex policies need to be given time to demonstrate achievements: costs and benefits will be unevenly distributed over time.
The rise of post-modernism and the era of "post-truth" politics (Baron 2018) arguably makes it more important than ever to establish a more robust evidence base for the design and delivery of policy. Although there is now a substantial literature on the policy-implementation gap and on discrete aspects of implementation support, such as policy design or performance monitoring, there is relatively little evidence on the use or effectiveness of coherent and comprehensive policy support programs. Indeed we know little about their very existence. A first useful step would be to commission research into the role of policy support programs themselves, to better understand what works for whom and in what circumstances. 1 It is not enough to be aware of the policy-implementation gap; we need better ideas on how to fix it.

Note
1. The authors are currently engaged in a research project to evaluate the implementation support arrangements created in England to support the introduction of one major piece of legislation, the Care Act 2014. The frameworks and ideas outlined in this article will be tested empirically to assess the role of implementation support in this specific context and to better understand the value of such interventions more generally.

Disclosure statement
No potential conflict of interest was reported by the authors.