Policy windows and multiple streams: an analysis of alcohol pricing policy in England

This article adopts a multiple streams approach to examine the failure to implement minimum unit pricing (MUP) for alcohol in England. It demonstrates that the multiple streams model provides a valuable conceptual tool for explaining how and why policies are, and are not, enacted. However, it finds that while problem streams and policy streams are useful heuristic devices, in practice they may overlap and be mutually constitutive. The case of MUP also highlights the potential for policy spillover between jurisdictions and different policy contexts, showing both limits to, and the complex nature of, these processes. It shows the need for high level political commitment in order to implement controversial policies, even when they are backed by strong supporting evidence. Furthermore, it highlights the importance of civil society actors not just in bringing policy issues onto the agenda, but in supporting governments in adopting measures to address them. This study shows that wider factors are important in and the Moreover, it is what key when whether


Introduction
In March 2012, the Conservative-Liberal Democrat coalition government announced its intention to introduce minimum unit pricing (MUP) for alcohol in England in The Government's Alcohol Strategy (GAS) (HM Government, 2012):

Theoretical perspective and methods
In Agendas, Alternatives and Public Policy, Kingdon (1984) attempted to understand the necessary conditions for policy change to occur; for a so-called 'policy window' to open. Kingdon's (1984) starting point is the fundamental ambiguity of the policy process; that is, the potential for issues to be identified and interpreted differently in different contexts and by different actors and, therefore, enter onto the policy agenda in multiple ways and forms (Zahariadis, 2003). In attempting to explain how, in the context of this ambiguity, certain issues become politicised while others do not, and certain policy projects succeed where others fail, Kingdon identified three 'streams' or elements of policy debates which are integral to these outcomes. First, the problem stream focuses on the identification of an issue, or collection of issues, which become framed as 'policy problems'; that is to say problems requiring governmental responses. Second, the policy stream relates to the development of potential policy responses to address the identified problem(s). Third, the political stream refers to the political factors that need to align in order for the proposed solution to a policy problem to be enacted. Factors influencing the political stream include the institutional setting and actors involved (that is, the right actors being in place in key 'gatekeeping' positions), the status of public opinion, the electoral cycle, the availability of resources and other factors which have an impact on the political commitments of key decision makers.
For policy change to occur all three streams need to converge to create a window of opportunity (Kingdon, 1984). However, while the convergence of these streams is a necessary condition for policy change, it is not sufficient. Policymakers must still act to bring about policy change when the conditions allow this. Thus, while the MSA recognises the structural factors shaping policy decisions, the model highlights the key role of political agency. Within each stream different actors are of principle relevance. In the problem and policy streams, researchers, policy advocates, civil society organisations, stakeholders and vested interests are to the fore. In the political stream, it is decision makers -civil servants, administrators, government ministers and members of parliament -who are the key actors. In each of these streams, public opinion, the salience of policy issues and the popularity of proposed measures are important contextual factors throughout the policy process. Kingdon (1984) identifies a key role for policy entrepreneurs, who may be located in government or civil society; who may bring problems and/or solutions onto the policy agenda; promote widespread public support for policy change; and/or influence key decision makers to take policies forward (see also Mintrom and Norman, 2009). Policy entrepreneurship is associated with the idea of (sometimes dramatic or seismic) policy change. Policy entrepreneurs, therefore, distinguish themselves from other policy actors through their attempts to bring about a fundamental reorienting of public policy, as well as potentially reshaping the policymaking context in which this change occurs. Policy entrepreneurs may be located in various contexts and institutional settings involved in the policy process. As Kingdon (1984) states, policy entrepreneurs: could be in or out of government, in elected or appointed positions, in interest groups or research organizations. But their defining characteristic, much as in the case of a business entrepreneur, is their willingness to invest This account of policy entrepreneurship focuses on individuals (or groups of individuals) within collective entities such as civil society organisations or government. Yet it is also possible for institutions (legal persons) to have political agency. Bodies such as Departments of Health are often identified as key policy actors within analyses of policy change in the domestic sphere in the same way that states are treated as collective actors within the discipline of international relations. Here, we view institutional actors such as the Department of Health, and even the Scottish government, as possessing agency -and thus as (potential) policy entrepreneurs -without attempting to disaggregate them into the key sub components (that is, units or directorates general) or the specific individuals within them that are driving policy change. This is appropriate for the level of analysis presented here and is also in keeping with the guarantees of anonymity given to interviewees.
Policy windows may open up suddenly and unexpectedly and close again just as quickly.
As policy windows open, policy actors must seize the opportunity presented for change to occur. At certain times, policy windows may open as a result of unexpected developments -which Kingdon terms focusing events -while at other times, they may result from more institutionalized aspects of the political cycle such as elections or budgets (see also Birkland (1997;1998)). Policy entrepreneurs play a key role in 'selling' policies, 'coupling' the different streams and thus driving governance change (Bache, 2013); a process which Ackrill and Kay (2011) term 'brokering' (in contrast to their 'advocacy' function in the agenda-setting phase). As will be examined below, the example of MUP in England highlights how, even when factors align to open a policy window, policy change may not occur and the moment may pass.
Kingdon's work represents a seminal contribution to the field of policy studies, which has been applied to an increasing range of policy issues and decision-making arenas. The MSA has been the subject of several recent reviews (Ackrill and Kay, 2011;Cairney and Jones, 2016;Jones et al, 2016;Rawat and Morris, 2016) and a special edition of a leading public policy journal mapping the state of the literature (see Béland and Howlett, 2016). Here we confine ourselves to discussing specific aspects of the MSA and its employment in the field of policy studies as it relates to the present case.
The empirical application of the MSA has been extended from the agenda-setting phase (where Kingdon saw its relevance) to the entire policymaking and implementation process (Zahariadis, 2003;Exworthy and Powell, 2004;Ridde, 2009;Howlett et al, 2015;Boswell and Rodrigues, 2016;Rawat and Morris, 2016). As it has expanded from its initial focus on US politics the approach has been refined and developed to take account of different policy contexts -for example, the greater openness of polities outside the US to external stimuli and events (Cairney and Jones, 2016). This includes studies of policymaking within the EU (Zahariadis, 2003;Ackrill and Kay, 2011;Ackrill et al, 2013;Bache, 2013). In complex, multi-level polities such as the EU, policy ideas may spread rapidly across different policy settings and decision-making forums. The application of the MSA to the EU has led scholars to engage with issues of overlapping or ambiguous policy competence and the policy 'spillover' from one setting to another (Exworthy and Powell, 2004;Ridde, 2009;Ackrill and Kay, 2011;Ackrill et al, 2013;Bache, 2013;Bache and Reardon, 2013;Zahariadis, 2013). Indeed, in the updated second edition of Agendas and Alternatives, Kingdon (2003) himself recognised the potential for Policy windows and multiple streams 319 'spillover' between policy settings via 'demonstration effects'. As Ackrill and Kay (2011) comment: 'If a policy issue occupies multiple institutionally connected policy arenas, a policy decision taken in one arena may impact directly on policy decisions in others, even forcing a decision when none otherwise would be made' (p 73). Boswell and Rodrigues (2016) argue that the focus of studies in this context has been largely top-down versus bottom-up, with the MSA being employed to explain how policies developed by a central government filter down to, or are implemented by, lower levels. This reflects the far larger number of studies employing the MSA which focus on the national or federal level of decision making versus lower (sub-national) levels of governance (Rawat and Morris, 2016). Others have suggested that policy entrepreneurship may be more effective at lower levels (Henstra, 2011;Oborn et al, 2011;Dudley, 2013) . Exworthy and Powell (2004), meanwhile, have highlighted not just the vertical connections between governance levels but horizontal connections and the need for alignment between policy actors at the same level (for example, different government ministries) to facilitate policy change.
In the context of UK public health debates, Cairney (2007a) identified both top-down (from the EU) and bottom-up (from the devolved administrations) influence on the UK government's tobacco policies relating to advertising and smoke free environments, with the Scottish government leading the rest of the UK toward smoke-free public spaces (see also Cairney (2007b)). These studies exist within a larger literature on the shifting dynamic of UK policy since devolution (see, for example, Bache and Flinders (2004); Keating et al (2009)).
Alongside the focus on policy 'spillover' and multi-level governance, scholars have highlighted the role of knowledge transfer within the process of policy development (see, for example, Shaxson et al (2012); Oliver et al (2014); Hawkins and Parkhurst (2016)) and argued for the relevance of this literature to the MSA. Cairney (2011) has suggested the concept of a 'policy transfer window' as a means of combining the insights from the two literatures in order to add greater depth to our understanding of the way in which external (knowledge) stimuli may affect policy development. While not working with the MSA framework, Mavrot and Sager (2018) have identified the way in which specific forms of vertical knowledge transfer occur in multi-level polities such as Switzerland.
This literature is of relevance to the present article given the specific constitutional context in which the MUP debates occurred; namely, the UK's asymmetric system of devolution in which policy developments in Westminster -affecting England and the UK as a whole -are closely linked to, and influenced by, developments in the devolved administrations (and vice versa). Furthermore, the debate on MUP was framed explicitly in terms of the underlying evidence base for alcohol related harms and the need for effective alcohol policy.
The analysis presented below draws on 26 semi-structured interviews conducted by the first author with policy actors in London and Edinburgh between February and October 2018. Respondents included current and former policymakers within the Scottish and UK governments, representatives of public health bodies, alcohol NGOs and other policy actors identified via stakeholder mapping and purposive and snowball sampling. Industry actors were not interviewed for this study for a number of reasons including uncertainties about access and the additional complexity anticipated in the data in light of our previous studies. An important implication of this decision is that the dataset is restricted to perceptions of industry actors as held by other actors. However, it is possible to understand the dynamics of the MUP debate through triangulation of interviewees from different sectors. Interviews were recorded, transcribed and analysed using established qualitative research methods (see Braun and Clarke (2006); Rubin and Rubin (2012); Brinkmann (2013)). A two-stage process of thematic coding was undertaken by the first author in liaison with the second author who reviewed transcripts independently to identify key topics and themes prior to their refinement in subsequent analyses.

The 'problem' stream
The entry of MUP on to the policy agenda in England was the result of a series of developments within the policy stream driven by important interventions from the research, advocacy and policy communities in both England and Scotland Hawkins and Holden, 2014). Alcohol harms have long been acknowledged as a policy problem requiring governmental intervention in a series of policy documents from governments of different political colours in England and Scotland. For example, most UK government administrations in the 1970s and 1980s published such documents in response to rising levels of alcohol harms (Baggott, 1990). More recently in England, the 'New' Labour government's Alcohol Harm Reduction Strategy for England (AHRSE) offered the first cross-departmental strategic response, and was followed by Safe, Sensible, Social (Room, 2004;Anderson, 2007). Both these documents eschewed the prevailing research consensus on the most effective, population level approaches to tackling harm (reductions in availability, price increases and regulation of advertising and promotion) (see Babor et al (2010)). Instead, measures such as public information campaigns and co-regulatory schemes, which are unsupported by evidence but strongly favoured by industry actors, were central to the policy approach they set out (Hawkins and Holden, 2013). In the period up to 2008, a similar policy approach was also evident in Scotland (see Scottish Executive (2007)). Thus, while successive governments in both contexts acknowledged the extent of alcohol related harms and accepted their responsibility to address these, they consistently failed to adopt those policy measures which would most effectively achieve their stated objectives . The prevailing policy regimes were seen instead to reflect industry preferences (Room, 2004;Anderson, 2007;McCambridge, 2012).
The policy documents reflect a conceptualization of alcohol related harm as limited to a minority of problematic drinkers, who should be targeted by policy. Furthermore, it identifies the responsibility for this problem as lying with the individual consumer who should be encouraged (via public information or advice) or coerced (via the criminal justice system) to change their drinking behaviour. The alcohol related harms most prominently targeted are those associated with heavy episodic drinking rather than the long-term health and social consequences of heavy drinking. This approach minimises perceptions that alcohol harms extend across society and that the population as a whole reducing consumption will reduce levels of a wide range of alcohol related health and social harms, as indicated by the available evidence (Babor et al, 2010). The long-term success of this approach to alcohol policy in England -notwithstanding the evidence available on both the nature of the problem and the responses -is indicative of the effectiveness of industry influencing strategies. These include industry actors developing long-term relationships with government and positioning themselves as key policy stakeholders (Baggott, 1990;Hawkins et al, 2012;Hawkins and Holden, 2014;McCambridge et al, 2014;Nicholls and Greenaway, 2015).

Policy windows and multiple streams 321
After devolution in 1999, this policy approach was largely mirrored in Scotland under the Labour-Liberal Democrat coalition administrations (Scottish Executive, 2007). This situation changed dramatically with the election of a Scottish National Party (SNP) government in 2007  and the subsequent publication of their alcohol strategy (Scottish Government, 2008). The strategy documents recognised the scale of alcohol related harms as a policy problem requiring population-level policy responses including MUP. MUP is a whole population measure in the sense that it affects the price of alcohol and influences alcohol consumption in the general population, because overall demand (and thus harm) reduces as price increases. However, MUP can also be said to be a targeted measure as not all products are affected by it, and inexpensive high strength products were of particular concern. For example, the prices of very few drinks sold in on-licence premises will change as a result of MUP. Instead, it affects those cheap drinks, which are high in alcohol content, and are sold in supermarkets and other off-licence premises (that is, for consumption elsewhere) which will be required to be sold at higher prices. MUP subtly changes the definition of the problem with which policy contends; it may also change the use of evidence and the balance of power between competing sets of policy actors.

The 'policy' stream
The publication of the 2008 Scottish alcohol strategy represented an important break from previous policy regimes which were insufficient to address alcohol related harms and to achieve the stated desire of governments across the UK of reducing them. Developments in the problem stream around the understanding of alcohol related harms went hand in hand with -indeed were dependent on -the identification of, and advocacy for, new forms of policy response able to address the alcohol problem so defined. It would be misleading to see MUP simply as a policy response to changing perceptions of the nature of alcohol related harms. Rather, new understandings of the alcohol problem and its putative solution were coproduced within concurrent and mutually imbricated developments within the policy and problem streams. The development of MUP as a policy intervention required alcohol related harms to be reframed and reconstructed as a different kind of policy problem requiring that particular solution (Katikireddi et al, 2014a).
The development of MUP as a viable, evidence-informed policy option was the result of developments in both England and Scotland and of cooperation between policy actors in each setting. The first modelling of the effects of the price-based interventions in the UK, including MUP, was commissioned by the Department of Health (DH) in London and undertaken by researchers at the University of Sheffield School of Health and Related Research (ScHARR) (Brennan et al, 2008) . This was later supplemented by additional analyses from the same team (Meng et al, 2012;Brennan et al, 2014;Angus et al, 2016). The Sheffield modelling played a vital role in bringing MUP onto the policy agenda in Scotland, providing policy advocates and policymakers with additional evidence to support their positions and justify decisions taken (Katikireddi et al, 2016). The Sheffield model also reflects interest in, and openness to, pricing as an important policy lever by officials within the DH alcohol policy team who commissioned the work, as well as their wider understanding of the limitations of previous policies in tackling the sheer scale of the alcohol problem.
Drawing on the Sheffield modelling, and in liaison with colleagues in England, alcohol policy advocates in Scotland succeeded in explicitly changing the framing of policy debates to the whole-population approach (Hawkins and Holden, 2013;Katikireddi et al, 2014a). In so doing, they reconceptualised the nature of alcohol policy problems facing the UK and the appropriate policy responses. These developments in the problem and policy streams were the first step in opening a window of opportunity for a fundamental reorientation of alcohol policy away from industryfavoured approaches and towards price-based and other population-level measures initially in Scotland, and subsequently elsewhere in the UK.
This laid the foundation for policy entrepreneurs to bring forward concrete proposals to adopt price-based interventions and created a window of opportunity for policy change to occur. The election of an SNP government for the first time since devolution served as a focusing event in the policy's development . This led to a rebalancing of the role of the industry and public health actors within the alcohol policy process and the subsequent adoption of MUP by the incoming administration . Following developments in Scotland, MUP soon emerged as the key alcohol policy focus elsewhere in the UK.

The 'political' stream
While the problem and policy streams developed symbiotically in Scotland and England -as a result of the close links between government, researchers and civil society actors in each setting -significant divergence was evident in the political stream. While the introduction of MUP in Scotland enjoyed sustained political commitment by key policy actors across the Scottish government, the policy lacked similar political support in England. In addition, the MUP debate in England took place within a complex and highly politicised policy context in which other policy agendas were being pursued in different parts of government.

The inclusion of MUP in the government alcohol strategy
The Conservative-Liberal Democrat coalition government was formed following the UK general election in May 2010, which the two parties had contested on the basis of very different approaches to alcohol policy. While the Liberal Democrats had committed to 'support a ban on below-cost selling', and were 'in favour of the principle of minimum pricing' (Liberal Democrats, 2010), the Conservatives had signalled their intention of reforming alcohol policy, and public health policy more generally (The Conservative Party, 2010). In opposition, Andrew Lansley, the incoming Secretary of State for Health, had convened a Public Health Commission involving health NGOs and industry actors, including the alcohol sector (Unilever, 2009). This partnership-based, co-regulatory approach formed the blueprint for the coalition's PHRD, introduced in 2011 (Department of Health, 2011). This flagship public health policy was launched as a policy priority in March 2011, just ten months after the general election, in the absence of an overarching alcohol strategy. The 2010 coalition agreement had included a commitment to ban below cost sales of alcohol and to 'review alcohol taxation and pricing', while underlining the need not to 'unfairly penalise' moderate drinkers (HM Government, 2010). This reference to alcohol pricing had appeared to reflect a compromise position closer to the Liberal Policy windows and multiple streams 323 Democrat perspective. The PHRD, however, placed voluntary, partnership-based approaches with industry at the very heart of the alcohol policy agenda.
The adoption of MUP by the UK government in Westminster was unforeseen by many actors across government and the NGO sector, as had been the earlier decision to bring forward an alcohol strategy in any form. Asked about how they became aware of plans for a new alcohol strategy, one civil servant observed:

John Healey [then Shadow Secretary of State for Health] challenged Andrew
Lansley about this in the House of Commons to say are you actually developing an Alcohol Strategy? When is it going to come out? And to everyone's surprise, Andrew Lansley said we're going to bring one out as soon as possible, and that was news to all civil servants.
Even when it became clear that there would be a new strategy, alcohol policy NGOs were circumspect about their chances of getting MUP onto the agenda: We were told, in the run up to the publication of the Government's Alcohol Strategy, that minimum pricing was just off the table. [...] MUP was definitely not being floated as a serious idea, until it was announced by David Cameron, and it took us all by surprise.
The fact the policy was not widely trailed before its announcement meant that not just policymakers, but the wider policy community were unprepared for its landing. Health NGOs and advocates which would be expected to support the policy and defend it publicly had not had the opportunity develop coherent media and influencing strategies around the announcement. This was crucial in a context in which the public still needed to be convinced about the merits of the policy. As one government actor commented: "to a certain extent the alcohol field weren't ready for [MUP's] inclusion and the public weren't either". It was consistently reported by interviewees that it was a personal decision of the Prime Minister, David Cameron, to include MUP in the alcohol strategy. One respondent suggested that the Prime Minister was motivated by concern for the underlying issues: I think that Cameron had been convinced by the evidence, and I think that some of the Special Advisors in Number 10, have been convinced by the evidence, as well […]. David Cameron, seemingly, took that decision on his own. He didn't have cross-Cabinet support for it.
However, the surprise nature of the inclusion of MUP in the strategy, and the lack of trailing in advance suggests that this was a decision taken without extensive consultation by the former Prime Minister, and without appreciating the degree of political commitment needed to see the policy through to implementation. In keeping with this analysis, MUP stood out from the remainder of the content of the new alcohol policy, as it alone embraced a whole population perspective on reducing alcohol harms (McCambridge, 2012). Despite much speculation about the timing and rationale for the inclusion of MUP in the GAS, one governmental actor candidly stated: "I am as baffled now as I was then that it suddenly got included".

324
A lack of political commitment MUP in England lacked political support within government and had no champions at senior ministerial levels prepared to drive through such a controversial measure against strong opposition. Despite the personal role played by David Cameron in getting MUP into the alcohol strategy, he failed to offer the ongoing public endorsement for the policy provided by Nicola Sturgeon in Scotland.
A similar lack of support was also evident at ministerial level within both DH and the Home Office: the departments jointly responsible for the delivery of the alcohol strategy, and thereby MUP. As one public health respondent familiar with the policy commented: I noticed that, after the day it was announced, I don't think I saw another government minister speak in favour of minimum pricing […] there was no senior member of the government speaking for the policy, which, when you've got something that controversial […] there was no one making that argument. That was why it felt to me this policy just didn't have a future because there was no one advocating for it.
Health Secretary, Andrew Lansley -the key architect of the PHRD -was vehemently opposed to MUP, making its introduction under the leadership of DH problematic. Partly as a result of Lansley's opposition to MUP, the Home Office assumed increasing responsibility for drafting the GAS and for the subsequent consultation for which Home Secretary Theresa May provided the ministerial foreword (as David Cameron had done for the GAS itself). However, May was little more positively inclined towards MUP than Lansley and provided negligible support for the policy. This is evident in the words of one Scottish government actor involved in liaising with Westminster on alcohol policy: We ended up having not nil contact with the Home Office […] but the political management of the Home Office […] were not sympathetic to minimum unit price to the point where [Theresa May] stopped her own officials going to meetings.
Unlike in DH, the lack of support for MUP, and the absence of a champion within the department, was evident not just at ministerial level, but among Home Office officials. This is due in part at least to the policy focus of the Home Office on public order, with the benefits of MUP more apparent in long-term health outcomes. As one government respondent commented: MUP as a policy, I don't think it's bought into.
[…] I don't think they understand what it actually is or what it's trying to achieve and who it will affect.
This confusion about the nature and purposes of MUP was evident in the framing of its objectives in the GAS in terms of binge drinking and public order issues, as well as in David Cameron's foreword. Appreciation of the effectiveness of MUP in addressing the long-term health consequences, as well as the short-term intoxication-related dangers, of drinking was weak outside DH. The gradual shift of responsibility for the Policy windows and multiple streams 325 alcohol strategy from DH to the Home Office meant that the policy emerged in the context of an absence of political ownership of the policy and considerable institutional uncertainty. Outside of the civil servants in DH who had been intimately involved in the policy's development, including in commissioning key research that underpinned it, levels of understanding of, and support for, the policy across government were weak.

The policy window closing
Following publication of the 2012 GAS, the Home Office held a public consultation on the level of MUP, with 45p per unit of alcohol initially proposed for consideration. The consultation ran for ten weeks from 28 November 2012, and the government's response was published in July 2013 following significant delays. The consultation was not designed to address the question of whether or not to introduce MUP in England, but the level at which it should be set. However, the submissions made by industry actors went far beyond the remit of the consultation -on the level of the MUP -and questioned the rationale behind the policy as part of a wider strategy to oppose its introduction (Gornall, 2014;Hawkins and McCambridge, 2014) . Moreover, the Home Office minister making the announcement used responses to the consultation as a rationale to discontinue its stated plan to introduce MUP (Browne, 2013).
During the consultation period, there was the apparent shift within government towards the introduction of a ban on below cost selling of alcohol as an alternative to MUP; a policy that was known to be less effective than the latter (Brennan et al, 2014). As one of the researchers at the University of Sheffield involved in the alcohol price modelling commented: And at some point [the Home Office] started asking questions about the ban on sales below VAT and duty. So, we knew at that point that there was certainly being a consideration that minimum pricing might be dropped.
[…] There was a gradual transition over time where it became apparent we knew that minimum pricing wasn't going to happen.
The decision to move forwards with a ban on below cost sales represented a subtle redefinition of the problem. For example, a Home Office source commented: I think the industry recognised that Ministers wanted to regulate, to remove the worst examples of loss leading sales of alcohol. It certainly didn't mean that if we did ban below cost sales then minimum unit pricing was definitively off the table and it was never presented as, we will do this but we won't do MUP.
This policy drift is indicative also of the general trend in government policy in this period away from 'progressive' issues (such as the environment, social exclusion and public health), which had characterised David Cameron's leadership of the Conservatives in opposition and his early years as Prime Minister, towards a more traditionally Conservative policy agenda. This shift in direction coincided with the recruitment by Mr Cameron of Australian political strategist, Lynton Crosby, to manage the Conservatives' electoral strategy in November 2012; after the adoption of MUP in the GAS in March 2012 and before the policy's discontinuation in July Benjamin Hawkins and Jim McCambridge 326 2013. This reorientation was encapsulated in the words attributed to Mr Crosby that his aim was to 'get the barnacles off the boat' (Watt, 2013). This included not just MUP but also standardised cigarette packaging, although the latter decision was ultimately reversed .
The absence of MUP measures from the government's legislative programme announced in the Queen's Speech in May 2013 fuelled speculation that the policy had been abandoned. In July 2013, it was the Liberal Democrat Home Office Minister Jeremy Browne (2013) who announced to Parliament that the government would not bring forward plans to implement MUP, citing the need for greater supporting evidence of its effectiveness and the need to protect the interests of moderate drinkers: There has been much speculation about the Government's plans in relation to minimum unit pricing. That policy will remain under consideration, but it will not be proceeded with at this time. We do not yet have enough concrete evidence that its introduction would be effective in reducing harms associated with problem drinking -this is a crucial point -without penalising people who drink responsibly. We will tackle the most egregious examples of cheap alcohol by banning sales of alcohol below the level of alcohol duty plus value-added tax.
The statement claiming insufficient evidence contradicted the earlier view of the evidence supporting MUP in the GAS, while the evidence in support of MUP had strengthened in the meantime (Stockwell et al, 2012;Zhao et al, 2013) . Browne's (2013) announcement was framed explicitly as an opportunity for industry actors to fill the policy void through self-regulatory measures, and was linked to the Department of Health's Responsibility Deal Alcohol Network (RDAN): Our decision not to proceed with the introduction of minimum unit pricing at this stage gives the alcohol industry an opportunity to demonstrate what more it can do to reduce the harms associated with problem drinking. Our challenge to the industry is to increase its efforts, building on what has already been achieved through the public health responsibility deal. That includes improving education to promote safer drinking, reducing the availability of the high-strength products that cause the most harm for problem drinkers, and responsible marketing and product placement.
Respondents identified a range of factors that contributed to the decision not to bring forward measures to implement MUP. One respondent attributed the decision to opposition among Liberal Democrat MPs, including Browne himself, whose constituencies included cider-producing regions in the South-West of England. Others identified key Home Office officials, and ministerial aides, as playing a key role in the form and content of the 2013 announcement. Specific claims that the statement was amended to downplay the supporting evidence for MUP following interventions from ministerial special advisors within the Home Office formed the basis of a complaint by a former civil servant to the Civil Service Commission, which was rejected in October 2017 (Civil Service Commission, 2017).

Policy windows and multiple streams 327
The government was also concerned about the legality of MUP in the context of ongoing challenges to the policy in Scotland. As one civil servant commented: I think Ministers understandably wanted to see how that case ran and played out before committing to introducing minimum unit pricing in England and Wales. That was the immediate reaction to that, you know that we should just pause, not take it off the table but just wait and see how that ran.
The uncertainty about the final outcome of the Scottish legal case was given also as the rationale for not setting out a timetable for revisiting the policy, a decision which meant that while the introduction of MUP formally remained government policy, there was little prospect of it being introduced in the foreseeable future. As the previous respondent continued: [W]e were quite careful at the time not to tie ourselves to a timeframe in terms of when we were going to then bring minimum unit pricing back on the table or, if indeed, we were going to do that at all, because there was still a chance that the policy would be struck down in the Courts, it would be taken off the agenda in Scotland. I think the other thing that it did was create an opportunity for the industry to step up and do more. This was something that was in our thinking at the time.
Following the final ruling in the Scottish MUP case by the UK Supreme Court, there have been no further developments on MUP in England brought about by government. This reflects the controversial nature of the policy and the lack of political commitment to it in the Conservative government. Moreover, by the time of the ruling, the UK's decision to leave the EU and the ongoing Brexit negotiations absorbed political focus and energy. This meant that even high priority issues have been left in stasis. A Health and Social Care Select Committee (2018) enquiry briefly reconsidered the issue in a one-day hearing in January 2018, but there appears to be little early prospect that measures to enact MUP will be brought forward.

Discussion and conclusion
The case presented here demonstrates the ability of 'spillover' to occur between policy settings within the UK, but also both limits to, and the complex nature of, these processes. The failure to enact MUP in England, following legislation in Scotland demonstrates the importance of high-level political commitment to controversial policy issues -even those with strong supporting evidence and a precedent for the introduction in related settings. In addition, it highlights the important role that civil society actors can play, not just by bringing policy issues onto the agenda and promoting policy responses, but by supporting government decisions to pursue these measures once announced. The Westminster government did not trail its decision to introduce MUP in advance and thus undermined the ability of civil society bodies to support their policy.
The multiple streams model of the policy process provides a valuable conceptual framework through which to explain the adoption of MUP in England in March 2012 and the subsequent decision in July 2013 not to move forward with implementation of the policy. The introduction of MUP in England was facilitated, in the problem and policy streams, by an iterative, symbiotic process of problem definition and policy development which occurred through a collaboration between researchers, advocates and policy actors in both England and Scotland. This included the policy entrepreneurship of officials in the DH in London bringing price-based measures onto the policy agenda by commissioning the initial modelling of price interventions by the University of Sheffield.
In the political stream, the policy process in England and Scotland diverged significantly. As documented previously, MUP enjoyed strong and enduring political commitment at the highest levels of government in Scotland in the face of significant opposition from powerful vested interests in the alcohol industry (see Holden and Hawkins (2012;; Katikireddi et al (2014a)), which was crucial in seeing the policy through to implementation in May 2018. This underlines the key role of political agency in policy development. While often unforeseen and unpredictable circumstances ('focusing events') may lead policy windows to open, the commitment of policy actors is needed to grasp the opportunities provided by the window and to effect policy change.
While developments in Edinburgh created normative pressure for the Westminster government to follow suit on MUP, public health actors saw little prospect of this occurring under the coalition government. Even key government actors were unprepared for its adoption in the 2012 alcohol strategy. The lack of forewarning about the policy announcement meant pro-MUP advocates were unable to give effective support to such a controversial measure. This support was needed to insulate it (and the government proposing it) from the obvious challenges which would come from the industry. Ministerial support for the policy was lacking in DH and the Home Office: the two government departments most closely involved with alcohol policy. Moreover, while officials in DH had played a crucial role in the emergence of MUP in the policy stream, their attention had been directed firmly towards other policy initiatives, namely the RDAN, which were associated personally with the Secretary of State for Health, Andrew Lansley. Within the Home office, the problem of alcohol related harm, and thus the relevant policy responses, were viewed through the lens of law and order, rather than health: a set of issues which MUP was not primarily designed to address .
Responsibility for the adoption of MUP was attributed to David Cameron personally, demonstrating the ability of key political actors to shape policy agendas. Yet the failure to see the policy through to implementation demonstrates the limitations of political agency at even the highest levels, particularly in the absence of key allies. Dropping MUP was one aspect of the wider shift in political orientation within David Cameron's government from late 2012, leaving the policy without a champion in government and thus its survival is unlikely.
The main contribution of this article is to offer an explanation of the failure of MUP in England informed by key theories from the field of policy studies. However, the analysis presented here also makes an important theoretical contribution to the existing literature on the MSA. It deepens our understanding of the MSA through its application to a new policy issue and policymaking context: alcohol policy in the context of the UK's system of asymmetric devolution. The MSA has offered important insights into developments at all stages of the policy process and helps explain both convergence and divergence between England and Scotland. In so doing, it builds on existing studies of the MSA in multi-level governance systems.